96467 Latin America and the Rising South Changing World, Changing Priorities Latin America and the Rising South Changing World, Changing Priorities Augusto de la Torre, Tatiana Didier, Alain Ize, Daniel Lederman, and Sergio L. Schmukler © 2015 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 1 2 3 4 18 17 16 15 This work is a product of the staff of The World Bank with external contributions. The findings, in- terpretations, 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 follow- ing conditions: Attribution—Please cite the work as follows: De la Torre, Augusto, Tatiana Didier, Alain Ize, Dan- iel Lederman, and Sergio L. Schmukler. 2015. Latin America and the Rising South: Changing World, Changing Priorities. Washington, DC: World Bank. doi:10.1596/978-1-4648-0355-0. 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 the Publishing and Knowledge Division, The World Bank, 1818 H Street NW, Washington, DC 20433, USA; fax: 202-522-2625; e-mail: pubrights@ worldbank.org. ISBN (paper): 978-1-4648-0355-0 ISBN (electronic): 978-1-4648-0356-7 DOI: 10.1596/978-1-4648-0355-0 Cover photo: © Inked Pixels/Bigstock.com. Used with permission. Further permission required for reuse. Cover design: Critical Stages, Inc. Library of Congress Cataloging-in-Publication Data has been requested. Contents Foreword . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xi Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xiii Abbreviations. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xv Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .1 Changes at the center of the world economy . . . . . . . . . . . . . . . . . . . . . . . . .2 How the rise of the South conditioned development in Latin America and the Caribbean: An interpretation . . . . . . . . . . . . . . . . . . . . . . . . . . . 13 Changing world, new priorities . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31 Structure of the report . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35 Annex OA . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35 Notes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38 1. Three Global Trends That Shaped Latin American and Caribbean Development at the Dawn of the Twenty-First Century . . . . . . . . . . . . . . . . . . . . . . . . . . 41 Set of Facts 1: The weight of the South in the global economy has risen, particularly after 2000, but its rise has not been even across sectors or types of flows. . . . . . . . . 42 Set of Facts 2: The rise of the South has had asymmetric effects on global trade and financial networks. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50 Set of Facts 3: The structure of bilateral trade and financial connections of the South has been generally different from that of the North, with geography and endowments arguably shaping their evolving structure. . . . . . . . . . . . . . . . . . . . . . . . . . 58 Notes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70 2. The Structure of Trade Linkages and Economic Growth . . . . . . . . . . . . . . . . . . 73 Trade and economic growth . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75 The nature of traded goods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81 The nature of trading partners . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 98 Potential frictions affecting trade and growth dynamics . . . . . . . . . . . . . . . . . .106 Annex 2A . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .117 Notes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .118 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .123 v vi CONTENTS 3. Big Emerging Markets, Big Labor Market Dislocations? . . . . . . . . . . . . . . . . . .133 The rise of the South and the restructuring of global markets in manufacturing, agriculture, and mining . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .135 A closer look at manufactures exports and the role of China through the lens of export similarity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .137 Recent trends in manufacturing employment in Latin America and the Caribbean . . . . .142 Labor market adjustment paths in response to the rise of China . . . . . . . . . . . . . .143 Potential distributional implications of China-induced labor market adjustments . . . . .148 Concluding remarks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .150 Notes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .151 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .152 4. The Changing Patterns of Financial Integration in Latin America and the Caribbean. . . .153 The role of Latin America and the Caribbean in international financial transactions . . . .156 Growth in the intensive and extensive margins . . . . . . . . . . . . . . . . . . . . . . .162 Financial flows and trade flows . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .176 Foreign direct investment and GDP growth. . . . . . . . . . . . . . . . . . . . . . . . .181 Annex 4A . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .189 Notes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .191 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .193 5. Ascending with the South Winds: Will Low Saving in Latin America and the Caribbean Be a Drag? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .197 Concepts and literature review: When does saving matter for trend growth? . . . . . . . .202 Looking back: Latin America and the Caribbean under the spell of the interest rate channel . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .205 Looking ahead: Growth-impairing effects of low saving through the exchange rate channel . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .214 Annex 5A The Benchmarking Approach . . . . . . . . . . . . . . . . . . . . . . . . . .219 Annex 5B . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .224 Notes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .225 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .228 Boxes 1.1 Differences in international trade integration: The case of Latin America and the Caribbean and East Asia . . . . . . . . . . . . . . . . . . . . . . . . . . . 65 2.1 Methodology of trade and growth regression estimations . . . . . . . . . . . . . . . 78 2.2 What has driven the dispersion of production tasks away from the North toward the South? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 89 2.3 Asymmetry in the use of temporary trade barriers . . . . . . . . . . . . . . . . . . .113 3.1 Construction of the China effect index . . . . . . . . . . . . . . . . . . . . . . . .140 4.1 How do bilateral data compare with balance of payments data? . . . . . . . . . . .158 4.2 How did the global financial crisis affect investment in and by the region? . . . . . .165 4.3 Model setup and identification strategy . . . . . . . . . . . . . . . . . . . . . . . .185 Figures O.1 The rise of the South. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2 O.2 The South’s share of global trade flows . . . . . . . . . . . . . . . . . . . . . . . . . 3 CONTENTS vii O.3 The South’s share of global capital inflows . . . . . . . . . . . . . . . . . . . . . . .4 O.4 The global trade network . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .5 O.5 Similarity and systemic importance in the global trade network . . . . . . . . . . . . .6 O.6 The global financial network for syndicated bank loans . . . . . . . . . . . . . . . . .7 O.7 Regional clustering in global value chains, 2011. . . . . . . . . . . . . . . . . . . . .8 O.8 Density maps of regional trade networks . . . . . . . . . . . . . . . . . . . . . . . 10 O.9 Composition of foreign assets and liabilities in the South, by region. . . . . . . . . . 12 O.10 Saving, investment, and the current account . . . . . . . . . . . . . . . . . . . . . . 13 O.11 Real U.S. interest rates. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14 O.12 Terms of trade within Latin America and the Caribbean . . . . . . . . . . . . . . . 15 O.13 Export similarity indexes in manufacturing in Brazil and Mexico . . . . . . . . . . . 16 O.14 Effects of the rise of China on gross exports from Latin America and the Caribbean, by sector, 2001–11 average . . . . . . . . . . . . . . . . . . . . 17 O.15 Sectoral composition of cross-border flows in Latin America and the Caribbean, 2003–2011 average . . . . . . . . . . . . . . . . . . . . . . . . 18 O.16 Exports of intermediate goods as share of total exports in three global value chains . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19 O.17 Backward and forward participation in global value chains, 2011 . . . . . . . . . . 21 O.18 Employment shares in the formal and informal manufacturing sectors of Argentina, Brazil, and Mexico . . . . . . . . . . . . . . . . . . . . . . . . . . . 22 O.19 Evolution of wages in Brazil relative to wages in Mexico . . . . . . . . . . . . . . . 23 O.20 Responses to a positive global supply shock in Latin America and the Caribbean and other emerging market regions . . . . . . . . . . . . . . . . 23 O.21 Responses to a global monetary easing in Latin America and the Caribbean and other emerging market regions . . . . . . . . . . . . . . . . . . . . . . . . . . 24 O.22 Domestic saving, real exchange rates, and sovereign risk ratings, 1990–2012 average . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25 O.23 Saving and real exchange rate gaps for higher-income countries in Latin America and the Caribbean . . . . . . . . . . . . . . . . . . . . . . . . . . 26 O.24 Country ratings for selected country groups . . . . . . . . . . . . . . . . . . . . . . 27 O.25 Sovereign risk rating, growth, and investment gaps, 1990–2012 . . . . . . . . . . . 30 1.1 Rise of the South: Share of world GDP, trade, and capital flows. . . . . . . . . . . . 43 1.2 Sectoral composition of trade flows . . . . . . . . . . . . . . . . . . . . . . . . . . 45 1.3 Sectoral composition of financial flows across regions . . . . . . . . . . . . . . . . 46 1.4 Composition of global financial flows across sectors . . . . . . . . . . . . . . . . . 47 1.5 Composition of foreign assets and liabilities in the South, by region . . . . . . . . . 48 1.6 Patterns of net integration into the global economy . . . . . . . . . . . . . . . . . . 50 1.7 Global trade and financial networks . . . . . . . . . . . . . . . . . . . . . . . . . 51 1.8 Similarity in global trade networks . . . . . . . . . . . . . . . . . . . . . . . . . . 54 1.9 Structural equivalence of trade connections . . . . . . . . . . . . . . . . . . . . . . 55 1.10 Extensive margin of South-South connections . . . . . . . . . . . . . . . . . . . . 57 1.11 Regional composition of cross-border connections of countries in Latin America and the Caribbean . . . . . . . . . . . . . . . . . . . . . . . . . . 58 1.12 Clusters in the global trade network . . . . . . . . . . . . . . . . . . . . . . . . . . 60 1.13 Regional composition of cross-border investments . . . . . . . . . . . . . . . . . . 61 1.14 Regional clustering in global value chains, 2011. . . . . . . . . . . . . . . . . . . . 63 1.15 Sectoral composition of bilateral cross-border flows. . . . . . . . . . . . . . . . . . 64 viii CONTENTS B1.1.1 Density maps of trade networks . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66 1.16 Sectoral composition of cross-border flows for Latin America and the Caribbean . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68 2.1 Intraindustry trade. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 83 2.2 Shares of traded goods of different factor intensities. . . . . . . . . . . . . . . . . . 85 2.3 Growth of global value chains . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 88 2.4 The rise of the South in selected global value chains . . . . . . . . . . . . . . . . . . 90 2.5 Technological composition of exports from the South, by region . . . . . . . . . . . 92 2.6 Participation in global value chains . . . . . . . . . . . . . . . . . . . . . . . . . . 94 2.7 Growth effects of the stage of the participation in global value chains. . . . . . . . . 97 2.8 The global trade network . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .100 2.9 Composition of trading partners. . . . . . . . . . . . . . . . . . . . . . . . . . . .102 2.10 Average cost of trading in 2013 . . . . . . . . . . . . . . . . . . . . . . . . . . . .107 2.11 Land transportation, by region, 2011 . . . . . . . . . . . . . . . . . . . . . . . . .108 2.12 Ship and port activity, second half of 2013 . . . . . . . . . . . . . . . . . . . . . .109 2.13 Liner shipping connectivity index in selected countries, 2013 . . . . . . . . . . . . .110 2.14 Share of world air freight transport by selected countries, 2013 . . . . . . . . . . . .111 B2.3.1 Foreign targets of temporary trade barriers imposed by selected countries in Latin America and the Caribbean . . . . . . . . . . . . . . . . . . . . . . . . . .114 3.1 Global export market shares of selected large economies, by sector, 2001, 2006, and 2011 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .136 3.2 Global import market shares of selected large economies, by sector, 2001, 2006, and 2011 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .138 3.3 Export similarity indexes in manufacturing for Argentina, Brazil, and Mexico, 1999–2011. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .139 3.4 Effects of the rise of China on gross exports of selected countries in Latin America and the Caribbean, by sector, 2001–11 . . . . . . . . . . . . . . .141 3.5 Employment shares in the formal and informal manufacturing sectors of Argentina, Brazil, and Mexico, before and after 2000 . . . . . . . . . . . . . . .143 3.6 Simulated short- and long-run impacts of the rise of China on wages in Argentina, Brazil, and Mexico, by sector . . . . . . . . . . . . . . . . . . . . . . .144 3.7 Simulated short- and long-run impacts of the rise of China on informal employment in Argentina, Brazil, and Mexico . . . . . . . . . . . . . . . . . . . . 145 3.8 Simulated short- and long-run impacts of the rise of China on the residual sector in Argentina, Brazil, and Mexico . . . . . . . . . . . . . . . . . . . . . . . .146 3.9 Evolution of relative wages in Brazil to Mexico, 2001–09 . . . . . . . . . . . . . . .147 B4.1.1 Comparison between bilateral and balance of payments account data on mergers and acquisitions and greenfield investment, 2003–11 . . . . . . . . . . .158 4.1 Cross-border investment shares by Latin America and Caribbean (LAC) countries in North, South, and other LAC countries, by type of investment, selected years . . . . . . . . . . . . . . . . . . . . . . . . .160 4.2 Cross-border investment shares by North, South, and Latin America and Caribbean (LAC) countries, by type of investment, selected years . . . . . . . . . . .161 4.3 Cross-border investment to and from countries in Latin America and the Caribbean, selected years . . . . . . . . . . . . . . . . . . . . . . . . . . .163 4.4 Cross-border holdings of and extensive margin for portfolio investments, 2001–11 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .166 4.5 Cross-border flows of and extensive margin for syndicated loans, 1996–2012 . . . .167 4.6 Cross-border flows of and extensive margin for mergers and acquisitions, 1990–2011 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .168 CONTENTS ix 4.7 Cross-border flows of and extensive margin for greenfield investment, 2003–11 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .170 4.8 Extensive margin of cross-border financial flows within Latin America and the Caribbean, by type of investment, selected years . . . . . . . . . . . . . . .174 4.9 Extensive margin of cross-border financial flows from Latin America and the Caribbean to countries in other regions of the South, by type of investment, selected years. . . . . . . . . . . . . . . . . . . . . . . . . .175 4.10 Sectoral composition of cross-border financial flows to and from Latin America and the Caribbean, by type of investment, selected years . . . . . . . . . . . . . . .177 4.11 Sectoral composition of cross-border financial flows to and from Latin America and the Caribbean, by type of investment, 2003–11 average . . . . . . . . . . . . .178 4A.1 Number of active cross-border connections, by type of investment and region . . . .189 4A.2 Sectoral composition of cross-border financial flows to and from Latin America and the Caribbean, by type of investment and subregion, 2003–11 average . . . . . .190 5.1 Growth paths of Latin America and the Southeast Asian Tigers, 1950–2014 . . . . .198 5.2 Growth rates in selected emerging economies, 2003–14 . . . . . . . . . . . . . . . .198 5.3 Domestic saving rates in selected economies, adjusted for per capita GDP, 2012 . . .199 5.4 External competitiveness (Big Mac index), adjusted for per capita GDP, 2012 . . . .200 5.5 The three channels linking saving and growth . . . . . . . . . . . . . . . . . . . . .203 5.6 Saving rates of higher-income countries in Latin America and the Caribbean and middle-income countries in Southeast Asia . . . . . . . . . . . . . . . . . . . .205 5.7 Domestic saving and real exchange rate gaps . . . . . . . . . . . . . . . . . . . . .206 5.8 Domestic saving and sovereign risk rating gaps . . . . . . . . . . . . . . . . . . . .207 5.9 Real exchange rate and growth gaps in selected country groups . . . . . . . . . . . .208 5.10 Saving and real exchange rate gaps for higher-income countries in Latin America and the Caribbean . . . . . . . . . . . . . . . . . . . . . . . . . .209 5.11 Saving and real exchange rate gaps in selected country groups . . . . . . . . . . . .210 5.12 Country ratings for selected country groups . . . . . . . . . . . . . . . . . . . . . .210 5.13 Policy-adjusted gaps for high-saver and low-saver higher-income countries in Latin America and the Caribbean, 1981–2012 . . . . . . . . . . . . . . . . . . .211 5.14 Real U.S. interest rate . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .212 5.15 Incidence of crises in Latin America and Southeast Asia, 1980–2010 . . . . . . . . .212 5.16 Impulse responses in Latin America and the Caribbean and other emerging market economies to positive global demand shocks . . . . . . . . . . . . . . . . .213 5.17 Impulse responses in Latin America and the Caribbean and other emerging market economies to global monetary easing . . . . . . . . . . . . . . . . . . . . .214 5.18 Impulse responses in Latin America and the Caribbean and other emerging market economies to positive global supply shocks . . . . . . . . . . . . . . . . . .215 5.19 Saving and exchange rate gaps for higher-income countries in Latin America and the Caribbean, 2011–12 averages . . . . . . . . . . . . . . . . . . . . . . . . .216 5.20 Composition of foreign assets and liabilities in selected countries in Latin America and the Caribbean, 1990–2011 . . . . . . . . . . . . . . . . . . .217 Tables OA.1 Country group composition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35 2.1 Regression results on the effect of the nature of traded goods on economic growth . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 80 2.2 Regression results on the effects of the composition of trading partners on economic growth . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .104 2A.1 Data description and sources . . . . . . . . . . . . . . . . . . . . . . . . . . . . .117 x CONTENTS 3.1 Percentage of workers in bottom 40 percent of income distribution in Argentina, Brazil, and Mexico, by sector . . . . . . . . . . . . . . . . . . . . . .149 4.1 Cross-border investment, by pairs of regions and type of investment (annual average, millions of 2011 U.S. dollars) . . . . . . . . . . . . . . . . . . . .157 4.2 Shares of cross-border investment by source and receiver region, normalized by GDP of Latin America and the Caribbean (annual average, percent) . . . . . . . .162 4.3 Intensive margin of financial connections across regions . . . . . . . . . . . . . . .171 4.4 Extensive margin of cross-border financial flows . . . . . . . . . . . . . . . . . . .172 4.5 Region-to-region financial flows . . . . . . . . . . . . . . . . . . . . . . . . . . .172 4.6 Global financial and trade flows. . . . . . . . . . . . . . . . . . . . . . . . . . . .180 4.7 Foreign direct investment and labor productivity growth in the host country . . . . .187 5A.1 Country group composition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .220 5A.2 Data description and sources . . . . . . . . . . . . . . . . . . . . . . . . . . . . .222 5A.3 Data definitions and sources. . . . . . . . . . . . . . . . . . . . . . . . . . . . . .223 5A.4 Signs and length restrictions on global and domestic shocks. . . . . . . . . . . . . .223 Foreword The dynamics of the world economy have average income per capita has remained changed radically and the once immutable barely 30 percent of that of the United States. assumptions of the global trade and finan- In other words, the region has been unable cial order no longer hold fast. In the last two to narrow a gaping income disparity with its decades alone, wealth has shifted so pro- northern neighbor. foundly that the simple, old North-South This is not to say that Latin America has hierarchy—where the North were the rich been unable to grow. In fact, during the com- few and the South were the many poor coun- modity boom of the 2000s, average growth tries of the world—is no longer a given. In rates reached nearly 5 percent. Moreover, fact, in 1990, the majority of the world popu- income growth of the poorest 40 percent was lation, 62 percent, lived in poor countries. As higher in Latin America and the Caribbean of 2010, 72 percent of the world’s population than in any other region of the world, relative lived in middle-income countries. to the total population, making growth also Such tremendous transformation is the equitable. inspiration for the World Bank’s latest Global economic activity, however, has regional flagship report for Latin America slowed and medium-term growth prospects and the Caribbean, Latin America and the have diminished. Latin America is now in its Rising South: Changing World, Changing fourth year of growth deceleration, and it is Priorities. As an in-depth look at the region’s expected to grow below 1 percent in 2015. expanding global connections in trade and This poses brand new challenges, particu- finance, and a sober assessment of its prom- larly as the conditions that led to the good ise and challenges, the report is an import- years of the 2000s are not with us anymore. ant contribution in and of itself; at the same Current global conditions pose similar time, as a report that tracks global trends, challenges to all middle-income countries, it also provides an invaluable analysis that not only those in Latin America. Indeed, the World Bank is uniquely positioned to disappointing growth in major emerging undertake. economies around the world raises import- While these global trends were the inspira- ant concerns, particularly considering that tion, the motivation behind this report is the two thirds of the extreme poor in the world urgent need to disentangle the complicated still live in middle-income countries. For knot of Latin America’s growth problem. the World Bank Group, a global institution For more than 100 years, Latin America’s committed to eradicate extreme poverty by xi xii FOREWORD 2030 and to boost prosperity for the bot- therefore our hope that a profound look at tom 40 percent of the population, these are the way Latin America—and the world— crucial challenges. have been integrating will help shed a light The web of connections that have multi- on the way forward. In other words, our plied throughout the world from the North expectation is that a clearer understanding to the South, from the South to the North, of how the South has been rising—and how and, perhaps more significantly, from the it has not—will help those countries break South to the South represents an import- out of their middle-income status and move ant change over the past two decades. It is closer to the group of rich nations. Jorge Familiar Vice President for Latin America and the Caribbean The World Bank Acknowledgments T his report was prepared by a core team Rodríguez-Clare, David Rosenblatt, and comprising Augusto de la Torre, Tati- Shahid Yusuf. We are also grateful for valu- ana Didier, Alain Ize, Daniel Leder- able comments and insights received from man, and Sergio L. Schmukler. Additional Paulo Bastos, Laura Chioda, Ana M. Fer- contributions were made by Erhan Artuç, nandes, Eduardo Fernández-Arias, Michael Chad Bown, Fernando Broner, Constantino Ferrantino, Margaret Ellen Grosh, Jose Hevia, Ha Nguyen, Samuel Pienknagura, Luis Irigoyen, Ayhan Kose, Gian Maria Luis Servén, and Ganesh Wignaraja. We Milesi-Ferretti, Marc A. Muendler, Ana L. thank Magali Pinat for invaluable help in Revenga, Sergio Urzua, and other partic- putting together and coordinating the doc- ipants in the authors’ workshop that took uments that constitute this 2015 Regional place on February 27–28, 2014. While we Flagship Report. We are particularly grate- are grateful for the guidance and com- ful for the truly outstanding research assis- ments received, the core team is responsi- tance provided by Matias Moretti (Chapter ble for all remaining errors, omissions, and 4), Magali Pinat (Chapters 1 and 2), and interpretations. Diego Rojas (Chapter 3), who in addition Book design, editing, and production were co-authored some of the background papers coordinated by the World Bank’s Publish- for this Report. We also benefitted from able ing and Knowledge department under the research assistance at different stages of the supervision of Patricia Katayama and Mark project provided by Diego Barrot, Julia Got- Ingebretsen. We also appreciate the assis- tlieb, Lucas Rusconi, Martin Sasson, Tanya tance provided by Mauro Lopes Mendes Taveras, and Shajuan Zhang. de Azeredo, Sergio Jellinek, and Marcela The team was fortunate to receive superb Sanchez-Bender on the report’s publication advice and guidance from the following and dissemination activities. Finally, we peer reviewers: Eduardo Cavallo, Tito Cor- thank Ruth Delgado Flynn and Jacqueline della, Barry Eichengreen, Caroline Freund, Larrabure Rivero for unfailing administra- Aart Kraay, William Maloney, Andrés tive support. xiii Abbreviations BoP balance of payments CPIS Coordinated Portfolio Investment Surveys EAP East Asia and Pacific ECA Europe and Central Asia ER exchange rate ES endogenous saving FDI foreign direct investment FVA foreign value added GDP gross domestic product GVC global value chain G-7 Group of Seven i.i.d. independent and identically distributed IFS International Financial Statistics IIT intraindustry trade IPR intellectual property rights IR interest rate ISIC International Standard Industrial Classification LAC Latin America and the Caribbean M&A mergers and acquisitions MENA Middle East and North Africa MNC multinational corporation OLS ordinary least squares PTA preferential trade agreement R&D research and development RCA relative comparative advantage RTA regional trade agreement SA South Asia xv xvi ABBRE VIATIONS S-GMM system generalized method of moments SITC Standard International Trade Classification SSA Sub-Saharan Africa SVAR structural vector autoregression TEU 20-foot equivalent unit TTB temporary trade barrier UNCTAD United Nations Conference on Trade and Development VAR vector autoregression WTO World Trade Organization Overview T he world economy is not what it used Made in China?” (De la Torre and others to be 30 or even 15 years ago. The 2011). While China was the sole focus then, rise of the South—that is, the growing the analysis here is deeper and broader, not economic influence of emerging economies— least because it covers the evolving role of has changed the global economic landscape.1 emerging economies more generally. The changes have been deep and most likely This report argues that as the world econ- permanent. They reflect not only the grow- omy has irreversibly changed, LAC has been ing economic heft of the South, given its sub- adjusting to the associated global economic stantially higher growth rates with respect to shocks, both commercial and financial. The the North (that is, the advanced economies), adjustment process has been conditioned by but also structural changes. The South has LAC’s trade and financial structures and become a driver of global economic trends reflected in the observed patterns of struc- by playing a role that is qualitatively different tural change. Key challenges have emerged from that of the North. At the epicenter of for the region, particularly because the these changes has been China. changes may not have improved the region’s This report focuses on the restructuring of prospects for long-term economic growth. the global economy and its implications for Simply put, economic policy priorities in the the development and policy priorities of Latin region have evolved in response to worldwide America and the Caribbean (LAC). It exam- changes even as these changes have exacer- ines how the global economy has changed, bated some of the region’s long-standing especially with regard to the patterns of development challenges, such as those associ- international trade and financial integration ated with its dependence on mineral and agri- as well as the differential roles played by the cultural commodities and its comparatively large emerging economies and the traditional low saving rates. The debate in the region economic powers. Some of these themes were over public policy priorities in the context of explored, in a preliminary fashion, in the a new global landscape will thus likely inten- September 2011 issue of the LAC Region’s sify, with the growth agenda at its core. semiannual report series, “Latin America The rest of this overview addresses the and the Caribbean’s Long-Term Growth: “what,” the “how,” and the “so what” 1 2 LATIN AMERICA AND THE RISING SOUTH questions associated with the rise of the and foreign investment as potential drivers South and its implications for LAC. The of growth and productivity; labor market overview is organized in three main sec- frictions, which make economic adjustments tions. The first documents salient features of sluggish and thus reduce the potential gains the new global economic order by focusing from globalization; and the region’s notori- on the rising prominence of emerging econ- ously low national saving rates, which may omies. It characterizes the tectonic shifts hamper long-term growth by undermining in the global economy, including by look- external competitiveness. ing at the data through the lens of network analysis. It then examines the fundamental change in the role of the South in the global Changes at the center of the economy and highlights key dimensions of world economy heterogeneity within the South. To fully understand the implications of the The second section provides an economic economic rise of the South, it is helpful to interpretation of how the changes at the distinguish between the economic weight of heart of the global economy are conditioning emerging economies, the extent of trade and growth and employment prospects in LAC. financial integration of these countries, and This narrative posits that, from the point of the different roles played by the North and view of LAC, the rise of the South manifested South countries that are systemically import- itself as a set of economic shocks working ant for the world economy. through commercial and financial channels. The impacts of these shocks varied across the Tectonic shifts in the global region, depending on countries’ initial trade economic landscape structures, resource endowments, degree of financial globalization, and saving patterns, For most of the 20th century, global eco- among other factors. nomic activity was concentrated in the The third section assesses broad policy developed North (composed of Canada, the areas that, given the rising South phenome- United States, the Western Europe coun- non, should find their way to the top of the tries, and Japan, which joined the pack region’s growth-oriented reform agenda. only after World War II). Since the dawn of Among these areas are the structure of trade the 21st century, the South (defined as all FIGURE O.1 The rise of the South a. World GDP b. World trade 100 100 90 90 Share of world trade (%) 80 80 Share of world GDP (%) 70 70 60 60 50 50 40 40 30 30 20 20 10 10 0 0 19 1 19 4 19 7 19 0 73 19 6 19 9 19 2 19 5 19 8 19 1 19 4 20 7 20 0 03 20 6 20 9 12 19 1 19 4 19 7 19 0 73 19 6 19 9 19 2 19 5 19 8 19 1 19 4 20 7 20 0 03 20 6 20 9 12 6 6 6 7 7 7 8 8 8 9 9 9 0 0 0 6 6 6 7 7 7 8 8 8 9 9 9 0 0 0 19 19 20 19 19 20 South China North Sources: Calculations based on data from World Development Indicators (WDI) and Direction of Trade Statistics (DOTS). Note: The North includes the G-7 members and Western Europe countries. The South includes all other economies. G-7 = Group of Seven; GDP = gross domestic product. OVERVIEW 3 developing economies not in the North), led 48 percent (figure O.2, panel a), and its share by China and other large emerging econ- of global imports of primary (agricultural omies, has risen with surprising speed. In and mineral) goods expanded from 32 per- fact, several South countries have become cent to 47 percent (figure O.2, panel b). An major, systemically important players in the acceleration of financial globalization accom- global economy. The gross domestic prod- panied the rise of the South in commercial uct (GDP) of the South, which represented flows. The South’s share of global capital about 20 percent of world GDP between inflows (including foreign direct investment the early 1970s and the late 1990s, doubled [FDI]) rose from about 18 percent in the to about 40 percent by 2012, with China 1970s to 25 percent in the 1990s and to more alone accounting for 12 percent of global than 50 percent by 2012 (figure O.3). GDP (figure O.1, panel a). The increase in the economic weight of The rising share of the South in global the South is likely here to stay: it is probably GDP was accompanied by increasing influ- neither short lived nor reversible. Although ence in international trade and finance. long-term economic forecasts are notoriously Indeed, although the secular process of uncertain, current projections suggest that globalization of the South had long been the South will continue to gain importance in advancing, the 2000s saw a notable intensi- the world economy. According to the World fication of this process. The South’s partici- Bank’s 2013 Global Development Horizons, pation in global trade rose from 24 percent in the share of the South in global GDP will 1970 to 35 percent in 2000 and 51 percent in reach 55 percent by 2025. A 2012 report by 2012 (figure O.1, panel b). This advance was the U.S. National Intelligence Council proj- associated with major transformations in the ects this share to reach 70 percent by 2030. structure of world trade, as the weight of the The Asian Development Bank forecasts South varied across sectors. Between 2000 that the share of exports from the South and 2012, the South’s share of global exports will rise to 64 percent of global exports by of manufactures increased from 32 percent to 2030 (Anderson and Strutt 2011). The 2013 FIGURE O.2 The South’s share of global trade flows a. Share of exports of manufactured goods b. Share of imports of primary goods 50 50 45 Share of world imports of Share of world exports of manufactured goods (%) 40 40 primary goods (%) 35 30 30 25 20 20 15 10 10 5 0 0 00 01 02 03 04 05 06 07 08 09 10 11 12 00 01 02 03 04 05 06 07 08 09 10 11 12 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 Other South China India Russian Federation Other South China India Korea, Rep. Korea, Rep. Poland Turkey Czech Republic Thailand Australia Lithuania Singapore Vietnam Malaysia Source: Calculations based on data from Comtrade database. Note: The eight South countries that gained the most in market share between 2000 and 2012 are shown separately from the rest of South countries. The North includes the G-7 mem- bers and Western Europe countries. The South includes all other economies. G-7= Group of Seven. 4 LATIN AMERICA AND THE RISING SOUTH FIGURE O.3 The South’s share of global capital inflows have yet to be linked to a wide set of other countries, especially in terms of financial 100 connections. Indeed, only 18 percent of the 90 potential South-South connections related to portfolio flows were active in 2011.4 80 Share of world capital inflows (%) 70 The fundamental change in the global 60 role of the South 50 Changes in relative economic weight provide a bird’s-eye view of the rise of the South. But, 40 impressive as they are, they do not illustrate 30 the full scale of the economic shifts in the global landscape. Further insights into the 20 nature of the rise of the South emerge when 10 trade and financial connections are viewed from a global network perspective. Four key 0 stylized facts arise from this approach (for a 19 0 72 19 4 19 6 19 8 19 0 19 2 19 4 19 6 19 8 19 0 19 2 19 4 19 6 20 8 20 0 20 2 20 4 20 6 20 8 20 0 12 7 7 7 7 8 8 8 8 8 9 9 9 9 9 0 0 0 0 0 1 19 19 more detailed analysis, see chapter 1 of this South China North report). First, the North is no longer the center of Source: Calculations based on data from Balance of Payments Statistics (BOPS). Note: Gross capital inflows include portfolio, banking, and foreign direct investment flows. The the global trade network and the South is no North includes the G-7 members and Western Europe countries. The South includes all other econ- longer its periphery. Indeed, several econo- omies. G-7= Group of Seven. mies from the South have become part of what can be empirically characterized as the “cen- Global Development Horizons projects that ter” of global trade. This momentous change by 2025 the South will account for 63 per- is highlighted in figure O.4, which shows the cent of world capital inflows and 80 percent global trade network in 1980 and 2012. Each of world capital outflows. node in the graphs represents a country, and As the South gained weight in the global each link corresponds to exports from one economy, the number of its bilateral economic country to another (indicated by the arrows). connections proliferated. These ties increased Connections that are trivial in magnitude are in every direction, but new South-South con- not graphed, but once graphed, each con- nections rose more rapidly than North-South nection has the same weight. The greater the linkages in both trade and finance. In 1980, number of its connections to other countries, the number of active South-South trade con- the more centrally located a country is. nections was 40 percent of all possible con- The change has been remarkable. In 1980, nections (the number of connections that only a few North countries—the United would exist if every South country were con- States, some Western Europe countries, and nected to every other South country). This Japan—stood at the center of the global trade figure rose to 46 percent in 1990 and 70 per- network. In contrast, by 2012, several South cent in 2012. Trade linkages between North countries—including not only China but also and South countries expanded less rapidly Brazil, India, the Russian Federation, South (from 92 percent in 1980 to 96 percent in Africa, and Turkey—had moved to the center. 1990 and 98 percent in 2012), at least in part Second, at the center of the global trade because they had been almost fully exploited network, the role played by countries from since the 1980s.2 Similar trends are observed the South and countries from the North across different types of financial flows.3 To differs. This stylized fact is illustrated in be sure, this process is far from mature, as a figure O.5, which shows the relative (rather significant number of countries in the South than absolute) importance of each country OVERVIEW 5 FIGURE O.4 The global trade network a. 1980 b. 2012 BRA North countries Latin America and the Caribbean Other South countries Source: Calculations based on data from DOTS. Note: Networks are drawn using the Kamada-Kawai algorithm. Each node represents a country. Each link corresponds to an active connection between a pair of countries. Arrows indicate the direction of these connections. The North includes the G-7 members and Western Europe countries. Other South includes all other economies except Latin America and Caribbean countries. Only trade flows (exports) greater than $10 million in 1980 or greater than $100 million in 2012 are shown. The figure thus ignores very small countries. It would show similar results if these connections were reported. G-7 = Group of Seven. FIGURE O.5 Similarity and systemic importance in the global trade network a. 1980 Similarity in trade structures Systemic relevance in global trade b. 2012 Similarity in trade structures Systemic relevance in global trade Source: Calculations based on data from DOTS. Note: Each node represents a country. Each link corresponds to an active trade connection between a pair of countries. Arrows at the end of each link capture the direction of these connections. Trade connections are measured as exports as a share of total exports of the source country. Only shares greater than 1 percent are reported. The distance between countries reflects similarity in the structure of their trade connections: the closer countries are to one another, the more alike they are in terms of export shares. Countries capturing a larger share of other countries’ exports and connected with a larger number of trading partners appear on the right-hand side of the figure (more systemically relevant countries in global trade). The smaller the distance between two countries along the vertical dimension, the more similar the structure of their trade connections across other members of the network. OVERVIEW 7 in the global trade network. The vertical dis- across trading partners. The right side of the tance between countries in the figure reflects figure resembles a star, with small groups of the degree of similarity in the structure of central countries placed at a certain verti- their trade connections, whereby more sim- cal distance from one another. The Russian ilar countries are grouped closer together. 5 Federation and Turkey, for example, are not The farther to the right of the figure a coun- located near any North core country from try is located, the greater its importance to Europe, and Japan is not close to either China the global trade network.6 or the Republic of Korea. The implication is Panel a of figure O.5 shows that in 1980 that systemically important South countries only North countries were clustered toward play a different role from the role played by the right of the graph, thus indicating that North countries in the global trade network. they were of greatest systemic importance to These different roles seem to be inherently the global trade network. In addition, these linked to fundamental differences in factor countries were very close to one another endowments, trade, production, and aggre- along the vertical dimension, reflecting a gate demand structures, as discussed below. high degree of similarity in the structure of Third, there is a notable asymmetry in the their trade connections with other countries patterns of change in global trade and finan- in the network. cial networks. In the sphere of trade, the tra- The global trade network in 2012 shifted ditional overlap between the North and the dramatically (figure O.5, panel b). Sev- “center” (and the South and the “periphery”) eral countries from the South appeared on no longer holds. In contrast, in the sphere of the right side of the figure, indicating their finance, countries from the North still stand increased systemic relevance to world trade. alone at the center, as illustrated in figure O.6 However, they remained somewhat distant for syndicated bank loans. A similar picture (along the vertical dimension) from the other emerges for portfolio investments, merg- (North) countries on the right side of the ers and acquisitions (M&A), and greenfield figure, reflecting differences in trade shares investment flows. Whether this asymmetry FIGURE O.6 The global financial network for syndicated bank loans a. 1996 b. 2012 North Latin America and the Caribbean Other South Source: Calculations based on data from SDC Platinum. Note: Networks are drawn using the Kamada-Kawai algorithm. Each node represents a country. Each link corresponds to an active connection (a positive flow of investments) between a pair of countries. Arrows indicate the direction of these connections. The North includes the G-7 members and Western Europe countries. Other South includes all other economies except Latin America and Caribbean countries. G-7 = Group of Seven. 8 LATIN AMERICA AND THE RISING SOUTH proves transitory is debatable, although most of production activities belonging to the observers agree that it is unlikely to be dis- same production processes across countries. lodged soon, for several reasons. For start- As GVCs have gained prominence on the ers, there is broad recognition that the U.S. international trading scene, exports of final dollar continues and will continue to have a products have become increasingly composed stronghold as both the privileged currency of imports of intermediate inputs. To date, for international contracts and the safe haven GVCs are mostly regional, not global. The in times of global risk aversion. In addition, foreign value added (FVA) content in exports the scale and network effects associated with typically originates in neighboring countries the dominance of the advanced financial cen- (figure O.7).7 For example, about 56 percent ters (including New York, London, Frank- of the FVA in the exports of East Asian coun- furt, Tokyo) will not be easy for the South tries come from other East Asian economies, to overcome. This trade-finance asymme- and more than 72 percent of the FVA in the try in global networks stands in sharp con- exports of European countries come from trast to broad historical developments since other European economies. There is also the Industrial Revolution and throughout clustering—albeit less intense—across coun- most of the 20th century, when countries tries within LAC subregions. For instance, that became important trading powers also imports from other South American coun- became important international financial tries represent about 30 percent of the FVA in centers. the exports of South America. Fourth, despite an increase in the number of connections around the world, there is a The heterogeneity of the South significant degree of regional (geographic) clustering within global trade and financial The rise of the South in global economic networks. Underpinning these clustering pat- affairs conceals important differences across terns has arguably been the development of South countries. Four types of heterogeneity global value chains (GVCs)—the distribution are noteworthy. The first is differences in FIGURE O.7 Regional clustering in global value chains, 2011 45 40 35 Share of exports (%) 30 25 20 15 10 5 0 East Asia Europe North and Central South America America Intraregional East Asia Europe North and Central America South America Sources: Calculations based on data from Eora MRIO and WDI. Note: Figure shows the geographical composition of sources of foreign value added used in a country’s exports, scaled by the country’s exports. OVERVIEW 9 the changes in export and import shares of India, Korea, Poland, and Turkey) are outside the South (recall figure O.2). The rise of the LAC. As such, LAC gained global relevance South implied a growing share of the South as a major commodity exporting region even (as a whole) in global manufacturing exports. though it lost relevance as a manufacturing But only a subgroup of South countries car- exporter. ried the load in this regard, with China the A second important dimension of het- leader by a wide margin. China’s share in erogeneity within the South is the contrast global manufacturing exports increased by between LAC and the East Asian economies more than 10 percentage points, from slightly in terms of the density of their regional trade less than 5 percent in 2000 to more than 15 networks. Figure O.8 highlights this feature percent in 2012. In contrast, the other top 20 by providing snapshots of the regional trade South countries in terms of their increases networks of these two regions in 1980 and in global shares—a group that includes 2012. Each regional trade network includes Brazil and Chile—increased their share of (as nodes) all countries of the region plus the global manufacturing exports as a group by five countries from the rest the world that are only about 8 percentage points. The shares the largest trading partners for each regional of world manufacturing exports of several network.8 large South countries (for example, Malay- In 1980 the trade networks of LAC and sia, Mexico, and the Philippines) actually East Asia were similar: they were thin, declined. unbalanced, and centered on a few domi- The rise of the South also featured a sub- nant North economies. Japan and the United stantial increase in its share of trade (exports States were the only two dense nodes in the and imports) of primary (mineral and agricul- 1980 snapshot of the East Asian network, tural) products. But cross-country differences and the United States was the sole dense node within the South are stark. In particular, the in the 1980 LAC network. set of South countries whose shares in com- By 2012 the two regional networks had modity exports rose most significantly has diverged. The East Asian network had little overlap with the set of South countries become substantially denser and more bal- whose shares of commodity imports rose. anced, with high-density connections distrib- In contrast, the set of South countries whose uted rather evenly across numerous countries shares of manufacturing exports rose signifi- (nodes), including not just Japan, the United cantly (virtually all of which are outside LAC) States, and China but also Korea, Malaysia, has greater overlap with the set of South Singapore, and Thailand. In contrast, the countries whose shares of commodity imports 2012 snapshot of the LAC trade network rose. Australia, Brazil, and the Russian Fed- was almost as thin as it was in 1980, and it eration jointly accounted for the largest gains remained dominated by the United States, in the shares of global primary exports (their with Brazil a very distant second. A signif- share rose from 13 percent in 2000 to 23 icant change between 1980 and 2012 was percent in 2012). Other top 20 commodi- that China joined the LAC network, albeit at ty-exporting countries from the South include a comparatively low density.9 Azerbaijan, India, Kazakhstan, and several The large difference in regional network LAC countries (Bolivia, Chile, Colombia, densities in 2012 reflects trade connections Ecuador, Peru, and Uruguay). China stands within East Asia that became multidirec- out as a giant commodity importer: its share tional (that is, intense in the direction of of global imports of agricultural and mineral virtually every country within the network). commodities rose from less than 4 percent in In contrast, connections within the LAC net- 2000 to more than 15 percent in 2012. All work have remained largely bi-directional, other South countries with rising manufac- linking LAC countries mainly with the turing export shares that also increased their United States and secondarily with China shares of imports of commodities (such as (and, within the South America subregion, 10 LATIN AMERICA AND THE RISING SOUTH FIGURE O.8 Density maps of regional trade networks a. The Latin American network, 1980 b. The Asian network, 1980 (ccontinued) (continued) OVERVIEW 11 FIGURE O.8 Density maps of regional trade networks (continued) c. The Latin American network, 2012 d. The Asian network, 2012 Sources: De la Torre, Didier, and Pinat 2014 and DOTS. Note: Figure shows the density maps of two regional trade networks based on bilateral exports, measured as a share of total exports of the sending country in 1980 and 2012. The density of a country in these maps depends on the number of neighboring countries and the economic distance between countries. The node density is translated into colors using a red-green-blue scheme in which red indicates the highest density and blue the lowest. Each country is represented by its three-letter acronym. See box 1.1 in chapter 1 of this report for technical details. 12 LATIN AMERICA AND THE RISING SOUTH FIGURE O.9 Composition of foreign assets and liabilities in Brazil). The density of connectivity in the the South, by region East Asian network also suggests strong feedback effects, whereby tighter trade con- a. LAC-7 nections within East Asian emerging econ- 10 omies boost trade with advanced countries 5 in the North and vice versa. In contrast, 0 LAC countries (with the possible exceptions –5 of Mexico and Costa Rica) seem to signifi- Percent of GDP –10 cantly underexploit the potential for comple- –15 mentarities and mutually reinforcing effects –20 between intraregional trade and global –25 trade. These different patterns may be linked –30 to the fact that East Asian countries partici- –35 pate much more actively in GVCs than LAC 1990 1993 1996 1999 2002 2005 2008 2011 countries do. Year A third salient dimension of heteroge- b. Asia-7 neity concerns the asymmetric shifts in the 20 net debtor-creditor positions with respect 10 to the rest of the world for different emerg- ing regions in the South. LAC and East Asia 0 followed a similar pattern in this respect, in Percent of GDP –10 sharp contrast with countries from Eastern Europe and Central Asia (figure O.9). During –20 the 2000s, there was a major shift from debt –30 to equity in the external net liability posi- –40 tions of East Asia and LAC (in the context of the rise of the South). In contrast, Eastern –50 1990 1993 1996 1999 2002 2005 2008 2011 Europe and Central Asia shifted its position Year toward debt liabilities. Regarding debt contracts, East Asia and c. ECA-7 5 LAC went from being large net debtors with 0 respect to the rest of the world in the 1990s –5 to significant net creditors during the 2000s. This change reflected a strengthening of –15 Percent of GDP macrofinancial policy frameworks, which –25 entailed a process of external debt reduction by governments coupled with self-insurance –35 through accumulation of international –45 reserves by central banks.10 It also reflected the continued presence of large current –55 account surpluses, particularly among the 1990 1993 1996 1999 2002 2005 2008 2011 Year high-saving East Asian economies. Over the same period, both East Asia and Net debt position Net equity position LAC became more active users of foreign equity finance, which led to rising net debtor Source: Calculations based on updated and extended version of dataset constructed by Lane and positions in risk-sharing equity contracts Milesi-Ferretti 2007. Note: Ratios are calculated at the country level and then averaged across countries (simple average) (particularly FDI) with respect to the rest of between 1990 and 2011. LAC-7: Argentina, Brazil, Chile, Colombia, Mexico, Peru, and Uruguay. Asia-7: the world. The equity-laden position LAC China, India, Indonesia, the Republic of Korea, Malaysia, Philippines, and Thailand. ECA-7: Croatia, the Czech Republic, Hungary, Lithuania, Poland, the Russian Federation, and Turkey. GDP = gross and East Asia achieved in the 2000s arguably domestic product. represents a more resilient form of integrating OVERVIEW 13 into often volatile international financial mar- FIGURE O.10 Saving, investment, and the current account kets than the debt-laden external net liability position of Eastern Europe and Central Asia. a. LAC-7 A fourth dimension of heterogeneity that 40 is key to understanding the implications of 30 the rise of the South is the differences in 20 Percent of GDP the relative importance of domestic versus 10 external demand in macroeconomic aggre- gates. The contrast is sharpest between 0 LAC and East Asia. While in LAC domestic –10 demand largely drives the economy, in East –20 Asia external demand is a dominant force. –30 That LAC exhibits domestic demand–driven –40 macroeconomic patterns implies an excess 80 82 84 86 88 90 92 94 96 98 00 02 04 06 08 10 12 19 19 19 19 19 19 19 19 19 19 20 20 20 20 20 20 20 of aggregate demand over national income and, hence, typically low saving rates and b. EAP-5 a penchant for current account deficits 40 (figure O.10). The external demand–driven 30 patterns of East Asia imply an excess of 20 Percent of GDP national income over aggregate demand and, hence, typically high domestic saving 10 rates and current account surpluses. The 0 macroeconomic patterns of the emerging –10 economies of Eastern Europe and Central –20 Asia are more similar to LAC than to East –30 Asia. As argued below, a macroeconomic pattern that relies on external demand, and –40 80 82 84 86 88 90 92 94 96 98 00 02 04 06 08 10 12 therefore high national saving rates, may 19 19 19 19 19 19 19 19 19 19 20 20 20 20 20 20 20 be more conducive to seizing the potential c. ECA-6 growth benefits associated with the rise of 40 the South. 30 20 Percent of GDP How the rise of the South 10 conditioned development in Latin America and the 0 Caribbean: An interpretation –10 –20 The rise of the South has left a noticeable mark upon the world economy. The pre- –30 ceding discussion highlights the heteroge- –40 80 82 84 86 88 90 92 94 96 98 00 02 04 06 08 10 12 neity of structural economic characteristics 19 19 19 19 19 19 19 19 19 19 20 20 20 20 20 20 20 within the South before and during its rise, Investment Saving Current account especially since 2000. This section interprets these global and regional trends, based on the Source: Calculations based on data from the IMF’s International Financial Statistics (IFS). evidence presented in this report. Note: Simple regional averages are presented. LAC-7 includes Argentina, Brazil, Chile, Colombia, Mexico, Peru, and Uruguay. EAP-5 includes Indonesia, the Republic of Korea, Malaysia, the Philip- From the viewpoint of small open- pines, and Thailand. ECA-6 includes Croatia, the Czech Republic, Hungary, Lithuania, Poland, and economies, including LAC countries, the Turkey. GDP = gross domestic product. rise of the South can be understood as having set three types of global shocks in motion: a supply shock, a demand shock, 14 LATIN AMERICA AND THE RISING SOUTH and a financial shock. Both the demand and of manufactures, led by but not limited to supply shocks have been associated with China. This shock presumably lowered the the asymmetric rise of the South across (quality-adjusted) prices of manufactured industries and trade flows (exports ver- goods and thus dampened global inflationary sus imports). The financial shock has been pressures. The shock can be interpreted as related to the recycling of savings from the emanating from an increase in the number of emerging South. manufacturing workers engaged in interna- LAC countries responded differently to tional trade, whose labor services were previ- these shocks as a result of differences in ously not integrated into the global economy initial conditions, including factor endow- (arguably the case of China before it joined ments, initial trade structures, and macro- the World Trade Organization in 2001). economic frameworks. As it is difficult to For LAC economies, this shock implied precisely identify the direction of causality, increased international competition for var- this narrative provides an interpretation of ious manufacturing industries. It thus insti- the facts and statistical findings rather than gated structural changes across sectors as a model of how the world economy has been well as within LAC’s manufacturing sector. operating. The resulting decline in the relative prices of This section thus characterizes the rise manufactured goods was also associated with of the South from the viewpoint of LAC improved terms of trade for economies that as a combination of external shocks. Sub- were net importers of manufactured goods. sequently, it examines the heterogeneous A demand shock was associated with an responses to such shocks across countries in increase in global demand for primary goods. the region and discusses the potential impli- It reflected the relatively high commodity cations for LAC’s long-term growth and (to a intensity of imports of the larger rising South lesser extent) employment. countries, particularly China. The result was a rise in commodity prices—an unusu- ally vigorous upswing phase of a veritable The rise of the South as external shocks commodity supercycle.11 For commodity for Latin America and the Caribbean exporters, including in LAC, this shock was A global supply shock was related to the huge associated with terms of trade gains. expansion in South-originated production The effects of the global supply shock may have dominated the effects of the global demand shock to the extent that large cur- FIGURE O.11 Real U.S. interest rates rent account surpluses were observed at the epicenter of the shock (China and other East 12 Asian economies). Consequently, the com- 10 bination of the global supply and demand 8 shocks engendered a global financial shock. Percentage points 6 This shock was associated with the inter- 4 national recycling of net savings from the 2 South, particularly from the Asian and Mid- 0 dle Eastern countries, and changes in relative –2 prices in financial markets around the world, –4 including exchange and interest rates. These –6 South countries integrated into the global economy with persistent current account 80 19 2 84 86 19 8 90 92 19 4 96 20 8 00 02 04 06 08 10 12 14 8 8 9 9 19 19 19 19 19 19 19 20 20 20 20 20 20 20 surpluses that were accumulated mainly in Sources: Calculations based on data from the Board of Governors of the U.S. Federal Reserve System the form of international reserves, most of and the Federal Reserve Bank of Cleveland databases. Note: Series was constructed by deflating the (effective) monthly federal funds rate by the inflation which were recycled through the North. The rate for the previous 12 months. result was a “global savings glut” that eased OVERVIEW 15 financial constraints in countries with exter- a deterioration of their terms of trade because nal and fiscal deficits, particularly the United of their export dependence on light manufac- States, and exerted significant downward tures and high level of imports of commod- pressure on world interest rates.12 Accom- ities. In addition, in some LAC economies, modative monetary policy in the North con- low domestic saving rates further reduced the tributed to the maintenance of unusually competitiveness of the manufacturing sector, low global interest rates (figure O.11). With and in economies with large agricultural and low interest rates in the North, a search for mining sectors, wages were pushed up, as yield among investors triggered capital flows explained below. to the South, including LAC, where borrow- Illustrative of the differences within LAC ing spreads fell to historically low levels and as a whole, figure O.13 shows the evolution currencies experienced strong appreciation of indexes of manufacturing export simi- pressures. larity for Brazil and Mexico. Brazil’s highly diversified export structure (spanning from agricultural commodities to automobiles) has Heterogeneity of impacts as a result been more similar to that of the United States of initial sectoral trade weights and the European Union than that of China. The combination of these supply and demand In contrast, Mexico’s manufacturing export shocks affected the LAC countries’ patterns basket has been consistently more similar to of trade differently, depending on their nat- ural endowments, geographical character- istics, economic size, and initial production FIGURE O.12 Terms of trade within Latin America and the and trade structures. The shocks were chan- Caribbean neled through changes in the terms of trade starting in the early 2000s and reflected the 170 extent to which initial trade structures were similar to those of China, at the epicenter of 160 these shocks, and the United States. Only a few countries in the region—chiefly 150 Mexico and, to a lesser extent, countries in Central America—maintained an export 140 Index (2000 = 100) structure similar to that of China. The trade structures of most countries in the region 130 were quite different from that of China. For the economies of South America, where the 120 dominant resources are land and mining endowments, the combination of external 110 supply and demand shocks translated into unequivocal and significant improvements in 100 their terms of trade (figure O.12). In contrast, Mexico’s diversified economy—which com- 90 bined an initially broad and relatively strong manufacturing base with substantial produc- 80 00 01 02 03 04 05 06 07 08 09 10 11 12 13 tive capacity in commodities (such as fossil 20 20 20 20 20 20 20 20 20 20 20 20 20 20 fuels, coffee, and iron ore)—experienced South America Mexico Central America stagnant terms of trade.13 In Mexico, the sup- Sources: Calculations based on data from the Economic Commission for Latin America and the ply shock that kept manufacturing prices in Caribbean (CEPAL). check was compensated for by the demand Note: Simple average across countries within each LAC subregion are presented. South America includes Argentina, Bolivia, Brazil, Chile, Colombia, Ecuador, Paraguay, Peru, Uruguay, and República shock that increased commodity prices. Cen- Bolivariana de Venezuela. Central America and Caribbean includes Costa Rica, the Dominican tral America and the Caribbean experienced Republic, El Salvador, Guatemala, Haiti, Honduras, Nicaragua, and Panama. 16 LATIN AMERICA AND THE RISING SOUTH China’s. Approximately 60 percent of Mex- the Caribbean, Central America, and Mexico, ico’s exports of manufactures were simi- where initial export structures were similar lar to those of China, compared with only to China’s (panel a). In contrast, the negative 30 percent in the case of Brazil.14 The global impact of the rise of China on manufacturing manufacturing supply shock dampened exports was significantly weaker for South the potential growth of LAC’s manufactur- American economies. The positive impact ing exports in general, with the effect most on their exports of agricultural and mineral acute in countries whose export structures commodities was substantial (panels b and were most similar to China’s at the outset (in c).15 In fact, South American countries repre- 2000). LAC countries that benefited the most sent all the observations in the three panels of from the Asia-led global commodity demand figure O.14 that were above the LAC average. shock were countries that were rich in nat- ural resources and had a commodity-ori- Weak participation of Latin America and ented initial export structure that matched the Caribbean in global value chains the structure of commodity (agricultural and mineral) imports of China. The sectoral composition of trade conditioned Empirical attempts to gauge the impact the within-LAC heterogeneity of export and of the rise of the South on LAC exports are import responses to the global supply and consistent with differences in the evolution demand shocks. These shocks boosted LAC’s of the terms of trade and the variance in the share in world commodity exports while degree of similarity between the LAC region’s undercutting the region’s share in global initial trade structures and the trade structure manufacturing exports. Financial flows to of China. Figure O.14 illustrates these pat- LAC countries seem to have reinforced these terns by presenting indexes of the quantitative trends. Specifically, LAC’s cross-border finan- impact of the rise of China on the growth rate cial inflows from the South have been more of manufacturing, mineral, and agricultural biased toward the primary sector than flows exports for a large sample of LAC countries from North countries. For example, during between 2000 and 2011. The heterogeneity of the 2000s, 92 percent of the total cross-border the estimated impacts across countries in the M&A investments from the South in LAC region is pronounced. The negative impact on went to the primary sector, whereas only the exports of manufactures was stronger for 48 percent of the same type of investments FIGURE O.13 Export similarity indexes in manufacturing in Brazil and Mexico a. Brazil b. Mexico 0.7 0.7 Manufacturing exports similarity index Manufacturing exports similarity index 0.6 0.6 0.5 0.5 0.4 0.4 0.3 0.3 0.2 0.2 0.1 0.1 0 0 1999 2003 2006 2009 2011 1999 2003 2006 2009 2011 China World United States European Union Korea, Rep. Japan Russian Federation South Africa Sources: Calculations based on data from World Integrated Trade Solution (WITS) and Comtrade; index proposed by Finger and Kreinin 1979. Note: The higher the index, the greater the similarity between the manufacturing export baskets of two economies. FIGURE O.14 Effects of the rise of China on gross exports from Latin America and the Caribbean, by sector, 2001–11 average a. Manufacturing exports Haiti Honduras El Salvador Mexico Dominican Republic Costa Rica Guatemala St. Lucia Panama Nicaragua LAC St. Kitts and Nevis St. Vincent and the Grenadines Dominica Grenada Colombia Brazil Peru Jamaica Belize Ecuador Uruguay Chile Argentina Bolivia Suriname Venezuela, RB Paraguay Guyana Cuba Ϫ20 Ϫ18 Ϫ16 Ϫ14 Ϫ12 Ϫ10 Ϫ8 Ϫ6 Ϫ4 Ϫ2 0 Percentage change b. Agricultural exports c. Mining exports Paraguay Brazil Argentina Chile Guyana Honduras Brazil Peru Uruguay Cuba Bolivia Jamaica LAC Guyana Cuba LAC Nicaragua Bolivia Venezuela, RB Uruguay St. Kitts and Nevis Haiti Dominican Republic Grenada Suriname Dominica Jamaica Belize Mexico Paraguay Chile Colombia Guatemala Argentina Peru Dominican Republic Belize Venezuela, RB Costa Rica Mexico El Salvador Suriname St. Vincent and the Grenadines Guatemala Haiti St. Lucia Panama Ecuador Honduras Nicaragua Dominica Panama Grenada St. Vincent and the Grenadines Ecuador Costa Rica St. Lucia El Salvador Colombia St. Kitts and Nevis 0 2 4 6 8 10 12 14 16 0 5 10 15 20 25 30 35 Percentage change Percentage change Source: Artuç, Lederman, and Rojas 2015, based on data from WITS and Comtrade. Note: Sectoral classification of trade flows is based on the ISIC classification, Revision 3. Agriculture corresponds to ISIC codes 0111–0500, mining to ISIC codes 1010–1429, and manufacturing to ISIC codes 1511–3699. See box 3.1 in chapter 3 of this report for technical details. LAC = Latin America and the Caribbean. 17 18 LATIN AMERICA AND THE RISING SOUTH from the North in LAC went to the primary evidence to support a nuanced yet positive sector (figure O.15). Large, albeit less striking, answer to the first question.17 The second differences are also observed in cross-border question is examined in a subsequent section. greenfield investments and syndicated loans.16 New forms of cross-border trading These trends suggest that the proliferation emerged alongside the rise of the South. One of LAC’s ties with the South was driven to manifestation of this phenomenon was the a larger extent by natural endowment–based proliferation of GVCs. These chains entail comparative advantages than by integration the offshoring and international distribution into manufacturing GVCs. Two key ques- of specialized activities that are part of an tions may be raised in this regard. First, is integrated production process. They typically LAC indeed characterized by weaker inte- involve a group of firms located in different gration into GVCs than other South regions? countries that operate at different stages of Second, are some types of trade structures the same production process in a coordinated (such as structures associated with partici- fashion, all under the aegis of a lead firm, with pation in GVCs) more conducive to growth the goal of enhancing the overall efficiency than others? The rest of this section provides of the chain. The GVC-based globalization FIGURE O.15 Sectoral composition of cross-border flows in Latin America and the Caribbean, 2003–2011 average LAC’s exports North South Syndicated Financial flows to LAC loans from the North Mergers and acquisitions Greenfield investment Syndicated Financial flows to LAC loans from the South Mergers and acquisitions Greenfield investment 0 10 20 30 40 50 60 70 80 90 100 Percent of total cross-border flows Primary Light manufacturing Heavy manufacturing Source: Calculations based on data from Comtrade, SDC Platinum, and fDi Markets. Note: The primary sector includes agriculture, hunting, forestry, and fishing; mining; and crude petroleum and natural gas. The light manufacturing sector includes food, beverages, and tobacco; textiles and apparel (including leather); and wood and paper-related products. The heavy manufacturing sector includes refined petroleum and related products, chemicals and plastics, nonmetallic minerals, metals, machinery and equipment, and transport equip- ment. The North includes the G-7 members and Western Europe countries. The South includes all other economies. Figure excludes offshore centers. G-7 = Group of Seven; LAC = Latin America and the Caribbean. OVERVIEW 19 pattern is thus driven more by firms’ global Within LAC, Central America and Mex- strategies than by traditional country-based ico gained relative importance during the comparative advantages. The resulting mul- early 1990s, probably as a result of the North ticountry production process calls for a finer American Free Trade Agreement (NAFTA). analysis of trade patterns that goes beyond the They peaked around 2000 and then lost traditional focus on broad sectors and skill ground, even as Eastern Europe rose, until categories (see, for instance, Baldwin 2012). about 2009. Since then, Central America and Measuring the intensity and quality of Mexico seem to have experienced a rebound. integration of a country into GVCs is a chal- The contrast with South America is stark: it lenge. Given the paucity of suitable data, did not experience a relative surge in terms of proxies must be used.18 One way to do so is to focus on exports of GVC-relevant interme- diate goods, as these fragmented production FIGURE O.16 Exports of intermediate goods as share of processes require that parts and compo- total exports in three global value chains nents cross borders before finished goods are shipped to final markets. Figure O.16 doc- a. By North and South uments the rise of exports of intermediate Share of world exports in 3 GVCs (%) 14 goods that are relevant for GVCs in three 12 industries: apparel and footwear, electronics, 10 and automobiles and motorcycles. The North started visibly losing its dom- 8 inance in the exports of these intermediates 6 (measured as share of total exports of GVCs 4 in the three industries) in the late 1980s, 2 when the South’s activity appears to have taken off (figure O.16, panel a). This process 0 1962 1970 1980 1990 2000 2012 accelerated in the 1990s; by 2009 the South’s From North countries From South countries exports of intermediate goods for these GVCs had surpassed the exports of the North. The b. By region in the South North’s relative importance in GVC-relevant 2.5 intermediate exports began to decline around Share of world exports in 3 GVCs (%) 2000—yet another piece of evidence that a 2.0 major global restructuring broadly coincided with China’s accession to the World Trade 1.5 Organization. Participation in GVC-relevant exports 1.0 of intermediate goods varied widely across countries and regions within the South (figure 0.5 O.16, panel b). The first economies from the South that picked up sizable shares of global 0 1962 1970 1980 1990 2000 2012 trade in intermediates were the East Asian East Asian Tigers Other East Asia MENA SSA China Tigers (Hong Kong SAR, China; Korea; Sin- ECA South Asia South America Central America, the gapore; and Taiwan, China), whose surge Caribbean, and Mexico began in the 1970s. They were followed by Sources: Calculations based on data from Comtrade; classification of intermediate goods other Asian countries (Indonesia, Malaysia, into three major global value chains (apparel and footwear, electronics, and automobiles the Philippines, and Thailand), which picked and motorcycles) is from Sturgeon and Memevodic 2010. Note: The North includes the G-7 members and Western Europe countries. The South up sharply in relative importance during the includes all other economies. East Asian Tigers include Hong Kong SAR, China; the 1990s but then lost ground precipitously Republic of Korea; and Singapore. Other East Asia includes Indonesia, Malaysia, the Phil- ippines, and Thailand. All other regions follow the World Bank classification of countries. after 2000, when China rose to a dominant ECA = Europe and Central Asia; G-7 = Group of Seven; GVC = global value chain; MENA = position. Middle East and North Africa; SSA = Sub-Saharan Africa. 20 LATIN AMERICA AND THE RISING SOUTH exports of GVC–relevant intermediates, and larger countries and China? The conse- it never had as large a share as many other quences were indeed asymmetric across LAC South regions. This evidence suggests that countries and tradable industries, as could be geography (that is, proximity to the United expected. States and distance from East Asian coun- In Argentina, Brazil, and Mexico, the tries) played a key role within LAC as a con- share of manufacturing employment, espe- ditioning factor for the region’s participation cially formal employment, has declined since in GVCs. roughly 2000 (figure O.18). The fact that it Another way of gauging a country’s inte- was most apparent in Mexico—one of the gration into GVCs is to focus on GVC-related countries in the region hardest hit by the rise forward and backward linkages. From this of China in global markets of manufactured perspective, even raw commodity exporters products—suggests that the employment can participate in GVCs, albeit in the forward impact of China was particularly intense linkage space, by, for instance, exporting where the trade effects were largest. inputs (such as crude oil) for the manufacture Evidence from the simulation models pre- of intermediate goods with greater degrees of sented in this report indicates that the impact processing or final goods (such as gasoline and of China on labor market dynamics in Argen- other oil derivatives). Figure O.17 shows the tina, Brazil, and Mexico (through global mar- differences between regions and subregions kets of manufactured goods, agriculture, and around the world in terms of their backward- mining) was substantial in the short run but, and forward-linkage participation in GVCs. perhaps contrary to expectations, relatively Mexico and Central America relate to weak in the longer run (for technical details, GVCs mainly as manufacturers of final see chapter 3 of this report). Labor market goods, hence predominantly in the backward frictions appear to have significantly increased linkage part of GVCs. Moreover, they have the short-run pain of the adjustment for work- integrated toward the final stages of GVCs ers in the manufacturing industry. However, with North countries, particularly the United these effects were counterbalanced in Argen- States. South American countries, by con- tina and Brazil by the positive employment trast, being net commodity exporters, are effects of rapidly rising agriculture and min- inserted mainly in the forward-linkage seg- ing imports from China. Mexico fared a bit ments of GVCs. worse: the simulation estimates suggest that The East Asian countries show equal par- the negative effects on labor demand in man- ticipation in the forward and backward seg- ufacturing were too large to be compensated ments of GVCs, implying that about half of for by the relatively small positive effects on their GVC-related trade is from imports of Mexico’s labor demand in agriculture and intermediate goods and half from exports of mining. This China-led rise of the South can final goods. This benchmark of 50 percent thus plausibly and at least partially explain may be relevant for growth, as it could be a why wages (adjusted for purchasing power sweet spot for the maximization of certain parity) rose faster in Brazil than in Mex- learning spillovers, as, for instance, produc- ico since the early 2000s (figure O.19). The ers of tradables can learn as much from their evidence on the seemingly small longer-run suppliers of imported goods as from the buy- employment impacts should be interpreted ers of their exports. cautiously, however. Evidence from other sources discussed in this report suggests that labor market frictions that inhibit labor migra- Differential employment effects tion within countries may result in significant How did the economic shocks emanating long-term losses in areas that had high levels from the restructuring of global trade affect of manufacturing employment before the rise employment in LAC, especially given the sim- of China (see, for instance, Autor, Dorn, and ilarity in the trade structures of the region’s Hanson 2013; Chiquiar 2014). OVERVIEW 21 FIGURE O.17 Backward and forward participation in global value chains, 2011 a. Across regions b. By country in Latin America and the Caribbean EAP Mexico El Salvador North Panama ECA Aruba SA Barbados Honduras MENA Bahamas, The Dominican SSA Republic Mexico and Uruguay Central America Belize The Caribbean Nicaragua South Costa Rica America 0 10 20 30 40 50 60 70 80 90 100 Guatemala Share of GVC participation (%) Jamaica Argentina Chile Haiti Antigua and Barbuda Cuba Ecuador Paraguay Bolivia Colombia Brazil Trinidad and Tobago Peru Venezuela, RB 0 10 20 30 40 50 60 70 80 90 100 Share of GVC participation (%) Backward GVC linkages Forward GVC linkages Sources: Calculations based on data from Eora-MRIO and WDI. Note: Participation in global value chains (GVCs) is proxied by the share of a country’s export that is part of a multistage trade process. This measure is constructed by adding the foreign value added used in a country’s own exports (backward GVC linkages) to the value added supplied to other countries’ exports (forward GVC linkages) and scaling the total by the country’s total exports of goods and services. Panel a reports cross-country averages. The North includes the G-7 members and Western Europe countries. The South includes all other economies. All other regions follow World Bank classification of countries. EAP = East Asia and Pacific; ECA = Europe and Central Asia; G-7 = Group of Seven; GVC = global value chain; MENA = Middle East and North Africa; SA = South Asia; SSA = Sub-Saharan Africa. into the world economy. This seldom explored Low saving rates in Latin America structural dimension of globalization is based and the Caribbean on the composition of demand—that is, the LAC’s response to the global shocks was also relative importance of domestic versus exter- conditioned by the net integration of countries nal demand relative to the country’s income. 22 LATIN AMERICA AND THE RISING SOUTH FIGURE O.18 Employment shares in the formal and national saving could be related to external informal manufacturing sectors of Argentina, Brazil, competitiveness, balance of payments sustain- and Mexico ability, investment, and growth, among other factors. This section documents key relevant a. Argentina facts regarding the patterns of saving, invest- 1991–99 14.7 ment, and real exchange rates in LAC relative Formal 2000–05 to other middle-income South regions. The 8.8 effects of (low) saving on growth are dis- 2006–12 9.9 cussed further below. 1991–99 5.8 Figures 20 and 21, which come from an Informal 2000–05 5.5 econometric model discussed in this report, show the comparative dynamics of saving, 2006–12 4.5 investment, the current account, and the real 0 2 4 6 8 10 12 14 16 exchange rate resulting from global shocks for Share of employment in manufacturing sectors (%) LAC and non-LAC emerging economies.19 As b. Brazil discussed earlier, the supply shock in the first decade of the 2000s seems to have dominated 1990–99 10.2 the demand shock. Hence, the focus is on the Formal 2001–05 9.3 response to an increase in global supply and to a decline in world interest rates (equivalent 2006–11 9.9 to a shock from monetary easing). 1990–99 3.6 Assuming no major institutional or struc- Informal 2001–05 4.7 tural change during the entire period, a posi- tive supply shock (an increase in global supply) 2006–11 4.3 boosts LAC’s investment, appreciates its real 0 2 4 6 8 10 12 exchange rate, and widens its current account Share of employment in manufacturing sectors (%) deficit more and more persistently than in c. Mexico other emerging economies (figure O.20). At the same time, such a shock depresses LAC’s 2000–04 13.4 saving rates for a prolonged period (in con- Formal 2006–12 9.8 trast with other emerging economies). Consistent with the earlier discussion, a Informal 2000–04 8.2 favorable global monetary shock that took 2006–12 8.3 place over the same period accentuated the macroeconomic effects of the global supply 0 2 4 6 8 10 12 14 16 Share of employment in manufacturing sectors (%) shock in LAC. In fact, the econometric exer- cise finds that a decline in the U.S. interest Sources: Calculations based on data from Encuesta Permanente de Hogares-Continua rate led to a rise in LAC’s investment rate, an (EPHC) surveys in Argentina, Pesquita Nacional por Amostra de Domicilios (PNAD) surveys appreciation of its exchange rate, and a fall in Brazil, and Encuesta Nacional de Ingresos y Gastos de los Hogares (ENIGH) surveys in Mexico. in its saving rate (figure O.21). These effects Note: Informal workers are defined as workers without social security benefits. were also more durable than in other emerg- ing economies. The patterns of low saving rates and The patterns of net integration of LAC coun- appreciating real exchange rates that pre- tries are undisputedly related to the region’s vailed in many LAC countries over the past historically low savings rates. Indeed, the dif- decade can thus be at least partially explained ference between aggregate domestic demand as region-specific responses to global shocks and income is the external current account, emanating from the rising South. The dif- which is also equal to the difference between ferences in macroeconomic responses to domestic saving and investment. For its part, the global shocks between LAC and other OVERVIEW 23 emerging South regions seem to have dimin- FIGURE O.19 Evolution of wages in Brazil relative to wages ished during the past decade, however, at in Mexico least in part thanks to improvements in 0.86 macroeconomic policy management. In par- 0.84 ticular, evidence from the econometric exer- 0.82 cise suggests that the adoption of inflation- targeting-cum-exchange-rate-flexibility and 0.80 improved fiscal rules in several LAC countries 0.78 Ratio appears to have led to significantly smoother 0.76 responses of output, consumption (hence sav- 0.74 ing), and investment to global shocks. This 0.72 smoothing was counterbalanced, at least 0.70 in inflation-targeting countries, by larger 0.68 responses in the real exchange rate. 0.66 L AC ’s pat ter ns of macroeconom ic 2001 2002 2003 2004 2005 2006 2007 2008 2009 responses to the global shocks, and the Source: National average wages in local currency are from the International Labour Office. They change in such patterns over the past decade, were converted to international purchasing power parity constant 2005 U.S. dollars using the con- are arguably influenced by LAC’s reliance on version factor from World Development Indicators (WDI). FIGURE O.20 Responses to a positive global supply shock in Latin America and the Caribbean and other emerging market regions a. Investment b. Real exchange rate 1.5 0.002 1.0 Percentage points Percentage points 0 0.5 –0.002 0 –0.5 –0.006 1 11 21 31 41 1 11 21 31 41 Quarters Quarters c. Saving rate d. Current account 0.20 0.05 0.15 0.10 0 Percentage points Percentage points 0.05 0 –0.05 –0.05 –0.10 –0.10 –0.15 –0.20 –0.15 1 11 21 31 41 1 11 21 31 41 Quarters Quarters LAC Non-LAC emerging market economies Source: Hevia and Servén 2014. Note: Lines represent the accepted median model deviation from the trend from a global demand shock, in terms of the sign restrictions defined in Hevia and Servén (2014). See table 5A.4 in chapter 5 of this report for technical details on the sign restrictions. Non-LAC emerging market economies include Hungary, India, Indonesia, the Republic of Korea, the Philippines, Poland, the Russian Federation, South Africa, Thailand, and Turkey. LAC = Latin America and the Caribbean. 24 LATIN AMERICA AND THE RISING SOUTH FIGURE O.21 Responses to a global monetary easing in Latin America and the Caribbean and other emerging market regions a. Investment b. Real exchange rate 1.5 0.000 –0.001 1.0 Percentage points Percentage points –0.002 0.5 –0.003 –0.004 0 –0.005 –0.5 –0.006 1 11 21 31 41 1 11 21 31 41 Quarters Quarters c. Saving rate d. Current account 0.1 0.05 0 0 Percentage points Percentage points –0.1 –0.05 –0.2 –0.10 –0.3 –0.15 1 11 21 31 41 1 11 21 31 41 Quarters Quarters LAC Non-LAC emerging market economies Source: Hevia and Servén 2014. Note: Solid lines represent accepted model median deviation from the trend from a global demand shock , in terms of the sign restrictions defined in Hevia and Servén (2014). See table 5A.4 in chapter 5 of this report for technical details on the sign restrictions. Non-LAC emerging market economies include Hungary, India, Indonesia, the Republic of Korea, the Philippines, Poland, the Russian Federation, South Africa, Thailand, and Turkey. LAC = Latin America and the Caribbean. domestic demand (associated with low saving of payment vulnerability effect, which can rates and a penchant for current account defi- also hinder growth. 21 Where the ER chan- cits). Some evidence to back this statement nel dominates, one would expect to observe was provided earlier, in connection with fig- a pattern in which countries that save less ure O.10, which shows that current account grow less and have appreciated real exchange deficits tend to emerge systematically in LAC, rates. Where the IR channel dominates, one even during the recent times of favorable would also expect to see that countries that terms of trade. save less grow less. Yet, real exchange rates Low saving rates arguably condition mac- would be undervalued in this case, reflecting roeconomic outcomes and responses to exter- low sovereign ratings and vulnerable balance nal shocks through one of two channels. The of payments trajectories. first is a real exchange rate (ER) channel—a The patterns observed in figure O.22 are competitiveness-reducing effect caused by consistent with these expectations. The vari- appreciating real exchange rates that can ables of interest in the scatter plots reflect hinder growth. 20 The second is an interest medium-term equilibrium relations that are rate (IR) channel, associated with a balance presented in the form of deviations from the OVERVIEW 25 benchmark.22 The size of the deviations can FIGURE O.22 Domestic saving, real exchange rates, and sovereign be attributed largely to differences in policies risk ratings, 1990–2012 average and policy-driven institutions.23 a. Domestic saving and real exchange rate gaps Panel a of figure O.22 shows that an ER 2 pattern is consistent with the entire analyzed sample: on average countries that save more have more competitive real exchange rates, 1 relative to benchmark. However, LAC coun- National saving EAP tries (divided into two groups, higher-income MENA countries [LAC1] and lower-income coun- 0 HI ECA tries [LAC2]) tend to be located in the SSA LAC2 LAC1 lower-left quadrant, where exchange rates are undervalued. In contrast, East Asia and –1 Pacific countries tend to occupy the upper- left quadrant, where oversaving is associated with undervaluation. These patterns suggest –2 –1 –0.5 0 0.5 1 that low saving rates have historically influ- Real exchange rate enced macroeconomic outcomes in LAC mainly through the IR channel—that is, b. Domestic saving and sovereign risk rating gaps through adverse balance of payments vul- 2 nerability effects reflected in low country ratings. This finding is consistent with the scatter diagram in panel b of figure O.22, 1 which shows that worldwide data also sup- National saving EAP port an IR pattern (countries that save less MENA ECA tend to have lower sovereign risk ratings). 0 SSA LAC1 HI LAC is located closer to the fitted line, LAC2 although it still appears as an undersaving and underrated region. –1 Two key caveats have to be made in this regard. First, there has been considerable heterogeneity within LAC, as shown in panel –2 –1 –0.5 0 0.5 1 a of figure O.23. Between 1990 and 2012, Sovereign risk rating the region started to break free from the LAC1 countries per period spell of low sovereign ratings (figure O.24) LAC1 countries 1990–2012 average and hence started to transition from an IR LAC2 countries per period to an ER pattern. Chile, Mexico, Panama, LAC2 countries 1990–2012 average and Peru appear as oversavers with under- Other countries per period valued real exchange rates (all relative to Other groups of countries 1990–2012 average benchmark), whereas the Bahamas, Barba- dos, Brazil, Costa Rica, and Uruguay appear Sources: Calculations based on data from United Nations (UNSTAT), WDI, and Institutional Investor as undersavers with overvalued exchange database. Note: The linear fit was calculated for the period version of the complete country sample for rates. These country cases thus conform 1990–2012. LAC1 countries are countries in Latin America and the Caribbean (LAC) with annual per to the ER pattern. In contrast, Colombia, capita GDP of more than $5,000; LAC2 countries are those with annual per capita GDP of $5,000 or less; see table OA.1 for list of countries in all groups). GDP = gross domestic product; EAP = East Asia Ecuador, and Trinidad and Tobago sit in the and Pacific; ECA = Europe and Central Asia, HI = high income; MENA = Middle East and North Africa; lower-left quadrant, with low domestic sav- SSA = Sub-Saharan Africa. See annex 5A in the main report for details on how the benchmarks are calculated. ing and undervalued exchange rates. These patterns suggest that these latter countries have remained more persistently under the grip of the IR channel. Perhaps surprisingly, 26 LATIN AMERICA AND THE RISING SOUTH FIGURE O.23 Saving and real exchange rate gaps for higher- were generated to effect the transfer of capi- income countries in Latin America and the Caribbean tal abroad.24 a. 1990–2012 Second, consistent with the suggestion 0.5 stemming from the dynamic analysis referred ER high savers IR high savers to earlier, the benchmarking exercise identi- CHL PAN VEN fies an accelerated migration of LAC1 coun- ARG PER MEX tries toward the ER pattern during the first decade of the 2000s, as real exchange rates National saving 0 ECU COL BHS appreciated substantially and sovereign risk BRA URY ratings rose steeply. Country ratings actu- TTO ally converged in this period to those of the –0.5 BRB middle-income countries of Southeast Asia IR low savers (figure O.24), with several countries in LAC CRI joining the investment-grade asset class. 25 ER low savers This migration reflected improvements in –1 –0.4 –0.2 0 0.2 macrofinancial policy frameworks and, at Real exchange rate least in South America, the powerful forces of the global shocks associated with the rise b. 2011–12 of the South. In fact, as shown in panel b of 0.5 PAN* figure O.23, many LAC1 countries moved ARG* significantly closer to the ER pattern that VEN is observed for the entire sample (the fitted CHL MEX PER line) during the 2011–12 period. Particularly 0 National saving TTO* ECU strong real appreciations took place in Brazil, BHS* CRI COL Colombia, Costa Rica, and Uruguay. URY BRA –0.5 Implications for growth: Trade structure, foreign direct investment, BRB* and the composition of aggregate –1 demand –0.4 –0.2 0 0.2 0.4 Real exchange rate Do the LAC-specific trade and aggregate demand structures really matter for growth? Sources: Calculations based on data from UNSTAT and WDI. This section summarizes the main findings Note: The linear fit (shown in both panels) was calculated for the complete country sample for 1990–2012. Higher-income countries in Latin America and the Caribbean are countries with annual of a battery of econometric tests conducted per capita GDP of more than $5,000 (see annex table OA.1 for list of countries). See annex 5A in to shed light on this question, with special chapter 5 of this report for technical details on the calculation of the benchmarks. Three-letter country groupings correspond to ISO 3166 standard. * = due to missing data for the 2011–12 period, attention on the relevance for growth of trade the latest available period was used. IR=countries affected by the interest rate channel. ER= coun- structure, FDI, and domestic saving. The key tries affected by the real exchange rate channel. GDP = gross domestic product. message is that economic structures matter for growth. A reassessment of the region’s growth- and productivity-oriented reform Argentina and República Bolivariana de agenda from the angle of structure would Venezuela appear as high savers with over- therefore be useful. valued currencies. As these countries have The role of trade structure. The litera- had sovereign ratings well below the average ture supports the notion that trade openness of the LAC1 group, a plausible explanation can raise growth rates, at least temporarily, for their location in the figure is the repeated during the transition to a higher steady-state occurrence of exchange controls and epi- path of GDP per capita. 26 There is much sodes of massive capital flight, during which debate, however, regarding the channels excess saving and current account surpluses through which this transition may operate. OVERVIEW 27 The traditional answer, dating back to the FIGURE O.24 Country ratings for selected country groups neoclassical theories of trade, has been that 80 trade lifts growth (at least transitionally) 75 70 Sovereign risk rating index through the efficiency gains of specializa- 65 tion based on comparative advantage. This 60 channel hinges on differences in either fac- 55 tor endowments (labor, capital, land, natural 50 resources) or average productivities across 45 27 40 countries. 35 More recently, the focus has been on a dif- 30 ferent (and arguably complementary) mecha- 25 nism, whereby trade boosts growth by serving 20 1980 1984 1988 1992 1996 2000 2004 2008 2012 as a conduit for learning spillovers and tech- nology diffusion (see Keller 2004 for an early LAC1 Argentina and Venezuela, RB Southeast Asian middle-income countries review of the literature). One implication is that when it comes to its impact on growth, Source: Calculations based on data from Institutional Investor database. not all trade is created equal. The question is Note: Middle-income countries in Southeast Asia include Indonesia, the Republic of Korea, Malaysia, the less about whether and how much an econ- Philippines, and Thailand. LAC1 countries are countries in Latin America and the Caribbean (LAC) with annual per capita GDP of more than $5,000 (see annex table OA.1 for a list of countries). GDP = gross omy trades but rather how much it learns domestic product. from its international trade. This realization naturally shifts the debate toward questions such as how and with which partners a coun- try trades. Empirically, these questions point can thus surmise that to the extent that trade to measurable dimensions that can be used flows embody technology and knowledge, as proxies for learning-intensive trade. 28 As producers can benefit more from exports and such, this report adds to the growing evi- imports that are part of the same industry dence that suggests that certain features of a or a GVC than they can from exports and nation’s trade structure matter for economic imports that correspond to unrelated or dis- development and growth. Some of these connected activities. The composition of features include the degree of intraindustry trading partners may also play an important trade, participation in GVCs, the composition role in how much countries learn and how of trading partners, and the degree of export quickly they adopt new technologies. concentration. These features shed light on The econometric evidence in this report the extent to which technology diffusion and suggests that trade linkages with the North the learning intensity of trade can positively could indeed yield higher growth payoffs affect growth and other economic outcomes, than trade with the South. The results, based such as macroeconomic volatility (see, for on data for 1960–2010, indicate that a 1 per- instance, Lederman and Maloney 2007; Alva- centage point increase in the degree of trade rez, Buera, and Lucas 2013; and Pinat 2015). openness with North countries is associated This report analyzed the relationship with a 1.6 percent increase in GDP per capita between several characteristics of trade per year over a five-year period, followed by structure and growth, given that there is no potentially longer-lasting effects. In contrast, overarching consensus in the literature as to the estimated effect of trade with the South which ones are most influential. Two partic- is much lower: a 1 percentage point increase ularly interesting characteristics—intrain- in the degree of trade openness with South dustry trade and participation in GVCs—are countries is associated with an increase in likely to be related to international technol- GDP per capita of only about 0.3 percent. ogy and knowledge flows because they tightly The difference in the estimated effects link trade to domestic factor and input mar- when trading with the North versus the South kets, logistics, and production processes. One seems to be associated with differences in the 28 LATIN AMERICA AND THE RISING SOUTH structure of trade along several dimensions, M&A flows are positively (and significantly) arguably including the extent and manner associated with the recipient country’s labor in which countries participate in GVCs (see productivity within manufacturing indus- chapter 2 of this report and Didier and Pinat tries, North-South, South-North, and South- 2015 for technical details and a deeper analy- South flows are not (for technical details, see sis of the structure of trade linkages and eco- chapter 4 of this report and Didier, Nguyen, nomic growth). Controlling for the overall and Pienknagura 2015). These findings sug- volume of trade flows, increases in participa- gest that LAC and other South economies tion in GVCs, especially the middle segments have yet to benefit in terms of labor produc- of these chains, yield additional gains in GDP tivity increases within manufacturing indus- per capita. An increase in the share of total tries from their flourishing connections with trade that comes from intraindustry trade has the rest of the South or the North. a positive and statistically significant associa- Other evidence, however, suggests that tion with income growth. Trading with coun- LAC has benefited from the presence of mul- tries at the center of the global trade network tinational corporations through different is associated with higher growth, arguably channels, including by accelerating the exit of because these types of connections expose the low-productivity domestic firms and enhanc- country to the frontier of ideas and technol- ing the productivity of domestic firms across ogies. The econometric results also suggest all industries (see, for instance, Lederman that countries benefit more from interna- and others 2014). tional trade connections when they have a The new evidence on FDI presented more educated labor force, which points to in this report suggests that aggregate the importance of human capital formation industry-specific labor productivities in the for the absorption of foreign technology and South so far appear to be unaffected by knowledge. foreign firms’ mergers with or acquisitions Intraindustry trade and insertion into the of domestic firms. Future research could core of GVCs thus appear to be more condu- attempt to ascertain the features in North cive to higher long-term growth rates. Except economies that allow them to benefit from possibly in Mexico, Costa Rica, and Uru- M&A flows within industries, with an eye guay, the rise of the South has not systemat- toward understanding whether these positive ically yielded these types of growth-inducing effects depend on public policies (as imped- changes in trade structures in LAC. iments to or propagators of learning spill- The role of foreign direct investment. The overs), the quality of institutions, the quality increase in financial flows across countries, of human capital, or other factors. The sec- especially FDI, could be driven by com- tion on policy priorities below addresses panies seeking to capitalize on efficiency these issues. improvements made possible through the The role of the composition of aggregate fragmentation of production stages across demand. Do low national saving rates—a countries. Therefore, the rising participation trademark of LAC economies—hamper of the South in global financial flows could growth? Mainstream open-economy growth be a potential driver of economic growth. models typically assume that foreign and Such flows may not only ease financing con- domestic saving are perfect substitutes. straints in recipient economies but also be a Implicit in these models is the notion that conduit for technology diffusion and learn- what really matters for growth are invest- ing spillovers. Indeed, policymakers from the ment (and profit) prospects, but not how South, including LAC, see the attraction of investment is financed. This view is con- FDI and multinational corporations as a pol- sistent with the assumption that factors of icy priority. production (particularly capital) respond The empirical findings presented in this to small differences in relative returns by report indicate that although North-North flowing into their most productive uses, OVERVIEW 29 both across countries and industries or surprising, given that factor mobility is lower firms within countries. The implication is (and factor misallocation higher) in emerg- that domestic saving, and more broadly the ing than advanced economies. Foreign and composition of aggregate demand, is not a domestic saving are thus less perfect substi- determinant of the equilibrium real exchange tutes, as far as growth is concerned, in these rate. Rather, the latter would be determined emerging economies. The result also suggests only by productivity differentials across trad- that saving rates can in some sense com- able and nontradable industries driven by pensate for market imperfections and pol- supply-side characteristics, such as the capital icy obstacles that get in the way of efficient intensity of production. Consequently, saving resource allocation. As the allocative func- and the real exchange rate would not affect tion of markets improves, saving should be growth, as small increases in returns to cap- less of a constraint on growth. ital would immediately attract capital to the Finally, important asymmetries seem to countries, industries, or firms that temporar- characterize the effects of saving on growth. ily offer higher returns. The real exchange In particular, a higher domestic saving rate rate would adjust back to its equilibrium level has a greater positive impact on growth when accordingly. countries experience current account deficits. This view clashes with certain well- This finding should not be surprising, as it established stylized facts. For example, coun- stands to reason that the benefits of a saving tries that rely on foreign saving grow less effort that help to avoid unviable balance of (see, for instance, Prasad, Rajan, and Subra- payments trajectories outweigh the benefits manian 2007); countries whose productivity of a saving effort that increase an already falls behind are countries that “tax” saving strong current account surplus. (see, for instance, Gourinchas and Jeanne When the data are explored in ways that 2012); and there is considerable misalloca- identify the underlying mechanisms, the rel- tion of factors of production, which shows up evance of both the external competitiveness in large and persistent dispersion of produc- (ER channel) and the balance of payment tivities across firms, sectors, and countries. vulnerability (IR channel) effects of saving is This report provides evidence in support borne out. Figure O.25, which uses the entire of the alternative hypothesis that national sample, shows deviations from benchmark in saving matters for growth, implying that the domestic saving–real exchange rate space. domestic and foreign saving are imperfect For all observations in each quadrant (that substitutes. Econometric evidence suggests is, for all the dots plotted in figure O.22, that national saving rates have an impact on panel a), figure O.25 shows the average of the growth (for technical details, see chapter 5 of corresponding deviations from benchmark this report and De la Torre and Ize 2015). It for other key variables (namely, sovereign rat- shows that, on average, a 10 percentage point ings, growth rates, and investment rates). increase in the saving rate (which would Four key messages emerge from figure bring the average LAC saving rate to the level O.25. First, countries with undervalued real in Southeast Asia) would increase GDP per exchange rates grow faster than countries capita by 1–2 percentage points a year for with overvalued currencies. This finding at least three years, followed by potentially is a restatement of the well-known finding long-lasting effects of similar magnitudes of Rodrik (2008). Second, the ER pattern thereafter. The evidence is preliminary and strongly emerges from the world data: coun- thus should be interpreted with caution. tries that oversave typically have under- However, it does strengthen the argument valued real exchange rates and grow faster that saving matters for long-term growth. than other countries, whereas countries that The findings also suggest that the sav- undersave typically have overvalued curren- ing-to-growth link is stronger for middle-in- cies and grow more slowly. Third, the IR come countries. This result should not be pattern also emerges from the data: countries 30 LATIN AMERICA AND THE RISING SOUTH FIGURE O.25 Sovereign risk rating, growth, and investment gaps, 1990–2012 1.0 1.1 0.7 0.2 Oversaving –0.1 –0.5 National saving gap Undervalued Overvalued Undersaving 0.2 0.1 –0.7 –1.1 –0.3 –1.3 Real exchange rate gap Sovereign risk rating Growth Investment Sources: Calculations based on data from UNSTAT and WDI. Note: Each bar in the figure represents the simple average of sovereign risk rating, growth, or investment gaps for the observations located in each quadrant of the scatter plot. The scatter plot is a reproduction of figure O.22, panel a. Each point represents a country for a given time period. See chapter 5 for addi- tional details. that undersave and face balance of payments ratings for much of LAC, even as the region viability problems (that is, countries in which boosted growth and reduced systemic vul- sovereign risk ratings are well below bench- nerabilities. However, LAC adapted and mark) also have undervalued real exchange responded to these shocks with its traditional rates. Fourth, saving affects future growth domestic demand–reliant (low saving) mac- through investment: countries that oversave roeconomic structure, which led to strong relative to benchmark typically outperform real appreciations, especially in countries their peers in terms of investment rates, that save less.29 The force of the external tail- especially where the real exchange rate is winds was such that they more than offset undervalued. (and actually concealed) the adverse growth During the past decade or so, LAC was effects of low saving. Now that the tailwinds caught up in the forces of real and mone- of commodity prices no longer blow, one can tary global shocks precisely at a time when hypothesize that, given the vastly improved significant improvements in macrofinancial country ratings, low saving rates in LAC may policy frameworks were materializing. The hinder growth less through balance of pay- confluence of these external and internal fac- ments vulnerability effects and more through tors promoted rapid improvement in country external competitiveness effects. OVERVIEW 31 Changing world, new priorities The root causes of such labor market fric- tions remain unclear. The policy agenda is The rise of the South has affected at least therefore far from obvious. Regulatory rigid- three major policy areas, all of which have ities, which are often bypassed by voluntary implications for employment and growth. In shifts to informality, are unlikely the only some respects, the global shocks may have source of friction (although they are undoubt- temporarily dimmed the urgency of such old edly important). Other sources could include policy challenges as commodity dependence, skills mismatches (including mismatches aris- labor market frictions, and low saving rates. ing from information asymmetries or limited However, as the pull of the rise of the South skill portability) and transport costs within tapers off and the tailwinds recede, the pol- countries. icy agenda should turn even more forcefully The role of skills mismatches is evidenced toward the issues highlighted below. by the well-known finding that the estimated costs of moving to a new job varies signifi- cantly across industries, which implies that Reducing labor market frictions skills are to a large extent industry or firm Labor market frictions made the process of specific. LAC’s experience over the past adjustment to the global supply and demand decade, as well as the powerful forces of tech- shocks unnecessarily costly, especially for the nical change, calls for a policy agenda aimed net commodity importing countries in LAC. at facilitating and enhancing skills develop- They explain why China was once the scape- ment, skills matching, and the formation of goat of choice for LAC policymakers.30 more flexible human capital, so that workers Especially since 2001, when China accel- can more easily adjust to production innova- erated its pace of growth in global trade, tions and shifting market realities by chang- workers in LAC could have benefited from ing jobs and careers over their working lives the declining prices of manufactures and the at lower personal (and social) costs. This pol- employment opportunities in agriculture, icy agenda naturally puts a premium on suit- mining, and nontraded domestic industries if able reforms to educational systems, labor they had been able to switch jobs easily. How- market rules and contracts, social protection ever, the evidence in this report, as well as the benefits (to make them more portable and public’s tendency to worry about competition compatible with labor mobility), and training from China, suggests that labor market fric- and retraining programs. tions prevent workers from easily transition- The potential role of transport costs (and ing to industries where they could be most hence transport-related policies) in interin- productive. The evidence indicates that work- dustry labor mobility has received little atten- ers behave as if they have “sticky feet,” the tion to date. The costs of moving labor across title of a recent World Bank report on trade industries may reflect the concentration of and jobs (Hollweg and others 2014). As Chi- industries across territories. In Brazil, for nese competition in manufactured goods mar- example, most manufacturing is concentrated kets became tough, manufacturing industries around São Paulo and the southeastern coast, had to adjust, partly by shedding workers whereas agriculture is located in the interior and partly by retooling to regain competitive- of the country. The costs of moving workers ness. Workers stuck in “senescent” (declining) and their families across vast geographical manufacturing industries bore a heavy price, regions may help explain the sluggishness of in the form of unemployment or informality. labor market adjustments within countries. They would have been better off had they In fact, a growing body of academic litera- been able to adapt their skills and more easily ture argues that transport costs may play an move within countries to take advantage of inhibiting role in the integration of domestic better employment opportunities. labor markets. 32 LATIN AMERICA AND THE RISING SOUTH There is, however, persistent, albeit rel- and are undistorted. From this viewpoint, atively low-level, rural-to-urban migration removing policy distortions that get in the within LAC countries, including Brazil and way of market-driven resource allocation and Mexico. It is thus also plausible that the reducing the costs of doing business will nat- choice of migration by workers across vast urally attract corporations from around the distances is driven not just by transport costs world. Trade structures would then special- but also by workers’ specific circumstances ize and respond endogenously to comparative and preferences, some of which may be unre- advantages and a business-friendly environ- lated to market signals. For instance, being ment. Whether the efficient outcome is a close to family may be an overriding consid- knowledge-intensive type of export growth eration for workers unless they face extreme will depend on factor endowments and rela- circumstances (shocks) or belong to commu- tive returns, but the outcome would move the nities with a historical inclination for migrat- economy to its production possibilities fron- ing to specific destinations. tier. This paradigm emphasizes public policy The objective here is not to prescribe spe- failures that hinder market forces rather than cific policies but rather to argue that policy market failures. It thus puts a premium on makers need to rethink broad priorities. reforms that seek to maximize the operation Infrastructure is one area that may be prime of the Invisible Hand. for reconsideration, not just because of its The alternative view is that by itself, the relationship with competitiveness (through market may not automatically bring knowl- its impact on firms’ cost structures) but edge from abroad and will thus underexploit also because poor infrastructure may make opportunities for boosting technology-driven domestic labor markets less nimble and thus endogenous growth dynamics. From this less able to absorb permanent shocks. perspective, some form of industrial policy will be required to induce market players to internalize the positive externalities associ- Fostering trade, foreign investment, ated with the exploitation of knowledge spill- and knowledge spillovers overs. A 2014 report by the Inter-American For some LAC countries, the rise of the South Development Bank, Rethinking Productive brought some benefits, such as lower bor- Development, provides a set of organizing rowing costs and better terms of trade for net principles to discipline thinking about choos- exporters of agriculture and mining prod- ing industrial policy interventions to target ucts. However, the structure of trade between specific types of market failures. LAC and the South seems to be less growth Looking through the prism of the rising inducing than its trade with the North. Like- South phenomenon, this debate boils down wise, FDI into LAC (in the form of M&A) to a balancing act. On the one hand are the that originates in other South countries does potential benefits of improvements in the not seem to be raising labor productivity market-enabling environment that reduce within industries in the region. Labor pro- trade costs for domestic agents, who in turn ductivity appears to more clearly benefit from are guided by competition and relative price North-North M&A activity. Both sets of signals in enhancing their trade and financial results suggest that some rethinking is called linkages with both the South and North. On for in the area of structural change and the the other hand are the coordinating roles of scope for learning and technology diffusion the state, including through the provision through ties with global partners. of specific tax or subsidy incentives, or tar- There have been two extreme paradigms geted loans and loan guarantees, for firms about policy challenges in this area. One is and workers to move into preselected activ- the laissez-faire view, which posits that learn- ities that have a good chance of becoming ing from foreign knowledge will take place part of GVCs or fostering intraindustry trade as long as domestic markets function well patterns. OVERVIEW 33 A safe approach is one that strikes a Fourth, both vertical and horizontal sensible balance between the laissez-faire industrial policies need to be put on the table, and industrial policy approaches. First particularly for countries that have advanced and foremost, policy should do no harm: on the laissez-faire front, so that old policy policy-induced distortions that get in the distortions do not get in the way of the poten- way of efficient resource allocation and tial success of new industrial policies. Coun- unnecessarily raise the costs of international tries throughout LAC already have some transactions should be reduced. The report industrial policies in place, such as invest- highlights one such distortion: the region’s ment and trade promotion that targets cer- increasing reliance on temporary trade barri- tain types of firms and industries over others. ers (such as antidumping, countervailing, and An extension of this debate could encompass safeguard import duties), which appear to be policy-based incentives, including tax and overused, especially against China and other expenditure policies, with an eye on helping South economies. Many other actions can markets internalize large positive external- be considered in this regard, including elim- ities associated with research and develop- inating or redesigning government programs ment (R&D) and technology adoption and that unintentionally subsidize informality or adaptation. Given that industrial policies can unduly encourage firms to remain small. have significant downsides, it is important Second, there is plenty of room for pos- that they be designed and implemented in itive policy actions aimed at improving ways that generate information and learning the market-enabling environment—by, for (so that impacts can be assessed and mistakes instance, raising information transparency corrected promptly along the way) and com- and disclosure standards and strengthening plement and crowd in market forces (in order contract rights. In general, horizontal pol- to widen the scope for efficiency gains). icies of this nature can only help, although they may not necessarily remove the most Raising national saving rates binding constraints to the development of growth-friendly globalization patterns. Pol- A reform agenda in LAC focused exclusively icies aimed at improving the functioning of on the sorely needed enabling environment labor markets while maintaining adequate and supply-side reforms may not be suffi- labor protections are worthy of special atten- cient to avoid the downsides of globalization tion in this regard. while fully reaping its upsides. A demand- Third, it is time to get serious about side component focused on raising national assessing deficits in the formation of human saving rates, intended to prevent persistent and physical capital (particularly transport, currency overvaluations and balance of pay- energy, and telecommunications infrastruc- ments vulnerabilities, is also a crucial ele- ture), which may be constraining the abil- ment of the growth-oriented reform agenda. ity of individuals and firms to engage in This demand-side component is particularly cross-border transactions efficiently. On the important for LAC countries that exhibit human capital side, educational systems need chronic low saving rates. It is also key in the upgrading, particularly in ways that allow context of market imperfections that limit them to foster the type of skills modern econ- the scope of factors to quickly and smoothly omies demand. Workers need to be trained move to their more productive uses. and retrained, on and off the job, through- Keeping these considerations on the out their working lives. On the infrastruc- policy radar screen may not be easy, given ture side, closing gaps is essential to reducing that the region’s historical low-saving/low- international trade costs, a key determinant growth syndrome may be shifting in the of the emergence of, and incorporation into, context of the rising South and the region’s GVCs and other types of international com- more resilient macrofinancial policy frame- mercial relations. works. The greatly improved sovereign risk 34 LATIN AMERICA AND THE RISING SOUTH ratings that now characterize much of LAC Third, careful social safety net reforms can may facilitate external borrowing, which (in strengthen domestic saving. The region made the best of cases) can conceal the adverse progress in the past decade in mainstreaming growth consequences of uncompetitive real and targeting social assistance to the poorer exchange rates or (in the worst of cases) and most vulnerable segments of the popu- rekindle LAC’s traditional tendency to suf- lation, including through highly successful fer from balance of payments sustainability conditional cash transfer programs. Several problems. LAC countries complemented these efforts Although economists often resist treating with improvements in noncontributory social saving as a policy variable, a saving-boosting benefits, especially through minimum pen- reform agenda is within reach, although it sion pillars (so-called social pensions) and will require patience and persistence and is the provision of health services at very low likely to be fraught with tensions. There are or no cost to poor households and informal at least four entry points for a comprehensive workers. Given the social benefits of higher policy approach. saving rates, however, as the region consid- First, raising public sector saving can raise ers second-generation reforms to the health, national saving, because it is unlikely that pensions, and unemployment safety nets, it the private sector will completely offset such should ensure that such reforms should not efforts by reducing its saving. Raising public only improve fairness and financial sustain- saving through fiscal tightening (by raising ability but also promote self-reliance (instead revenues, reducing expenditures, or both) of excessive reliance on the state), especially would not be easy in the current global eco- among the elites and upper social echelons. nomic environment. Fiscal reforms that boost Fourth, in designing short-run mac- public saving, and hence tilt public outlays in roeconomic interventions, policy makers favor of investment, would have to confront should take more explicit account of the the difficult and sensitive question of who growth-boosting saving agenda. Doing so would consume less today. Tensions would militates in favor of shifting toward a tighter thus arise over the distribution of taxes and fiscal, looser monetary macroeconomic policy expenditures across space, households, and mix—something that is politically difficult to firms as well as between current and future achieve, especially in the current environment generations. Deft political leadership would of weak world demand, which puts a pre- be needed to increase frugality and foster mium on spending rather than saving. The asset building (which implies a sacrifice of current international financial environment, some consumption today) in a way that pro- characterized as it is by low interest rates and tects the basic consumption needs of the poor. abundant liquidity, could encourage policy Second, there may be openings for imple- makers to borrow imprudently and hence risk menting saving-enhancing policies in the fiscal and balance of payments sustainability financial sector. Since the late 1990s, finan- problems in the future. To reconcile short-run cial development in LAC has been strongly aggregate demand management with lon- biased in favor of consumer finance when ger-run growth objectives, it is crucial that contrasted with other regions, as De la Torre, LAC maintain robust saving rates. Ize, and Schmukler (2011) show. Reforms of The rise of the South has deeply changed financial regulations could help promote sav- the global economy, and irreversibly so. Poli- ing, investment, and production rather than cies and reform agendas have to adapt to this consumption. Financial inclusion could be momentous change. The challenge is great, expanded on the deposit-taking and payment but it provides LAC’s political leadership with side rather than the lending side. Macropru- an opportunity to shine. It is time for cold- dential regulatory policy aimed at prevent- headed rethinking of policy priorities that can ing credit-fueled consumption booms is also unleash growth potential of an immensely called for. diverse and in many ways rich region. OVERVIEW 35 Structure of the report especially the changes brought about the emergence of China, has affected labor The five chapters that make up the rest of markets in the region. It provides a discus- this report provide a more detailed analy- sion of how social protection policies can sis of the rise of the South and the nature of help reduce labor market adjustment costs LAC’s evolving external connections. They when economies face long-lasting structural draw implications from these changes for the changes emanating from the reconfiguration region’s economic development. of the global economy. Chapter 1 sets the stage for the rest of the Chapter 4 provides a detailed analysis of report by characterizing the rise of the South the degree of financial connectivity of LAC and outlining a set of relevant trends that are countries with the North and the South. It shaping LAC’s economic prospects. investigates the extent to which LAC’s finan- Chapter 2 explores the notion that the cial integration is related to its trade inte- structure of trade matters for economic gration and the degree to which financial development. It analyzes the extent to which flows are associated with increases in labor the trade connections of countries in LAC— productivity. particularly with other South countries—can Chapter 5 studies the evolving connectiv- lead to a virtuous cycle of thriving trade and ity between LAC and the rising South based economic growth. It focuses on the potential on the relative importance of domestic ver- for technology diffusion and learning spill- sus external demand. It evaluates whether overs from the region’s international trade the low domestic saving rates in the region linkages. impaired its growth potential in the past and Chapter 3 assesses whether and how the may continue to do so in the future given ongoing restructuring of the global economy, changes in the world economy. Annex OA TABLE OA.1 Country group composition Region Countries Higher-income countries in Latin America Argentina, the Bahamas, Barbados, Brazil, Chile, Colombia, Costa Rica, Ecua- and the Caribbean (LAC1) dor, Mexico, Panama, Peru, Trinidad, Uruguay, Venezuela, RB. Lower-income countries in Latin America Belize, Bolivia, the Dominican Republic, El Salvador, Guatemala, Guyana, Hon- and the Caribbean (LAC2) duras, Nicaragua, Paraguay East Asia and Pacific (EAP) Bangladesh; Bhutan; Cambodia; China; Fiji; Hong Kong SAR; China; India; Indonesia; the Republic of Korea; Malaysia; Pakistan; Papua New Guinea; the Philippines; Sri Lanka; Thailand; Tonga; Vietnam Europe and Central Asia (ECA) Albania, Armenia, Azerbaijan, Belarus, Bosnia and Herzegovina, Bulgaria, Cro- atia, the Czech Republic, Estonia, Georgia, Greece, Hungary, Kazakhstan, the Kyrgyz Republic, Latvia, Lithuania, the former Yugoslav Republic of Macedo- nia, Moldova, Mongolia, Romania, Slovenia, Tajikistan, Turkmenistan, Ukraine High income Australia, Belgium, Canada, Cyprus, Denmark, Finland, France, Germany, Ice- land, Ireland, Israel, Italy, Japan, New Zealand, Norway, Portugal, Spain, Swe- den, Switzerland, the United Kingdom, the United States Middle East and North Africa (MENA) Algeria, the Islamic Republic of Iran, Jordan, Lebanon, Morocco, Syria, Tunisia, Turkey Sub-Saharan Africa (SSA) Angola, Benin, Botswana, Burkina Faso, Burundi, Cameroon, Chad, Côte d’Ivoire, Equatorial Guinea, Ethiopia, Gabon, The Gambia, Ghana, Guinea, Guinea-Bissau, Kenya, Lesotho, Madagascar, Malawi, Mali, Mauritania, Mau- ritius, Mozambique, Namibia, Niger, Rwanda, Senegal, South Africa, Sudan, Swaziland, Togo, Uganda, Zambia Note: The dividing line between LAC1 and LAC2 countries is per capita income of $5,000 a year. 36 LATIN AMERICA AND THE RISING SOUTH account the relative (rather than the absolute) Notes importance of each country in its regional 1. In this report, the North includes the Group of trade network. The distance between countries Seven (G-7) members (Canada, France, Ger- reflects the degree of similarity in the structure many, Italy, Japan, the United Kingdom, and of their trade connections (“similarity” is mea- the United States) plus the following Western sured in terms of the relative importance that Europe countries: Andorra, Austria, Belgium, a country has in other countries’ exports and Denmark, Finland, Greece, Iceland, Ireland, the relative importance that countries have Liechtenstein, Luxembourg, Monaco, Nether- in a given country’s exports). Countries with lands, Norway, Portugal, San Marino, Spain, similar trade structures are clustered together Sweden, and Switzerland. The South includes in figure O.8. Unlike figure O.5, however, all other economies, including all countries in figure O.8 depicts the density of connections, Latin America and the Caribbean (LAC). hence the systemic importance of countries in 2. As a share of the total number of possible con- their respective regional network, in terms of nections, the number of LAC trade connec- colors. The systemic importance of countries tions with North countries remained almost increases as colors shift from green to yellow to stable, at about 98 percent, between 1990 and red. Distance between countries is defined by 2012, whereas the number of LAC-South con- the sum in absolute value of the differences in nections increased from about 40 percent in trade shares between countries for a given des- 1990 to 62 percent in 2012. tination. The density captures the average dis- 3. LAC’s financial connections with other South tance per number of connections; the smaller countries also grew faster than its connections the distance, the higher the density (see De la with North countries, especially during the Torre, Didier, and Pinat 2014 and Van Eck and second half of the 2000s. For a deeper analysis Waltman 2010 for more technical details). of the degree of financial connectivity of LAC 9. The contrast between the two regional net- countries with the North and the South, see works in 2012 is captured by measures of Didier, Moretti, and Schmukler (2015) and average node density, defined as the average chapter 4 of this report. across nodes of the number of links over the 4. The share of active financial connections total number of possible connections. The within the South in 2011 was even smaller average node density in 2012 was 0.99 for for mergers and acquisitions (1.4 percent), East Asia and just 0.89 for LAC. The disper- syndicated loans (2.0 percent), and greenfield sion of node centrality (the standard deviation investments (3.6 percent). of the node density) was 0.09 for the East Asia 5. This similarity in export shares captures two network and 0.31 for the LAC network. distinct dimensions: the relative importance a 10. Various issues of the semiannual report series given country has in other countries’ exports produced by the World Bank’s Chief Econo- and the relative importance that other coun- mist Office for LAC (http://go.worldbank.org/ tries have in a given country’s exports. WTVI133GT0) examine the improvement in 6. The importance of a country to the global LAC’s macrofinancial policy management, trade network rises with its share in other beginning with the April 2008 issue, enti- countries’ exports and the number of its bilat- tled “Latin America’s New Immune System: eral trade connections. How Is It Coping with the Changing External 7. The measure of FVA of exports captures only Environment?” backward linkages (the imports a country uses 11. In contrast with other commodity cycles expe- in producing its exports). It does not capture rienced by LAC in the post–World War II era, forward linkages (the exports of a country that the rise of the South was associated with the are used by other countries as inputs to pro- simultaneous surge in the international prices duce their exports). The patterns of regional of virtually all commodities exported by LAC clustering in forward linkages are qualitatively economies for an extended period of time. In similar to the ones reported here. Chapter 2 this sense, it was a supercycle (see Sinnott, of this report provides a detailed analysis of Nash, and De la Torre 2010). GVCs. 12. Bernanke (2005) argues that a confluence of 8. The algorithm underlying figure O.8 is simi- factors led to the emergence of a global saving lar to that of figure O.5, in that it takes into glut, including policy interventions to boost OVERVIEW 37 exports in Asia, higher oil prices in the Middle participation in GVCs is lower than that of East, and a dearth of investment opportunities other South regions, even though there has and an aging population in advanced indus- been an increase in its participation since the trial countries. Mendoza, Quadrini, and Rios- 1990s. There is, however, considerable het- Rull (2007) attribute high saving in emerging erogeneity across LAC countries. market countries to relatively low levels of 18. FDI data, for instance, do not typically dif- financial development, which generate greater ferentiate between affiliates that provide precautionary saving. Caballero, Farhi, and inputs to parent companies and affiliates Gourinchas (2008) instead emphasize the lack that produce the same good or service as its of investment opportunities in these countries parent. and the associated shortage of financial assets 19. Non-LAC emerging market economies include as the main source of the global saving glut. Hungary, India, Indonesia, Korea, the Philip- Similarly, the IMF (2005) stresses low invest- pines, Poland, the Russian Federation, South ment rates following the Asian crisis rather Africa, Thailand, and Turkey. The economet- than an increase in saving rates. ric exercise entailed the estimation of struc- 13. In fact, between 2000 and 2011 Mexico was tural vector auto-regressive models (SVARs) a net exporter of mining products every year, (see Hevia and Servén 2014 and chapter 5 of a net exporter of agricultural commodities in this report for technical details). some years, and a net importer of manufac- 20. Low domestic saving implies an excess of tured products every year. Its gross exports of domestic expenditure over income. For small manufactured goods faced stiff competition open economies, which cannot influence from China, however, as discussed later in the international prices, the excess expenditures overview. that flow out of the country are satisfied by 14. Exports are disaggregated at the four-digit higher imports at unchanged international level of the International Standard Industrial prices. The excess of expenditure that falls on Classification (ISIC). the nontradable sector of the economy raises 15. Brazil, Chile, and Peru were among the coun- domestic prices, particularly if the economy is tries that benefited most from China’s rising near full employment. The rise in the prices imports of mineral commodities. Some Cen- of nontradables relative to tradables is a real tral American and Caribbean economies also exchange rate appreciation. It can become seem to have received a boost in their mineral durable to the extent that factors (especially exports (such as zinc from Honduras and alu- capital) are sticky and reallocate sluggishly to minum and bauxite from Jamaica), confirm- more productive uses across sectors and bor- ing that natural resource endowments were ders, a fact that is borne out by the observed important determinants of the impact of the large and persistent differentials in factor rise of China. productivities across firms, sectors, and coun- 16. There are, however, significant differences tries (see Hsieh and Klenow 2010; Svyerson within countries in LAC. Chapter 4 of this 2011; and Artuç, Lederman, and Rojas 2015, report explores the link between trade and among others). financial flows. 21. Low saving leads to a systematic tendency 17. The empirical literature on the extent of inte- toward current account deficits, which imply gration of LAC countries into GVCs is sparse, a buildup of external liabilities over time. Such but it has been expanding. Useful references a buildup can make the balance of payments are UNCTAD’s 2013 report Global Value more vulnerable to shocks and raise the risk Chains: Investment and Trade for Develop- of default, which would be reflected in a bias ment and the Inter-American Development toward higher risk premiums. Bank’s report Synchronized Factories (Blyde 22. Each point in the scatter plot represents a 2014). Chapter 2 of this report expands on country for a given time period. As the aim of this literature by providing more detailed this figure is to capture medium-term equilib- evidence on LAC’s participation in GVCs, rium relationships, each period is a three-year including its integration into GVCs with average. North and South countries. The general mes- 23. The benchmark is calculated based on regres- sage of this literature is consistent with the sion analysis for the entire sample. It indicates message of this report—namely, that LAC’s where a country is expected to be, controlling 38 LATIN AMERICA AND THE RISING SOUTH for its stage of development (as proxied by References per capita income); structural features that are largely beyond the control of policy (for Alcalá, F., and A. Ciccone. 2004. “Trade and Pro- example, demographic structure, natural ductivity.” Quarterly Journal of Economics resource endowment, economic size); and the 119 (2): 612–45. average policies of its peers (see De la Torre Alvarez, F., F. Buera, and R. Lucas. 2013. “Idea and Ize 2015 for technical details). Flows, Economic Growth, and Trade.” NBER 24. Associated with capital flight episodes were Working Paper 19667, National Bureau of multiple exchange rate regimes, which tend to Economic Research, Cambridge, MA. show up in the data as overvaluations, given Anderson, K., and A. Strutt. 2011. “Asia’s Chang- that the official exchange rate is typically used ing Role in World Trade: Prospects for South- to measure the purchasing power parity index. South Trade Growth to 2030.” ADB Working 25. Chile, Colombia, Mexico, Peru, Trinidad and Paper 264, Asian Development Bank, Manila. Tobago, and Uruguay were in the elite group Artuç, E., D. Lederman, and D. Rojas. 2015. of “investment-grade” countries. “The Rise of China and Labor Market Adjust- 26. See Frankel and Romer (1999) and Alcalá and ments in Latin America.” World Bank Policy Ciccone (2004), among many others. Singh Research Working Paper No. 7155, Washing- (2010) reviews this literature. ton, DC. 27. Endowments determine the structure of pro- Autor, D. H., D. Dorn, and G. H. Hanson. 2013. duction, employment, and trade in neoclassical “The China Syndrome: Local Labor Market models of trade and development in the tradi- Effects of Import Competition in the United tion of Hecksher-Ohlin. Relative national aver- States.” American Economic Review 103 (6): age productivities matter in Ricardian models. 2121–68. 28. Some studies, notably Hausmann, Hwang, Baldwin, R. 2012. “Trade and Industrialisation and Rodrik (2007), put the emphasis on what after Globalisation’s Second Unbundling: How a country trades as a means of identifying the Building and Joining a Supply Chain Are Dif- productivity embedded in the traded good. ferent and Why It Matters.” In Globalization However, the empirical approach in such in An Age of Crisis: Multilateral Economic studies suffers from important limitations, as Cooperation in the Twenty-First Century, ed. Lederman and Maloney (2012) note. R. Feenstra and A. Taylor. Chicago: University 29. Both high and low savers in the LAC1 group of Chicago Press. of countries had substantially undervalued Bernanke, B. 2005. “The Global Saving Glut and currencies in the 1980s and 1990s, and both the U.S. Current Account Deficit.” Speech 77, groups experienced a substantial appreciation Board of Governors of the Federal Reserve Sys- during the 2000s. However, the real apprecia- tem, Washington, DC. tion was much more pronounced (and invest- Blyde, J., ed. 2014. Synchronized Factories: ment and growth lower) among low savers, Latin America and the Caribbean in the Era which became significantly overvalued rela- of Global Value Chains. Washington, DC: tive to benchmark by the end of the period. Inter-American Development Bank. In contrast, the high savers were able to retain Caballero, R.J., E. Farhi, and P.-O. Gourinchas. somewhat undervalued currencies by the end 2008. “Financial Crash, Commodity Prices of the period. and Global Imbalances.” NBER Working 30. As an example, at a summit meeting of the Paper 14521, National Bureau of Economic Asia-Pacific Economic Cooperation in 2002, Research, Cambridge, MA. President Vicente Fox remarked, “It is not Chiquiar, D. 2014. Presentation at the con- clear whether or not China is actually com- ference on “The Global Insertion of Asian petitive. Perhaps it is, but perhaps its current Economies,” Center for Latin American success is based on the fact that they do not Monetary Studies (CEMLA)–Central Bank respect a series of rules that other countries, of Mex ico, Ju ne 5 – 6. w w w.cem la.org such as Mexico, do respect” (cited in Leder- /actividades/2014/2014-06-InsertionAsian man, Olarreaga, and Perry 2009, 4). Economies.html. OVERVIEW 39 De la Torre, A., T. Didier, C. Calderón, T. Cord- Labor Market Frictions Shape the Impact of ella, and S. Pienknagura. 2011. “Latin America International Trade on Jobs and Wages. Wash- and the Caribbean’s Long-Term Growth: Made ington, DC: World Bank. in China?” World Bank LAC Semi-Annual Hsieh, C.-T, and P. J. Klenow. 2010. “Develop- Report, September, Washington, DC. ment Accounting.” American Economic Jour- De la Torre, A., T. Didier, and M. Pinat. 2014. nal: Macroeconomics 2 (1): 207–23. “Can Latin America Tap the Globalization IADB (Inter-American Development Bank). 2014. upside?” World Bank Policy Research Working Rethinking Productive Development: Sound Paper 6837, Washington, DC. Policies and Institutions for Economic Trans- De la Torre, A., and A. Ize, 2015. “Saving for formation. Washington, DC. Growth: New Views on an Old Debate from IMF (International Monetary Fund). 2005. a Latin American Perspective.” World Bank, “Global Imbalances: A Savings and Investment background paper for this report, Washing- Perspective.” World Economic Outlook, Sep- ton, DC. tember, Washington, DC. De la Torre, A., A. Ize, and S. Schmukler. 2011. Keller, W. 2004. “International Technology Dif- Financial Development in Latin America and fusion.” Journal of Economic Literature 42 the Caribbean: The Road Ahead. Washington, (3): 752–82. DC: World Bank. Lane, P. R., and G. M. Milesi-Ferretti. 2007. “The Didier, T., M. Moretti, and S. Schmukler. 2015. External Wealth of Nations Mark II: Revised “The Changing Patterns of Financial Integra- and Extended Estimates of Foreign Assets And tion in Latin America.” World Bank Policy Liabilities, 1970–2004.” Journal of Interna- Research Working Paper 7190, Washington, tional Economics 73 (2): 223–50. DC. Lederman, D., and W. F. Maloney. 2007. “Trade Didier, T., H. Nguyen, and S. Pienknagura. 2015. Structure and Growth.” In Natural Resources: “FDI and Growth: Does the Source Country Neither Curse nor Destiny, ed. D. Lederman Matter?” World Bank, background paper for and W. Maloney. Palo Alto, CA: Stanford Uni- this report, Washington, DC. versity Press. Didier, D., and M. Pinat. 2015. “The Nature of ———. 2012. Does What You Export Matter? In Trade and Growth Linkages.” World Bank, Search of Empirical Guidance for Industrial background paper for this report, Washington, Policies. Washington, DC: World Bank. DC. Lederman, D., J. Messina, S. Pienknagura, and Finger, J. M., and M. Kreinin. 1979. “A Measure J. Rigolini. 2014. Latin American Entrepre- of Export Similarity and Its Possible Uses.” neurs: Many Firms, Little Innovation. Wash- Economic Journal 89 (356): 905–12. ington, DC: World Bank. Frankel, J. A., and D. Romer. 1999. “Does Trade Lederman, D., M. Olarreaga, and G. Perry, eds. Cause Growth?” American Economic Review 2009. China’s and India’s Challenge to Latin 89 (3): 379–99. America: Opportunity or Threat? Washing- Gourinchas, P.-O., and O. Jeanne. 2012. “Global ton, DC: World Bank. Safe Assets.” BIS Working Paper 399, Bank for Mendoza, E. G., V. Quadrini, and J.-V. Ríos-Rull. International Settlements, Basel. 2007. “Financial Integration, Financial Deep- Hausmann, R., J. Hwang, and D. Rodrik. 2007. ness and Global Imbalances.” CEPR Discus- “What You Export Matters.” Journal of Eco- sion Paper 6149, Centre for Economic Policy nomic Growth 12 (1): 1–25. Research, London. Hevia, C., and L. Servén. 2014. “The Macroeco- Pinat, M. 2015. “Diffusion of Ideas and Central- nomic Impact of Global and Domestic Shocks ity in the Trade Network.” World Bank, Wash- on Latin America.” Background paper for this ington, DC. report, World Bank, Office of the Chief Econ- Prasad, E., R. Rajan, and A. Subramanian. 2007. omist for Latin America and the Caribbean, “Foreign Capital and Economic Growth.” Washington, DC. Cornell University, Department of Economics, Hollweg, C. H., D. Lederman, D. Rojas, and Ithaca, NY. E. Ruppert-Bulmer. 2014. Sticky Feet: How 40 LATIN AMERICA AND THE RISING SOUTH Rodrik, D. 2008. “The Real Exchange Rate and 05/2010, United Nations Industrial Develop- Economic Growth.” Brookings Papers on Eco- ment Organization, Vienna. nomic Activity 39 (2): 365–439. U NC TA D (United Nations Conference on Singh, T. 2010. “Does International Trade Cause Trade and Development). 2013. Global Value Economic Growth? A Survey.” World Econ- Chains: Investment and Trade for Develop- omy 33 (11): 1517–64. ment. World Investment Report, Geneva. Sinnott, E., J. Nash, and A. de la Torre. 2010. U.S. National Intelligence Council. 2012. Global Natural Resources in Latin America and the Trends 2013: Alternative Worlds. Washing- Caribbean: Beyond Booms and Busts? Wash- ton, DC. ington, DC: World Bank. Van Eck, N. J., and L. Waltman. 2010. “Software Sturgeon, T., and O. Memevodic. 2010. “Map- Survey: VOSviewer, a Computer Program for ping Global Value Chains: Intermediate Goods Bibliometric Mapping.” Scientometrics 84 (2): Trade and Structural Change in the World 523–38. Economy.” UNIDO Development Policy and World Bank. 2012. Global Development Hori- Strategic Research Branch Working Paper zons. Washington, DC. Three Global Trends That Shaped Latin American and Caribbean 1 Development at the Dawn of the Twenty-First Century T he world economy is not what it used that the South has arrived at the center of to be 30 or even 15 years ago. For the global economy with surprising speed, most of the twentieth century, the especially since the dawn of the 21st cen- developed North dominated the global econ- tury. This reconfiguration of the global omy.1 This dominance led to the emergence landscape suggests the need to go beyond of various strands of “dependency” theory, the static North-South paradigm toward a which found green pastures in Latin Amer- dynamic center-periphery one. ican development thinking. 2 The essence of This report argues that the economic Latin American structuralism was pessi- shocks emanating from the rise of South mism: the dominance of the North, acting countries as central players in global eco- as “center” to a “periphery” of developing nomic relations have brought significant South countries, would be ever rising, at least changes to economies in Latin America in part because of the secular trends in the and the Caribbean (LAC), with notable dif- prices of exports from the South relative to ferences within the region depending on exports from the North. the economic structures that each country The world economy has evolved in the inherited from the twentieth century. LAC past several decades, rendering this central is an increasingly globalized region, and its tenet of Latin American dependency the- economic future depends a great deal on the ory obsolete. Several South economies are extent and quality of its external connections. now part of what can be empirically char- It is likely that not only the incidence of inter- acterized as the “center” of global commer- national trade and financial connections but cial relations. This chapter documents this also the nature of these international linkages empirical regularity through network anal- matter for its future economic growth and yses based on bilateral trade and financial for the generation of good-quality jobs. This data that show how countries are part of report therefore places significant emphasis global networks. Being at the center of a on the consequences of the changing nature global network entails having numerous and of LAC’s external connections, analyzing quantitatively important bilateral connec- particularly their trade, financial, macroeco- tions. It is in this (narrow) technical sense nomic, and labor market aspects. 41 42 THE RISE OF THE SOUTH As a starting point for this analysis, this As of the writing of this report, the world chapter outlines three sets of facts related to seems to be entering a phase that many the rise of the South that are shaping LAC’s observers have called “the new normal,” economic prospects: characterized by a slower global growth. The second half of the 2010s is thus poised 1. The weight of the South in the global to have different dynamics from the first economy has risen, particularly after decade of the 2000s. 2000, but its rise has not been even Despite the swiftness of these changes, across sectors or types of flows. projections suggest that these patterns are not 2. Several South countries are now at the temporary and that the South will continue center of the global trade network, but to gain space in the years to come.3 This out- none is at the center of global financial look partly reflects the broad reach of the rise networks. of the South, a phenomenon that goes well 3. The structure of bilateral trade and beyond the emergence of China as a giant in financial connections of the South has the global economy. Indeed, these trends are been generally different from that of not driven by a small set of South countries; the North, with geography and endow- they are observed across a vast number of ments arguably shaping their evolving countries. During the 2000s, 69 of 164 South structure. countries in the sample grew faster than the average South country, 130 more rap- idly than the fastest-growing North country Set of Facts 1: The weight of the (Luxembourg), and 154 more rapidly than South in the global economy has the average North country. risen, particularly after 2000, but Although China is not the only South its rise has not been even across economy behind these trends, it has played sectors or types of flows. a particularly important role. In the span The South has been growing faster than the of less than 30 years, it transformed itself North. The gross domestic product (GDP) in from a rural, inward-looking, slow-growing current dollars of the South remained at about economy to a fast-growing and increasingly 20 percent of world GDP between the 1970s urban and industrial one. Between 1978, and 1990s (figure 1.1, panel a). By the late when economic liberalization began, and 2000s, this share had doubled to 40 percent. 2012, China’s economy expanded more Moreover, the globalization of the South, than 20-fold in real terms. In 1978 Chi- which picked up in the late 1980s and con- na’s nominal GDP represented about 1.7 tinued apace during the 1990s, accelerated percent of world GDP; by 2012 China had and intensified substantially in the 2000s. become the world’s second-largest econ- The South accounted for 51 percent of global omy in terms of nominal GDP, representing trade flows in 2012, up from just 24 percent about 51 percent of the United States’ GDP in 1970 and 35 percent in 2000 (see figure and 11.3 percent of global GDP in current 1.1, panel b). The South received less than 20 dollars. China also gained prominence in percent of global capital inflows in the 1970s global trade, becoming the world’s largest and about 26 percent in the 1990s, whereas exporter in absolute terms and one of the by the end of the 2000s it received almost world’s largest importers. Its rise in global 55 percent (see figure 1.1, panel c). South finance was more modest but also signifi- countries also became more representative cant: China represented about 8 percent of as sources of capital flows, sending about 55 global capital inflows (9 percent of global percent of global capital outflows between capital outflows) in 2012, up from 1 per- 2008 and 2012, up from 14 percent in 1990. cent (1 percent) in 2000. THREE GLOBAL TRENDS THAT SHAPED LATIN AMERICAN AND CARIBBEAN DE VELOPMENT 43 The rise of the South has changed FIGURE 1.1 Rise of the South: Share of world GDP, trade, the composition of global trade flows and capital flows across sectors and between exports and imports within sectors. a. World GDP 100 The rise of the South in the global economy 90 reflects not only higher growth rates in the 80 Share of world GDP (%) South than in the North but also differences 69 60 70 74 in structural features. The patterns of glo- 60 78 78 76 balization of the North and the emerging 50 South differ in important ways. In particular, 40 9 there is significant heterogeneity in the sec- 30 toral composition of trade flows of the North 2 2 4 5 20 3 31 26 when contrasted with the South as well as 10 20 21 21 21 in the sectoral composition of trade flows 0 across South countries. The export baskets 3 7 2 9 9 9 –0 –0 –1 –8 –9 –7 00 04 08 80 90 70 of South countries typically include a larger 20 20 20 19 19 19 share of primary goods than those of North Years countries (figure 1.2, panel a). Between 2000 and 2012, for example, the share of primary b. World trade 100 goods in total goods exports was 57 percent 90 in Sub-Saharan Africa (SSA), 29 percent in Share of world trade (%) 80 LAC, and just 8 percent in the North. 52 54 There are also differences in the sectoral 70 68 64 58 68 66 61 60 75 73 76 75 composition of imports of North and South 50 countries (see figure 1.2, panel b). The share 9 40 7 8 of primary goods in imports averaged about 4 6 30 2 2 4 10 percent in the South and 14 percent in the 0 1 0 1 20 39 37 North between 2000 and 2012. China in 32 35 30 33 10 25 26 29 24 25 30 particular and East and South Asian econ- 0 omies more broadly seem to be exceptions 19 –79 19 –89 19 –79 19 –89 2 20 –07 20 –99 20 –07 20 –99 20 –03 2 20 –03 –1 –1 among South economies: the composition of 70 80 70 80 08 04 90 04 90 00 08 00 19 19 their trade baskets is on average more simi- Exports Imports lar to that of North countries than to other South countries. c. World capital flows In light of these differences, changes 100 in the weight of the South in global trade, 90 Share of world capital flows (%) especially during the 2000s, differed across 80 45 46 sectors and between exports and imports 70 65 69 within a given sector. The weight of the 60 75 68 68 82 80 85 84 82 North in global trade declined substantially 50 9 7 during the 2000s in both the primary (agri- 40 culture and mining) and manufacturing sec- 1 1 30 2 1 tors, though rankings across sectors were 2 45 48 20 0 1 31 30 0 0 1 34 30 10 18 19 24 15 16 17 0 19 –79 19 –89 19 –79 19 –89 2 20 –99 20 –07 20 –07 20 –99 20 –03 2 20 –03 –1 –1 70 80 70 80 08 90 04 04 90 00 08 00 Sources: Calculations based on data from World Development Indicators 19 19 (WDI), Direction of Trade Statistics (DOTS), and Balance of Payments Statistics (BOPS). Capital inflows Capital outflowsa Note: The North includes the G-7 members and Western Europe countries. The South includes all other economies. G-7 = Group of Seven; South (excluding China) China North GDP = gross domestic product. a. Capital outflows exclude international reserves. 44 THE RISE OF THE SOUTH broadly maintained (see figure 1.2, panels countries generally receive a larger share of c and d). The flipside of this trend is an financial flows in the primary sector than increase in the shares of the South: between North countries, though there is significant 2000 and 2012, its share of global manu- variation in the magnitude of these differences factures exports increased from 30 percent across countries. For example, between 2003 to 46 percent and its share of global com- and 2012, on average countries in Europe and modities exports rose from 62 percent to 68 Central Asia (ECA), LAC, and SSA received percent. at least 50 percent of syndicated loans and There is also significant heterogeneity merger and acquisition (M&A) inflows in the within the South across both countries and primary sector. The share in North countries sectors.4 China is by a wide margin the most was about 20 percent. Foreign investments by important country behind the expansion of South countries are also tilted toward the pri- the South in global exports of manufactur- mary sector on average. For example, between ing: its share increased more than 10 percent- 2003 and 2012, the share of greenfield invest- age points, from slightly less than 5 percent ments abroad that went to the primary sector in 2000 to about 16 percent in 2012 (see was much larger in LAC (44 percent) than in figure 1.2, panel c). Together the other top the North (19 percent). 20 South countries increased their share in As the weight of South countries in global global manufacturing exports by no more financial flows changed, so did the sectoral than 9 percentage points.5 At the same time, composition of global financial flows, espe- the share of world manufacturing exports of cially during the 2000s. The share of global some South countries (for example, Malay- inflows in the primary sector increased for sia, Mexico, and Philippines) declined. syndicated loans (from 25 percent to 35 per- The rise of the South in global primary cent of global flows) and for M&A (from exports features a different set of countries, 26 percent to 33 percent), whereas it fell with Australia, Brazil, and the Russian Fed- slightly for global greenfield flows (from eration registering the largest gains in global 22 percent to 19 percent) between 2003–07 shares. Among the top 20 South countries are and 2008–12. There was also a recomposi- India, Nigeria, South Africa, and some LAC tion of senders and receivers of global flows countries (Bolivia, Chile, Colombia, Ecua- across sectors (figure 1.4). The weight of dor, and Peru). China experienced the largest North countries as senders and receivers of increase in weight on the receiving end: its financial flows generally declined during share of global imports of (agricultural and this period, especially in the primary sector, mining) primary products rose from about 3 where the share of North countries in global percent in 2000 to 14 percent in 2012 (see M&A fell 23 percentage points as senders figure 1.2, panel d). Several South countries and 21 percentage points as receivers. Con- with increases in manufacturing exports, versely, the weight of the South in global such as India, Poland, the Republic of Korea, capital flows increased, though different and Turkey, also increased their imports of regions of the South gained space in differ- commodities. ent sectors and in different types of flows. For example, countries in East Asia and Pacific (EAP) typically increased their share The rise of the South has also led to as receivers of global syndicated loan flows a significant recomposition of global in the primary sector, whereas countries in financial flows across sectors and types ECA and the Middle East and North Africa of flows. (MENA) lost global participation. LAC and The sectoral composition of global gross EAP countries almost tripled their global financial flows (capital account–related share as receivers of M&A flows in the pri- flows by foreign and domestic agents) for the mary sector, whereas China and EAP coun- South and the North differ (figure 1.3). South tries became large senders of these flows. THREE GLOBAL TRENDS THAT SHAPED LATIN AMERICAN AND CARIBBEAN DE VELOPMENT 45 FIGURE 1.2 Sectoral composition of trade flows a. The sectoral composition of b. The sectoral composition of exports across regions imports across regions North 3 5 92 North 3 11 86 China 11 98 China 6 18 76 ECA 3 21 75 ECA 3 7 90 EAP 3 11 86 EAP 2 8 90 LAC 9 20 71 LAC 4 6 90 MENA 2 61 37 MENA 5 4 91 South 6 4 90 South Asia Asia 4 21 75 SSA 11 46 43 SSA 3 5 92 Central America 4 13 82 Central America and Mexico and Mexico 4 2 94 The Caribbean 3 24 73 The Caribbean 5 13 81 South 12 26 62 South America America 4 8 89 0 10 20 30 40 50 60 70 80 90 100 0 10 20 30 40 50 60 70 80 90 100 Percent Percent Agriculture Mining Manufacturing c. The composition of global exports d. The composition of global imports across sectors and regions across sectors and regions 2000 3 10 6 2 17 54 2000 4 9 6 7 8 64 Agriculture Agriculture 2005 3 10 8 3 17 51 2005 8 8 7 7 6 61 2012 3 13 91 19 43 2012 14 11 8 8 7 47 2000 1 10 8 27 13 32 2000 3 7 4 2 3 77 Mining Mining 2005 1 10 13 34 12 27 2005 6 10 523 70 2012 1 17 22 7 14 28 2012 14 10 6 14 58 2000 5 12 41 6 70 2000 2 12 5 3 6 69 Manufacturing Manufacturing 2005 10 12 6 1 5 64 2005 4 13 7 4 5 64 2012 16 12 815 54 2012 6 15 10 6 7 53 0 10 20 30 40 50 60 70 80 90 100 0 10 20 30 40 50 60 70 80 90 100 Percent Percent China EAP South Asia ECA MENA SSA LAC North Source: Calculations based on data from Comtrade. Note: Panels a and b show the average sectoral composition of exports and imports between 2000 and 2012 across regions. The sectoral classifica- tion of trade flows is based on the International Standard Industrial Classification (ISIC) grouping, Revision 3. Agriculture corresponds to ISIC codes 0111–0500, mining to ISIC codes 1010–1429, and manufacturing to ISIC codes 1511–3699. The North includes the Group of 7 (G-7) members and Western Europe countries. EAP = East Asia and Pacific; ECA = Europe and Central Asia; LAC = Latin America and the Caribbean; MENA = Middle East and North Africa; SSA = Sub-Saharan Africa. 46 THE RISE OF THE SOUTH FIGURE 1.3 Sectoral composition of financial flows across regions a. Syndicated loans by receiver region by sender region North North China China ECA ECA EAP EAP LAC LAC MENA MENA South Asia South Asia SSA SSA Central America Central America and Mexico and Mexico The Caribbean The Caribbean South America South America 0 10 20 30 40 50 60 70 80 90 100 0 10 20 30 40 50 60 70 80 90 100 Percent Percent b. Mergers and acquisitions by receiver region by sender region North North China China ECA ECA EAP EAP LAC LAC MENA MENA South Asia South Asia SSA SSA Central America Central America and Mexico and Mexico The Caribbean The Caribbean South America South America 0 10 20 30 40 50 60 70 80 90 100 0 10 20 30 40 50 60 70 80 90 100 Percent Percent c. Greenfield investments by receiver region by sender region North North China China ECA ECA EAP EAP LAC LAC MENA MENA South Asia South Asia SSA SSA Central America Central America and Mexico and Mexico The Caribbean The Caribbean South America South America 0 10 20 30 40 50 60 70 80 90 100 0 10 20 30 40 50 60 70 80 90 100 Percent Percent Primary sector (LAC) Manufacturing sector (LAC) Primary sector (other) Manufacturing sector (other) Sources: Data on syndicated loans and mergers and acquisitions are from SDC Platinum. Data on greenfield investments are from fDi Markets. Note: The sectoral classification of financial flows is based on the International Standard Industrial Classification (ISIC), Revision 3. The primary sector corresponds to ISIC codes 0111–0500 and 1010–1429. The manufacturing sector corresponds to ISIC codes 1511–3699. The North includes the G-7 members and Western Europe countries. The South includes all other economies. EAP = East Asia and Pacific; ECA = Europe and Central Asia; LAC = Latin America and the Caribbean; MENA = Middle East and North Africa; SSA = Sub-Saharan Africa. THREE GLOBAL TRENDS THAT SHAPED LATIN AMERICAN AND CARIBBEAN DE VELOPMENT 47 FIGURE 1.4 Composition of global financial flows across sectors a. Syndicated loans by receiver region by sender region 2003–07 10 10 45 2003–07 6 92 Primary Primary 2008–12 19 11 46 2008–12 7 87 Manufacturing Manufacturing 2003–07 4 2 86 2003–07 6 92 2008–12 6 5 81 2008–12 9 89 0 10 20 30 40 50 60 70 80 90 100 0 10 20 30 40 50 60 70 80 90 100 Percent Percent b. Mergers and acquisitions by receiver region by sender region 2003–07 8 5 71 2003–07 5 4 7 74 Primary Primary 2008–12 20 15 49 2008–12 14 17 2 53 Manufacturing Manufacturing 2003–07 6 4 83 2003–07 6 4 83 0 2008–12 6 4 78 2008–12 7 3 80 2 0 10 20 30 40 50 60 70 80 90 100 0 10 20 30 40 50 60 70 80 90 100 Percent Percent c. Greenfield investments by receiver region by sender region 2003–07 2 15 9 14 21 25 11 2003–07 3 14 6 66 Primary Primary 2008–12 3 23 8 10 21 19 15 2008–12 6 10 4 66 Manufacturing Manufacturing 2003–07 18 14 15 10 3 10 23 2003–07 3 10 1 75 2008–12 12 15 11 11 6 15 21 2008–12 6 9 2 69 0 10 20 30 40 50 60 70 80 90 100 0 10 20 30 40 50 60 70 80 90 100 Percent Percent China EAP South Asia ECA MENA SSA LAC North Sources: Data on syndicated loans and mergers and acquisitions are from SDC Platinum. Data on greenfield investments are from fDi Markets. Note: The sectoral classification of financial flows is based on the International Standard Industrial Classification (ISIC), Revision 3. The primary sector corresponds to ISIC codes 0111–0500 and 1010–1429. The manufacturing sector corresponds to ISIC codes 1511–3699. The North includes the G-7 members and Western Europe countries. The South includes all other economies. EAP = East Asia and Pacific; ECA = Europe and Central Asia; LAC = Latin America and the Caribbean; MENA = Middle East and North Africa; SSA = Sub-Saharan Africa. 48 THE RISE OF THE SOUTH The composition of global net financial external net liability positions from debt to flows also experienced important changes. equity (figure 1.5). Countries from the South In particular, there was a recomposition of that had been large net debtors became net net equity and net debt flows within coun- creditors with respect to the rest of the world tries in both the North and South. Since the in debt contracts. This change reflected in late 1990s, partly in response to the painful large part the significant accumulation of lessons learned from the recurrent crises suf- international reserves that followed the crises fered during the late twentieth century, many of the late 1990s. At the same time, countries countries in the South, especially in Asia and from the South became more active users of Latin America, have steadily changed the foreign equity finance, which led to a rising structure of their external assets and liability net debtor position in risk-sharing equity positions. Many countries in the South, espe- contracts (particularly foreign direct invest- cially in EAP and LAC, have switched their ment [FDI]) with respect to the rest of the FIGURE 1.5 Composition of foreign assets and liabilities in the South, by region a. LAC-7 b. Asia-7 10 20 5 10 0 0 –5 % of GDP % of GDP –10 –10 –15 –20 –20 –30 –25 –30 –40 –35 –50 1990 1993 1996 1999 2002 2005 2008 2011 1990 1993 1996 1999 2002 2005 2008 2011 Year Year c. ECA-7 d. SSA-7 5 20 0 0 -5 –20 –15 –40 % of GDP % of GDP –60 –25 –80 –35 –100 –120 –45 –140 –55 –160 1990 1993 1996 1999 2002 2005 2008 2011 1990 1993 1996 1999 2002 2005 2008 2011 Year Year Net debt position Net equity position Source: Updated and extended version of dataset constructed by Lane and Milesi-Ferretti (2007). Note: Ratios are calculated at the country level and then averaged across countries (simple average) between 1990 and 2011. LAC-7: Argentina, Brazil, Chile, Colombia, Mexico, Peru, and Uruguay. Asia-7: China, India, Indonesia, the Republic of Korea, Malaysia, Philippines, and Thailand. ECA-7: Croatia, the Czech Republic, Hungary, Lithuania, Poland, the Russian Federation, and Turkey. SSA-7: Angola, Ghana, Kenya, Nigeria, South Africa, Sudan, and Zambia. GDP = gross domestic product. THREE GLOBAL TRENDS THAT SHAPED LATIN AMERICAN AND CARIBBEAN DE VELOPMENT 49 world. In contrast, countries from the North demand, and current account deficits have became net creditors in equity contracts and been persistent. For all of the debate about net debtors in debt contracts. These patterns the commodity boom in the 2000s, current reflect to some extent the dynamics of the account surpluses among LAC’s commodity recomposition of net savers and net borrow- exporters were in most cases short-lived: the ers in the global economy. surpluses were virtually gone by mid-2008 and only temporarily recovered in 2009, as an undesired consequence of the global trade The net integration of countries into the collapse. Indeed, by 2012 current account global economy reflects uneven growth deficits were the norm in these countries, rates of imports and exports, capital with only República Bolivariana de Vene- inflows and outflows, or both. zuela displaying a current account surplus Another dimension of globalization is the (figure 1.6). In contrast, East Asian coun- net integration of countries into the global tries consistently generated relatively large economy, based on the relative importance current account surpluses during most of the of countries’ domestic and external demands. past two decades. Moreover, LAC economies To the extent that external demand reflects typically integrated with relatively appreci- the excess of national income over absorp- ated real exchange rates compared with East tion (comprising both consumption and Asian countries. The Economist ’s Big Mac investment spending), countries with external Index provides some evidence that currencies demand–driven integration patterns typically in East Asia have been relatively underval- run systematic current account surpluses or ued, whereas currencies in LAC have been systematic excesses of domestic saving over relatively overvalued. investment (reflecting on the one hand the The rise of China and other South players difference between exports and imports and (especially in Asia and among oil-exporting on the other hand the difference in capital countries) with persistent current account inflows and outflows). In contrast, countries surpluses and large accumulation of inter- with a connection to the rest of the world national reserves has led to a heated debate based on domestic demand generally have over their contribution to the “global imbal- systematic excesses of domestic investment ances” in trade and finance and the “global over domestic saving and therefore typically saving glut,” which has accumulated largely run current account deficits. The pattern of in U.S. Treasury bonds. A prominent view is globalization can thus differ across countries that an excess of saving over investment in as a result of uneven growth rates of imports these South economies, invested in U.S. dol- versus exports or of capital inflows and out- lar assets, eased financial conditions in defi- flows. These persistent current account defi- cit countries, particularly the United States, cits are usually accompanied by consistently and exerted significant downward pressure overvalued currencies.6 on world interest rates.7 With low interest This seldom explored aspect of global- rates in North economies, a search for yield ization is particularly important for many among investors triggered capital flows to LAC countries, as it reflects a dependence LAC and other South countries, where bor- on external saving and a reliance on domes- rowing spreads fell to historically low lev- tic demand that sets them apart from many els and currencies appreciated significantly. other South economies, especially in East Indeed, for most of the 2000s, the strong Asia. In several LAC countries, notably Mex- tailwinds coming from commodity prices, ico and most South American countries, along with large volumes of capital inflows, aggregate demand has been clearly tilted reinforced the broad appreciation pressures in favor of domestic rather than external in LAC. 50 THE RISE OF THE SOUTH FIGURE 1.6 Patterns of net integration into the global economy 25 Latin America and the Caribbean Current account balance as a share of GDP, 2012 (%) 20 East Asia and Pacific Other countries Singapore 15 10 5 Korea, Rep. Venezuela, RB China Philippines Hong Kong, Argentina 0 Thailand SAR, China Mexico Brazil Chile Indonesia Colombia –5 Uruguay Costa Rica –10 –80 –60 –40 –20 0 20 40 60 80 100 Big Mac Index as of July 2012 (at market price valuations) Sources: Calculations based on data from WDI and Economist Intelligence Unit. Set of Facts 2: The rise of the European countries), and Japan were at the South has had asymmetric core of the network. By 2012 several coun- effects on global trade and tries from the South, including not only China financial networks. but also Brazil, India, the Russian Federation, South Africa, Turkey, and others, had moved Several South economies have joined to the center. As a result of these changes, the the North at the center of the global South is no longer a synonym for periphery trade network. (and the North no longer a synonym for cen- This momentous change stands out clearly in ter) in global trade. panel a of figure 1.7, which shows the global trade network in 1980 and 2012. Each node The roles of North and South countries represents a country, and each link corre- at the center of the global trade sponds to an active bilateral connection network have differed. that exceeds a minimum threshold (in panel a, exports from one country to another, as Figure 1.8 illustrates the differing roles of indicated by the arrows). Connections that North and South countries. It takes into are trivial in magnitude are not graphed, but account the relative (rather than absolute) once graphed each connection has the same importance of each country in the global weight.8 Countries with a larger number of trade network. The distance between coun- connections are more centrally located in the tries reflects similarity in the structure of figure. their trade connections—the closer countries In 1980 a set of North countries stood at are to one another, the more alike they are what can be empirically characterized as the in terms of export shares. This similarity in center of the global trade network; the United export shares captures two distinct dimen- States, Germany (and a few other Western sions: the relative importance a given country THREE GLOBAL TRENDS THAT SHAPED LATIN AMERICAN AND CARIBBEAN DE VELOPMENT 51 FIGURE 1.7 Global trade and financial networks a. 1980 b. 2012 BRA North countries Latin America and the Caribbean Other South countries (continued) 52 THE RISE OF THE SOUTH FIGURE 1.7 Global trade and financial networks (continued) b. Global network of portfolio investmentsb 2001 2011 c. Global banking network based on syndicated loansc 1996 2012 d. Global FDI network based on mergers and acquisitions 1990 2012 North countries Latin America and the Caribbean Other South countries Continue (continued) THREE GLOBAL TRENDS THAT SHAPED LATIN AMERICAN AND CARIBBEAN DE VELOPMENT 53 FIGURE 1.7 Global trade and financial networks (continued) e. Global FDI network based on greenfield investments 2003 2012 North countries Latin America and the Caribbean Other South countries Sources: Calculations based on data on trade are from Direction of Trade Statistics (DOTS), on portfolio investments from Coordinated Portfolio Investment Survey (CPIS), on syndi- cated loans and mergers and acquisitions from SDC Platinum, and on greenfield investments from fDi Markets. Note: Networks are drawn using the Kamada-Kawai algorithm. Each node represents a country. Each link corresponds to an active connection (a positive flow or stock of investments) between a pair of countries. Arrows indicate the direction of these connections. For each dataset, the left-hand column shows the networks in the first year of the sample and the right-hand column shows the networks in the last year of the sample. The North includes the G-7 members and Western Europe countries. Other South includes all other economies except Latin America and Caribbean countries. All new syndicated loans on a given year are reported. FDI = foreign direct investment; G-7 = Group of Seven. a. Only trade flows (exports) greater than $10 million in 1980 or greater than $100 million in 2012 are reported. b. Only positive holdings of foreign portfolio assets (equity and bonds) are reported. c. All new syndicated loans on a given year are reported. has in other countries’ exports and the rela- the countries on the right are very close to one tive importance that other countries have in a another on the vertical dimension, reflecting given country’s exports. Countries that cap- a high degree of similarity in the structure of ture a larger share of other countries’ exports their trade connections with other countries and that are connected with a larger number in the network. The global trade network in of trading partners (that is, countries that 1980 thus tended to display a sort of “single are more important in the global network) polarity,” with some North countries acting appear to the right in figure 1.8. Along the as a single pole (that is, playing the same role) vertical dimension, the smaller the distance for world trade. between two countries, the more similar the The global trade network in 2012 reveals structure of trade connections across mem- a tectonic shift: several countries from the bers of the network. South appear on the right side of panel b of During the 1980s and 1990s, only North figure 1.8, indicating their increased rele- countries were clustered toward the right vance to world trade. However, they remain of the graph, indicating that only they were somewhat distant (along the vertical dimen- systemically important to the global trade sion) from the other (North) countries on the network. For example, for 1980 the United right side of the figure. This side of the figure States, Germany, and Japan appear at the far resembles a star, with small groups of central right side of panel a in figure 1.8. In addition, countries placed at a certain distance from FIGURE 1.8 Similarity in global trade networks a. 1980 Similarity in trade structures Systemic relevance in global trade b. 2012 Similarity in trade structures Systemic relevance in global trade Source: Calculations based on data from Direction of Trade Statistics (DOTS). Note: Each node represents a country. Each link corresponds to an active trade connection between a pair of countries. Arrows at the end of each link capture the direction of these connections. Trade connections are measured as exports as a share of total exports of the source country. Only shares greater than 1 percent are reported. The distance between coun- tries reflects similarity in the structure of their trade connections: the closer countries are to one another, the more alike they are in terms of export shares. Countries capturing a larger share of other countries’ exports and connected with a larger number of trading partners appear on the right-hand side of the figure (more systemically relevant countries in global trade). The smaller the distance between two countries along the vertical dimension, the more similar the structure of their trade connections across other members of the network. THREE GLOBAL TRENDS THAT SHAPED LATIN AMERICAN AND CARIBBEAN DE VELOPMENT 55 one another. Russia and Turkey, for example, FIGURE 1.9 Structural equivalence of trade connections are not located near any North core country a. 1980 b. 2012 from Europe, and Japan is not close to either China or Korea. Among the systemically 100 100 important countries, South countries thus play a different role from North countries in international trade. It is in this empirically 90 90 well-defined sense that the global trading landscape has become more heterogeneous and “multipolar.” This rising heterogeneity at the center of 80 80 global trade is also apparent when coun- tries are grouped according to the structural equivalence of their trade connections (see, for example, Burt 1976). Two countries play 70 70 the same role in the network (that is, have exact structural equivalence) when they have the same connections to all other countries. 60 60 In figure 1.9, countries are grouped by dif- Share of the world (%) Share of the world (%) ferent threshold levels of similarity in their trade structure (based on the value of trade flows between countries and the composi- 50 50 tion of trading partners). In 1980 there were basically three dominant groups of countries. In 2012, for the same threshold level of sim- ilarity in trade structures, there were many 40 40 more than three groups, and fewer countries belonged to each of the top three groups. These patterns suggest that as the South 30 30 gained space in global trade, the diversity of trade structures increased around the world. Intrinsically related to this diversity are argu- ably differences in the sectoral composition 20 20 of the trade flows of South and North coun- tries, as discussed in Set of Facts 1. 10 10 Unlike global trade, global finance has not been fundamentally restructured: the North still stands alone at the center of the global financial networks. 0 0.75 0.45 0.32 0 0.75 0.45 0.32 A key feature of the new dynamics of the Similarity Similarity global economy has been the asymmetry in the pattern of change in global trade and Source: Calculations based on data from Direction of Trade Statistics (DOTS). Note: Countries are grouped according to different threshold levels of similarity in their trade struc- financial networks.9 In the sphere of trade, ture (based on the volume of trade flows between countries and the composition of partners). the traditional correspondence between the Within each bar, the share of countries that belong to the same structurally equivalent group are shown in different colors. Each bar shows these grouping of countries at different threshold levels, North and the center (and the South and the reported in the x-axis. The structural equivalence of trade connections is based on the similarity of periphery) has been reconfigured. In con- the correlation matrix of trade flows. trast, in the sphere of finance, North coun- tries still stand alone at the center of the 56 THE RISE OF THE SOUTH global financial networks, though the South For all types of financial flows, the number has increased its connectivity within these of active South-South connections as a share networks (see figure 1.7, panels b, c, d, and of all active connections in the world increased e for portfolio investments, syndicated loans, more than the number of North-North, M&A, and greenfield investment flows). North-South, and South-North connections. Whether this asymmetry proves transitory The growing number of connections among is a matter of hot debate. Not only is there the relatively small countries of the South is a broad recognition of the U.S. dollar as an significant driver of these patterns. Although international currency, but the scale and net- larger South countries (such as Brazil, China, work effects associated with financial centers India, and Russia) increased the number of will not be easy for the South to overcome. their connections with other countries in the Moreover, the trade-finance asymmetry South—and in terms of volume these connec- stands in sharp contrast with broad historical tions typically dominate—they accounted for developments since the Industrial Revolution a relatively small fraction of the total number and throughout most of the twentieth cen- of South-South connections. The breadth of tury, when countries that became important the reach of the rise of the South phenome- economic powers also became international non is thus key to these patterns of financial financial centers. London, New York, and flows. To be sure, many countries in the South Tokyo, for example, became financial cen- have yet to be connected with a wide set of ters as their nations strengthened their roles countries in the world, especially in terms of as gravity poles for regional and even global financial connections with other countries in economic activity. the South, suggesting that there is still signifi- cant scope for the continued expansion of the South in cross-border flows. The growth of the South has been widespread, with new South-South as well as South-North and North-South LAC is increasingly connected with connections developing. other South countries in both trade and finance. As the South gained prominence in the global economy, the number of its bilateral inter- Countries in LAC broadened and deepened national connections proliferated. New con- their connections with other South countries, nections were established not only between though the value of such connections is still the South and the North but also within the relatively small, especially in finance, when South (figure 1.10). In 1990 only 46 percent contrasted to LAC-North connections (figure of the total number of possible South-South 1.11). For instance, the share of the South in trade connections were active; by 2012 this total trade flows to and from LAC countries proportion had risen to 70 percent.10 Similar increased about 70 percent (from 26 percent trends are observed across types of financial to 45 percent), during the 2000s. The expan- flows. In 2001 South countries had portfo- sion of syndicated loans and M&A between lio investments in 10 percent of the countries LAC and other South countries was also strik- of the South; by 2011 this share had more ing, albeit from lower bases, with syndicated than doubled, to 21 percent. The increase in loans rising almost 180 percent (from about this extensive margin for FDI flows within 4 percent to 12 percent of total flows) and the South was also considerable, albeit from M&A increasing more than 140 percent (from much lower starting points. The share of the about 15 percent to 37 percent of total flows). total number of South-South connections for LAC countries also became increasingly M&A that were active rose from 0.1 percent integrated with a wider set of other South in 1990 to 1.3 percent in 2011, and the share countries; intraregional integration has deep- of active greenfield investments rose from ened, and linkages with other South coun- 2.2 percent in 2003 to 3.4 percent in 2011. tries have expanded. These patterns have FIGURE 1.10 Extensive margin of South-South connections a. Tradea 1980 2012 b. Portfolio investmentsb 2001 2011 c. Syndicated loans 1996 2012 d. Mergers and acquisitions 1990 2012 e. Greenfield investments 2003 2012 Sources: Calculations based on data on trade are from Direction of Trade Statistics (DOTS), on portfolio investments are from Coordinated Portfolio Investment Survey (CPIS), on syn- dicated loans and mergers and acquisitions are from SDC Platinum, and on greenfield investments are from fDi Markets. Note: Each line represents a positive flow or stock of investments between two South countries. For each dataset, the left-hand column shows the network in the first year of the sample and the right-hand column shows the network in the last year of the sample. South countries comprise all countries that do not belong to the G-7 or are not located in West- ern Europe. G-7 = Group of Seven. a. Only connections worth more than $10 million are included. b. Only positive holdings of foreign portfolio assets (equity and bonds) are reported. 58 THE RISE OF THE SOUTH been widespread across flow types (trade, FIGURE 1.11 Regional composition of cross- portfolio investments, loans, and FDI). China border connections of countries in Latin America in particular has emerged as an important and the Caribbean partner for some LAC countries, especially South American countries on the trade front. 2000-2002 In 1990 virtually no trade existed between Trade LAC and China. By the late 2000s, LAC- 2003-2005 China trade represented 12 percent of total trade flows to and from LAC countries. On 2010-2012 the financial front, China’s role has been more limited, though its importance has been 2000-2002 rising, especially for FDI. investmentsa Portfolio 2003-2005 Set of Facts 3: The structure of 2010-2012 bilateral trade and financial connections of the South 2000-2002 has been generally different from that of the North, with Syndicated loans 2003-2005 geography and endowments arguably shaping their evolving 2010-2012 structure. Despite the increased diversification of 2000-2002 connections around the world, there is Mergers and acquisitions significant regional clustering in both 2003-2005 trade and financial relations. The South has broadened and deepened its 2010-2012 connections not only with North countries but also with other South countries. How- 2000-2002 investments Greenfield ever, the strongest trade ties for countries in both the North and the South are with neigh- 2003-2005 boring countries, suggesting that geographi- cal proximity has played an important role 2010-2012 in the evolution of these connections. Most 0 10 20 30 40 50 60 70 80 90 100 Central American and Caribbean countries, Percent for example, belong to a single cluster with North American countries, centered on the Other South Latin America and the Caribbean United States (figure 1.12). South American China North countries form a smaller cluster, centered on Brazil, made up mostly of countries in Sources: Calculations based on data on trade are from Direction of Trade Statistics (DOTS), on portfolio investments are from Coordinated Portfolio Mercosur. Other large clusters include one Investment Survey (CPIS), on syndicated loans and mergers and acquisi- consisting of European countries, centered tions are from SDC Platinum, and on greenfield investments are from fDi Markets. on Germany, and another comprising Asian Note: The figure considers both inflows and outflows. The North includes economies, including Japan and most East the G-7 members and Western Europe countries. Other South includes all other countries except China and countries in LAC. G-7 = Group of Seven. Asian economies, centered on China. a. The composition of portfolio investments is based on the holdings of Similar patterns are observed in global cross-border portfolio (equity and bonds) assets. Because of data limita- tions, these data cover only the following periods: 2001–02, 2003–05, and finance. South countries generally send the 2010–11. THREE GLOBAL TRENDS THAT SHAPED LATIN AMERICAN AND CARIBBEAN DE VELOPMENT 59 majority of their financial investments to factory North America, factory Europe, and the North, but neighboring South countries factory Asia. come in second as a share of these invest- As GVCs have gained prominence, ments. Countries in LAC typically invest in exports of final products have become other LAC countries, Asian countries invest increasingly composed of imports of inter- largely in other Asian countries, Eastern mediate inputs: more intermediate goods European countries invest mostly in other are traded across borders, and more parts Eastern European countries, and so on (fig- and components are imported for use in ure 1.13). These patterns hold for portfolio exports. Data on the sources of foreign investments and syndicated loans as well value added (FVA) in exports point to the as FDI (both M&A and greenfield invest- regional nature of GVCs, showing that the ment). The largest non-North recipients FVA content in exports typically originates of investments from LAC countries during in neighboring countries (figure 1.14, panel the 2000s, for example, were other LAC a).14 For example, almost 40 percent of the countries, which accounted for 7 percent of FVA in the exports of EAP economies comes total portfolio investments, 24 percent of from other economies in EAP, and more new syndicated loans, 34 percent of M&A than 75 percent of the FVA in the exports of flows, and 61 percent of new greenfield ECA countries comes from other ECA and investments. Western Europe countries. The degree of regional clustering in the sources of FVA in exports is much less pro- The development of global value chains nounced in LAC than in other South regions, has arguably played an important role though there is some clustering within LAC in the regional clustering in trade and subregions (see figure 1.14, panel b). Imports financial connections. from other South American countries repre- Underpinning to some extent these cluster- sent about 35 percent of the FVA in exports ing patterns has been the development of of South America on average, with the rest global value chains (GVCs)—the dispersion of LAC adding only another 3 percent of of production stages and processes across imported FVA. Similar patterns are observed countries.11, 12 GVCs are more regional than for Central America and Mexico. global.13 In Central America, for example, There is, however, a striking contrast in the dairy sector has crossed borders, and a the relative importance of other regions in GVC encompassing producers in El Salvador the FVA of exports across LAC countries. and Nicaragua has developed. Local compa- For Mexico, Central America, and the Carib- nies in El Salvador have forged local partner- bean, the United States and Canada are prom- ships with small industries in Nicaragua to inent sources of imported inputs used in their produce their national cheese (“el quesillo”), exports. GVCs in South America (as proxied which is then sold in the United States (see by the FVA in exports) seem much less tied Martinez-Piva and Zúñiga-Arias 2012). to North America than GVCs in Mexico, In Sub-Saharan Africa, the recent entry of Central America, and the Caribbean. In fact, South African clothing manufacturers into the three major centers of global production neighboring countries (such as Lesotho and (North America, Western Europe, and East Swaziland) has led to the rise of regional Asia) provide a more balanced contribution value chains driven by South African retail- to the exports of South America. For exam- ers (see Morris, Staritz, and Barnes 2011). ple, in 2011 the United States and Canada More broadly, Baldwin and Lopez-Gonzalez provided about 40 percent of the FVA in the (2013) highlight the importance of three exports of Mexico, Central America, and the large production clusters around the world: Caribbean but only 19 percent of the FVA in 60 THE RISE OF THE SOUTH FIGURE 1.12 Clusters in the global trade network a. Main clusters of the network in 2012 United States Brazil South Africa China Australia Germany b. The U.S. cluster c. The Brazilian cluster ABW SUR BOL CAN ARG DMA TCD KNA GUY MEX JAM CHL KHM PER HTI DOM TTO BRB Brazil NIC COL HND United States LCA NER URY VCT VEN GTM SLV CRI PRY ECU GAB BHS NGA BLZ GRD PAN d. The German cluster e. The Chinese cluster DZA MDG CPV CMR NPL GHA MAR JOR SEN GNQ MDV AZE ALB GNB CAF SYC TUN CYP MLI IRQ MUS SLB SLE HRV AGO IND FRA ESP HKG BEN GBR PRT ITA GRC TKM GMB BEL LUX Germany SVN MNG KWT IRN QAT BIH IRL ISL HUN NL POL BGR MNE SDN NOR BGD ZMB MRT ARE PAK CIV ETH SVK MKD MDA KOR DNK BDI CZE SRB COG China OMN SWE PHL ARM TUR SGP JPN FIN GEO BFA YEM LVA AUT RUS MYS RWA BRN UGA EST TJK VNM LAO SAU KGZ BHR TZA UZB UKR BLR THA LTU COM VUT KEN KAZ Source: Calculations based on data from Direction of Trade Statistics (DOTS). Note: This figure shows the results of clustering analysis on the global trade network in 2012. Panel a shows the most central countries for each of the main clusters (in the other panels) of the global trade network. Panels b through e show the composition of countries that belong to each of these individual clusters. Each node represents a country. Each link corresponds to an active trade connection between a pair of countries. The thickness of the link indi- cates the strength of these connections. For clarity purpose, panels only display the top 10 percent of links in the U.S. and Chinese clusters, and the top 5 percent of links in the German cluster. All the links are displayed for the Brazilian cluster. THREE GLOBAL TRENDS THAT SHAPED LATIN AMERICAN AND CARIBBEAN DE VELOPMENT 61 FIGURE 1.13 Regional composition of cross-border investments a. Portfolio investments b. Syndicated loans North North LAC LAC South and South and Sender regions Sender regions Southeast Southeast Asia Asia Eastern Eastern Europe Europe Middle East Middle East and Central and Central Asia Asia Africa Africa 0 10 20 30 40 50 60 70 80 90 100 0 10 20 30 40 50 60 70 80 90 100 % of investments to receiver regions % of investments to receiver regions c. Mergers and acquisitions d. Greenfield investments North North LAC LAC South and South and Sender regions Sender regions Southeast Southeast Asia Asia Eastern Eastern Europe Europe Middle East Middle East and Central and Central Asia Asia Africa Africa 0 10 20 30 40 50 60 70 80 90 100 0 10 20 30 40 50 60 70 80 90 100 % of investments to receiver regions % of investments to receiver regions Receiver regions North LAC South and Southeast Asia Eastern Europe Middle East and Central Asia Africa Others Sources: Calculations based on data on trade are from Direction of Trade Statistics (DOTS), on portfolio investments are from Coordinated Portfolio Investment Survey (CPIS), on syn- dicated loans and mergers and acquisitions are from SDC Platinum, and on greenfield investments are from fDi Markets. Note: Each bar in each graph corresponds to the sender region and each group of countries within a given bar corresponds to the receiving region. Offshore centers are excluded. The North includes the G-7 members and Western Europe countries. G-7 = Group of Seven; LAC = Latin America and the Caribbean. 62 THE RISE OF THE SOUTH the exports of South American countries. For South are also strikingly different. The share South American countries, about 16 percent of financial inflows in the primary sector is of FVA originated in Asia, and 28 percent larger in the South than in the North, inde- originated in Western Europe. pendent of whether the flows are from South Despite the increase in the importance or North countries (see figure 1.15, panel of the South in GVCs over the past decade, b). For example, the primary sector’s share North countries still remain a significant of syndicated loans averaged 45 percent of source of imported inputs used in the exports South-South flows and just 19 percent of of LAC countries.15 Also notable is the lim- North-North flows between 2003 and 2012. ited participation of LAC countries as sources South-South M&A flows are also tilted of FVA for the exports of other countries, toward the primary sector when contrasted especially South countries.16 South American with North-North flows. Flows to the pri- countries are more present in this regard than mary sector, for example, accounted for other LAC countries, albeit mostly because of 54 percent of South-South flows but just 20 their commodities exports. percent of North-North flows. Similar pat- Overall, although some regional clustering terns are also observed for greenfield invest- is observed within LAC, the patterns of trade ments. Relative to North-North financial integration in the region are different from flows, North-South and South-South flows those observed in other South economies, include a larger share of investments in the East Asia in particular. Box 1.1 examines primary sector. some of these differences. The overall patterns of bilateral con- nections of the South and the North sug- gest that the South’s trade and financial Endowments have also played a role linkages are to some extent rooted in the in the structure of trade and financial forces of comparative advantage associated linkages. with relative endowments. The evidence There is significant heterogeneity in the presented above on the dynamics of trade sectoral composition of global trade and connections suggests triangular trading financial flows not only of the North when relationships between some South and contrasted with the South but also within the North economies. An example of this tri- South (as discussed in Set of Facts 1). There angularity is the trade connections between is also heterogeneity in the sectoral compo- China and South America, whereby China sition of bilateral connections: South-South imports commodities from LAC (espe- connections are different from North-South cially South America) and exports manu- connections, which in turn are different from facturing goods to LAC and the rest of the North-North connections. world, including the North. In contrast, One characteristic that reflects the dif- the North’s trade linkages, especially link- ferences in bilateral trade connections is ages with other North economies, embed the degree of intraindustry trade (IIT). The components of product differentiation and degree of IIT, measured by the Grubel-Lloyd economies of scale.17 index, ranges from 0 (pure interindustry trade) to 1 (pure IIT trade). The degree of The sectoral composition of LAC’s IIT varies across South and North countries trade and financial connections with as well as within the South. North-North other South countries is different from connections are typically characterized by the composition of its ties with North a higher degree of IIT than South-North countries. and South-South connections (figure 1.15, panel a). Trade and financial flows to LAC countries The sectoral compositions of bilateral from the South include a larger share of flows financial connections of the North and the in the primary sector than do flows from the THREE GLOBAL TRENDS THAT SHAPED LATIN AMERICAN AND CARIBBEAN DE VELOPMENT 63 FIGURE 1.14 Regional clustering in global value chains, 2011 a. Sources of FVA in exports across regions b. Sources of FVA in exports across LAC countries Argentina EAP Brazil Chile Colombia Japan South America Bolivia Ecuador SA Peru Paraguay Uruguay ECA Venezuela, RB Aruba Western Europe Antigua and Barbuda Bahamas, The The Caribbean MENA Barbados Cuba Dominican Republic SSA Haiti Jamaica North Trinidad and America Tobago Belize Central America Costa Rica Central America and Mexico and Mexico Guatemala Honduras The Caribbean Mexico Nicaragua South America Panama El Salvador 0 5 10 15 20 25 30 35 40 45 50 0 5 10 15 20 25 30 35 Share of exports (%) Share of exports (%) Intra- EAP Japan South ECA Western MENA SSA North Mexico and The South regional Asia Europe America Central America Caribbean America Sources: Calculations based on data from Eora MRIO and World Development Indicators. Note: Figure shows the regional composition of sources of foreign value added used in a country’s exports, scaled by the country’s exports. In panel b, the intraregional category cap- tures the share of foreign value added sourced from the LAC subregion to which each country belongs. North America excludes Mexico. EAP = East Asia and Pacific; ECA = Europe and Central Asia; LAC = Latin America and the Caribbean; MENA = Middle East and North Africa; SA = South Asia; SSA = Sub-Saharan Africa; FVA = foreign value added. 64 THE RISE OF THE SOUTH FIGURE 1.15 Sectoral composition of bilateral cross-border flows a. The degree of intraindustry trade 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0 1962–1969 1970–1979 1980–1989 1990–1999 2000–2011 Years North-North North-South South-South b. Bilateral financial flows 100 90 35 39 30 80 42 52 70 57 59 60 7 75 15 60 79 88 75 Percent 50 20 13 11 40 30 25 21 58 20 54 6 45 5 20 41 38 8 10 19 20 19 20 3 20 13 9 0 ns A eld ns A eld ns A eld ns A eld M& M& M& M& loa loa loa loa nfi nfi nfi nfi ee ee ee ee ted ted ted ted Gr Gr Gr Gr ica ica ica ica nd nd nd nd Sy Sy Sy Sy to North to South to North to South From North… From South… Primary Light manufacturing Heavy manufacturing Sources: Data on intraindustry trade are from Comtrade, on syndicated loans and mergers and acquisitions are from SDC Platinum, and on greenfield investments are from fDi Mar- kets. Note: Panel a shows trade-weighted averages of country pairs. Panel b presents the sectoral composition of bilateral financial flows (namely syndicated loans, M&A, and green- field) over 2003–2012. The primary sector corresponds to SIC codes 0–1500, light manufacturing to SIC codes 2000–2800 and 3100–3200, and heavy manufacturing to SIC codes 2800–3100 and 3200–3800. The North includes the G-7 members and Western Europe countries. The South includes all other economies. G-7 = Group of Seven; M&A = mergers and acquisitions. THREE GLOBAL TRENDS THAT SHAPED LATIN AMERICAN AND CARIBBEAN DE VELOPMENT 65 North (figure 1.16, panels a and b). Partic- and 2012, twice the average 5 percent of ularly striking is the large share of M&A imports from the North. investments from the South to LAC in the Trade and financial flows from LAC primary sector (92 percent during the 2000s). countries to the South are also tilted toward In contrast, only 48 percent of M&A invest- the primary sector when compared with ments from the North were in the primary flows to the North (see figure 1.16, panels sector. Large but less marked differences are a and c). The share of natural resources also observed in greenfield investments and in exports from LAC to the South during syndicated loans. Regarding trade flows, the 2003 –12 was about 60 percent larger share of natural resources in LAC imports than the share of exports to the North from the South was 10 percent between 2003 (46 percent versus 29 percent on average). BOX 1.1 Differences in international trade integration: The case of Latin America and the Caribbean and East Asia The analysis of the evolution of the connections persisted over time. The trade connections within within East Asia and LAC trade networks shows the East Asian clustering form a much denser net- considerable differences between the two regions. work than those within the LAC cluster. In 2012 The density maps of the trade connections within almost all countries were fully connected with all each region are particularly telling (figure B1.1.1).a other countries within the East Asian network (as One contrast in the nature of the trade connec- indicated by a network density measure of 0.99 for tions of LAC and East Asian economies is the evo- this cluster). Countries in the LAC network were not lution of the relative importance of different coun- as fully integrated with one another (the network tries within the networks. In 1980 trade networks density measure for this cluster was 0.89). in both regions centered on countries of the North, Figure B1.1.1 also suggests that trade connections especially the United States for LAC and Japan for within the East Asian network are multidirectional Asia. By 2012 many countries from both the North and intense in every direction, whereas those within and the South were central players in the East Asian the LAC network tend to be mainly bidirectional, network, appearing as very dense nodes in the map. especially with the United States. For instance, tri- These countries included not only China and Japan ads of trade connections are typically observed at but also Republic of Korea, Malaysia, Singapore, and a higher frequency in the East Asian network than Thailand. In contrast, in the LAC network no node in the LAC network. In 2012 the number of triads was as dense as the United States. Brazil was the clos- as a share of the maximum number of triads within est node in density, but it was far less dense than the a network was 0.99 in the East Asian network and United States. China entered as a new player in the 0.92 in the LAC network. This type of connectivity LAC network in 2012, though its density was low. observed in the East Asian network suggests strong Differences in the dispersion of the centrality measure feedback effects, whereby the tight trade connec- associated with each node within the two networks tions within the region boost trading with the rest support these patterns. For example, in 2012 the dis- of the world and vice versa. In contrast, LAC coun- persion of the node centrality was significantly lower tries do not seem to leverage intraregional trade to in the East Asian network (0.09) than in the LAC net- enhance their overall level of connectivity within work (0.31), indicating that there is less variation in the global trade network. These patterns may be the density of nodes in the former than the latter. linked, at least in part, to the more active partici- Another contrast is the degree of connectivity of pation of East Asian countries in GVCs relative to countries within the networks, a feature that has LAC countries. a. The density of a node depends on the number of neighboring countries and the economic distance between countries. The sample of countries included affects the maps, making it hard to directly compare node density across panels. Still, some features are comparable across maps. The set of countries in each of the two trade clusters analyzed includes all South countries within each region. This set of countries was expanded to include the five largest trading partners (measured by the total volume of trade flows) for countries in the region located outside the region. (continued) 66 THE RISE OF THE SOUTH BOX 1.1 Differences in international trade integration: The case of Latin America and the Caribbean and East Asia (continued) FIGURE B1.1.1 Density maps of trade networks a. The Latin American network, 1980 b. The Asian network, 1980 (continued) (continued) THREE GLOBAL TRENDS THAT SHAPED LATIN AMERICAN AND CARIBBEAN DE VELOPMENT 67 BOX 1.1 Differences in international trade integration: The case of Latin America and the Caribbean and East Asia (continued) FIGURE B1.1.1 Density maps of trade networks (continued) c. The Latin American network, 2012 d. The Asian network, 2012 Source: De la Torre, Didier, and Pinat 2014. Note: Density is based on the number of neighboring countries and the economic distance between countries. See box 1.1 for details on the mapping meth- odology. The node density is translated into colors using a red-green-blue scheme—from the highest density (red) to the lowest (blue). 68 THE RISE OF THE SOUTH FIGURE 1.16 Sectoral composition of cross-border flows Financial flows in the primary sector are for Latin America and the Caribbean also generally larger when destined to South a. Trade flows countries. For example, greenfield invest- ments in the primary sector represented LAC’s exports North 60 percent of all LAC greenfield invest- ments in the South but only 27 percent of South investments in the North. The differences in the composition of syndicated loans from LAC’s imports North LAC countries to the rest of the world were also large: between 2003 and 2012, 13 per- South cent of flows to the North were in the pri- mary sector versus 92 percent of flows to 0 10 20 30 40 50 60 70 80 90 100 the South. Outflows associated with M&A Percent b. Financial flows to LAC … were the only exception to this pattern: the share of flows from LAC to the North in Syndicated the primary sector was larger than the share … from the North loans of flows to the South. M&A Overall, LAC’s cross-border flows to and from the South are tilted toward the primary Greenfield sector when contrasted with flows to and from the North. There are, however, signif- Syndicated icant differences within countries in LAC, … from the South loans which are explored in chapters 2 and 4. These M&A patterns of integration in the world economy suggest that as connections with South coun- Greenfield tries deepened during the 2000s, both trade and financial flows became, to some extent, 0 10 20 30 40 50 60 70 80 90 100 more rooted in comparative advantage forces Percent and endowments. c. Financial flows from LAC … Syndicated loans Notes … to the North 1. In this report, the North includes the Group M&A of Seven (G-7) members (Canada, France, Germany, Italy, Japan, the United Kingdom, Greenfield and the United States) plus the following Western Europe countries: Andorra, Austria, Syndicated Belgium, Denmark, Finland, Greece, Ice- loans land, Ireland, Liechtenstein, Luxembourg, … to the South M&A Monaco, Netherlands, Norway, Portugal, San Marino, Spain, Sweden, and Switzer- land. The South includes all other economies, Greenfield including all countries in Latin America and 0 10 20 30 40 50 60 70 80 90 100 the Caribbean (LAC). Percent 2. The Latin American school of dependency theory was born in 1949 with the publication Primary Light manufacturing Heavy manufacturing of two articles, one by the German develop- Sources: Data on trade flows are from Comtrade, on syndicated loans and mergers and acqui- ment economist Hans Singer and the other by sitions are from SDC Platinum, and on greenfield investments are from fDi Markets. the Argentine economist and former director Note: Averages for the 2003–2012 period are reported. The primary sector corresponds to SIC codes 0–1500, light manufacturing to SIC codes 2000–2800 and 3100–3200, and heavy of the United Nations Economic Commis- manufacturing to SIC codes 2800–3100 and 3200–3800. The North includes the G-7 mem- sion for Latin America and the Caribbean bers and Western Europe countries. The South includes all other economies. G-7 = Group of Raúl Prebisch. The Prebisch-Singer hypothesis Seven; LAC = Latin America and the Caribbean; M&A = mergers and acquisitions. THREE GLOBAL TRENDS THAT SHAPED LATIN AMERICAN AND CARIBBEAN DE VELOPMENT 69 postulated that the developing countries of 7. Bernanke (2005) argues that a confluence of LAC would experience immiserizing growth factors led to the emergence of a global saving because of a secular deterioration in their terms glut, including policy interventions to boost of trade as manufactured goods (exported by exports in Asia, higher oil prices in the Middle the North) gained in value relative to agricul- East, and a dearth of investment opportunities tural and mining commodities (exported by and an aging population in advanced indus- many LAC economies, including Prebisch’s trial countries. Mendoza, Quadrini, and Rıos- homeland, Argentina). Events did not confirm Rull (2009) attribute high saving in emerging these predictions during the early 21st century, market countries to relatively low levels of a period characterized by improving terms of financial development, which generate greater trade for many countries in LAC and the rela- precautionary saving. Caballero, Farhi, and tive decline of the North. See Love (1980) on Gourinchas (2008) emphasize instead the lack the role of Prebisch in shaping Latin American of investment opportunities in these coun- structuralism and Love (2005) for a review of tries and the associated shortage of financial economic structuralism in Latin America. assets as the main source of the global saving 3. The World Bank’s 2012 Global Development glut. The International Monetary Fund (IMF) Horizon (GDH) forecasts that the share of the (2005) also stresses low investment rates fol- South in global GDP will reach 55 percent in lowing the Asian crisis rather than an increase 2025. A 2013 report by the National Intel- in saving. ligence Council projects that this share will 8. For ease of exposition, the figures show only reach 70 percent in 2030. The Asian Devel- the connections above a certain valued thresh- opment Bank forecasts that the share of South old. Hence, the graphical representation of exports will rise to 64 percent of global exports small countries in particular is not entirely by 2030, while the share South imports will accurate, as connections smaller than the remain at about 46 percent. Regarding finan- adopted thresholds are not shown. The results cial flows, the 2013 GDH projects that the are qualitatively similar if these connections share of the South in world capital will rise to are reported. 63 percent of inflows and 80 percent of out- 9. The analyses of global networks based on flows in 2025. bilateral connections use data on trade, port- 4. There is great heterogeneity in trade struc- folio investments, FDI, and (syndicated) bank tures across LAC economies in particular loans. The trade data are from the IMF’s Direc- and within the South more broadly. This tion of Trade Statistics (DOTS), covering the cross-country heterogeneity plays a particu- period 1980–2012. The portfolio investments larly important role in the analyses presented data are from the IMF’s Coordinated Portfo- in chapter 3 of the labor market implications lio Investment Surveys (CPIS), which provide of the rise of the South in global markets. data on the stock of portfolio assets between 5. These top 20 South countries include Brazil, 2001 and 2011. For FDI, firm-level transac- Chile, the Czech Republic, Hungary, India, tion data on M&A from Thomson Reuters’ Korea, Poland, the Slovak Republic, Thai- SDC Platinum cover the period 1991–2012, land, Turkey, and Vietnam. and (announced) firm-level greenfield invest- 6. The causal link going from the current ments from the Financial Times’ FDi Markets account to the equilibrium real exchange rate cover the period 2003–12. For syndicated has been amply studied in the open economy loans, the Thomson Reuters’ SDC Platinum macroeconomic literature with two-sector transaction-level data on syndicated loans (tradables and nontradables) models. Excess is used for the period 1996–2012, covering demand for saving over investment raises the more than 150 source and recipient countries. demand for both tradables and nontradables. All firm-level transaction data are aggregated For a small price-taking economy, the excess at the bilateral country level. Chapter 4 pro- demand for tradables is resolved solely vides additional details on these data. through quantities (a widening of the current 10. The total number of possible South-South account deficit) at given world prices, but the connections is defined as the number of active excess demand for nontradables raises their connections that would exist if each country price relative to the price of tradables, appre- of the South country were connected to every ciating the real exchange rate. See, for exam- other country of the South country in the ple, Dornbusch (1980). world. 70 THE RISE OF THE SOUTH 11. The rise of a diverse set of South economies and Zhu (2010); and Johnson and Noguera with relatively large pools of relatively low- (2012a), among others. wage workers, abundant raw materials, siz- 17. See Hanson (2012) for a discussion of the able domestic markets, and/or highly capable empirical determinants of export specializa- export-oriented manufacturers is tightly con- tion and why South-South trade looks so dif- nected to changes in the dynamics of produc- ferent from North-North trade. tion and demand in the global economy. The development of GVCs is thus arguably linked to the rise of the South. Nonetheless, the spe- References cific roles South countries play in individual BOPS (Balance of Payments Statistics). Interna- GVCs vary according to their degree of open- tional Monetary Fund, Washington, DC. http:// ness to trade and foreign investment, infra- elibrary-data.imf.org/DataExplorer.aspx. structure and logistics capabilities, and other Baldwin, R., and J. Lopez-Gonzalez. 2013. strategic considerations. Chapter 2 analyses “Supply-Chain Trade: A Portrait of Global the development of GVCs. See also Gereffi Patterns and Several Testable Hypotheses.” and Luo (2014) and OECD (2013). NBER Working Paper 18957, National Bureau 12. GVCs have expanded to a wide range of of Economic Research, Cambridge, MA. industries—from manufacturing to ser- Baldwin, R., and A. Venables. 2013. “Spiders and vices—covering not only final goods but also Snakes: Offshoring and Agglomeration in the components, subassemblies, research and Global Economy.” Journal of International development (R&D), and innovation. See, for Economics 90 (2): 245–54. example, Gereffi (2014), Baldwin and Ven- Bernanke, B. 2005. “The Global Saving Glut ables (2013), and UNCTAD (2011). and the US Current Account Deficit.” Speech, 13. Sturgeon and Van Biesebroeck (2010) argue Board of Governors of the Federal Reserve Sys- that on the production side of the automo- tem, March 10. bile industry, the dominant trend has been of Burt, R. 1976. “Positions in Networks.” Social regional integration, a pattern that has been Forces 55: 93–122. intensifying since the mid-1980s. In North Caballero, R., E. Farhi, and P. O. Gourinchas. America, South America, Europe, Southern 2008. “An Equilibrium Model of ‘Global Africa, and Asia, the production of parts tends Imbalances’ and Low Interest Rates.” Ameri- to take place regionally to feed final assembly can Economic Review 98 (1): 358–93. plants that produce largely for regional mar- Comtrade (U.N. Trade Statistics) (database). United kets. See also Johnson and Noguera (2012b). Nations, New York. http://comtrade.un.org. 14. The analysis of the foreign content of exports De la Torre, A., T. Didier, and M. Pinat. 2014. captures mostly “backward supply chains.” “Can Latin America Tap the Globalization It does not capture “forward supply chains” Upside?” World Bank Policy Research Work- in which countries export parts that are inte- ing Paper 6837, Washington, DC. grated into other countries exports of final Dornbusch, R. 1980. Open Economy Macroeco- goods. The analysis of these forward linkages nomics. New York: Basic Books. reveals that these patterns of regional clus- DOTS (Direction of Trade Statistics) (database). tering are also present if one considers the International Monetary Fund, Washing- destination of value added supplied to other ton, DC. http://elibrary-data.imf.org/Data countries’ exports (these unreported results Explorer.aspx. are available upon request). See chapter 2 for Economist Intelligence Unit (data tool). London, a more detailed analysis of GVCs. England. http://www.eiu.com/. 15. Chapter 2 provides a more detailed analysis Eora MRIO (database). http://worldmrio.com/. of the nature of LAC’s integration into GVCs See Lenzen and others (2012, 2013). with North and South countries. fDi Markets (database). Financial Times, London. 16. For a detailed analysis of the role of individ- http://www.fdimarkets.com/. ual countries in GVCs around the world, see Gereffi, G. 2014. “Global Value Chains in a Post– Baldwin and Lopez-Gonzalez (2013). For Washington Consensus World.” Review of empirical analyses of the value added content International Political Economy 21 (1): 9–37. of trade, see Hummels, Ishii, and Yi (2001); Gereffi, G., and X. Luo. 2014. “Risks and Oppor- Koopman, Wang, and Wei (2008); Trefler tunities of Participation in Global Value THREE GLOBAL TRENDS THAT SHAPED LATIN AMERICAN AND CARIBBEAN DE VELOPMENT 71 Chains.” World Bank Policy Research Work- Dimensions.” Latin American Research ing Paper 6847, Washington, DC. Review 40 (3): 100–25. Hanson, G. 2012. “The Rise of Middle Kingdoms: Martinez-Piva, J., and G. Zúñiga-Arias. 2012. Emerging Economies in Global Trade.” Journal Economic Integration and Value Chain: Case of Economic Perspectives 26 (2): 41–64. Study from Central America Dairy. ECLAC Hummels, D., J. Ishii, and K. Yi. 2001. “The report, Economic Commission for Latin Nature and Growth of Vertical Specialization America and the Caribbean, Santiago, Chile. in World Trade.” Journal of International Mendoza, E., V. Quadrini, and J. Rıos-Rull. Economics 54 (1): 75–96. 2009. “Financial Integration, Financial Devel- IMF (International Monetary Fund). 2005. World opment, and Global Imbalances.” Journal of Economic Outlook. Washington, DC. Political Economy 117 (3): 371–416. Johnson, R., and G. Noguera. 2012a. “Account- Morris, M., C. Staritz, and J. Barnes. 2011. ing for Intermediates: Production Sharing and “Value Chain Dynamics, Local Embedded- Trade in Value Added.” Journal of Interna- ness, and Upgrading in the Clothing Sectors of tional Economics 86 (2): 224–36. Lesotho and Swaziland.” International Jour- ———. 2012b. “Fragmentation and Trade in nal of Technological Learning, Innovation Value Added over Four Decades.” American and Development 4 (1): 96–119. Economic Review 102 (3): 407–11. OECD (Organisation for Economic Co-operation Koopman, R., Z. Wang, and S. J. Wei. 2008. “How and Development). 2013. Interconnected Much of Chinese Exports Is Really Made in Economies: Benefiting from Global Value China? Assessing Domestic Value-Added When Chains. Paris. Processing Trade Is Pervasive.” NBER Working SDC Platinum (database). Thomson Reuters, Paper 14109, National Bureau of Economic New York. http://thomsonreuters.com/sdc Research, Cambridge, MA. -platinum/. Lenzen, M., K. Kanemoto, D. Moran, and A. Sturgeon, T., and J. Van Biesebroeck. 2010. Geschke. 2012. “Mapping the Structure of the “Effects of the Crisis on the Automotive World Economy.” Environmental Science and Industry in Developing Countries: a Global Technology 46 (15): 8374–381. Value Chain Perspective.” World Bank Policy Lenzen, M., D. Moran, K. Kanemoto, A. Research Working Paper 5330, Washington, Geschke. 2013. “Building Eora: A Global DC. Multi-regional Input-Output Database at High Trefler, D., and S. C. Zhu. 2010. “The Structure Country and Sector Resolution.” Economic of Factor Content Predictions.” Journal of Systems Research 25 (1): 20–49. International Economics 82 (2): 195–207. Lane, P. and G.M. Milesi-Ferretti. 2007. “The UNCTAD (United Nations Conference on Trade External Wealth of Nations Mark II: Revised and Development). 2011. World Investment and Extended Estimates of Foreign Assets and Report: Non-Equity Modes of International Liabilities, 1970–2004.” Journal of Interna- Production and Development. Geneva. tional Economics 73: 223–250. World Development Indicators. World Bank, Love, J. L. 1980. “Raúl Prebisch and the Origins Washington, DC. http://data.worldbank.org of the Doctrine of Unequal Exchange.” Latin /data-catalog/world-development-indicators. American Research Review 15 (3): 45–72. ———. 2005. “The Rise and Decline of Eco- nomic Structuralism in Latin America: New The Structure of Trade Linkages and Economic Growth 2 This chapter examines the extent to which the trade connections of countries in Latin America and the Caribbean (LAC)—particularly connections with other South countries—can lead to a virtuous cycle of thriving trade and economic growth. It briefly overviews the literature that shows that the nature of trade connections may play an important role in the trade-growth nexus, especially through technological diffusion and learning spillovers. This finding is par- ticularly important for countries in the region as the connections LAC has forged with other South countries over the past decade are different from its connections with North countries. The chapter presents new evidence on the importance of the nature of LAC’s trade connec- tions for economic growth. Although the implications vary widely across countries, the analy- sis indicates that the structure and quality of trade baskets merits attention. The extent to and manner in which countries participate in global value chains (GVCs) also affect the dynamics of trade and growth. In this regard, LAC countries seem to lag other South economies, espe- cially East Asia. The composition of trading partners, especially connections with countries that are central to the global trade network, also appears to matter for growth. The evidence presented suggests that growth effects associated with trade openness are related to strong ties with countries that are more exposed to the frontiers of ideas and technologies. Other factors that may be reducing the growth potential of LAC’s trade connections include the quality of the region’s transport networks and its trade agreements. L atin America and the Caribbean (LAC) with North countries in many ways, from is an increasingly globalized region. Its the sectoral composition of its export baskets economic future depends a great deal (see Set of Facts 3 in chapter 1) to the degree on the quality of its external connections. of intraindustry trade (IIT) and the intensity LAC countries have broadened and deep- of factors embedded in exports. LAC coun- ened their trade connections with other South tries also differ from other South countries in countries (see Set of Facts 2 in chapter 1). their trade integration patterns. The region The nature of LAC’s connections with other has its own peculiarities, which are high- South countries differ from its connections lighted throughout this chapter. 73 74 LATIN AMERICA AND THE RISING SOUTH More broadly, the chapter explores the in these GVCs facilitates economic upgrad- implications for LAC’s economic develop- ing—for instance, the movement of firms or ment of the rise of the South and the chang- more generally countries toward activities ing nature of its cross-border linkages. The that require more skilled work or processes key question it addresses is whether such that make use of more advanced technologies. changes are evolving toward higher-quality The novel evidence presented in this chapter connections. Do the linkages arising go suggests that participation in GVCs does not beyond the generation of static efficiency automatically translate into additional gains gains (such as gains associated with compar- from trade beyond the gains associated with ative advantages)? Are they more likely to increased export volumes: location within deliver dynamic gains, particularly through GVCs matters in this regard. Being part of learning spillovers and technology diffusion, GVCs, especially in its middle segments, is thereby further boosting the long-run growth associated with higher per capita income of per capita income? growth rates. The composition of trading This chapter reports some new empirical partners in these production chains also regularities that suggest that the structure of seems to play an important role. trade connections affects the trade-growth LAC seems to have just boarded this new nexus. This evidence indicates that on aver- train on the global economic landscape, at age, after controlling for the overall volume least compared with some economies in other of trade flows, the degree of IIT and the fac- regions, including East Asia. Countries in tor intensity embedded in traded goods affect South America and the Caribbean typically the dynamics of trade and growth linkages. have a greater forward participation in GVCs; There is wide variation in how these fac- that is, they tend to join GVCs at their initial tors operate in individual countries in LAC. stages, by supplying inputs to other countries. Many of their effects on economic growth In contrast, Mexico and Central American are nonlinear and depend on countries’ level economies have more backward participa- of trade openness and labor force educa- tion rates, as they tend to be at the end of tion. Nonetheless, a conclusion that emerges GVCs: the share of foreign inputs used in consistently from the empirical analysis is their exports is much higher than the share of that the structure and quality of trade con- their exports used in other countries’ exports. nections merit attention. The key issue for This pattern partly reflects the exports of LAC is thus to understand how countries their goods to the United States for final can develop trade linkages in ways that more consumption rather than further transfor- effectively foster learning spillovers and tech- mation. Countries in the region do not seem nology diffusion in particular and economic to be as well integrated into GVCs as other development more broadly. South economies, although there has been an The extent of and manner in which increase in their participation since the 1990s. countries participate in global value chains This chapter also provides some evidence (GVCs) also affect trade and growth link- that the role of trading partners in the global ages. The development of GVCs—that is, trade network matters. For sufficiently inte- the dispersion of production stages and grated countries, an increase in trade link- processes across countries—has been an ages with countries at the center of the global important pillar of the global economy since trade network is typically accompanied by the late 1980s. Production in these chains strong growth effects. In addition, there comprises economic activities of all levels, seems to be some form of complementarity from small-scale, household-based work between trade openness and the share of to high-skilled, technology-intensive, and trade with these countries. The results also knowledge-intensive work. An important indicate that countries need educated enough issue is thus the extent to which participation labor forces to be able to benefit most from THE STRUCTURE OF TRADE LINKAGES AND ECONOMIC GROWTH 75 trading with central countries, suggesting led to an increase in the level of IIT, and the that human capital development is import- share of skilled labor–intensive goods in the ant for the absorption of foreign technology trade basket rose. Yet, these countries have and knowledge. These findings are consistent the lowest participation rates in GVCs among with the idea that the growth effects asso- LAC countries, and most of their GVC link- ciated with trade openness are not related ages are forward linkages, suggesting inser- simply to the development of strong trade tion in the initial stages of GVCs (supplying ties with a single country, but instead to the other countries with inputs). In contrast, establishment of strong ties with countries Mexico is one of the most integrated coun- that are more exposed to the frontiers of tries in the region, but it has developed pro- ideas and technologies. duction chains mainly with North countries The chapter identifies a set of frictions (the United States in particular); its placement that are particularly important for LAC is at the end of these chains, as indicated by countries and that could act as trade barri- its strong backward participation rate. Mex- ers, thereby affecting the dynamics between ico has increased its trade connections with trade integration and economic growth. The South countries, but these connections are quality of transport networks is arguably characterized by a lower degree of IIT and a one such barrier. On average, LAC seems to smaller share of skill-intensive goods than its underperform not only North countries but connections with North countries. also some other South countries on a range Several factors affect the opportunities of indicators capturing accessibility to and and challenges associated with trade rela- the quality of transport networks. There is tions. One is the ability of policy makers to also some evidence suggesting that the region craft effective policies. This ability depends is not spending sufficiently or effectively on in part on their sector-specific knowledge, infrastructure. which affects how well they understand the The quality of trade policies can also ham- potential constraints firms face. Another is per economic ties. Evaluation of a wide range the presence of market failures—from the of trade agreements suggests that progress provision of infrastructure to the accumu- has been made in reducing overall trade and lation of human capital, the creation and regulatory barriers between LAC and other management of ideas, and the resolution South countries. Coverage of goods and ser- of coordination failures within existing vices is generally good, although more needs industries—suggesting an important role to be done to reduce residual barriers and for policies. The evidence in this chapter is intensify integration with South countries. also suggestive of positive externalities from Much less progress has been made on more the way countries integrate into the global difficult regulatory issues, such as investment, economy. All of these factors suggest that the intellectual property rights, and competition. active engagement of policymakers may be Overall, the empirical evidence is mixed: called for. it is not clear that LAC trade connections are evolving toward higher quality, which could boost the long-run growth of per capita Trade and economic growth income. Increased trade linkages with South Trade integration has advantages and disad- countries have been a mixed bag for countries vantages. Whether the net effects are positive in LAC. Although the structure and qual- or negative depends on many factors, such as ity of trade connections improved in many initial conditions, structural features (includ- aspects, there is great variation within the ing geography and natural endowments), and region. For example, in some South Amer- policy frameworks, as well as the interactions ican countries, such as Colombia and Peru, and interdependencies among these factors the new connections with South countries and the evolving nature of globalization. 76 LATIN AMERICA AND THE RISING SOUTH The key is to distinguish the more beneficial the signal to noise ratio in relative prices and forms of integration and identify the condi- increasing the incentives for firms to continu- tions under which the upsides of globaliza- ously optimize production, management, and tion can be seized and maximized and the marketing practices, in order to avoid falling downsides minimized or avoided. behind their competitors, remain profitable, Laying out the basic theoretical arguments and keep or expand market share. Compe- on the channels through which trade integra- tition can also stimulate quality upgrading.4 tion may affect economic growth can help In addition, contestability stemming from in organizing and interpreting the evidence foreign competitors may spur innovation presented in this chapter.1 Trade integration by local firms, especially if these firms are affects economic growth through four chan- not too far from the technological frontier.5 nels: larger markets, competition, technol- Changes in competitive forces associated ogy diffusion and learning spillovers, and with increased trade integration can also volatility. Accessibility to larger markets can alter product markups and markup distor- enable classical efficiency gains from trade tions, with consequent growth effects. For through specialization (which can involve example, increased competition can lead to some reallocation of production as well as losses in market power for individual firms, labor) in line with comparative advantages thereby leading to reductions in markups and and relative endowments. It can also facili- markup dispersion. This effect in turn could tate the development of scale economies and lead to a decline in productivity losses as a Marshallian-type positive externalities that result of misallocation.6 A key question is, may be unavailable within small local mar- then, whether the benefits of tougher com- kets. In addition, international trade can facil- petition from abroad can be achieved while itate access for both producers and consumers mitigating its negative effects (see Aghion and to goods that are cheaper, of higher quality, Griffith 2008). and of greater variety. 2 For producers trade Integration into global markets can greatly provides access to better-functioning input accelerate technological progress, a key markets. These effects imply that as countries driver of long-run growth. Technology diffu- integrate into global markets, they can trade sion and learning spillovers in particular can larger quantities of each good (the intensive foster technological progress. These effects margin), a wider set of goods (the extensive can take place through imports, especially margin), and/or higher-quality goods.3 of intermediate goods, which can embed Access to larger markets may also be asso- technologies that may be unavailable in the ciated with increased competition. Industrial importing countries. Use of these goods can organization models predict a negative effect reduce the costs of product development and of increasing competition on innovation lead to the production of new products.7 and growth to the extent that it reduces the Knowledge spillovers are also possible. For monopoly rents that reward successful inno- instance, when imports of advanced equip- vators. Because reaping adequate returns for ment are put to use in production processes, one’s innovation requires some form of tem- the importer can benefit from various forms porary monopoly power, weak competition of support, training, and advice provided by can entice producers to innovate (Schum- the supplier. Learning can also occur through peter 1942). Tougher competition can also exporting, not least through the upgrading be associated with positive growth effects. needed to meet international product qual- For instance, greater trade integration can ity standards and certification requirements. deepen and widen the monopoly-busting, Exporters may gain access to new technol- efficiency-enhancing effects of competition. ogy, and they can learn a great deal from Increased competition from abroad can also feedback from global buyers, including on enhance resource allocation by improving how to innovate and improve production THE STRUCTURE OF TRADE LINKAGES AND ECONOMIC GROWTH 77 processes and managerial practices to bet- geography-based instrument used is likely ter satisfy demand niches, consistently attain to be correlated with other geographic vari- high quality, and more ably adapt to chang- ables that affect income through nontrade ing circumstances.8 Of course, in practice the channels. Acemoglu, Johnson, and Robinson extent of technology diffusion and learning (2001) note that these geographic instru- spillovers varies greatly, depending on the ments are closely correlated with countries’ nature of a country’s trade linkages, as dis- experiences during colonial times, which in cussed below. turn help explain international differences Greater integration into global markets in governance and institutions. Although the implies more exposure to external shocks debate persists, many papers, including the (especially price shocks, such as terms of most recent ones, find a causal effect from trade shocks), which in turn can increase trade to income levels and growth.10 macroeconomic uncertainty and volatility. The rest of this chapter sheds light on Conventional wisdom indicates that there this debate by going deeper into the global is a negative relationship between volatil- trends highlighted in chapter 1. In particu- ity and growth (for theoretical analyses, see lar, it explores the implications for the eco- Caballero and Hammour 1994 and Aghion nomic development of LAC that the changing and Saint-Paul 1998. For some empirical nature of its cross-border linkages entails. evidence, see Ramey and Ramey 1995 and The key question is whether such changes Servén 2003). Even if macroeconomic volatil- are evolving in a direction of higher-quality ity does not directly affect economic growth, connections—that is, do the linkages arising it entails direct welfare costs, which can be go beyond the generation of static efficiency particularly large in South economies (see, gains (such as those associated with compar- for example, Loayza and others 2007 and ref- ative advantages)? Are they more likely to erences therein). At the other extreme, partic- deliver dynamic gains, particularly through ipation in global markets through trade can learning spillovers and technology diffusion, help dampen volatility generated by domestic thereby providing a further boost to the long- shocks. For instance, greater trade integra- run growth of per capita income? A bench- tion can attenuate shocks in domestic mar- mark regression specification allow us to kets for agricultural products, such as shocks revisit the literature on trade and growth. Per associated with local climate conditions. Didier and Pinat (2015), a country’s growth The impact of increased trade integration on rate is assumed to be a linear function of the output volatility depends on various factors, degree of its trade openness (captured by the related but not restricted to the composition value of its trade flows) and the level of its of these flows, patterns of specialization, the human capital. This benchmark specifica- degree of concentration, the degree of finan- tion controls for conditional convergence cial development, and the sources of shocks.9 effects—that is, more developed countries Whether and how trade leads to growth typically grow less than less developed ones. is thus an empirical issue. A large body of Also included in the estimations is a set of literature discusses the empirical evidence control variables that capture not only their on the role of trade in fostering economic potential effect on growth rates but also development and growth. In a seminal arti- whether they can affect the relation between cle, based on 1985 data, Frankel and Romer trade openness and growth. In particular, (1999) show that cross-border trade, instru- the control variables considered are proxies mented with countries’ geographic character- for the development of public infrastructure istics, has a quantitatively large and robust and for relative price stability and exchange positive effect on income. Many researchers rate fluctuations. Box 2.1 provides additional have questioned these findings. For example, details on the methodology employed and the Rodríguez and Rodrik (2001) argue that the sample analyzed here. 78   latin americ a and the rising South Box 2.1  Methodology of trade and growth regression estimations Drawing on Didier and Pinat (2015), the main ques- where TOc,tTCc,t represents the interactions tion this chapter addresses is whether and how the between trade openness and the nature of trade con- structural features of countries’ trade connections nections (in terms of both partners and products) at affect economic growth. The chapter explores the country c time t. role of the nature of trade in different goods and Potential nonlinearities between the proxy for with different partners. Equation B2.1.1 is the human capital and the features of trade connections benchmark regression specification: is also considered. The idea is that the effects of human capital on growth could vary with the nature yc,t 2 yc,t 2 1 5 b0yc,t 2 1 1 b1CVc,t 1 b2TOc,t of trade relations. For example, to the extent that certain features of trade connections are associated 1 b3HKc,t 1 mt 1 hc 1 Pc,t (B2.1.1) with greater technology diffusion and learning spill- over, their effects on growth depend on the develop- where yc,t is gross domestic product per capita for ment of human capital. Equation B2.1.3b provides country c at time t ; TOc,t is trade openness; HKc,t this alternative specification: is human capital; CVc,t are control variables; mt are (unobserved) time-specific effects; hc are (unob- yc,t 2 yc,t 2 1 5 b0yc,t 2 1 1 b1CVc,t 1 b2TOc,t served) country-specific effects; and Pc,t is the error term. 1 b3HKc,t 1 b4TCc,t 1 b5HKc,tTCc,t All of the regressions presented in this chapter expand on this benchmark specification by includ- 1 b6 1 HKc,tTCc,t 2 2 1 mt 1 hc 1 Pc,t ing a set of proxies for the features of trade charac- teristics in order to infer whether they are associated (B2.1.3b) with additional growth effects. Equation B2.1.2 is the expanded regression specification: where HKc,tTCc,t represents the interaction between the level of human capital and the nature yc,t 2 yc,t 2 1 5 b0yc,t 2 1 1 b1CVc,t 1 b2TOc,t of trade connections (in terms of both partners and products) at country c time t. 1 b3HKc,t 1 b4TCc,t 1 mt 1 hc 1 Pc,t Equations B2.1.3a and B2.1.3b are used to esti- mate the total growth effects of changes in the struc- (B2.1.2) tural features of trade relations that are shown in the figures throughout this chapter. This methodology where TCc,t represent proxies for the features of does not identify the mechanisms through which trade connections (in terms of both partners and trade structure affects growth, but it provides sug- products). gestive empirical evidence on the extent to which To capture potential nonlinearities in the effects externalities such as those associated with technol- of trade openness on growth, this specification is ogy diffusion and learning spillovers matter. further expanded by adding terms capturing the These trade-growth regressions pose several chal- interaction between trade openness and, in turn, the lenges for estimation. A number of empirical papers different proxies for the nature of trade relations. in the growth literature adopt the system generalized Equation B2.1.3a displays this extended regression method of moments (S-GMM) procedure developed specification: in Arellano and Bover (1995) and Blundell and Bond (1998) to overcome the endogeneity issue. Dollar yc,t 2 yc,t 2 1 5 b0yc,t 2 1 1 b1CVc,t 1 b2TOc,t and Kraay (2004); Loayza and Fajnzylber (2005); and Chang, Kaltani, and Loayza (2009), for exam- 1 b3HKc,t 1 b4TCc,t 1 b5TOc,tTCc,t ple, use this methodology to estimate trade-growth regressions. Beck and Levine (2004); Beck, Levine, 1 b6 1 TOc,tTCc,t 2 2 1 mt 1 hc 1 Pc,t and Loayza (2000); and Rajan and Subramanian (2008) use it in the finance-growth literature. (B2.1.3a) The S-GMM procedure estimates a system of equations that combines the regression specification (continued) THE STRUCTURE OF TRADE LINKAGES AND ECONOMIC GROWTH 79 BOX 2.1 Methodology of trade and growth regression estimations (continued) in levels, as described above, and the same spec- The S - GM M procedure relies on four key ification in differences. This method deals with assumptions: (a) the error terms are not serially both the unobserved country-specific effects in this correlated, (b) shocks to growth are not predictable dynamic setup and the potential biases arising from based on past values of the explanatory variables, the endogeneity of explanatory variables. Differ- (c) the explanatory variables are uncorrelated with encing the regressions controls for the unobserved future realizations of the error term, and (d) the cor- country-specific effects, but it creates the additional relation between the explanatory variables and the problem that the error term of the differentiated equa- country-specific effects is constant over time. Not- tion is correlated with the lagged dependent variable. withstanding these assumptions, the method allows Taking advantage of the panel structure of the data- current and future values of the explanatory vari- set, the S-GMM procedure uses so-called internal ables to be affected by growth shocks—it is exactly instruments to address this issue as well as the poten- this type of endogeneity that the method is designed tial endogeneity of the explanatory variables. More to handle. In addition, the consistency of the specifically, for the equation in levels, the instruments S-GMM estimates of the parameters of interest and are given by the lagged differences of the explanatory their asymptotic variance-covariance matrix depend variables, whereas for the equation in differences, on whether lagged values of the explanatory vari- the instruments are lagged observations of both the ables are valid instruments in the growth regression. explanatory and the dependent variables. Three specification tests are used to evaluate these The set of instruments grows with the number potential issues: (a) the “full Hansen” test of over- of explanatory variables and time periods. As the identifying restrictions on the full set of instruments time dimension of the sample size is limited, only a (which tests the validity of the instruments by ana- restricted set of moment conditions is used in order lyzing the sample analog of the moment conditions to avoid overfitting bias. More specifically, only the used in the estimation process), (b) the “incremental first appropriate lag of each time-varying explan- Hansen” test of overidentifying restriction on the atory variable is used as an internal instrument. additional instruments that are introduced in the lev- For the variables measured as period averages, the els equations (which tests the stationarity assumption instruments correspond to their average in period on which these instruments are based), and (c) a sec- t Ϫ 2 ; for the variables measured as initial values ond-order serial correlation test (which tests whether within a given period, the instruments correspond the error term is serially correlated). The results of to their observation at the start of period t Ϫ 1. the Hansen and serial correlation tests (not reported) As a consequence, in the estimations of equations indicate that the null hypothesis of correct model B2.1.2 and B2.1.3a and b, the proxies for the nature specification cannot be rejected, lending support to of trade connections are interacted one at a time in the estimation results shown in this chapter. order to simplify the interpretation of the results This two-step S-GMM procedure is adopted and to avoid overextending the number of required throughout this chapter to estimate the trade-growth instruments (and hence the number of estimated relations for an unbalanced panel dataset covering 118 parameters). Even with this restricted set of instru- countries, including 24 in LAC. For each country, the ments, there are specifications in which the actual dataset comprised at most 10 observations of nonover- number of instruments is close to or even larger lapping five-year panels spanning 1960–2010. Annex than the number of countries in the sample. In these table 2A.1 provides details on data used for each vari- cases, a restricted sample of control variables is able included in these estimations. (All the estimated used to reduce the number of explanatory variables. regressions use variables represented in logs.) This benchmark regression specifica- gross domestic product (GDP) per capita tion, shown in the first column of table 2.1, has a negative and statistically significant yields estimates comparable to the estimates coefficient, which is typically interpreted in the empirical literature relying on the as evidence in favor of conditional con- cross-country variation of within-country vergence. The coefficient associated with changes. Trade openness is positive and sta- the degree of human capital development tistically significant, indicating its positive is not statistically significant in this speci- impact on average economic growth. Initial fication, though it is typically positive and 80 LATIN AMERICA AND THE RISING SOUTH significant in other specifications.11 The income growth, and the terms of trade are estimated coefficients on the control vari- negatively associated with growth, a result ables are also statistically significant and that captures the adverse effects of rela- with the expected sign: public infrastruc- tive price and exchange rate instability on ture is positively associated with per capita growth outcomes. TABLE 2.1 Regression results on the effect of the nature of traded goods on economic growth Dependent variable: Growth rate of real GDP per capita (1) (2) (3) (4) (5) Initial GDP per capita –2.318*** –2.888*** –2.620*** –0.422*** 0.068 [0.174] [0.062] [0.139] [0.081] [0.074] Labor force education –0.035 0.240** 0.077 0.898*** 1.014*** [0.204] [0.106] [0.128] [0.048] [0.158] Terms of trade –0.941*** –1.404*** –0.937*** [0.164] [0.071] [0.154] Public infrastructure 2.036*** 2.160*** 1.701*** [0.151] [0.044] [0.097] Trade openness 1.571*** 1.390*** 0.515*** 1.133*** [0.234] [0.129] [0.133] [0.119] Trade linkages with North countries 1.670*** [0.073] Trade linkages with South countries 0.228*** [0.079] Intraindustry trade (IIT) 7.841*** [0.707] Share of trade in: Primary products 5.572*** [0.784] Unskilled labor–intensive goods 12.756*** [0.807] High-tech-intensive goods 11.726*** [0.626] Skilled labor–intensive goods 26.141*** [1.076] Participation in GVCs: Participation in middle stages 1.301** [0.585] Participation in initial stages –5.440*** [0.514] Number of observations 846 800 806 806 806 Number of countries 117 114 117 117 117 Source: Didier and Pinat 2015. Note: This table reports the regressions of real GDP per capita growth on a number of indicators capturing the nature of products traded. See text and annex table 2A.1 for details on indicators used. In column (2), the Wald test on the difference of coefficients of the trade linkages with North and South countries is statistically significant. Robust standard errors are shown in brackets. All regressions include time dummies. Significance level: * = 10 percent, ** = 5 percent, *** = 1 percent. THE STRUCTURE OF TRADE LINKAGES AND ECONOMIC GROWTH 81 Not all trade connections are the same: Facts 3 in chapter 1). The patterns of connec- trading with North countries seems to be tions between LAC and other South countries associated with larger effects than trading suggest that trade flows during the 2000s with South countries (column 2 of table 2.1). were rooted in comparative advantage forces LAC countries have broadened and deepened and endowments, especially when contrasted their connections with other South countries with the nature of its connections with North (as documented in Set of Facts 2 in chapter countries. For example, during 2003–11, the 1). The share of the South in total trade flows average share of natural resources in exports to and from LAC countries increased about from LAC to the South (29 percent) was 70 percent, from 26 to 45 percent, during almost 50 percent larger than the share to the the 2000s. These regressions seem to imply North (46 percent). that these tighter connections to South coun- One strand of the literature has studied tries may have negatively affected the trade- whether certain types of traded goods have growth dynamics in LAC. Therefore, key greater growth-enhancing effects than others. questions are the extent to which the struc- For LAC the issue is whether greater reliance tural features of trade connections affect the on traded primary goods has the same growth trade-income nexus and whether there is a potential as greater reliance on trading other role for policy actions. goods. The “natural resource curse” claim Differences in the nature of trade connec- put forward by Sachs and Warner (1995, tions may play a key role in explaining the 1997) would suggest a negative answer to trade and growth dynamics; they may be one this question, based on empirical evidence of the reasons why the debate on whether that, on average, commodity-reliant countries trade causes growth remains unsettled. The grow more slowly than other countries.12 estimations presented above, like most of The debate over this issue has been heated. the regressions in the literature, capture At its core is the idea that price mechanisms whether the total value of trade flows matters do not fully capture the benefits of trading for growth or if the imposed linear relation a good when its production involves (Mar- between trade and growth simply reflects an shallian) externalities and rents. Externali- average effect. ties offer perhaps the strongest argument for The hypothesis of this chapter is that some the assertions that some goods are superior features of cross-border trade linkages are to others.13 These externalities can be either more likely to foster the growth-enhancing local (for example, if productivity increases effects of trade than others. To explore the with the overall size of the industry) or inter- issue, the analysis expands the benchmark industry (if production in one industry leads regression specification to include a set of to spillovers in other industries). Hence mar- proxies for the features of trade characteris- ket forces may not provide an economy with tics in order to infer whether they are asso- the appropriate set of incentives for the opti- ciated with additional growth effects and mal allocation of resources. therefore alter the average effect estimated Important for the discussion of trade- in the benchmark specification. The chapter growth dynamics are the externalities intrin- explores two broad aspects of the structural sically related to the technology diffusion and features of trade connections—the nature of learning spillovers channel discussed above. the products traded and the composition of The potential and intensity of these spillover trading partners. effects may vary with the type of industry involved in international trade. In particu- lar, some industries may offer greater poten- The nature of traded goods tial for the upgrading of production to more The sectoral composition of LAC’s connec- differentiated, higher-quality, higher-value tions with other South countries is different products or for the development of linkages from the composition of its connections with within and across industries. It has long North countries (as highlighted in Set of been recognized that industries with forward 82 LATIN AMERICA AND THE RISING SOUTH and backward linkages tend to be good for Growth and the composition growth. For example, Hausmann, Hwang, of traded goods and Rodrik (2007); Hidalgo and others (2007); Hidalgo and Hausmann (2009); and The first relevant trade characteristic is the Haussman and Hidalgo (2011) argue that degree of IIT, which ranges from 0 (pure countries specializing in products that can interindustry trade) to 1 (pure intraindus- serve as “launching pads” for other products try trade).16 LAC countries typically have (or even industries) are likely to have better lower levels of IIT than other emerging growth prospects. They liken the “prod- regions (figure 2.1, panel a). Although the uct space” to a forest in which the distance share of IIT rose in most LAC countries between products depends on the similar- between 1990 and 2011, the gap with other ity of their production capabilities. A short regions did not decline, except in Mexico. distance indicates that it is easier for firms In 2011 the degree of IIT in LAC was about to learn how to enter into proximate busi- 0.25—far lower than the figures for the ness activities and for technology to diffuse East Asian Tigers (Hong Kong SAR, China; more rapidly to nearby products. The authors Singapore; and Republic of Korea) (0.49), provide evidence that economies located in China (0.47), and Europe and Central Asia the denser parts of the forest are thus more (ECA) (0.45). Within LAC, Caribbean and likely to experience greater technology dif- Central American countries show a higher fusion and learning spillovers—and conse- average degree of IIT than South American quently economic growth.14 Being located in countries. a high-density segment of the product space The degree of IIT for LAC countries var- could also result in greater export diversifi- ies across trading partners (see figure 2.1, cation, which in turn could reduce macroeco- panel b). On average, it is slightly higher nomic volatility. when LAC countries trade with North part- Other researchers, such as Lederman and ners, though there is significant heterogeneity Maloney (2012), focus on how goods are across countries. In Brazil, Costa Rica, and produced rather than on what goods are pro- Mexico, for example, the level of IIT is much duced. They argue that the literature offers higher in their trade linkages with North little reliable evidence on the superiority of countries, whereas in Argentina, Colombia, one type of good over another and hence gives and Guatemala, trade with South countries little support to the selection of products or has substantially higher levels of IIT. industries for special treatment. They note A variety of arguments has been put for- that the empirical evidence reveals an extraor- ward suggesting that the degree of IIT is dinary heterogeneity of country experiences linked to differentiated growth outcomes and within product categories.15 Underlying the that differences in IIT with respect to North evidence is the notion that the same produc- and South countries may have played a role tion process for a given traded good in two in the dynamics of trade and growth in LAC. different firm and country setups may entail A high level of IIT within broadly defined very different degrees of technology diffusion industries indicates that the adoption, adap- and learning spillovers. These differences may tation, and mastery of foreign technologies arise from identically classified goods being available through imported goods may be produced at very different levels of productiv- easier to the extent that they are directly ity, quality, and technological sophistication, applicable to a countries’ export basket. High as well as from the nature of trading relations, IIT thus increases the probability that knowl- including the fact that in an evolving global edge and technology gained from imports production system, countries increasingly can be applied to exports. trade in tasks rather than in goods. The rest Alvarez, Buera, and Lucas (2013) adopt of this section, which draws on Didier and a similar concept. They argue that improve- Pinat (2015), discusses the role of these two ments in technology can arise from interac- aspects for countries’ growth prospects. tions among firms that are brought together FIGURE 2.1 Intraindustry trade a. Average level of intraindustry trade, by region, 1990 and 2011a b. Average level of intraindustry trade in Latin America and the Caribbean, by country, 2011 Mexico LAC (28) Dominican Republic Bahamas, The Central America, the Caribbean, Costa Rica and Mexico (17) Jamaica Barbados South America (11) Brazil Belize Guyana North (21) Bolivia Suriname St. Vincent and China the Grenadines Venezuela, RB Chile ECA (25) Peru Nicaragua EAP (20) Paraguay Ecuador MENA (15) Colombia Uruguay South Asia (7) Guatemala El Salvador Argentina SSA (29) Panama 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0 0.1 0.2 0.3 0.4 0.5 0.6 Degree of IIT Degree of IIT 1990 2011 Trade with North countries Trade with South countries c. Estimated total growth effects of increasing intraindustry d. Estimated total growth effects of increasing intraindustry trade by 10 percentage points from sample mean trade by 10 percentage points from sample mean (interaction with trade openness) (interaction with labor force education) 0.8 LAC-7 EAP-7 EE-7 0.8 EAP-7 LAC-7 EE-7 0.7 0.7 0.6 0.6 Percentage points Percentage points 0.5 0.5 0.4 0.4 0.3 0.3 0.2 0.2 0.1 0.1 0 0 25 50 75 100 125 150 160 20 40 60 80 100 Trade openness (%) Labor force education (%) Sources: Calculations based on Comtrade data (panels a and b) and Didier and Pinat (2015) (panels c and d). Note: Intraindustry trade (IIT) is measured by the Grubel-Lloyd index. See annex table 2A.1 for details on how this indicator was constructed. See Didier and Pinat (2015) for details on how the total growth effects were calculated. The North includes the Group of Seven (G-7) members and other Western Europe countries; the South includes all other economies. EAP = East Asia and Pacific; ECA = Europe and Central Asia; LAC = Latin America and the Caribbean; MENA = Middle East and North Africa; SSA = Sub-Saharan Africa. LAC-7: Argen- tina, Brazil, Chile, Colombia, Mexico, Peru, and República Bolivariana de Venezuela. EAP-7: Cambodia, China, Indonesia, Malaysia, the Philippines, Thailand, and Vietnam. EE-7: Bulgaria, the Czech Republic, Hungary, Lithuania, Poland, the Russian Federation, and Turkey. Other regions follow World Bank classifications. a. Numbers in parentheses are number of countries in each region. 83 84 LATIN AMERICA AND THE RISING SOUTH by the prospects of gains from trade and that region, ranging from 0.9 to 1.1 percentage get new ideas by adapting better technolo- points; for Central America, the Caribbean, gies used in the production of other goods. and Mexico, the increase in growth would In fact, other researchers highlight that average about 0.3 percentage points. the extent of IIT may be a direct proxy for Another relevant characteristic related technological diffusion and knowledge spill- to the way goods are produced is factor overs.17 The economies of scale associated intensity, such as the degree of technol- with larger markets and the product differen- ogy and skilled and unskilled human capi- tiation that is possible with a higher level of tal embedded in traded goods. On average, IIT are thought to lead to more rapid produc- the share of high-technology-intensive tivity gains and hence faster growth. goods is larger in LAC countries’ trade bas- The estimations reported in column 3 of kets with North countries than with South table 2.1 show that for a given level of trade countries. This pattern is particularly evi- openness and human capital development, an dent in Argentina, Brazil, and Chile (figure increase in the level of IIT is associated with 2.2, panel a). There are also differences in a positive and statistically significant impact the patterns of trade of skilled labor–inten- on growth.18 The effect is sizable: a 10 per- sive goods. On average in LAC, the share of centage point increase in IIT from its sample skilled labor–intensive goods is larger in trade mean is associated with an increase of about with South countries than with North coun- 0.6 percentage points in growth. Although tries, especially in South America (see, for an increase in IIT always has positive and instance, Argentina and Colombia in figure large effects on growth in income per cap- 2.2, panel b). There is nonetheless great het- ita, the magnitude of this effect is nonlinear; erogeneity in the region. Mexico in particular the growth effects associated with greater stands as a notable exception to these patterns. IIT are smaller for very low (or very high) These differences may affect the likelihood levels of trade openness and human capital that trade connections deliver dynamic gains, development. particularly through learning spillovers and Figure 2.1 shows the total growth effects technology diffusion, thereby changing the of a 10 percentage point increase in IIT potential for the long-run growth of per from its sample mean as a function of trade capita income. More specifically, goods that openness (panel c) and labor force educa- require a larger share of skilled labor or high tion (panel d). These effects are obtained in technology in their production processes may unreported estimations, presented in Didier provide greater potential for upgrading and and Pinat (2015), with interacted coeffi- improvements. Moreover, their production cients between trade openness and IIT and may involve positive human capital exter- between the level of human capital develop- nalities. Increasing the production of certain ment and IIT.19 In countries in which exports goods (such as goods intensive in high tech- plus imports represent 50 percent or more of nology and skilled labor) may provide greater GDP, the growth effect is about 0.6 percent- incentives for accumulating high-level human age points. For countries with secondary or capital and thus be associated with greater tertiary enrollment rates of more than 20 per- growth effects. Exporting these goods may cent, the growth effects can be as large as 0.7 provide even greater incentives. For emerging percentage points. The effects of an increase economies, selling goods to consumers with in IIT can be sizable in LAC, where levels of higher incomes than domestic consumers— IIT are typically relatively low, though they and thus potentially higher valuation of vary significantly across countries in the quality—may require quality upgrading, region. The effects could be particularly large marketing, and other types of knowledge for South American countries, which gen- that skilled workers provide. Indeed, the erally register the lowest levels of IIT in the empirical evidence indicates that exporting THE STRUCTURE OF TRADE LINKAGES AND ECONOMIC GROWTH 85 FIGURE 2.2 Shares of traded goods of different factor intensities a. High-technology-intensive goods, 2011 b. Skilled labor–intensive goods, 2011 Panama Bahamas, The Argentina Mexico Brazil Bolivia Uruguay Guyana Guyana Jamaica St. Vincent and the Grenadines Paraguay Dominican Chile Republic Dominican Republic Chile Bolivia Costa Rica Jamaica Barbados Suriname Panama Peru Brazil Barbados Suriname Belize Belize Nicaragua Peru Venezuela, RB Venezuela, RB Colombia Uruguay El Salvador Nicaragua Ecuador Guatemala Guatemala El Salvador Paraguay Ecuador Costa Rica Argentina Bahamas, The Colombia Mexico St. Vincent and the Grenadines 0 10 20 30 40 50 60 70 80 0 10 20 30 40 50 60 Percent Percent Trade with North countries Trade with South countries c. Estimated total growth effects of increasing the shares of traded goods d. Estimated total growth effects of increasing the shares of traded goods of different factor intensities by 10 percentage points from sample mean of different factor intensities by 10 percentage points from sample mean (interaction with trade openness) (interaction with labor force education) LAC-7 EAP-7 EE-7 EAP-7 LAC-7 EE-7 3.0 3.0 2.5 2.5 2.0 2.0 Percentage points Percentage points 1.5 1.5 1.0 1.0 0.5 0.5 0 0 -0.5 -0.5 -1.0 -1.0 25 50 75 100 125 150 160 20 40 60 80 100 Trade openness (%) Labor force education (%) Share of traded goods intensive on: Primary products Unskilled labor-intensive goods High-tech-intensive goods Skilled labor-intensive goods Sources: Calculations based on Comtrade data (panels a and b) and Didier and Pinat (2015) (panels c and d); classifications of the goods intensity are from Hinloopen and van Mar- rewijk (2001) and Didier and Pinat (2015). Note: Figures report trade-weighted averages. See annex table 2A.1 for details on how this indicator was constructed. See Didier and Pinat (2015) for details on how the total growth effects were calculated. The North includes the G-7 members and other Western Europe countries; the South includes all other economies. LAC-7: Argentina, Brazil, Chile, Colombia, Mexico, Peru, and República Bolivariana de Venezuela. EAP-7: Cambodia, China, Indonesia, Malaysia, the Philippines, Thailand, and Vietnam. EE-7: Bulgaria, the Czech Republic, Hun- gary, Lithuania, Poland, the Russian Federation, and Turkey. 86 LATIN AMERICA AND THE RISING SOUTH firms tend to hire more skilled labor and some role in the trade-growth dynamics and pay higher wages than firms that sell only to that not all goods are expected to bring the domestic consumers (Brambilla, Lederman, same benefits to all economies. This finding and Porto 2012). is particularly relevant for LAC, where there The estimation in column 4 of table 2.1 is great heterogeneity in the composition of indicates that the factor intensity embedded the export basket across countries. On aver- in traded goods affects the nature of trade- age, South American countries export a larger growth linkages. The coefficient on trade share of primary goods, and the countries of openness is positive and statistically signif- Central America and the Caribbean export icant, as are all the coefficients associated more unskilled labor–intensive goods. In with the variables capturing the relative South America an increase of 10 percentage factor intensity of the traded basket. These points in the share of skilled labor–intensive results indicate additional growth effects goods (and a similar decline in the share of relative to the omitted baseline category of natural resource–intensive goods) would be the share of traded goods intensive in nat- accompanied by an increase in income growth ural resources, although the magnitude of of about 0.9 percentage points on average. the growth effects varies. A larger share of For Mexico, Central America, and the Carib- skilled labor–intensive goods is typically bean countries, the effects may be even larger, associated with the largest growth effects. about 1.1 percentage points. Increasing the There are significant changes in the relative share of high-tech-intensive goods would also ranking of goods at different levels of trade be associated with different growth effects in openness and labor force education, suggest- different LAC countries, with South Amer- ing that not all goods bring the same growth ican countries benefitting most (enjoying benefits to all economies. increased growth of about 0.9 percentage Figure 2.2 shows the total growth effects points). The magnitude of these effects can for different categories of products of a be traced back to the relatively small shares 10 percentage–point increase in the shares of these goods in these countries’ trade bas- of traded goods (from their sample means, kets, especially with South countries. accompanied by a decline of the same mag- The literature provides inconclusive evi- nitude in the share of traded goods in natural dence on the superiority of one type of good resources). It shows how these effects vary over another and hence on the selection of with the level of trade openness (panel c) and products or industries for special treatment. human capital development (panel d) (these What emerges consistently in empirical anal- effects are in addition to the direct effects of yses is that the structure and quality of trade trade openness and education on growth). baskets merits special attention. The evidence An increase in the share of traded goods that in this section is also suggestive of positive are intensive in skilled labor yields the larg- externalities, such as those associated with est effects on economic growth for almost technology diffusion and knowledge spill- all levels of trade openness and labor force overs associated with the structure of trade education. The second-largest growth effect relations. The key issue for LAC is thus to is associated with an increase in the share of understand how countries can develop trade high-tech-intensive goods, especially as trade linkages in ways that more effectively foster integration increases. In fact, for economies learning spillovers and technology diffusion with trade openness of 75 percent or higher, in particular and economic development the effects are even larger than the effects more broadly. Because economically large associated with skilled labor–intensive goods. growth effects appear to occur only when These changes in the relative ranking of changes in the structure of trade dynamics different types of goods at different levels are substantial, the development of a long- of trade openness and human capital devel- term policy agenda on trade and growth opment suggest that externalities may play issues is critical. THE STRUCTURE OF TRADE LINKAGES AND ECONOMIC GROWTH 87 South. Trade in intermediate goods is indic- Growth and insertion into global ative of GVCs, as fragmented production value chains processes require that parts and components A complementary way of examining the cross borders, sometimes more than once, scope for international trade–related growth before finished goods are shipped to final effects, especially effects associated with markets. South countries have indeed shown technology diffusion and learning spillovers, a remarkable expansion in their exports of is to focus on GVCs. The development of intermediate goods, especially when con- GVCs, which is characterized by the dis- trasted with the North. 24 The value of persion of production stages and processes exports of intermediate goods from South across countries, is an important aspect of the countries now exceeds the value of interme- changing patterns of economic globalization, diate goods exports from North countries what Baldwin (2006, 2012b) calls globaliza- (figure 2.3, panel a). Moreover, as a share of tion’s second unbundling. The technological world exports, exports of intermediate goods revolution, especially in information technol- have gradually expanded for South countries ogy, communications, and inventory man- since the mid-1980s (see figure 2.3, panel b). agement, facilitated development of these This growth in intermediate goods trade as production chains. Moreover, large wage dif- a share of world trade is noteworthy, espe- ferentials across countries and declining trade cially as total world exports can increase for costs made the geographical fragmentation of reasons other than increases in intermediate production profitable. 20 A large body of lit- goods trade.25 erature documents the importance of foreign There is, however, great heterogeneity in direct investment (FDI) flows in forging these the South’s participation in GVCs. While the global production chains.21 value of intermediate goods trade expanded The fragmentation of the production across all South regions, different trends process means that individual countries emerge if the ratios of intermediate goods no longer need to develop the full range of exports to total world exports are analyzed capabilities required to create a product or (see figure 2.3, panels c and d). Perhaps not provide a service.22 They can contribute par- surprisingly, China has displayed the most ticular components of the final good, becom- impressive growth in intermediate goods ing specialized in “tasks” that contribute to exports, both in dollar values and relative the overall production process. As GVCs to world trade. Other East Asian econo- have gained in prominence, “trade in tasks,” mies, especially the East Asian Tigers, also where value is added along the production expanded their exports of intermediate goods chain, has led to a significant increase in the throughout most of the 1980s and 1990s. value of global trade of intermediary goods More recently, countries in ECA expanded (WTO and IDE-JETRO 2011). Indeed, as their intermediate goods exports as a share Grossman and Rossi-Hansberg (2008) note of world exports. Within LAC, on average, “It’s not wine for cloth anymore.” Individual Mexico and countries in Central America products are no longer produced entirely in and the Caribbean export more intermedi- a single country; production chains are now ate goods than South American countries. spread out across many countries. Although the value of these exports in Cen- The development and establishment of tral America, the Caribbean, and Mexico GVCs as a pillar of the global economy and generally increased between 1990 and 2012, the consequent increase in trade in tasks that it remains small relative to East Asia and has been taking place since the 1980s are Eastern Europe. tightly linked to the rise of a diverse set of The change in the production processes of South countries.23 Box 2.2 discusses the driv- apparel and footwear, automotive goods, and ers of this dispersion of production stages electronics is particularly striking (see Ger- (or tasks) away from the North toward the effi 2014 and references therein). In 1962 the 88 LATIN AMERICA AND THE RISING SOUTH FIGURE 2.3 Growth of global value chains a. Value of exports of intermediate goods in b. Intermediate goods exports as share of total exports 2,250 global value chains, by the North and the South in global value chains, by the North and the South 14 Share of world exports in 3 GVCs (%) 2,000 12 1,750 Constant US$ (billions) 1,500 10 1,250 8 1,000 6 750 4 500 250 2 0 0 1962 1970 1980 1990 2000 2012 1962 1970 1980 1990 2000 2012 From North countries From South countries c. Value of exports of intermediate goods in d. Intermediate goods exports as share of total exports global value chains, by region in the South in global value chains, by region in the South 900 2.5 Share of world exports in 3 GVCs (%) 800 700 2.0 Constant US$ (billions) 600 1.5 500 400 1.0 300 200 0.5 100 0 0 1962 1970 1980 1990 2000 2012 1962 1970 1980 1990 2000 2012 East Asian Tigers Other East Asia MENA SSA China ECA South Asia South America Central America, the Caribbean, and Mexico Sources: Calculations based on data from Comtrade; classification of intermediate goods into three major global value chains (namely apparel and footwear, electronics, and automo- biles and motorcycles) is from Sturgeon and Memevodic (2010). Note: The North includes the G-7 members and other Western Europe countries; the South includes all other economies. ECA = Europe and Central Asia; MENA = Middle East and North Africa; SSA = Sub-Saharan Africa. East Asian Tigers: Hong Kong SAR, China; Singapore; and Republic of Korea. East Asia: Indonesia, Malaysia, Philippines, and Thailand. All other regions follow World Bank classifications. North accounted for more than 80 percent of of this process. In the automobile industry, world exports of apparel and footwear and for example, Brazil, India, Mexico, and the more than 90 percent of exports of electron- Republic of Korea are important exporters, ics and automobiles (figure 2.4). Since then having relied to varying degrees on FDI from the North has steadily lost ground to South lead firms in the North to jump-start their countries, accounting for only about half of industries. 26 The East Asian economies are exports in each market by 2012. Leading leading exporters of electronics. India, Indo- the expansion by South economies, China nesia, Pakistan, Turkey, and Vietnam occupy accounted for 16 percent of world exports prominent positions in apparel and footwear. of apparel and footwear and 24 percent of The value added to goods at different world exports of electronics in 2012. China stages of production has not shifted propor- did not entirely drive the shift in production tionally with the fragmentation of produc- from the North to the South, however; other tion across locations. Value added along the South countries have been an integral part production chain has actually shifted away THE STRUCTURE OF TRADE LINKAGES AND ECONOMIC GROWTH 89 BOX 2.2 What has driven the dispersion of production tasks away from the North toward the South? Starting in the 1960s—and especially since the headquarters in the United States with a large R&D early 1990s—the combination of large actual and center in North Carolina; an advanced notebook com- potential domestic market growth with a large sur- puter development facility in Japan; three final assem- plus of low-cost, adequately skilled labor; capable bly plants in China and one in India; regional distri- manufacturers; and abundant raw materials in the bution facilities in the Netherlands, Dubai, Florida, largest countries in the South (such as Brazil, China, Australia, and India; and a large corporate planning, India, the Republic of Korea, Mexico, Turkey, and finance, and business process development group in Vietnam) has led to waves of investment from the Singapore. Similarly, India’s Tata Motors acquired North.a This investment has financed both flourish- Jaguar and Land Rover in a deal that included pro- ing domestic markets in the South and exports back duction, design, and engineering facilities. to the North.b Key for this renewed wave of invest- A few Latin American and Caribbean firms have ments was the decline in trade and investment bar- also followed this path. In 2009, for example, the riers in the South. Many South countries underwent Mexican Group Bimbo acquired Weston Foods, the deep processes of trade and financial liberalization, U.S. fresh bread and baked goods division of Cana- with many countries unilaterally liberalizing tariffs dian conglomerate George Weston. The acquisition (see, for example, Johnson and Noguera 2012 and encompassed a premium brand portfolio in bakery Lopez-Gonzalez and Holmes 2011). goods as well as 22 bakeries and more than 4,000 Perhaps surprisingly, the 2008 global crisis exac- distribution routes. erbated these trends (Cattaneo, Gereffi, and Staritz The explicit efforts of lead firms in global value 2010). The rapid growth of productive capabilities chains (GVCs) to simplify their supply chains in in China, India, and other large South economies order to deal with a smaller number of highly capa- led to a significant shift in global demand, from ble and strategically located suppliers is also part goods (both finished goods and intermediates) pro- of the story behind the shift in production from the duced in the North to goods produced in the South. North to the South in the global economy. In fact, As demand in the North stagnated, the large South the trend toward specialization and fragmentation economies turned inward, redirecting production in GVCs is evolving as firms place more emphasis to their domestic markets and regional neighbors on strategic collaboration among companies within (Kaplinsky and Farooki 2010). Thanks to their GVCs (Gereffi 2014). For instance, as both lead large domestic markets, many South countries also firms and suppliers gain market share, they become became attractive for exporters in the North. increasingly aware of the strategic vulnerabilities Another important factor behind the steady shift with respect to access to supplies of raw materials of production away from the North to the South has (Lynn 2005). Consumer goods firms such as Cad- been the internationalization of firms from the South, bury, Coca-Cola, and Unilever, for example, are which have sought global expansion through mergers expanding their direct involvement in the procure- and acquisitions of established global brands. The ment and sustainability of the raw material sides of Chinese company Lenovo, for example, acquired their production chains. Manufacturers of automo- IBM’s personal computer division, in a deal that biles and electronic goods worry about the availabil- bought the company not only the brand but also a new ity of raw materials such as lithium and coltan. a. Krugman (1991) developed an important theoretical framework for understanding the dispersion of production tasks away from the North and toward the South. He argues that the location of industrial firms depends on both dispersion and agglomeration forces. Dispersion forces are related to actual costs, such as factor prices and potential production subsidies, as well as to gains from specialization, such as gains related to economies of scale and learning-by-doing. Agglomeration forces are related to separation costs; they include transmission and transportation costs, risks and managerial time, and knowledge spillovers. The dynamics of this trade-off between dispersion and agglomeration forces varies across firms/industries, leading to different locational outcomes for different firms/industries. See Baldwin and Venables (2013) for a theoretical analysis of this trade-off in GVCs. b. See Sturgeon and Van Biesebroeck (2011); Morris, Staritz, and Barnes (2011); UNCTAD (2011); and Baldwin (2012a), among many others. from offshore stages, especially stages asso- discrepancy between where final goods are ciated with the manufacturing and assembly produced and exported and where value is of products (Baldwin 2012a). In fact, the created and captured seems to have grown.27 90 LATIN AMERICA AND THE RISING SOUTH FIGURE 2.4 The rise of the South in selected global of which $80 went to Apple for its invention value chains and overall coordination of production and $75 went to distribution and retail. In con- a. Apparel and footwear 100 trast, assembly in China added at most a few Share of world exports in apparel 90 dollars to the product’s value (Dedrick, Krae- 80 mer, and Linden 2010). and footwear (%) 70 Overall, GVCs have created many oppor- 60 50 tunities for South countries. Participation in 40 GVCs can be a synonym for accessibility to 30 larger markets, which can enable the classi- 20 cal efficiency gains from trade through spe- 10 0 cialization. The potential for employment 1962 1970 1980 1990 2000 2012 and export generation can be large. GVCs can facilitate the capturing of scale econo- mies and positive externalities that may be b. Electronics Share of world exports in electronics (%) 100 unavailable within local markets. This bene- 90 fit may be particularly important for smaller 80 economies, which may be able to specialize 70 in tasks in which they have comparative 60 50 advantage rather than in goods that need to 40 be fully developed and produced internally. 30 The international division of labor (or tasks) 20 in the production process can also lead to 10 productivity increases that generate import- 0 1962 1970 1980 1990 2000 2012 ant welfare gains that can ultimately drive economic growth. These gains can arise through learning-by-doing effects, direct c. Automobiles and motorcycles technology transfers, and increased effi- Share of world exports in automobiles 100 90 ciency and productivity as a result of inter- 80 national competition. Involvement in GVCs and motorcycles (%) 70 can also yield indirect benefits, by providing 60 mechanisms for technology and knowledge 50 spillovers in particular and capacity build- 40 30 ing and economic development more widely, 20 thus having the potential to lead to virtuous 10 circles. 28 0 While greater integration into GVCs 1962 1970 1980 1990 2000 2012 implies greater complementarity of domestic Total exports from: and foreign productions, it also implies more North LAC China Other South exposure to external shocks through both Sources: Calculations based on data from Comtrade; classification of intermediate forward and backward linkages. It leaves goods into three major global value chains is from Sturgeon and Memevodic firms vulnerable to shocks in access to inter- (2010). Note: The North includes the G-7 members and other Western Europe countries; mediate inputs and demand for their final other South includes all other economies except Latin America and the Carib- output. In fact, GVCs have helped reshape bean (LAC) and China. the elasticity of international trade. The 2008 global crisis revealed a higher trade Apple’s iPod illustrates the allocation of value elasticity to external shocks through trade. added along the production chain. Of the These changes in elasticities at least partly $299 retail price in the United States in 2005, reflect the so-called “bullwhip effect”—the U.S. firms and workers captured $155, out fact that the farther away firms are from the THE STRUCTURE OF TRADE LINKAGES AND ECONOMIC GROWTH 91 final customer, the more affected they are by Upgrading can occur in many ways. The shocks in final demand. 29 Greater exposure standard form typically cited in the innova- to external shocks may be associated with tion literature is upgrading of products and greater macroeconomic uncertainty and vol- processes of production. One way to assess atility—and hence lower economic growth.30 this type of upgrading is to look at shifts in In addition, not all of the benefits dis- the technology content of countries’ exports. cussed above accrue automatically. Indeed, From the 1960s to the 1980s, countries there are growing concerns that the economic in LAC typically exported primary prod- gains of participating in global supply chains ucts and resource-based manufactures (fig- have not yielded the hoped for benefits. South ure 2.5, panels a and b).34 Beginning in the economies in particular seem uneasy about 1990s, significant changes took place in working conditions, employment instabil- Central America, the Caribbean, and Mex- ity, extended reliance on low value-added ico. By the late 2000s, a significant share of activities, and technological dependence on their exports were medium-technology man- foreign firms. Firms and entire industries ufactures (this share reached 30 percent in within countries may be locked into seg- 2010). The share of high-technology manu- ments of GVCs that do not require upgrading factures also increased, from practically zero of human capital and are relatively unprofit- in the late 1960s to 15 percent in 2011. Over able, greatly limiting the potential benefits of the same period, the share of low-technology participation. manufactures, resource-based manufactures, The extent to which the benefits of par- and primary products declined significantly. ticipation in GVCs materialize depends on These changes in the technology intensity of the capabilities of local firms, the structure manufactures occurred largely since the late of individual GVCs, and the general policy 1990s, during which time the export struc- framework in host countries. 31 Production ture of these countries was transformed from chains include economic activities of all lev- one based on raw materials to one in which els, from small-scale household-based work medium- and high-technology manufac- to high-skilled, technology-intensive, and tures are more representative. In contrast, knowledge-intensive work. A key issue is thus the export structure of South American whether participation in GVCs is inclusive countries remains concentrated in primary or exclusive in terms of facilitating economic goods and resource-intensive manufactures; upgrading, especially of lower-level firms in it changed little between the 1980s and the the chain.32 2000s. If anything, South American coun- Not all firms in all South countries face tries increased the share of exported primary the same opportunities and challenges asso- products during the 2000s. Similar patterns ciated with economic upgrading in GVCs. As are observed in their connections with both the literature shows, the recent trend toward North and South countries. consolidating suppliers may have different To put these trends in perspective, fig- effects on different firms depending on their ure 2.5 shows the export structure of other capabilities to meet higher standards. Quality South regions. A major shift in the export standards, one of the key mechanisms gov- structure of the East Asian economies took erning supply chains, can push out suppliers place beginning in the late 1960s, when low-, unable to comply, or they can trigger upgrad- medium-, and high-technology goods started ing. Moving up the chain toward more for- to replace primary goods and raw material– mal and skill-intensive work means that the intensive manufactures in their export bas- likelihood of enforceable standards probably kets. Since the late 1990s, low-technology rises. Meeting these higher standards is not manufactures have given some ground to just costly, it also requires a literate and com- more technology-intensive ones. By 2011 petent labor force that may be beyond the medium- and high-technology–intensive reach of many small-scale enterprises.33 goods accounted for almost 70 percent of 92 LATIN AMERICA AND THE RISING SOUTH FIGURE 2.5 Technological composition of exports from the South, by region 80 a. Central America, the Caribbean, and Mexico 80 b. South America 70 70 Share of regional exports (%) Share of regional exports (%) 60 60 50 50 40 40 30 30 20 20 10 10 0 0 1981 1985 1990 1995 2000 2005 2011 1981 1985 1990 1995 2000 2005 2011 80 c. China 80 d. East Asia and Pacific Share of regional exports (%) Share of regional exports (%) 70 70 60 60 50 50 40 40 30 30 20 20 10 10 0 0 1981 1985 1990 1995 2000 2005 2011 1981 1985 1990 1995 2000 2005 2011 80 e. Eastern Europe and Central Asia 80 f. Middle East and North Africa Share of regional exports (%) Share of regional exports (%) 70 70 60 60 50 50 40 40 30 30 20 20 10 10 0 0 1981 1985 1990 1995 2000 2005 2011 1981 1985 1990 1995 2000 2005 2011 80 g. South Asia 80 h. Sub-Saharan Africa Share of regional exports (%) Share of regional exports (%) 70 70 60 60 50 50 40 40 30 30 20 20 10 10 0 0 1981 1985 1990 1995 2000 2005 2011 1981 1985 1990 1995 2000 2005 2011 Resource-based products Primary products Low-tech manufacturing Medium-tech manufacturing High-tech manufacturing Sources: Calculations based on Comtrade database; classification of the technological composition of exports is from Lall (2000). Note: See annex table 2A.1 for details on how the indicator on the share of traded goods with different technological intensities was constructed. Regions follow the World Bank classification. THE STRUCTURE OF TRADE LINKAGES AND ECONOMIC GROWTH 93 the total exports of the East Asian Tigers trade and economic growth remains largely and about 33 percent of the exports of other unexplored. economies in East Asia and Pacific (EAP) GVCs are mostly regional, not global region. A similar transition has taken place in (as discussed in Set of Facts 3 in chapter 1). China. Medium- and high-technology goods, They may have played an important role in which represented just 23 percent of China’s the regional bias in trade and financial con- exports in 1990, accounted for 60 percent nections. An aspect that has not yet been of its exports by 2011. 35 Across ECA coun- explored is how countries are integrating into tries, low- and medium-technology-intensive GVCs. One indicator measures countries’ goods have gained space, accounting for participation in GVCs by separately taking 45 percent of the region’s exports in 2011. into account the participation of countries in A more subdued expansion has taken place GVCs as users of foreign inputs (backward for high-technology-intensive goods, which linkages or upstream component) and as sup- accounted for 10 percent of total exports. pliers of intermediate goods and services used Not all South regions have undergone these in other countries’ exports (forward linkages transformations. For instance, there is little or downstream component). 38 Input-output evidence of a significant shift toward these matrices provide a comprehensive assess- higher-technology-intensive exports in South ment of the extent to which countries are Asia. integrated vertically in global production Another form of upgrading, which is par- chains. Backward linkages are measured by ticularly relevant in the context of participa- the foreign value added in exports, whereas tion in GVCs, is the upgrading of function. It forward linkages are measured by the share occurs when firms in particular and countries of exported goods used as imported inputs to more broadly move toward activities requir- produce other countries’ exports. ing more skilled work, within or across pro- The aggregate GVC participation index, duction chains.36 Hence the location of firms constructed based on the Eora trade in value and even countries in GVCs matters for pro- added database, is consistent with the other ductivity and growth prospects. The distribu- indicators shown above. There has been a tion of profits and risks is intrinsically related global trend toward greater participation to the positioning of a firm/country within the into GVCs, especially by South countries GVC and the organizational arrangement of during the 2000s (figure 2.6, panel a). There the GVC. For instance, importing intermedi- is significant heterogeneity in the ways in ates in order to export final goods (backward and extent to which countries integrate into supply chain) may have very different effects GVCs around the world.39 from exporting parts so that other coun- First, there are large differences in the tries can export final goods (forward supply extent to which countries have joined chain).37 In backward chains, an increase in GVCs. Although there has been a signifi- overall productivity can be achieved as non- cant increase in the South’s participation, productive sectors get outsourced to other North countries, especially European coun- countries and the newly freed resources are tries, still have the largest GVC participa- allocated toward more productive sectors. In tion shares (figure 2.6, panel b). Within the forward chains, the creation of tighter link- South, economies in East Asia, ECA, Central ages can result in finer specialization, with America (especially Costa Rica), and Mex- gains arising through learning-by-doing ico are the most integrated into GVCs. South effects and technology transfers. Moreover, American countries also have relatively high given the importance of complementarities in participation ratios. Countries in Sub-Saha- supply chains, it is possible that increasing the ran Africa (SSA), the Middle East and North efficiency of one segment of the value chain Africa (MENA), and the Caribbean are the will increase the productivity of the chain as a least integrated into cross-country produc- whole. This mechanism linking supply chain tion networks. FIGURE 2.6 Participation in global value chains a. World participation in global value chains b. Regional participation in global value chainsa 70 North (24) Share of world exports ofgoods and services (%) 60 ECA (32) 50 22% EAP (24) 17% 14% 40 12% SA (8) 9% 30 SSA (43) 20 31% 34% 35% 36% 36% MENA (21) 10 Mexico and Central America (7) 0 1990 1995 2000 2005 2011 South America (11) With North countries With South countries The Caribbean (11) c. Participation in global value chains in 0 10 20 30 40 50 60 70 80 Latin America and the Caribbean, 2011 Share of regional exports of goods and services (%) Venezuela, RB 1990 2000 2011 Mexico Chile Argentina Bolivia Trinidad and Tobago Brazil Costa Rica Ecuador Belize Antigua and Barbuda Aruba d. Backward and forward participation in Jamaica global value chains by region, 2011a El Salvador EAP (24) Guatemala North (24) Haiti Barbados ECA (32) Uruguay SA (8) Peru Colombia MENA (21) Dominican Republic SSA (43) Honduras Mexico and Bahamas, The Central America (7) Nicaragua South America (11) Paraguay The Caribbean (11) Panama 0 10 20 30 40 50 60 0 10 20 30 40 50 60 70 80 90 100 Share of country’s exports of goods and services (%) Share of GVC participation (%) With North countries With South countries Backward GVC linkages Forward GVC linkages (continued) 94 THE STRUCTURE OF TRADE LINKAGES AND ECONOMIC GROWTH 95 FIGURE 2.6 Participation in global value chains (continued) e. Backward and forward participation in global value chains in Latin America and the Caribbean, 2011 Mexico El Salvador Panama Aruba Barbados Honduras Bahamas, The Dominican Republic Uruguay Belize Nicaragua Costa Rica Guatemala Jamaica Argentina Chile Haiti Antigua and Barbuda Cuba Ecuador Paraguay Bolivia Colombia Brazil Trinidad and Tobago Peru Venezuela, RB 0 10 20 30 40 50 60 70 80 90 100 Share of GVC participation (%) Backward GVC linkages with North countries Forward GVC linkages with North countries Backward GVC linkages with South countries Forward GVC linkages with South countries Sources: Calculations based on data from Eora MRIO and World Development Indicators (WDI). Note: Participation in global value chains (GVCs) is proxied by the share of a country’s export that is part of a multistage trade process. This measure is constructed by adding the foreign value added used in a country’s own exports (backward GVC linkages) to the value added supplied to other countries’ exports (forward GVC linkages) and scaling this total by the country’s total exports of goods and services. Panels b and d report cross-country averages. The North includes the G-7 members and other Western Europe countries; the South includes all other economies. EAP = East Asia and Pacific; ECA = Europe and Central Asia; MENA = Middle East and North Africa; SA = South Asia; SSA = Sub-Saharan Africa. a. Numbers in parentheses are number of countries in each region. LAC countries’ participation in GVCs rose 68 percent in South America, 90 per- increased dramatically between 1990 and cent in Mexico and Central America, and 2011. Participation in the North grew 224 percent in the Caribbean, making LAC about 62 percent. In contrast, participation the fastest-growing region in the world on 96 LATIN AMERICA AND THE RISING SOUTH this measure (the next fastest-growing South ECA (28 percent). Mexico and countries in region was EAP, where GVC participation Central America, along with some East Asian expanded almost 50 percent). There is great Tigers and MENA countries, have the lowest heterogeneity in LAC, with growth rates downstream component of the GVC partici- ranging from less than 10 percent in Uruguay pation index. and Peru to more than 200 percent in some Mexico and Central America have joined Caribbean countries. GVCs mostly through backward linkages— Second, there is significant cross-regional that is, their exports contain a significant and cross-country variation in the ways in share of foreign value added. In 2011 Mex- which countries integrate into GVCs. One ico (33 percent value added), El Salvador source of variation is the composition of part- (18 percent), and Costa Rica (17 percent) ners within GVCs (see figure 2.6, panel c). were among the most integrated countries in On average, Mexico and countries in Central LAC in this regard. Their backward partic- America and the Caribbean have higher GVC ipation (29 percent) is on a par with that of participation rates with North countries than East Asian economies (26 percent). In both with other South countries (Jamaica and regions, processing industries account for a Trinidad and Tobago are notable examples). significant share of exports. The backward In contrast, Bolivia, Paraguay, and Uruguay participation rates of countries in South have joined GVCs mostly with other South America (12 percent of exports) and the countries. On average, South American coun- Caribbean (10 percent) are among the low- tries display similar participation rates with est in the South. Countries in SSA, South North and South countries. Asia, and MENA also have low downstream Third, the placement of countries within components of GVC participation, averaging GVCs is heterogeneous. South American and about 10 percent or less of exports. Caribbean countries typically have higher This decomposition of GVC participation forward participation rates, whereas Mexico into forward and backward linkages can and Central American countries have higher shed light on the position of countries within backward participation rates (see figure 2.6, GVCs. South American countries tend to panels d and e). On average, South Ameri- join GVCs at their initial stages, by providing can countries are mostly suppliers of inputs: inputs to other countries. Brazil, Chile, Peru, about 26 percent of their exports are used and Trinidad and Tobago send a larger share as inputs in other countries’ exports. In con- of inputs to North than to South countries, trast, this figure is just 12 percent in Mexico whereas Bolivia and Paraguay send a larger and Central America. The countries in the share to South countries (see figure 2.6, panel region with the highest forward GVC partic- e). Mexico and Central America seem to be at ipation rates are Argentina, Bolivia, Brazil, the end of GVCs, given the high share of for- Chile, Ecuador, and Trinidad and Tobago. eign inputs used in their exports relative to the This finding is not surprising, given the large share of their exports used in other countries’ share of primary products in their exports. exports. This pattern reflects in large part the These goods typically require few foreign direction of their exports toward the domes- inputs for their production and are gener- tic market in the United States for final con- ally used at the initial stages of GVCs. South sumption rather than further transformation. America’s forward GVC linkages are among Countries with equally high backward the most extensive in the world, comparable and forward GVC participation rates can to the linkages of countries in ECA. Only also appear toward the middle of GVCs. North economies have greater forward par- Notable examples are economies in East Asia ticipation in GVCs. In 2011 about 35 percent that import a large fraction of the inputs of their exports were used as inputs in other embedded in their exports, which in turn are countries’ exports. This figure is higher than used as intermediate inputs in other coun- the figure for South America (26 percent) or tries’ exports. THE STRUCTURE OF TRADE LINKAGES AND ECONOMIC GROWTH 97 New empirical evidence at the country associated with the last stages of GVCs) is level suggests that being part of GVCs, espe- associated with positive and significant cially the middle of GVCs, is associated with effects on growth. In contrast, increasing higher growth rates and thus plays a role in the share of goods in the initial stages of the way international trade fosters economic GVCs (accompanied by a similar decline growth (see, for example, IMF 2013 and in the share of traded goods related to the UNCTAD 2013). This analysis approximates last stages of GVCs) is associated with neg- participation in GVCs using the degree of ative and statistically significant effects on upstreamness embedded in goods traded in growth. different industries, constructed according to The total growth effect of an increase of the measure developed in Antràs and others 10 percentage points in the share of traded (2012). In column 5 of table 2.1, this mea- goods in the middle segments of GVCs is sure considers insertion into three phases of positive when trade openness is superior to GVCs: beginning (exports of primary goods), 40 percent of GDP (figure 2.7, panel a). Gains middle (exports of intermediate goods), and in per capita income growth can be as large end (exports of final goods). The last cate- as 0.9 percentage points when a country is gory is omitted; the estimated effects should highly integrated into global markets. In therefore be interpreted as relative to inser- contrast, for levels of trade openness below tion at the final segments of GVCs. 100 percent, the point estimates indicate that The estimation results show that an increasing the share of the most upstream increase in the share of traded goods that traded goods is generally accompanied by typically belong to the middle of GVCs negative growth outcomes. (accompanied by a decline of the same mag- Nonlinear effects between participation at nitude in the share of traded goods typically the different stages of GVCs and the degree FIGURE 2.7 Growth effects of the stage of the participation in global value chains a. Estimated total growth effects of increasing the share of b. Estimated total growth effects of increasing the share of exported goods in different stages of the GVC by exported goods in different stages of the GVC by 10 percentage points from sample mean 10 percentage points from sample mean (interaction with trade openness) (interaction with labor force education) 1.0 LAC-7 EAP-7 EE-7 1.0 EAP-7 LAC-7 EE-7 0.8 0.8 0.6 0.6 0.4 0.4 Percentage points Percentage points 0.2 0.2 0 0 –0.2 –0.2 –0.4 –0.4 –0.6 –0.6 –0.8 –0.8 –1.0 –1.0 25 50 75 100 125 150 160 20 40 60 80 100 Trade openness (%) Labor force education (%) Participation in the initial stages of GVCs Participation in the middle stages of GVCs Source: Didier and Pinat 2015. Note: Indicator is based on the degree of upstreamness embedded in exported goods in different industries. See annex table 2A.1 for details on how it was constructed; see Didier and Pinat (2015) for details on how the total growth effects were calculated. LAC-7: Argentina, Brazil, Chile, Colombia, Mexico, Peru, and República Bolivariana de Venezuela. GVC = global value chain. EAP-7: Cambodia, China, Indonesia, Malaysia, the Philippines, Thailand, and Vietnam. EE-7: Bulgaria, the Czech Republic, Hungary, Lithuania, Poland, the Russian Federation, and Turkey. 98 LATIN AMERICA AND THE RISING SOUTH of labor force education are also observed. development and design (Kaplinsky, Terheg- For countries with secondary or tertiary gen, and Tijaja 2011). On the other hand, enrollment of more than 25 percent of the these firms may be locked in intense competi- active population, increasing the share of tion and face tight profit margins. traded goods in the middle of GVCs is asso- Overall, economic development today is ciated with positive effects on per capita to some extent inherently linked to upgrad- income growth (see panel b of figure 2.7). ing within rather than independently of The more educated the labor force, the GVCs. Yet participation in GVCs does not greater the growth effects accompanying the automatically translate into additional gains increase in the share of traded goods in the from trade beyond the gains associated with middle of GVCs. This increase reaches about increased export volumes. As discussed 0.5 percentage points for countries with above, several factors, most intrinsic and highly educated labor forces. In contrast, particular to individual GVCs, affect the the effects of increasing the share of traded dynamics of opportunities and challenges goods that fall in the initial stages of GVCs to thrive in these supply chains. Firms’ and is associated with a negative growth impact, countries’ competitiveness in GVCs reflects whatever the level of labor force education. not only their capacity to join and remain For LAC-7 countries, a 10 percentage–point part of GVCs over time but also their ability increase in the share of exported goods in the to upgrade within or across GVCs. In turn, middle of GVCs is associated with growth this ability to upgrade reflects the capacity of effects of about 0.25 percentage points. producers to generate, import, and apply new These results indicate that insertion into technologies.40 the middle of GVCs is associated with the The role of governments may seem limited largest increases in growth. Moreover, the in this context, as policymakers may not have growth effect appears to be larger the greater sufficient knowledge about the intricacies the level of trade openness; it is particularly of individual industries, GVCs, and market strong for countries with high levels of labor dynamics. But the ease with which countries force education. The underlying notion is that can design adequate policies is not indepen- the more the economic activities of a country dent of the extent of their sector-specific are connected to global production chains— knowledge, including knowledge of potential particularly the middle range of such chains— constraints domestic firms face. Moreover, and the more capable the country’s labor force the evidence presented in this section is sug- is, the more productivity-enhancing learning gestive of positive externalities associated and innovation effects can take place. with the way in which firms and countries The end market of GVCs also affects the integrate into GVCs. Market failures—from potential for upgrading opportunities, inde- the provision of infrastructure to the accu- pendent of firms’ or countries’ placement mulation of human capital and the resolu- within GVCs (Palpacuer, Gibbon, and Thom- tion of coordination failures within existing sen 2005; Gibbon 2008). In particular, con- industries—also abound. Therefore, there sumer preferences and government standards seems to be some scope for policy interven- are typically different in developing countries tion. In particular, policymakers can play a and more developed economies. Price is typi- key role in providing an appropriate set of cally the central consideration in South coun- incentives and support policies to help firms tries; product differentiation based on variety prosper in this new global economic land- and quality are less important (Kaplinski scape, in which GVCs are an integral part. and Farooki 2010). On the one hand, firms in GVCs targeting South economies as their end market may face lower entry barriers and The nature of trading partners impose looser standards for their products, Trading with North countries is associated making it easier for South firms to engage in with larger growth effects than trading with higher value-added activities, such as product South countries, as shown above. Hence an THE STRUCTURE OF TRADE LINKAGES AND ECONOMIC GROWTH 99 important aspect of the trade-growth nexus lead to more economic and political depen- is the composition and identity of trading dency (Dolan and Tomlin 1980; Packenham partners. The four channels through which 1992). The extent to which the composition trade and growth are related (discussed ear- and identity of trading partners matter for lier) may work in different ways depending economic growth is thus an empirical ques- on the composition of trade partners. Trad- tion, which the rest of this section assesses. ing with fast-growing and/or more advanced Whether trading partners are at the cen- economies may be associated with positive ter of the global trade network or on its growth effects partly as a result of aggregate periphery may affect the growth prospects demand effects for the goods in which the associated with their trade connections. The country has a comparative advantage. Arora channel of technology diffusion and knowl- and Vamvakidis (2005), for example, pro- edge spillovers may be particularly import- vide empirical evidence that trading partners’ ant in this regard. Independent of their level growth and relative income levels have strong of economic development or technological positive effects on domestic growth. Greater sophistication, the central countries in the integration with more advanced economies global trade network, which are more closely can also open up and enhance communica- connected to a wider range of countries, are tion channels that facilitate greater technol- more exposed to the technology and knowl- ogy diffusion and learning spillovers. edge frontiers. To the extent that firms get Trading with less developed countries new production-related ideas and technol- can also be growth enhancing, to the extent ogy by learning from firms with which they that it may lead to specialization in sectors do business (or compete), the establishment or tasks that are prone to technological of strong ties with countries more exposed and knowledge spillover effects. Increased to the frontiers of ideas and technologies competition from greater trade integration may lead to stronger growth effects.41 The can encourage entrepreneurs to pursue new quality and intensity of the feedback effects and distinctive ideas and technologies, but between buyers and sellers engaged in global under certain circumstances, competition trade, for example, may be greater if one of by dissimilar countries can hurt growth the countries involved is at the center of the outcomes. Intense competition from larger network. Trade with central countries may trading partners may reduce the profitabil- also be associated with a selection effect of ity of investments in knowledge in relatively putting domestic producers in contact with smaller economies if knowledge spillovers are the most efficient (subject to trade costs) for- national in reach. Increased competition with eign producers. All these factors enhance the a more technologically advanced trading likelihood of technology diffusion and learn- partner can slow innovation and growth in a ing spillovers. For a given country, then, the country that begins with some disadvantage potential for exposure to a wider set of ideas in research productivity if spillover effects are and technologies increases with the strength geographically concentrated (Grossman and of its trade ties with more central countries. Helpman 1991a). Core countries—countries with strong The composition of trading partners may connections to a large number of countries— also be associated with growth outcomes are more centrally located in the graphical through the volatility channel. Export bas- representation of the global trade network kets concentrated in few destinations may shown in figure 2.8. Each node in the fig- lead to increased volatility—as a result of ure represents a country, and each link cor- fluctuations in trading partners’ economies, responds to an active connection between a import-export patterns, or relative prices— pair of countries. As discussed in Set of Facts and hence be associated with worse growth 2 in chapter 1, during most of the 1980s outcomes (Loayza and Raddatz 2007; and 1990s, the global trade network cen- Haddad, Lim, and Saborowski 2013; Di tered on a small set of developed countries: Giovanni and Levchenko 2012). It may also the United States, Germany (as well as a few 100 LATIN AMERICA AND THE RISING SOUTH FIGURE 2.8 The global trade network a. 1980 b. 2012 North countries Latin America and the Caribbean Other South countries Source: Calculations based on data from IMF Direction of Trade Statistics (DOTS). Note: Networks drawn using the Kamada-Kawai algorithm. Each node represents a country. Each link corresponds to an active connection (positive trade flow) between a pair of countries. Arrows capture the direction of these connections. Only trade flows greater than $10 million in 1980 and greater than $100 million in 2012 are reported. The North includes the G-7 members and other Western Europe countries; the South includes all other economies. THE STRUCTURE OF TRADE LINKAGES AND ECONOMIC GROWTH 101 other Western Europe countries), and Japan. mostly by its strong ties with the United During the 2000s, several South economies, States and Canada). The share of trade with including Brazil, China, India, the Russian inner-periphery countries ranges from about Federation, and Turkey, among others, joined 15 percent in Honduras and Mexico to this club. almost 60 percent in Bolivia and Uruguay. Few countries occupy central places in On average, South American countries have the global trade network, although there is larger trade shares with inner-periphery no widely accepted definition of how many countries than do countries in Central Amer- and which countries can be considered core ica and the Caribbean. countries. Two alternative definitions of these To what extent are stronger trade ties with countries are adopted here. Using network countries in the center of the global trade analysis, countries are ranked according to network associated with higher growth? their share of world trade, their number of Column 1 of table 2.2 reports the estimations trading partners, and the position of their associated with the share of trade with the top partners in the global trade network.42 This three countries in the global trade network. To ranking changes over time to reflect the contrast the effects of trading with these center changes in the global trade network discussed countries with simply more concentrated trad- above. Based on this ranking, two proxies to ing relations, the regressions also include an characterize countries’ composition of trad- analogous proxy to capture countries’ share of ing partners are constructed: (a) the total trade with their main partners. The coefficient share of trade with the top three countries in on the share of trade with the most central the network and (b) the share of trade with countries in the global trade network is posi- countries in the top 5 percent of the net- tive and statistically significant; the coefficient work (the so-called core countries). To put on the share of trade with a country’s main the results in perspective, the analysis uses trading partners is negative and statistically as benchmarks the share of a country’s trade significant. The differential effect is econom- with its top three trading partners in terms ically large—about 0.8 percentage points. An of the total value of trade and the share of its increase of 10 percentage points in the share trade with countries in the 6th–30th percen- of trade with the top three most central coun- tiles (the so-called inner-periphery countries) tries is associated with an increase in growth of the network. of about 0.3 percentage points, whereas a LAC countries are generally as connected similar increase in the share of trade with the to countries in the center of the global trade top three main partners leads to a decline in network as other South regions, but the growth of about 0.5 percentage points. degree of connectivity to inner-periphery Figure 2.9 shows the total growth effect countries is more limited (figure 2.9, panels associated with an increase of 10 percentage a and b). The average share of trade with points in the share of trade with the most core countries is almost 50 percent in LAC— central countries and with the main trading similar to the shares observed in most other partners. It reveals how these effects vary regions. Only in countries in ECA and South with the degree of trade openness (panel c) Asia is the trade share with core countries and the level of human capital development below 40 percent. The average trade share (panel d). For low enough levels of trade with inner-periphery countries is just 35 per- openness, increasing trade ties with a coun- cent—well below the 54 percent in South try’s main trading partners is accompanied Asia, the 45 percent in East Asia, and the by a positive effect on per capita income 43 percent in ECA. growth, though the effect becomes nega- There is significant heterogeneity across tive at about 35 percent of trade openness. countries in LAC though. The share of trade In contrast, the total growth effect associ- with core countries ranges from 33 percent ated with an increase in the share of trade in Argentina to 81 percent in Mexico (driven with the most central countries in the global 102 LATIN AMERICA AND THE RISING SOUTH trade network increases, albeit only slightly, nonlinearity in the growth effect related to at low levels of trade openness and remains the degree of human capital development. positive throughout the range of observed The total growth effect for an increase of 10 levels of trade openness. There is also some percentage points in the share of trade with FIGURE 2.9 Composition of trading partners a. Share of trade with core and inner-periphery countries, by regiona in 2011 LAC (33) Central America, the Caribbean, and Mexico (22) South America (11) North (21) MENA (20) EAP (21) SSA (41) China ECA (31) South Asia (7) 0 10 20 30 40 50 60 70 80 90 100 Percent b. Share of trade with core and inner-periphery countries, by country in Latin America and the Caribbean in 2011 Mexico Costa Rica Honduras Venezuela, RB Chile Dominican Republic Colombia Guatemala Brazil Peru Trinidad and Tobago El Salvador Ecuador Nicaragua Haiti Jamaica Guyana Paraguay Panama Uruguay Bolivia Argentina 0 10 20 30 40 50 60 70 80 90 100 Percent Core countries Inner-periphery countries (continued) THE STRUCTURE OF TRADE LINKAGES AND ECONOMIC GROWTH 103 FIGURE 2.9 Composition of trading partners (continued) c. Estimated total growth effects of increasing the share d. Estimated total growth effects of increasing the share of trade with top-3 main partners and top-3 most central of trade with top-3 main partners and top-3 most central countries by 10 percentage points from sample mean countries by 10 percentage points from sample mean (interaction with trade openness) (interaction with labor force education) LAC-7 EAP-7 EE-7 2.0 EAP-7 LAC-7 EE-7 2.0 1.5 1.5 Percentage points Percentage points 1.0 1.0 0.5 0.5 0 0 –0.5 –0.5 –1.0 –1.0 –1.5 –1.5 –2.0 –2.0 25 50 75 100 125 150 160 20 40 60 80 100 Trade openness (%) Labor force education (%) Share of trade with top-3 main partners Share of trade with top-3 most central countries e. Estimated total growth effects of increasing the share of trade f. Estimated total growth effects of increasing the share of trade with core and inner-periphery countries by 10 percentage with core and inner-periphery countries by 10 percentage points from sample mean (interaction with trade openness) points from sample mean (interaction with labor force education) LAC-7 EAP-7 EE-7 3.0 EAP-7 LAC-7 EE-7 3.0 2.5 2.5 Percentage points Percentage points 2.0 2.0 1.5 1.5 1.0 1.0 0.5 0.5 0 0 25 50 75 100 125 150 160 20 40 60 80 100 Trade openness (%) Labor force education (%) Share of trade with core countries Share of trade with inner-periphery countries Sources: Calculations in panels a and b are based on data from DOTS; calculations in panels c–f are based on Didier and Pinat (2015). Note: See annex table 2A.1 for details on how countries were classified as core or inner periphery. See Didier and Pinat (2015) for details on how total growth effects were calculated. LAC-7: Argentina, Brazil, Chile, Colombia, Mexico, Peru, and República Bolivariana de Venezuela. EAP-7: Cambodia, China, Indonesia, Malaysia, the Philippines, Thailand, and Vietnam. EE-7: Bulgaria, the Czech Republic, Hungary, Lithuania, Poland, the Russian Federation, and Turkey. The North includes the G-7 members and other Western Europe countries; the South includes all other economies. EAP = East Asia and Pacific; ECA = Europe and Central Asia; LAC = Latin America and the Caribbean; MENA = Middle East and North Africa; SSA = Sub-Saharan Africa. a. Numbers in parentheses are number of countries in each region. the most central countries is typically posi- share of trade with countries in the 6th–30th tive, though declining, for countries in which percentiles (inner-periphery countries) of the more than 30 percent of the labor force edu- global trade network. The coefficients on cation. The effect on growth associated with the trade shares with core countries are pos- trading with the top three main trading part- itive and statistically significant, reinforcing ners is negative, though increasing with the the previous findings. The effects associ- level of labor force education. ated with the share of trade with countries Column 2 of table 2.2 reports the esti- in the inner periphery are typically larger: mations on the share of trade with countries the average effect of an increase of 10 per- in the top 5 percent (core countries) and the centage points in the share of trade with core 104 LATIN AMERICA AND THE RISING SOUTH TABLE 2.2 Regression results on the effects of the composition of trading partners on economic growth Dependent variable: Growth rate of real GDP per capita (1) (2) (3) (4) (5) (6) Initial GDP per capita –0.276*** –0.873*** –0.634*** –0.961*** –0.839*** –0.893*** [0.080] [0.070] [0.073] [0.093] [0.070] [0.082] Labor force education 1.418*** 1.887*** 1.687*** 1.729*** 1.623*** 1.617*** [0.126] [0.124] [0.102] [0.125] [0.110] [0.126] Trade openness 1.656*** 2.088*** 1.804*** 1.257*** 1.522*** 1.501*** [0.126] [0.149] [0.155] [0.165] [0.148] [0.137] Share of trade with: Top three partners –6.946*** [0.738] Top three most central countries in the global 4.371*** trade network [0.568] Core countries 13.819*** 8.887*** 8.836*** 10.269*** 9.209*** [1.199] [1.526] [1.434] [1.987] [2.181] Inner-periphery countries 15.678*** 6.816*** 5.625*** 10.252*** 8.691*** [1.263] [1.565] [1.583] [1.833] [2.218] Growth of core countries (trade-weighted average) 0.273*** [0.035] Growth of inner-periphery countries 0.881*** (trade-weighted average) [0.028] Participation in GVCs (share of total trade) 8.595*** 6.637*** 6.330*** [0.830] [0.948] [0.880] Participation in GVCs: Intermediate goods traded with core countries –1.166*** (as share of GVC participation with core countries) [0.236] Intermediate goods traded with inner-periphery 1.937*** countries (as share of GVC participation with [0.354] inner-periphery countries) Final goods traded with core countries (as share 1.775*** of GVC participation with core countries) [0.306] Final goods traded with inner-periphery coun- –0.470** tries (as share of GVC participation with inner-pe- [0.229] riphery countries) Number of observations 809 809 809 744 744 744 Number of countries 114 114 114 113 113 113 Source: Didier and Pinat 2015. Note: This table reports the regressions of real gross domestic product (GDP) per capita growth on a number of indicators capturing the composition of trading partners. See text and annex table 2A.1 for details on indicators used. Robust standard errors are shown in brackets. GVC = global value chain. All regressions include time dummies. Significance level: * = 10 percent, ** = 5 percent, *** = 1 percent. THE STRUCTURE OF TRADE LINKAGES AND ECONOMIC GROWTH 105 countries (from its sample mean) is associ- show that there is actually some heterogene- ated with an increase in growth of about 0.8 ity in these results depending on the composi- percentage points for the average country, tion of partners in the production chain. The whereas the effect reaches almost 1.2 per- growth effects associated with participation centage points for a similar increase in the in GVCs with inner-periphery countries are share of trade with countries in the inner largest in the middle and initial stages. In periphery. contrast, for participation in GVCs with core This perhaps counterintuitive result is countries, the growth effects associated with explained, at least in part, by the differential participation in the final stages of the chain growth rates of inner-periphery countries. are greatest. If these countries typically grow faster than There is a strong nonlinearity in the total countries at the core of the global trade net- growth effects associated with increases in work, trading with them is more likely to trade shares with these central countries be accompanied by larger growth effects— on trade openness and the human capital associated, for instance, with direct aggregate development, as shown in panels e and f demand effects. Indeed, the weighted-growth of figure 2.9. These growth effects are not rates of core and inner-periphery countries only positive but actually increasing with have a positive impact on growth of per trade openness, albeit at different degrees. capita GDP (column 3 of table 2.2). When At lower levels of trade openness (below this growth differential is controlled for, 80 percent), an increase in trade shares with the effects associated with the share of trade inner-periphery countries is associated with with core countries become larger than the slightly larger (though not statistically sig- effects associated with the share of trade with nificant) growth effects than an increase inner-periphery countries—and the growth in the share of trade with core countries. differential is statistically significant. The opposite is observed for higher levels These results also reflect greater insertion of trade openness. Similar nonlinear pat- into GVCs with inner-periphery countries. terns are observed for the relation between The degree and manner in which countries the degree of labor force education and participate in GVCs affects the dynamics of the total growth effects accompanying an trade and growth. To the extent that coun- increase in the share of trade with core tries participate more in GVCs with inner-pe- countries in the global trade network. The riphery (rather than core) countries, part of differential in growth effects associated the growth differential actually reflects this with increases in the share of trade with insertion in GVCs. core and inner-periphery countries increases The regression in column 4 of table 2.2 with both the degree of trade openness and explores this possibility. Consistent with the labor force education. The higher the level results in the previous section, participation of trade openness and the greater the degree in GVCs is positively associated with growth of labor force education, the larger are the prospects. When this participation is con- growth effects associated with an increase trolled for, the growth effect associated to the in the share of trade with core countries rel- share of trade with inner-periphery countries ative to inner-periphery countries. is smaller than the effect associated with the Overall, the estimation results indicate share of trade with core countries—and the that for sufficiently integrated countries, an positive growth differential is statistically increase in trade links with countries at the significant. center of the global trade network is accom- The findings in the previous section also panied by strong growth in income per indicate that insertion into the middle seg- capita, even after controlling for the over- ments of a GVC is associated with the larg- all volume of trade flows and a country’s est improvement in the trade-growth nexus. trade share with its main trading partners. The results in columns 5 and 6 of table 2.2 Furthermore, the results are indicative of a 106 LATIN AMERICA AND THE RISING SOUTH differential impact on growth for different that may affect the linkages between trade levels of openness. They suggest some form integration and economic growth. of complementarity between trade open- ness and the share of trade with the central The quality of transport networks countries in the global trade network. They also indicate that countries need to have edu- The ability of economies to integrate effi- cated labor forces to be able to benefit most ciently into the global economy depends to a from trading with core countries, suggesting great extent on the quality of hard and soft that human capital development is key for infrastructure services, ranging from trans- the absorption of foreign technology and portation, telecommunications, and finan- knowledge. These results are consistent with cial services to border processes and customs the idea that the growth effects associated practices to the business and regulatory envi- with trade openness are not related simply ronments. 43, 44, 45 In fact, internal (domestic) to the development of strong trade ties with trade and transaction costs can have a large a single country but rather to the establish- impact on a country’s external (international) ment of such ties with countries that are competitiveness. The extent of red tape and more exposed to the frontiers of ideas and access to efficient transport networks fea- technologies. ture prominently among the cost factors that The results in this section may interact determine whether firms can meet external with and complement the results of the pre- demand in a competitive and timely fashion. vious section, which characterized the inter- The quality of transport infrastructure is actions between growth and the nature of increasingly perceived as a determinant of traded goods. The results on participation in participation in GVCs. This measure includes GVCs and the composition of trading part- not only the existence of physical assets but ners provide only a glimpse of these potential also the efficiency and availability of trans- interactions, because the S-GMM proce- port services, such as trucking and transpor- dure is limited to a relatively restricted set of tation, storage and packaging facilities, and explanatory variables in the estimated regres- consolidation centers. sions if overfitting bias is to be avoided (see The World Bank’s Doing Business data- box 2.1). This methodology constrains a more base captures the internal costs associated thorough analysis of these interactions, which with shipping goods from the factory gate to is therefore left for future research. ports (for exports) and from ports to retail outlets (for imports) through its “cost of trad- ing” index. This indicator measures the fees Potential frictions affecting (excluding tariffs and trade taxes) associated trade and growth dynamics with exporting and importing a standardized A variety of factors could act as barriers to cargo of goods by sea transport, accounting the efficient allocation of resources within for the time and cost necessary to comply and across countries and thus affect trade with every official procedure (the time and and growth dynamics. Distortive govern- cost for sea transport itself are not included) ment policies, such as policies embedded in (Djankov, Freund, and Pham 2010). trade agreements or direct trade barriers, The results show that on average, it is more could encourage the inefficient growth of a expensive to export and import in the South specific sector or change the mix of a coun- than in the North (East Asian economies try’s exports. High trading costs associated are a marked exception) (figure 2.10, panel with the transport of goods or clearance at a). On average, LAC countries are well posi- the border could also play a role. This sec- tioned with respect to other South economies, tion focuses on two sets of frictions that are with internal costs associated with cross-bor- particularly important for LAC countries and der trading lower than in all regions except THE STRUCTURE OF TRADE LINKAGES AND ECONOMIC GROWTH 107 MENA and EAP. There is, however, great FIGURE 2.10 Average cost of trading in 2013 heterogeneity within LAC (figure 2.10, panel a. Average cost of trading in the North and the South b). Panama is the least expensive country North (ranked 38th worldwide), followed by Peru EAP (52nd) and Chile (53rd). At the other extreme, MENA among the most expensive countries in the LAC world for trade are República Bolivariana de South Asia Venezuela (175th), Colombia (162nd), and ECA Brazil (156th). SSA Access to efficient and competitive inter- 0 500 1,000 1,500 2,000 2,500 3,000 3,500 4,000 national transport networks is also crucial US$ per container for integration into global markets. The b. Average cost of trading in Latin America and the Caribbean availability of effective transport connec- tions, including ancillary services, affects the Panama location decisions of production. Trade in Peru intermediate goods is especially sensitive to Chile transport costs (World Bank 2009). Trans- El Salvador St. Vincent and portation infrastructure may also play a role the Grenadines in facilitating knowledge diffusion and spill- Costa Rica Trinidad and Tobago overs (Agrawal, Galasso, and Oettl 2014). Suriname The relatively poor quality of transport Dominican Republic networks in LAC countries seems to act as a Nicaragua trade barrier, constraining the ability of econ- Barbados omies in the region to integrate efficiently Uruguay into the global economy. On average, LAC Dominica countries seem to underperform both North Antigua and Barbuda countries and some other South countries on Haiti a range of indicators capturing accessibility to Bahamas, The and the quality of transport networks. There Honduras is some evidence that the region is not spend- Belize ing sufficiently or effectively on infrastruc- Guatemala ture, even though infrastructure development Ecuador offers significant potential to speed the pace Bolivia of growth in the region (Calderón and Servén Mexico 2010; Fay and Morrison 2007). There is wide Santa Lucia heterogeneity within the region, however. Grenada Jamaica Land transport Detailed data on the value Argentina of trade by different modes of transportation Paraguay are sparse, but data on the United States and Brazil LAC indicate that trade with land neighbors Colombia occurs mostly by surface modes (such as Venezuela, RB truck, rail, and pipeline); only 10 percent of 0 500 1,000 1,500 2,000 2,500 3,000 3,500 4,000 trade takes place by air or ocean (Hummels US$ per container 2007). About 10–20 percent of total trade by Source: Calculations based on the World Bank Doing Business Indicators. LAC countries is with land neighbors. The Note: The cost of trading is measured by the average cost associated with exporting and importing a standardized cargo of goods by sea transport. It is measured by fees (in U.S. dollars) levied on a 20-foot development of the land transport network container (excluding tariffs). Fees cover costs associated with completing all procedures required to is therefore an important factor behind intra- export or import goods. For exporting goods, procedures range from packing the goods into the container at the warehouse to their departure from the port of exit. For importing goods, procedures regional integration. range from the vessel’s arrival at the port of entry to the cargo’s delivery at the warehouse. For land- locked economies, these include procedures at the inland border post, since the port is located in the transit economy. The North includes the G-7 members and other Western Europe countries; the South includes all other economies. EAP = East Asia and Pacific; ECA = Europe and Central Asia; LAC = Latin America and the Caribbean; MENA = Middle East and North Africa; SSA = Sub-Saharan Africa. 108 LATIN AMERICA AND THE RISING SOUTH Data on road and railway density reveal are effectively connected to markets or trade a gap between North and South countries. outlets. Adjusted by population density, these mea- Data on the quality of land transport sures indicate that LAC lags behind North infrastructure suggest some scope for countries, though the evidence is more improvement in LAC. The quality of the road nuanced with respect to other South regions. network, proxied by the share of unpaved (figure 2.11, panel a).46 On average, LAC roads, is relatively poor when contrasted with outperforms MENA and South Asia in both other South regions: almost 70 percent of the road and rail density and performs about the roads in LAC are unpaved—a far larger share same as SSA. LAC has denser railway net- than in EAP and MENA (less than 30 per- works but sparser road coverage than EAP. cent) and South Asia (less than 50 percent) A caveat of this analysis is that measures of (see figure 2.11, panel b). LAC also seems to road and railway density are imperfect indi- lag behind in the quality of its railway net- cators of the quantity of transport services, work. Panama is the highest-ranked LAC especially services relevant for the develop- country in terms of the quality of its railroad ment of cross-border linkages, because they infrastructure (ranked 30th in the Global do not indicate whether production centers Competitive Forum Index); no other LAC FIGURE 2.11 Land transportation, by region, 2011 a. Population density-adjusted measure of b. Composition of road transportation density of land transportation 80 5 100 4 90 60 3 80 40 2 70 Railway density (residuals) Road density (residuals) 20 1 60 Percent 0 0 50 -1 40 -20 -2 30 -40 -3 20 -60 -4 10 -80 -5 0 North EAP ECA LAC MENA South SSA North EAP ECA LAC MENA South SSA Asia Asia Roads Railways (right axis) Paved road Non-paved road Sources: Calculations based on WDI. Note: Panel a reports residuals of regressions of measures of density of land transportation (road density and railway density) against population density at the country level. Cross-country averages are reported. Density of land transportation is measured by the number of kilometers of roads or rails per 100 squared kilometers of land area. Rail lines are the length of railway route available for train service, irrespective of the number of parallel tracks. Paved roads are roads surfaced with crushed stone (macadam) and hydrocarbon binder or bituminized agents, with concrete or cobblestones. All other roads are considered unpaved. The North includes the G-7 members and other Western Europe countries. Singapore and Hong Kong SAR, China, are excluded from the EAP average because of the physical characteristics of these economies. EAP = East Asia and Pacific; ECA = Europe and Central Asia; LAC = Latin America and the Caribbean; MENA = Middle East and North Africa; SSA = Sub-Saharan Africa. THE STRUCTURE OF TRADE LINKAGES AND ECONOMIC GROWTH 109 country features in the top 50. Moreover, 10 routes are much more developed than others of the world’s 20 worst performers, including as most shipping companies adopt a hub-and- Brazil, Colombia, and Peru, are in LAC. spoke operating structure. This operating structure consists of hub ports, lateral ports, Maritime transport For trade with non- main lines (long haul lines that connect hub neighboring countries, which corresponds ports and involve a set of sequential port calls to about 80 percent of world trade by value, typically across the oceans), and branch lines nearly all goods trade moves by ocean and air (short-haul lines connecting several lateral (Hummels 2007). Most manufactured and ports in one region to serve the main lines), semimanufactured goods are transported in which together form a complex transporta- liner vessels, as are bulk commodities like oil tion network system (Rodrigue and Comtois and petroleum products, iron ore, coal, and 2006; Ducret and Notteboom 2012). This grains. The international shipping industry hub-and-spoke arrangement has led to an carries about 90 percent of world trade in unbalanced geographical distribution of hub terms of volume, according to the Maritime ports around the world, with most of them International Secretariat Services (2013). The located in Asia and Europe (Hu and Zhu quality of maritime shipping services is thus 2009). Ports in Hong Kong SAR, China; Sin- an important determinant of competitiveness. gapore; and Rotterdam (the Netherlands) are It directly affects countries’ engagement in central hubs in the global network. Panama global trade and indirectly increases per capita and Kingston (Jamaica) are hubs in LAC. income. A map of marine traffic for cargo ships The use of maritime transportation is not during the second half of 2013 shows this homogeneous across countries. Some freight heterogeneity (figure 2.12). The highest FIGURE 2.12 Ship and port activity, second half of 2013 Source: © marinetraffic.com. Used with permission. Further permission required for reuse. Note: The map follows a red-yellow-green scheme from high to low to show the intensity of marine traffic and port activity. Passenger as well as cargo vessels and tankers are consid- ered in this map. 110 LATIN AMERICA AND THE RISING SOUTH FIGURE 2.13 Liner shipping connectivity index in selected intensity of marine traffic is in Europe, the countries, 2013 United States, and the Pacific coast of Asia. Hong Kong SAR, China China Traffic along Latin American coasts is sig- Singapore Korea, Rep. nificantly less dense. Malaysia United States Germany Data from the World Shipping Council United Kingdom Netherlands (n.d.) confirm that LAC countries are not at Belgium France the center of the world’s main shipping routes. Spain United Arab Emirates Italy In 2012 only 3 million 20-foot equivalent Japan Taiwan, China units (TEUs, a standard measure of container Saudi Arabia Egypt, Arab Rep. ship capacity) were shipped between Asia and Morocco Turkey Malta South America, the most active route for LAC Oman Portugal countries. This volume is a fraction of the Greece Panama 22 million TEUs shipped along the main trad- India Vietnam Lebanon ing route between Asia and North America. South Africa Sri Lanka LAC countries have accessibility to this Sweden Mexico Denmark global network, through its branch lines. A Canada Thailand proxy for the ease of access to high-capacity Russian Federation Poland and high-frequency global maritime freight Colombia Brazil Argentina transport systems is the Liner Shipping Con- Chile Peru nectivity Index (figure 2.13).47 In 2013 the Israel Uruguay export-oriented economies of East Asia took Australia Pakistan Indonesia the top five spots: China and Hong Kong Ukraine Bahamas, The SAR, China, were the highest-ranking econ- Romania Dominican Republic omies, followed by the transshipment hub of Jamaica Mauritius Jordan Singapore. North countries, including Bel- Ecuador Nigeria gium, Germany, Japan, the Netherlands, the Iran, Islamic Rep. Slovenia United Kingdom, and the United States, took Croatia Djibouti Guatemala most of the other top 15 spots. Ghana Yemen, Rep. Within LAC only Panama features in the New Zealand Venezuela, RB top 30 (at 25th). Mexico is the second-highest Philippines Bahrain Côte d'Ivoire ranking country in the region (32nd), fol- Trinidad and Tobago Syrian Arab Republic lowed by Colombia (38th) and Brazil (39th). Cyprus Bermuda In general, Central America and Caribbean Congo Namibia Benin countries typically reveal more restricted Togo Costa Rica use of the liner shipping network than South Angola Ireland Fiji American countries. Adjusting the index for Madagascar Kenya country size (proxied by population and land Tanzania Senegal area) does not improve the rankings of LAC Cameroon Honduras Belize countries—the top countries in the region Mozambique French Polynesia actually move significantly down: Mex- Finland New Caledonia Latin America and the Caribbean ico falls to 80th place, Brazil to 76th, and Gabon Sudan Rest of the world Colombia to 86th. The top three East Asian El Salvador Nicaragua Curaçao economies remain at the top of the ranking. Maldives Seychelles The spatial design of the maritime trans- Guinea Bangladesh port network reflects an equilibrium outcome 0 20 40 60 80 100 120 140 160 180 200 Index in which both demand and supply effects Source: Calculations based on UNCTAD data. Underlying data come from Containerization Interna- are at play. Demand factors include demand tional Online. for containerized transport and demand for Note: Index is based on five components of the maritime transport sector: the number of ships, their container-carrying capacity, the maximum vessel size, the number of services, and the number of specific transport service characteristics. companies that deploy container ships in a country’s ports. The highest value (100) represents the Central to supply-side considerations are the value for the country with the highest average index in 2004. All reported values are relative to this country-year observation. Only the top 100 countries are reported. THE STRUCTURE OF TRADE LINKAGES AND ECONOMIC GROWTH 111 strategies of container shipping liners, which FIGURE 2.14 Share of world air freight transport by selected aim to maximize profits and take advantage countries, 2013 of increasing economies of scale through the United States 21% strategic choice of market coverage (the hub- China and-spoke operating structure is particularly United Arab Emirates Korea, Rep. important in this regard). Other important Hong Kong SAR, China factors are port infrastructure, port system Japan development, and internal transport and Germany logistics infrastructure in the hinterland for Singapore port access (see, for example, Notteboom United Kingdom 2009). Netherlands These factors may be a constraint in many Luxembourg LAC countries, where port performance is Qatar France typically poor, although there is wide hetero- Russian Federation geneity within the region.48 Panama is one Australia of the top 10 countries in the world in port Thailand efficiency, but Bolivia (ranked 142nd), Brazil Turkey (131st), and Costa Rica (128th) are among Malaysia the least efficient, according to the Global Canada Competitiveness Report 2013–14 (Schwab Saudi Arabia Brazil and Sala-i-Martin 2013). The determinants India of port efficiency include excessive regu- Belgium lation, the prevalence of organized crime, Chile congestion, and the general condition of the Switzerland country’s infrastructure. South Africa New Zealand Air transport Although the global air cargo Indonesia Spain industry is still relatively small compared Italy with the maritime shipping industry, it has Colombia become a viable alternative for high-value Israel and low-volume as well as time-sensitive Ethiopia products. A growing emphasis on speed in Mexico cross-border shipments—which has accom- Finland panied the expansion of just-in-time business Philippines Vietnam models—highlights the increased importance Sri Lanka of air freight transport.49 Egypt, Arab Rep. Global air cargo grew significantly Austria between 1990 and 2013, more than doubling Portugal in volume, from 56 billion ton km to almost Bahrain 175 billion ton km.50 According to the Inter- Pakistan national Air Transport Association (IATA), Oman Kuwait Latin America and the Caribbean 35 percent of world merchandise trade in Rest of the world Kenya value was transported by air in 2013. Jordan Air traffic is concentrated in North Argentina countries, which accounted for almost 0 1 2 3 4 5 6 7 8 9 10 50 percent of all air freight transport in Share of world air freight transport (%) 2013 (figure 2.14). Within the South, EAP Source: World Bank Doing Business Indicators. (20 percent of world air freight) and MENA Note: Air freight is measured by the volume of freight, express, and diplomatic bags carried at each (13 percent) captured the largest shares of flight stage (operation of an aircraft from takeoff to next landing), measured in metric tons times kilometers traveled. Only countries with at least 0.01 percent of world air transport are reported. world air freight. LAC accounted for just 112 LATIN AMERICA AND THE RISING SOUTH 3 percent. Brazil (21st), Chile (24th), and or were recently signed. The expansion of Colombia (31st) were the highest-ranked cross-regional RTAs may reflect the fact countries in the region (although once coun- that many prospects for agreements within try size, proxied by population and land area, regions have already been exploited (Fioren- is controlled for, these countries drop signifi- tino, Touqueboeuf, and Verdeja 2007). cantly in the rankings). The increase in the number of RTAs has Like the maritime transport network, the produced overlapping membership. Coun- air transport network is characterized by a tries typically negotiate different trading hub-and-spoke structure. This structure may terms in each RTA with every country (or explain at least in part the geographical het- group of countries), each agreement with its erogeneity in the concentration of air traffic. own loopholes, exceptions, and regulations. In turn, those countries negotiate their own agreements and exceptions with others, turn- Emerging asymmetry in the quality ing the geographical representation of these of trade agreements agreements into a “spaghetti bowl” (Bhag- Like the quality of transport networks, trade wati 1995, 2008). policies can hamper economic ties. They may The coexistence in a single country of dif- be thought of as cost factors (or subsidies) ferent trade rules applying to different RTA that affect the way in which countries inte- partners is a common feature of the global grate in global markets. economy. These multiple rules of origin, stan- The rise of the South in international trade dards, and trade rules in overlapping RTAs has brought significant changes in commercial raise transactions costs for business. Dealing policies around the world. In particular, the with this multiplicity of rules may be partic- number of regional trade agreements (RTAs) ularly problematic for small and medium-size has multiplied, especially among South coun- enterprises. tries, arguably at the expense of multilateral Countries in LAC have undergone sus- agreements under the World Trade Organi- tained episodes of trade liberalization since zation (WTO).51, 52 According to the WTO’s the early 1990s, when the region began a Regional Trade Agreements Information Sys- process of unilateral, multilateral, and pref- tem (RTA-IS), more than 260 RTAs were in erential trade reforms. Many countries have force in 2013, up from less than 20 in 1990. maintained relatively open trade regimes, This growth in RTAs can be traced to several particularly over the first decade of the factors, including the development of GVCs, 2000s. Although there is substantial hetero- geopolitical considerations, the rise in protec- geneity in the institutional commitments in tionist tendencies, the need to reduce trade the region, under the aegis first of the Gen- and investment barriers, and the slow prog- eral Agreement on Trade and Tariffs (GATT) ress in the WTO Doha Round of trade negoti- and later of the W TO, most countries ations after more than a decade.53 bounded their import tariffs. 54 Moreover, Regionalism was an important early fea- countries applied tariffs that were well below ture of this drive to expand and deepen the bounded levels. In addition, numerous economic integration, which began in the LAC countries gained market access for their mid-1980s. Efforts started in the United exports and agreed to follow certain com- States and Europe, but groups of South mercial policy disciplines embodied in RTAs. countries around the world established and The policy issues raised by these trade strengthened their own regional groupings reforms are varied and complex; there is little (WTO 2011). consensus on the effects of the proliferation During the 2000s, regionalism declined of this heterogeneous set of discriminatory and a trend toward a broader geographi- trade agreements on world trade or economic cal scope of RTAs began developing, espe- growth.55 One issue, discussed in box 2.3, is cially for RTAs that are under negotiation how LAC countries have managed their trade THE STRUCTURE OF TRADE LINKAGES AND ECONOMIC GROWTH 113 BOX 2.3 Asymmetry in the use of temporary trade barriers The increasingly complicated and overlapping array on Trade and Tariffs (GATT)/ WTO multilateral of multilateral agreements (including the World trading system; governments may use them to imple- Trade Organization [WTO] membership) and pref- ment new import restrictions in response to certain erential trade agreements may have constrained the types of economic shocks.60 conventional use of trade policy instruments, such Although each individual act of import protec- as import tariffs, to respond to political-economic tion under TTB policies may be relatively small shocks. Policymakers in Latin America and the in scale—in that it may affect only a small group Caribbean (LAC) have been pressed to adjust to the of imported products or a targeted set of trading changing patterns of exposure to external economic partners—the cumulative use of such policies can influences. LAC economies have faced the emer- become economically meaningful. Indeed, Bown gence of large South economies, especially China, (2011) shows that for the largest countries in LAC with their hands tied, especially as political pres- (including Brazil and Mexico), the trade coverage sures emerged to protect certain industries. and frequency of use of these policies increasingly Countries in the region have not responded to rivals the United States and countries in the Euro- negative economic shocks with major trade pol- pean Union. a Those countries have a much longer icy reversals, at least not major increases in applied history of applying low rates of tariff protection import tariffs. This response stands in contrast to overall and using TTB policies to manage their trade earlier periods, when negative external shocks led policy in order to ensure a continued level of relative to increases in import protection. 58 For most LAC openness in the face of political-economic shocks. countries, especially Brazil, Chile, Mexico, and Peru, A key feature of TTBs is that they have a great the applied most-favored-nation (MFN) import tar- capacity to be imposed in bilateral, or at least more iffs (the nondiscriminatory tariffs that are offered to targeted, ways. Indeed, LAC countries have used all WTO member countries with which a country TTBs in an asymmetric way, targeting other South does not have a preferential relationship) were higher countries more than North countries. In fact, with in the early 1990s than they were in 2010. the exception of Chile, they have disproportionately How have LAC countries managed their trade pol- targeted imports from China with TTBs, though the icy in light of these changes and constraints? Simply use of these measures declined in the second half because national tariff policies may no longer react of the 2000s (figure B2.3.1). At its peak in 1993, countercyclically does not imply that trade policy is Mexico imposed TTBs on nearly 45 percent of its no longer responsive to transitory economic shocks. imports from China.b At their peaks, Peru imposed A detailed study prepared for this report that TTBs on 20 percent, Argentina and Brazil on almost draws on a new database reveals that LAC coun- 13 percent, and Colombia on almost 8 percent of tries are still making frequent changes to their trade imports from China. Even as late as 2011, Argentina policies and that some of them have been conse- imposed TTBs on nearly 7 percent, Brazil on more quential in the aggregate (Bown 2014). These trade than 4 percent, and Colombia and Peru on more policy changes encompass a relatively new (for these than 3 percent of imports from China. Although the countries) set of policy instruments that are not typ- number of such measures imposed on China appears ically captured by classical measures of tariff pro- to have been disproportionately high, the volume of tection. In particular, since 1990 LAC countries trade affected by TTBs appears to have been low: have increasingly adopted temporary trade barrier LAC-imposed TTBs affected only 1.9 percent of (TTB) policies (a term coined by Bown 2012)— China’s exports to LAC countries in 2012.c such as antidumping, countervailing, and safeguard These patterns are not specific to LAC. The use duties—as instruments of protection. 59 TTBs are of TTBs has proliferated across South countries. applied as additional import protection above the More South countries had a significant share of their existing applied tariffs that would otherwise be due imports covered by TTBs in 2012 than in 1998. on imported products; they are often applied at rates Notable examples of non-LAC South countries that exceeding 100 percent. These policy instruments are increased their use of TTBs between 1990 and 2012 permitted under the rules of the General Agreement are China, India, Indonesia, and Turkey. South (continued) 114 LATIN AMERICA AND THE RISING SOUTH BOX 2.3 Asymmetry in the use of temporary trade barriers (continued) FIGURE B2.3.1 Foreign targets of temporary trade barriers imposed by selected countries in Latin America and the Caribbean a. Argentina b. Brazil 14 14 Share of trade-weighted imports (%) Share of trade-weighted imports (%) 12 12 10 10 8 8 6 6 4 4 2 2 0 0 1988 1992 1996 2000 2004 2008 2011 1988 1992 1996 2000 2004 2008 2011 c. Chile d. Colombia 14 14 Share of trade-weighted imports (%) Share of trade-weighted imports (%) 12 12 10 10 8 8 6 6 4 4 2 2 0 0 1988 1992 1996 2000 2004 2008 2011 1988 1992 1996 2000 2004 2008 2011 e. Mexico f. Peru 14 14 Share of trade-weighted imports (%) Share of trade-weighted imports (%) 12 12 10 10 8 8 6 6 4 4 2 2 0 0 1988 1992 1996 2000 2004 2008 2011 1988 1992 1996 2000 2004 2008 2011 High-income countries’ exports under any TTB Chinas’ exports under any TTB Other emerging countries’ exports under any TTB Source: Bown 2014. Note: Temporary trade barriers include antidumping measures, countervailing duties, global safeguards, and China-specific transitional safeguards. TTB = temporary trade barrier. (continued) THE STRUCTURE OF TRADE LINKAGES AND ECONOMIC GROWTH 115 BOX 2.3 Asymmetry in the use of temporary trade barriers (continued) countries, including China and the Russian Feder- arising from the implementation of trade agreement ation, are also the major targets of TTBs, especially commitments since the early 1990s. In particular, from other South countries. LAC countries are an declines in domestic economic growth or increases exception to these trends, at least partially because in domestic unemployment rates, real exchange rate of the nature of their exports (commodities are not appreciations, and surges in bilateral import growth frequently targeted with TTBs; the most targeted are associated with subsequent increases in import industries are steel, chemicals, and textiles/apparel). protection through these policy instruments. These Bown (2014) provides strong evidence that for a patterns suggest not only that countercyclical wide set of countries in LAC, increases in import import protection is still in use in LAC but also that protection through TTBs are associated with tran- the rise of the South in the global economy has had sitory aggregate shocks, even after controlling for a significant impact on the use of trade policies in important changes in the institutional environment the region. a. The study covers 11 LAC countries: Argentina, Brazil, Chile, Colombia, Costa Rica, the Dominican Republic, Ecuador, Jamaica, Mexico, Panama, and Peru. b. Beginning in 1993, Mexico used antidumping measures against China to protect more than 20 percent of its imported product lines. These import restrictions remained in place until 2008, when they were removed. This share is smaller on a trade-weighted basis, because Mexico applied these import restrictions before it had significant imports of these products from China. c. TTB statistics were constructed using the methodological approaches described in Bown (2011, 2013) applied to updated data provided in Bown (2012). policy against the backdrop of these changes North-South RTAs in Asia typically eliminate in the nature of trade agreements. The box tariffs much more rapidly than South-South suggests that there has been an asymmetry in RTAs. For example, about 55 percent of the use of trade barriers imposed on North North-South RTAs eliminated tariffs on trade and South countries. in virtually all goods within two to five years Have the RTAs negotiated by LAC coun- of their entry into force. In contrast, only 23 tries been comprehensive enough to cover percent of South-South RTAs did so. The issues that most analysts believe are directly majority of RTAs between EAP and LAC (15 related to commercial policy disciplines and of 22) in effect as of 2013 included relatively the international diffusion of knowledge and fast liberalization processes. technology? In a background paper for this Second, North-South RTAs are more report, Wignaraja (2014) reviews the qualita- comprehensive than South-South RTAs in tive information of RTAs, examining whether their liberalization of services sectors. About the trade agreements in which LAC and EAP 73 percent of North-South RTAs in Asia are countries have participated are “comprehen- deemed comprehensive in covering at least sive” in the sense of covering aspects that go five key services, and another 18 percent well beyond traditional commercial policies. provide coverage of two to four key sectors. He assesses the scope and depth of RTAs The remaining 9 percent have general provi- in three key areas: the speed and coverage sions for liberalizing services and are in the of tariff liberalization, the number of ser- process of negotiating their services com- vice sectors covered, and the coverage and mitments. North-South RTAs thus seem to depth of “new issues,” such as FDI, intellec- have progressively liberalized the services tual property rights, trade facilitation, and sectors of their participants and provided for competition.56 deeper regulatory cooperation in services. Overall, he finds that North-South RTAs In contrast, South-South RTAs provide far are deeper and more comprehensive than less coverage of services, with 47 percent of South-South RTAs in several ways. 57 First, all such RTAs either excluding or including they generally differ in their tariff schedules: only limited services sector coverage. About 116 LATIN AMERICA AND THE RISING SOUTH 36 percent of South-South RTAs provide more restricted, with only 13 percent cover- some coverage; only 17 percent include com- ing these four areas. EAP-LAC RTAs follow prehensive coverage. the same trend. The prevailing approach of RTAs between EAP and LAC appear to these RTAs to the deeper integration issues be more comprehensive than other South- remains moderate. Some EAP–LAC RTAs South RTAs: about half of these agreements adopt a somewhat cautious approach to lib- provide substantial coverage in services. The eralization of sensitive regulatory barriers key service sectors covered in the majority in areas such as investment, competition, of these RTAs are labor mobility and entry and government procurement, arguably of business persons, which are probably reflecting the influence of domestic business included to promote two-way FDI flows and interests and lobbies as well as geopolitical new business opportunities between the two issues. Six agreements are classified as low- regions. However, some subsectors of busi- depth and 12 as medium-depth RTAs. Only ness, communications, transport, financial four EAP–LAC RTAs are deemed to be of services, tourism, and education services are high depth: the Republic of Korea–Peru FTA excluded from coverage of key obligations, (2011), the Trans-Pacific Strategic Economic such as national treatment, local presence, Partnership Agreement (2006), the Austra- and market access. The LAC countries in lia–Chile FTA (2009), and the Singapore– these RTAs typically exclude from national Costa Rica FTA (2013). These RTAs share treatment subsectors in tourism services, rec- features with the best Asian RTAs, though reational services, and radio and television they may still fall short in some key areas of broadcast services. In contrast, Asian coun- deep integration. tries’ exclusion lists consist mostly of subsec- Evaluation of these agreements suggests tors in business, transport, distribution, and that progress has been made in using RTAs education services. to reduce overall trade and regulatory barri- Last, North-South RTAs in Asia tend to ers between EAP and LAC in particular and favor deeper integration among their mem- countries of the South more broadly, albeit to bers. South-South RTAs lag in this regard, varying degrees. Goods and services are gen- providing only traditional coverage of liber- erally well covered; there has been much less alization of trade in goods and services. More progress in more difficult regulatory issues. than half of North-South RTAs comprehen- More remains to be done to reduce residual sively cover four new areas (investment, com- barriers to trade in goods and services and to petition policy, government procurement, intensify deep integration between EAP and and trade facilitation), and all North-South LAC. A one-size-fits-all good practice tem- RTAs cover at least one area beyond trade plate of RTA provisions is difficult to develop, liberalization. Examples of these deeper but the provisions in the four deep EAP–LAC North-South RTAs include Japan’s bilat- agreements offer insights on good practices eral agreements with Indonesia, Mexico, for future interregional RTAs, particularly on the Philippines, Singapore, and Thailand. important issues such as investment, intellec- The coverage in South-South RTAs is much tual property rights, and competition. THE STRUCTURE OF TRADE LINKAGES AND ECONOMIC GROWTH 117 Annex 2A ANNEX TABLE 2A.1 Data description and sources Variable Description Source Growth in gross domestic Growth rate of GDP per capita based on real GDP per capita PPP Penn World Table 7.1 product (GDP) per capita measured in 2005 constant dollars Initial GDP per capita GDP per capita measured in 2005 constant dollars PPP in first year Penn World Table 7.1 of each five-year period Labor force education Percentage of population older than 15 years that attained sec- Updated database from Barro-Lee ondary or tertiary schooling (2010) Public infrastructure Average number of telephone lines per capita World Development Indicators Terms of trade Ratio of export unit value indexes to import unit value indexes, World Development Indicators measured relative to base year (2000) Trade openness Sum of exports and imports, scaled by GDP Penn World Table 7.1 Trade linkages with North or Sum of exports and imports with North or South countries, scaled Penn World Table 7.1 South countries by GDP Intraindustry trade (IIT) Calculated using the Grubel and Lloyd (1975) methodology; Calculations based on two-digit Stan- degree of IIT ranges from 0 (pure interindustry trade) to 1 (pure dard International Trade Classification intraindustry trade) (SITC) Revision 2 data of Feenstra and others (2005), updated with Com- trade data Classification of traded goods Calculated using the definition of Hinloopen and van Marrewijk Calculations based on three-digit SITC based on factor intensity (2001). Traded goods are classified into five categories: primary Revision 2 data of Feenstra and others products, natural resource–intensive manufactures, unskilled (2005), updated with Comtrade data labor–intensive goods, skilled labor–intensive goods, and high-technology-intensive goods. Shares of traded goods in each category are calculated based on both exports and imports. Classification of traded goods Calculated using the definition of Lall (2000). Traded goods are Calculations based on three-digit SITC based on technology intensity classified into five categories: primary products, natural resource– Revision 2 data of Feenstra and others intensive manufactures, low-technology-intensive goods, medi- (2005), updated with Comtrade data um-technology-intensive goods, and high-technology-intensive goods. Shares of traded goods in each category are calculated based on both exports and imports. Degree of upstreamness of Calculated using the upstreamness measure presented in Antràs Calculations based on four-digit SITC exports and others (2012) for the United States. This measure is applied Revision 2 data of Feenstra and others to the basket of exported goods of every country in the sample. (2005), updated with Comtrade data Goods are divided into three categories based on their degree of upstreamness: beginning of global value chains (GVCs) (exports of primary products), middle of GVCs (exports of intermediate goods), and end of GVCs (exports of final goods). Share of trade with top three Calculated as share of country’s exports and imports with top Calculations based on DOTS main trading partners three trading partners (partners with largest value of bilateral total trade in a given year) Share of trade with three most Calculated as share of a country’s exports and imports with three Calculations based on DOTS central countries in the global most central countries in the global trade network—the coun- trade network tries with the highest values of the random walk betweenness centrality measure developed by Newman (2005) and Fisher and Vega-Redondo (2006). This classification is made for every year in the sample period. Share of trade with core and Calculated as share of a country’s exports and imports with coun- Calculations based on DOTS inner-periphery countries tries in the core and in the inner periphery of the global trade net- work. Core countries are countries ranked in the top 5 percent of the cross-country ranking given by the random walk betweenness centrality measure developed by Newman (2005) and Fisher and Vega-Redondo (2006). Inner-periphery countries are those ranked between percentiles 70 and 95. This classification is made sepa- rately every year in the sample period. (continued) 118 LATIN AMERICA AND THE RISING SOUTH ANNEX TABLE 2A.1 Data description and sources (continued) Variable Description Source Participation in global value Calculated as the ratio of trade in three major GVCs to total trade. Calculations based on Broad Eco- chains (GVCs) The three major GVCs are apparel and footwear, electronics, and nomic Categories (BEC) SITC Revision automobiles and motorcycles and are defined as in Sturgeon and 1 classification Memevodic (2010). Participation in GVCs: Share of The share of intermediate goods traded is calculated as the ratio of Calculations based on BEC SITC Revi- intermediate goods traded and intermediate goods traded to total trade in the three major GVCs. sion 1 classification share of final goods traded Analogously, share of final goods traded is calculated as the ratio of intermediate goods traded to total trade in the three major GVCs. The share of intermediate plus final goods traded sum to 100 percent. The three major GVCs are apparel and footwear, elec- tronics, and automobiles and motorcycles and are defined as in Sturgeon and Memevodic (2010). This ratio is calculated separately for GVC trade with core and inner periphery countries. Notes U-shape relationship between competition 1. Theoretical papers that emphasize the chan- and innovation by considering counteracting nels through which trade affects growth “escape competition” effects versus “Schum- include Arrow (1962); Vernon (1966); Krug- peterian” effects on innovation depending on man (1979); Helpman and Krugman (1985); firm or industry distance to the technological Romer (1990, 1993); Grossman and Help- frontier. man (1991a); Rivera-Batiz and Romer (1991); 6. There is some debate over whether (and under Matsuyama (1992); Eaton and Kortum what conditions) these procompetitive gains (1999); and Hummels and Klenow (2005). from trade are positive. Models with variable 2. Seminal papers on the efficiency gains from markups have yielded contradicting predic- trade include Ricardo (1817), Heckscher tions. For recent discussions of these procom- (1919), and Ohlin (1933). On economies of petitive effects of trade, see, for example, scale and externality, see, for example, Mar- Arkolakis and others (2012) and Edmond, shall (1879, 1890); Caballero and Lyons Midrigan, and Xu (2013). (1990, 1992); Chan, Chen, and Cheung 7. Using a foreign intermediate good in the (1995); and Segoura (1998). On product production of a final output involves the diversity, see, for example, Dixit and Stiglitz implicit usage of the technology of that good (1977), Krugman (1980), Lancaster (1990), in embodied form. There is a spillover in this and Romer (1990). process of international technology diffusion 3. For instance, Armington’s (1969) models to the extent that the intermediate good costs emphasize the intensive margin, whereas less than its opportunity costs, which include monopolistic competition models (for exam- the research and development costs of product ple, Krugman 1981) focus on the extensive development. See, for example, Grossman and margin and vertical differentiation on the Helpman (1991b), Rivera-Batiz and Romer quality margin (for example, Flam and Help- (1991), and Eaton and Kortum (2002). Keller man 1987; Grossman and Helpman 1991b). (2004) provides a survey of the channels 4. See, for example, Fernandes and Paunov through which technologies can diffuse from (2009) for evidence for Chile and Iacovone one country to another. See also Goldberg and and Javorcik (2008) for evidence for Mexico. others (2010) and references therein for a dis- 5. Nickell (1996), Thoenig and Verdier (2003), cussion and some empirical evidence on how Ederington and McCalman (2008), Bus- changes in product mix represent a potentially tos (2011), and Bastos and Straume (2012), important channel through which resources among others, explore the positive effects of are reallocated from less to more efficient uses innovation. Miyagiwa and Ohno (1997), Mat- following trade shocks. subara (2005), and Dhingra (2013), among 8. Grossman and Helpman (1991c) provide a others, discuss the “Schumpeterian” effect. theoretical framework in which knowledge Aghion and others (2005) find an inverted accumulation by domestic industrial agents THE STRUCTURE OF TRADE LINKAGES AND ECONOMIC GROWTH 119 depends on the extent of contact with their case of interindustry externalities. See Harri- foreign counterparts and thus on their levels son and Rodríguez-Clare (2010) for a review of commercial exchange with foreign firms, so of the literature. that the evolutions of comparative advantage 14. Hausmann, Hwang, and Rodrik (2007) and technological progress are interlinked and develop an index of complexity (based on the jointly determined. See also Lucas (1988); basket of goods that higher-income countries Young (1993); Keesing and Lall (1992); Blun- typically export) to rank countries’ export dell, Griffith, and Van Reenen (1995); Piore baskets. They find a statistically significant and Ruiz Durán (1998); Clerides, Lach, and association between complexity and growth: Tybout (1998); Gereffi (1999); and Castellani countries whose export baskets rank high on (2002), among many others. their complexity index tend to grow more 9. See, for example, Easterly and Levine (2001); rapidly. Kose, Prasad, and Terrones (2004); Broner, 15. This point of view questions the tendency Martin, and Ventura (2006); Giovanni and to attribute special growth-enhancing vir- Levchenko (2012); Kose and others (2009); tues to certain type of goods (say, high-tech and Loayza and Raddatz (2007), among manufactures) over others (say, mineral com- many others. modities or services). In fact, Lederman and 10. Papers on trade and income include Irwin and Maloney (2012) provide evidence against the Tervio (2002); Alcalá and Ciccone (2004); natural resource curse. They argue that insti- Rodrik, Subramanian, and Trebbi (2004); Fel- tutions and policies mediate whether natu- bermayr (2005); Noguer and Siscart (2005); ral resources turn into a blessing or a curse. and Dufrénot, Mignon, and Tsangarides When adequate, institutions and policies (2010). Papers on trade and growth include can help maximize the dynamic upsides and Dollar (1992), Edwards (1992), Jones (2000), minimize the dynamic downsides of natural Rodríguez and Rodrik (2001), Wacziarg resources. It is these underlying fundamentals, (2001), Easterly and Levine (2001), Dollar rather than the products themselves, that help and Kraay (2003), and Lee, Ricci, and Rigo- explain the contrast between, say, oil-rich bon (2004). Singh (2010) provides a review of Venezuela, which is trapped in rent-seeking this literature. dynamics, and mineral-rich yet prosperous 11. This result is consistent with the empirical lit- Australia. erature. Studies typically find a lack of statisti- 16. A classification of goods at the two-digit cal significance, or even a negative coefficient, Standard International Trade Classification on the variable capturing the level of human (SITC) industry level is adopted. The IIT mea- capital development (see, for example, De sure based on this broad industry classifica- Gregorio 1992; Benhabib and Spiegel 1994; tion captures the effects of trade of related but Islam 1995; Caselli, Esquivel, and Lefort different goods rather than trade of products 1996; and Pritchett 2000). with some degree of horizontal differentiation, 12. For a broad discussion of the links between which would be captured by a more narrow commodities and economic growth and devel- definition of IIT at the four- or six-digit level. opment in Latin America, see Sinnott, Nash, This broader classification is more indicative and De la Torre (2010). of possible technology diffusion and learning 13. There are several caveats to the externalities spillovers than a narrower one, which can argument. One is that expanding a sector be associated with the love for variety, as in with potential externalities does not neces- Krugman (1979). For example, “optical glass sarily imply that those externalities will auto- and elements of optical glass” and “glass mir- matically occur if the sector is not organized rors, unframed, framed” belong to the same appropriately (Baldwin 1969). Another is two-digit SITC category (industry code 66, that if one country can explore an externality “nonmetallic mineral manufactures”) but are in a good, so can others. If this is the case, not in the same four-digit SITC category (the then the supply of that good will already have former is classified as industry code 6642, the expanded and prices fallen to the point where latter as industry code 6648). the benefit of the externality will have been 17. See, for example, Helpman and Krugman completely offset (Rodríguez-Clare 2010). (1985, 1989); Bernstein and Nadiri (1989); This argument is mitigated somewhat in the and Badinger and Egger (2008). 120 LATIN AMERICA AND THE RISING SOUTH 18. The impact of trade openness and IIT on main activities of individual GVCs. The other income growth reflects the net effects of larger participants in GVCs are supplier companies, markets, competition, technology diffusion which produce goods and services used at dif- and learning spillovers, and volatility, as dis- ferent stages of the production chain. cussed in the previous section. 27. The dynamics of the interplay of power among 19. The underlying regression specification the participants in GVCs determines the allo- includes both simple and quadratic interaction cation of profits and risks along the production terms; the same approach is taken throughout chain. Lead firms, such as large multinational this chapter, as indicated in box 2.1. There- corporations, have greater market power, as a fore, the total growth impact shown in the result of product differentiation and branding. figures in this chapter accounts for the effects The fierce competition across firms (located of both the interaction terms and the open- even in different countries) for a place in GVCs ness variable itself, taking as given the initial may give even more bargaining power to these level of income and the remaining explanatory lead firms, leaving other participants in the variables. chain with little leverage. Of course, the bal- 20. Several theoretical papers—including Ethier ance of bargaining power across participants (1982), Sanyal and Jones (1982), Jones and varies with the specific organization of indi- Kierzkowski (1990), Lüthje (2003), Yi (2003), vidual GVCs. In building these international Burda and Dluhosch (2002), Grossman and production chains, lead firms decide not only Rossi-Hansberg (2008), and Baldwin and about location but also about the governance Robert-Nicoud (2014)—analyze the under- structure of these chains, varying from own- pinnings of the fragmentation of productions. ership (through FDI) to no control through 21. See, for example, Hanson, Mataloni, and arms-length trade or licensing and including Slaughter (2005); Harrison and McMillan everything in between. The bargaining power (2011); and Becker and Muendler (2010). of different parts of the GVC varies with these 22. Baldwin (2012b) argues that since 1985, man- arrangements (see, for example, Gereffi, Hum- agerial and technical know-how have become phrey, and Sturgeon 2005). Timmer and oth- more mobile as offshore stages of production ers (2014) argue that in most GVCs, there is need to seamlessly merge into onshore ones. a strong tendency for value to be added by Hence countries have been able to industrial- capital and high-skilled labor rather than by ize by joining GVCs rather than by building less-skilled labor. They claim that North econ- entire supply chains at home. omies increasingly specialize in activities car- 23. Several papers document this structural break ried out by high-skilled workers. in global trade. See Feenstra (1998); Hum- 28. See, for example, Lall (2000), Humphrey and mels, Ishii, and Yi (2001); Brülhart (2009); Schmitz (2002), and Narula and Dunning Johnson and Noguera (2012); and Koopman, (2010). Wang, and Wei (2014), among many others. 29. See, for example, Forrester (1961); Escaith, 24. Trade in intermediate goods is far from an Lindenberg, and Miroudot (2010); Alessan- ideal measure of GVC participation; it is dria, Kaboski, and Midrigan (2011); and indicative only of participation in GVCs, Altomonte and others (2012). as fragmented production processes require 30. During the 2008 global financial crisis, lead that parts and components cross borders— firms and large intermediaries within GVCs sometimes more than once—before finished provided some support to smaller firms to goods are shipped to final markets. As such, mitigate the impact of the crisis. For example, GVCs can expand without significant growth some retailers and buyers in the apparel sec- in intermediate goods trade, as trade statistics tor offered financial support to their suppliers do not contain information about trade in ser- (Frederick and Gereffi 2011). vices or the ownership of assets. 31. For example, Acer subsidiaries in Taiwan, 25. Changes in the relative prices of intermedi- China, successfully applied knowledge learned ate goods can also affect the ratio. If prices from one part of their production process to of intermediates increase more slowly than supply customers in other markets. In contrast, prices of other goods, the ratio may decrease. very few firms in Mexico have been able to use 26. Lead firms (typically multinational corpora- their links to automotive GVCs to internalize tions) are firms that control and define the technology (UNCTAD 2013). THE STRUCTURE OF TRADE LINKAGES AND ECONOMIC GROWTH 121 32. The literature also discusses the concept of 38. This approach provides a more detailed and social upgrading, which refers to improve- accurate description of countries’ participation ments within a firm in employment condi- in GVCs than the share of intermediate goods tions, including remuneration, worker rights in exports. It is more difficult to use, however, and benefits, and workplace safety. The extent because data are available for a much shorter of social upgrading is tightly linked to the time period (typically only the 1990s and the extent of economic upgrading, but other insti- 2000s) and it requires input-output matrices tutional factors also affect it (see, for example, at the country level. Barrientos, Gereffi, and Rossi 2011). 39. This database is derived from the EORA 33. In food GVCs, for example, large manufactur- global multiregion input-output table (World ers and supermarkets have generally worked MRIO). It uses many data sources, interpo- with a small group of large-scale suppliers lating and estimating missing data points, to that are capable of meeting their stringent provide broad, consistent coverage of value and costly requirements to ensure food safety added trade data for about 180 countries and quality at all stages of the production from 1990 to 2011. For a detailed description chain. Small farms, typically unable to com- of this database, see, for example, Lenzen and ply with the rigorous standards and lack- others (2012, 2013) and UNCTAD (2013). ing the required skills, often find themselves 40. Maloney and Valencia Caicedo (2014) pro- outside these chains (Dolan and Humphrey vide an interesting discussion of innovative 2004; Maertens and Swinnen 2009). Higher capacities based on historical examples con- standards have also spurred participation, trasting the experiences of the United States however, with some firms developing niche and Latin America. markets for organic products, for example 41. Alvarez, Buera, and Lucas (2013) adopt a sim- (Humphrey 2008). ilar concept, in which the flow of ideas is an 34. Relative price effects can partly explain these engine of economic growth. In their model, trends. trade serves as a vehicle for technology dif- 35. For a detailed analysis of China’s upgrading fusion and learning spillovers and hence can strategy, see, for example, Lall and Albaladejo lead to increased economic growth. (2004) and Rodrik (2006). 42. The random walk betweenness centrality 36. These categories of upgrading are important, measure is used to rank countries. This mea- because buyers typically have their own inter- sure is widely used in network analysis and ests to protect and thus are generally inter- has been applied to global trade and financial ested in limiting the upgrading path of their networks. See, for example, Newman (2005); suppliers. In the furniture global value chain, Fisher and Vega-Redondo (2006); and Reyes, for example, large global buyers such as Ikea Garcia, and Lattimore (2009). encourage process upgrading by their suppli- 43. The relevance of the quality of transport net- ers that reduces costs, but they zealously guard works for the trade and growth dynamics the design and branding functions (Kaplinsky, would ideally be assessed with the regression Morris, and Readman 2002). framework adopted throughout this chapter. 37. For example, Costinot, Vogel, and Wang However, lack of data constrains such analy- (2012) develop a model in which the position sis, as most indicators are available only since of workers on production chains affects the the 1990s at best and only for the 2000s in degree of wage inequality. Lopez-Gonzalez most cases. This section therefore presents a and Holmes (2011) provide empirical evidence more qualitative assessment. of a hump-shaped relation between back- 44. The literature provides some evidence that ward supply chains and per capita income. domestic trading costs and the economic busi- As countries get richer, they tend to use more ness environment are significant determinants intensively imported intermediate inputs to of the volume of trade between countries. See, export up to a certain threshold, after which for example, Limao and Venables (2001); Wil- they diminish the imported content of their son, Mann, and Otsuki (2003); Anderson and exports. In contrast, a U-shaped relation is Marcouiller (2002); and Hoekman and Nicita observed for forward supply chains: after a (2011). certain per capita income threshold, countries 45. These infrastructure services, especially data tend to supply more parts. and telecommunication services, may also 122 LATIN AMERICA AND THE RISING SOUTH play a role as enablers and facilitators of and the Caribbean-Canada Trade Agreement knowledge exchange. As transmission costs (CARIBCAN). This section refers to trade have declined and speeds soared over the agreements broadly as RTAs, though it covers past decade, the methods and mechanisms for some PTAs as well. transmitting data and communicating have 53. See, for example, ADB and IDB (2009); WTO also proliferated. Mobile money and mobile (2011); ADB, IDB, and ADBI (2012); and agriculture are examples of mobile technology Kawai and Wignaraja (2013). applications developed and consumed espe- 54. For example, Argentina, Brazil, Paraguay, cially in the South. and Uruguay initially undertook deeper ties 46. Data on the quality of road and railway by creating first a free trade area and then the infrastructure are from the World Economic Mercosur customs union in the early 1990s. Forum’s Global Competitiveness Report Mexico undertook trade liberalization by ini- (2013–14) (Schwab and Sala-i-Martin 2013). tially joining the GATT in 1986 before free- Data on road and railway density are from the ing trade with high-income partners through World Bank’s World Development Indicators. formation of the North American Free Trade 47. The Liner Shipping Connectivity Index (LSCI) Agreement (NAFTA). Chile, Colombia, Peru, captures countries’ level of integration in the and several Central American countries cre- liner shipping network. Liner shipping is typ- ated regional preferential trade areas and ically used for general cargo on fixed trade signed agreements with the United States, routes and on fixed timetables. The higher the among others. index, the easier it is to access a high-capacity 55. See, for example, Cernat (2001); Venables and high-frequency global maritime freight (2003); Carrère (2006); Baier, Bergstrand, transport system. and Vidal (2007); and Baldwin (2008). 48. For discussions of access to liner shipping and 56. The vast majority of the agreements analyzed port infrastructure in LAC, see, for example, in this subsection are FTAs. For a list of the Clark, Dollar, and Micco (2004); Morales Sar- agreements studied, see Wignaraja and Lazaro riera and others (2013); Wilmsmeier (2014); (2010) and Wignaraja (2014). and ECLAC (2014). 57. The definition of North in this subsection is 49. Evans and Harrigan (2005) provide some evi- slightly different from the definition in the dence that the growing importance of speed in rest of the report. North-South RTAs have shipping to final markets has led to a resourc- at least one developed country member, such ing of U.S. imports from Asia to Mexico and as Japan, the United States, the European the Caribbean. Union, Australia, New Zealand, or mem- 50. About two-thirds (in weight) of all air cargo in bers of the European Free Trade Association LAC travels by passenger aircraft. Air cargo sta- (EFTA). tistics may therefore underestimate the impor- 58. Some pressures have arisen through acute tance of air transport for the cross-border flows economic shocks, such as the contagion trig- of goods. gered by the 2008 global financial crisis. Oth- 51. For details on the debate on regionalism and ers have emerged from continued exposure multilateralism, see, for example, Plummer to longer-term trends, such as the sustained (2007), Bhagwati (2008), and WTO (2011). increases in global commodity prices and Chi- 52. According to the WTO, RTAs are recipro- na’s continued export expansion and global cal trade agreements between two or more dominance in manufacturing. partners not necessarily belonging to the 59. These import taxes are in principle imposed same geographical region; preferential trade on a temporary basis to help economies deal agreements (PTAs) involve unilateral trade with import surges (safeguards), import surges preferences. RTAs include free trade agree- associated with cheap imports priced at below ments (FTAs) and customs unions. PTAs cost by foreign exporting firms (antidumping include the European Union’s Generalized duties), or cheap imports associated with sub- Scheme of Preferences, nonreciprocal prefer- sidies by foreign governments (countervailing ential schemes for products from least devel- duties). oped countries only, and other nonreciprocal 60. For example, domestic governments are preferential schemes that have been granted charged with conducting investigations and ver- a waiver by the General Council, such as the ifying evidence of injury on firms of an import- Africa Growth and Opportunity Act (AGOA) competing domestic injury caused by increases THE STRUCTURE OF TRADE LINKAGES AND ECONOMIC GROWTH 123 in imports that have either been sold at a low Production and Trade Flows.” American Eco- price (antidumping), subsidized (countervail- nomic Review 102 (3): 412–16. ing duties), or are simply surging (safeguards). Arellano, M., and O. Bover. 1995. “Another Look at the Instrumental Variable Estimation of Error-Components Models.” Journal of References Econometrics 68 (1): 29–51. Acemoglu, D., S. Johnson, and J. Robinson. 2001. Arkolakis, C., A. Costinot, D. Donaldson, and “The Colonial Origins of Comparative Devel- A. Rodríguez-Clare. 2012. “The Ellusive opment: An Empirical Investigation.” Ameri- Pro-Competitive Effects of Trade.” Working can Economic Review 91 (5): 1369–401. Paper, Department of Economics, MIT, Cam- A DB (Asian Development Bank), and IDB bridge, MA. (Inter-American Development Bank). 2009. Armington, P. 1969. “A Theory of Demand for Comparative Perspectives on Trans-Pacific Products Distinguished by Place of Produc- Trade, Integration, and Development. Tokyo tion.” IMF Staff Papers 16 (1): 159–78. and Washington, DC. Arora, V., and A. Vamvakidis. 2005. “The Impli- ADB (Asian Development Bank), IDB (Inter- cations of South African Economic Growth for American Development Bank), and ADBI the Rest of Africa.” South African Journal of (Asian Development Bank Institute). 2012. Economics 73 (2): 229–42. Shaping the Future of the Asia and the Pacific Arrow, K. 1962. “The Economic Implications and the Latin America and the Caribbean of Learning by Doing.” Review of Economic Relationship. Tokyo and Washington, DC. Studies 29 (3): 155–73. Aghion P., N. Bloom, R. Blundell, R. Griffith, and Badinger, H., and P. Egger. 2008. “Intra- and P. Howitt. 2005. “Competition and Innova- Inter-Industry Productivity Spillovers in tion: An Inverted U-Relationship.” Quarterly OECD Manufacturing: A Spatial Economet- Journal of Economics 120 (2): 701–28. ric Perspective.” CESifo Working Paper 2181, Aghion, P., and R. Griffith. 2008. Competition CESifo Group, Munich. and Growth. Cambridge, MA: MIT Press. Baier, S., J. Bergstrand, and E. Vidal. 2007. “Free Aghion, P., and G. Saint-Paul. 1998. “Virtues of Trade Agreements in the Americas: Are the Bad Times Interaction between Productivity Trade Effects Larger than Anticipated?” World Growth and Economic Fluctuations.” Macro- Economy 30 (9): 1347–77. economic Dynamics 2 (3): 322–44. Baldwin, R. 1969. “The Case against Infant- Agrawal, A., A. Galasso, and A. Oettl. 2014. Industry Tariff Protection.” Journal of Politi- “Roads and Innovation.” Social Science cal Economy 77 (1): 295–305. Research Network Working Paper. ———. 2006 “Multilateralising Regionalism: Alcalá, F., and A. Ciccone. 2004. “Trade and Pro- Spaghetti Bowls as Building Blocs on the Path ductivity.” Quarterly Journal of Economics to Global Free Trade.” World Economy 29 119 (2): 612–45. (11): 1451–518. Alessandria, G., J. P. Kaboski, and V. Midrigan. ———. 2008. “Managing the Noodle Bowl: The 2011. “US Trade and Inventory Dynamics.” Fragility of East Asian Regionalism.” Singa- American Economic Review 101 (3): 303–07. pore Economic Review 53 (3): 449–78. Altomonte, C., F. Di Mauro, G. Ottaviano, A. ———. 2012a. “Global Supply Chains: Why Rungi, and V. Vicard. 2012. “Global Value They Emerged, Why They Matter, and Where Chains during the Great Trade Collapse: A They Are Going.” CEPR Working Paper 9103, Bullwhip Effect?” European Central Bank Center for Economic Policy Research, Wash- Working Paper 1412, Frankfurt. ington, DC. Alvarez, F., F. Buera, and R. Lucas. 2013. “Idea ———. 2012b. “Trade and Industrialisation after Flows, Economic Growth, and Trade.” NBER Globalisation’s Second Unbundling: How Working Paper No. 19667, National Bureau of Building and Joining a Supply Chain Are Dif- Economic Research, Cambridge, MA. ferent and Why It Matters.” In Globalization Anderson, J. E., and D. Marcouiller. 2002. “Inse- in An Age of Crisis: Multilateral Economic curity and the Pattern of Trade: An Empirical Cooperation in the Twenty-First Century, ed. Investigation.” Review of Economics and Sta- Robert Feenstra and Alan Taylor. Chicago: tistics 84: 342–52. University of Chicago Press. Antràs, P., D. Chor, T. Fally, and R. Hillberry. Baldwin, R., and F. Robert-Nicoud. 2014. “Trade- 2012. “Measuring the Upstreamness of in-Goods and Trade-in-Tasks: An Integrating 124 LATIN AMERICA AND THE RISING SOUTH Framework.” Journal of International Eco- 1990 –20 09.” World Economy 34 (12): nomics 92 (1): 51–62. 1955–98. Baldwin, R., and A. Venables. 2013. “Spiders and ———. 2012. Temporary Trade Barriers Data- Snakes: Offshoring and Agglomeration in the base. World Bank, Washington, DC. Global Economy.” Journal of International ———. 2013. “Emerging Economies and the Economics 90 (2): 245–54. Emergence of South-South Protectionism.” Barrientos, S., G. Gereffi, and A. Rossi. 2011. Journal of World Trade 47 (1): 1–44. “Economic and Social Upgrading in Global ———. 2014. “Economic Shocks and Interna- Production Networks: A New Paradigm for tional Agreements: Latin America’s Trade a Changing World.” International Labour Policy in a Multi-Polar World.” World Bank, Review 150 (3–4): 319–40. background paper for this report. Barro, R., and J. W. Lee. 2010. “A New Data Brambilla, I., D. Lederman, and G. Porto. Set of Educational Attainment in the World, 2012. “Exports, Export Destinations, and 1950–2010.” NBER Working Paper 15902, Skills.” American Economic Review 102 (7): National Bureau of Economic Research, Cam- 3406–38. bridge, MA. Broner, F., A. Martin, and J. Ventura. 2006. “Sov- Bastos, P., and O. R. Straume. 2012. “Global- ereign Risk and Secondary Markets.” Ameri- ization, Product Differentiation, and Wage can Economic Review 100 (4): 1523–55. Inequality.” Canadian Journal of Economics Brülhart, M. 2009. “An Account of Global 45 (3): 857–78. Intra-Industry Trade, 1962–2006.” World Beck, T., and R. Levine. 2004. “Stock Markets, Economy 32 (3): 401–59. Banks, and Growth: Panel Evidence.” Journal Burda, M., and B. Dluhosch. 2002. “Cost Com- of Banking and Finance 28 (3): 423–42. petition, Fragmentation, and Globalization.” Beck, T., R. Levine, and N. Loayza. 2000. Review of International Economics 10 (3): “Finance and the Sources of Growth.” Journal 424–41. of Financial Economics 58 (1): 261–300. Bustos, P. 2011. “Trade Liberalization, Exports, Becker, S. O., and M. A. Muendler. 2010. “Margins and Technology Upgrading: Evidence on of Multinational Labor Substitution.” Ameri- the Impact of MERCOSUR on Argentinian can Economic Review 100 (5): 1999–2030. Firms.” American Economic Review 101 (1): Benhabib, J., and M. Spiegel. 1994. “The Role 304–40. of Human Capital in Economic Development Caballero, R., and M. Hammour. 1994. “The Evidence from Aggregate Cross-Country Cleansing Effect of Recessions.” American Data.” Journal of Monetary Economics 34 Economic Review 84 (5): 1350–68. (2): 143–73. Caballero, R., and R. Lyons. 1990. “Internal ver- Bernstein, J. I., and M. I. Nadiri. 1989. “Research sus External Economies in European Industry.” and Development and Intra-Industry Spillovers: European Economic Review 34 (4): 805–26. An Empirical Application of Dynamic Duality.” ———. 1992. “External Effects in US Procyclical Review of Economic Studies 56 (2): 249–67. Productivity.” Journal of Monetary Econom- Bhagwati, J. 1995. “US Trade Policy: The Infat- ics 29 (2): 209–25. uation with FTAs.” Columbia University, Calderón, C., and L. Servén. 2010. “Infrastructure Department of Economics, Discussion Paper in Latin America.” World Bank Policy Research 726, New York. Working Paper 5317, Washington, DC. ———. 2008. Termites in the Trading System: Carrère, C. 2006. “Revisiting the Effects of How Preferential Agreements Undermine Free Regional Trade Agreements on Trade Flows Trade. Oxford: Oxford University Press. with Proper Specification of the Gravity Blundell, R., and S. Bond. 1998. “Initial Condi- Model.” European Economic Review 50 (2): tions and Moment Restrictions in Dynamic 223–47. Panel Data Models.” Journal of Econometrics Caselli, F., G. Esquivel, and F. Lefort. 1996. 87 (1): 115–43. “Reopening the Convergence Debate: A New Blundell, R., R. Griffith, and J. Van Reenen. Look at Cross-Country Growth Empirics.” 1995. “Dynamic Count Data Models of Tech- Journal of Economic Growth 1 (3): 363–89. nological Innovation.” Economic Journal 105 Castellani, D. 2002. “Export Behaviour and (429): 333–44. Productivity Growth: Evidence from Italian Bown, C. 2011. “Taking Stock of Antidump- Manufacturing Firms.” Review of World Eco- ing, Safeguards and Countervailing Duties, nomics 138 (4): 605–28. THE STRUCTURE OF TRADE LINKAGES AND ECONOMIC GROWTH 125 Cattaneo, O., G. Gereffi, and C. Staritz. 2010. Vegetables between Africa and the United Global Value Chains in a Post-Crisis World: A Kingdom.” Environment and Planning 36 (3): Development Perspective. World Bank, Wash- 491–510. ington, DC. Dolan, M., and B. Tomlin. 1980. “First World– Cernat, L. 2001. “Assessing Regional Trade Third World Linkages: External Relations and Arrangements: Are South-South RTAs More Economic Development.” International Orga- Trade Diverting?” Global Economy Quarterly nization 34 (1): 41–63. 2 (3): 235–59. Dollar, D. 1992. “Outward-Oriented Developing Chan, V. L., B. L. Chen, and K. N. Cheung. 1995. Economies Really Do Grow More Rapidly: “External Economies in Taiwanís Manufac- Evidence from 95 LDCs, 1976–1985.” Eco- turing Industries.” Contemporary Economic nomic Development and Cultural Change 40 Policy 13: 118–30. (3): 523–44. Chang, R., L. Kaltani, and N. Loayza. 2009. Dollar, D., and A. Kraay. 2003. “Institutions, “Openness Can Be Good for Growth: The Trade, and Growth.” Journal of Monetary Role of Policy Complementarities.” Journal of Economics 50 (1): 133–62. Development Economics 90 (1): 33–49. ———. 2004. “Trade, Growth, and Poverty.” Clark, X., D. Dollar, and A. Micco. 2004. “Port Economic Journal 114 (493): 22–49. Efficiency, Maritime Transport Costs, and DOTS (Direction of Trade Statistics) (database). Bilateral Trade.” Journal of Development Eco- International Monetary Fund, Washing- nomics 75 (2): 417–50. ton, DC. http://elibrary-data.imf.org/Data Clerides, S. K., S. Lach, and J. R. Tybout. 1998. Explorer.aspx. “Is L earning by Exporting I mportant? Ducret, C., and T. Notteboom. 2012. “The World Micro-Dynamic Evidence from Colombia, Maritime Network of Container Shipping: Mexico, and Morocco.” Quarterly Journal of Spatial Structure and Regional Dynamics.” Economics 113 (3): 903–47. Global Networks 12 (3): 395–423. Costinot, A., J. Vogel, and S. Wang. 2012. “Global Dufrénot, G., V. Mignon, and C. Tsangarides. Supply Chains and Wage Inequality.” Ameri- 2010. “The Trade-Growth Nexus in the can Economic Review 102 (3): 396–401. Developing Countries: A Quantile Regression De Gregorio, J. 1992. “Economic Growth in Approach.” Review of World Economics 146 Latin America.” Journal of Development Eco- (4): 731–61. nomics 39 (1): 59–84. Easterly, W., and R. Levine. 2001. “What Have Dedrick, J., K. L. Kraemer, and G. Linden. 2010. We Learned from a Decade of Empirical “Who Profits from Innovation in Global Value Research on Growth? It’s Not Factor Accu- Chains? A Study of the iPod and Notebook mulation: Stylized Facts and Growth Mod- PCs.” Industrial and Corporate Change 19 (1): els .” World Bank Economic Review 15 (2): 81–116. 177–219. Dhingra, S. 2013. “Trading away Wide Brands for Eaton, J., and S. Kortum. 1999. “International Cheap Brands.” American Economic Review Technology Diffusion: Theory and Measure- 103 (6): 2554–84. ment.” International Economic Review 40 (3): Di Giovanni, J., and A. Levchenko. 2012. “The 537–70. Risk Content of Exports: A Portfolio View ———. 2002. “Technology, Geography, and of International Trade.” NBER International Trade.” Econometrica 70 (5): 1741–79. Seminar on Macroeconomics 8 (1): 97–151. ECLAC (Economic Commission for Latin Amer- Didier, D., and M. Pinat. 2015. “The Nature of ica and the Caribbean). 2014. “Latin America Trade and Growth Linkages.” World Bank, and the Caribbean: Port System Evolution, background paper for this report. 1997–2013.” FAL Bulletin 2 (330). http:// Dixit, A. K., and J. E. Stiglitz. 1977. “Monop- www.cepal.org/Transporte/noticias/bolfall olistic Competition and Optimum Product /6/53166/FAL-330-WEB-ENG.pdf. Diversity.” American Economic Review 67 Ederington, J., and P. McCalman. 2008. “Endog- (3): 297–308. enous Firm Heterogeneity and the Dynamics Djankov, S., C. Freund, and C. Pham. 2010. of Trade Liberalization.” Journal of Interna- “Trading on Time.” Review of Economics and tional Economics 74 (2): 422–40. Statistics 92 (1): 166–73. Edmond, C., V. Midrigan, and D. Y. Xu. 2013. Dolan, C., and J. Humphrey. 2004. “Changing “Competition, Markups, and the Gains Governance Patterns in the Trade in Fresh from International Trade.” NBER Working 126 LATIN AMERICA AND THE RISING SOUTH Paper 18041, National Bureau of Economic Forrester, J. 1961. Industrial Dynamics, vol. 2. Research, Cambridge, MA. Cambridge, MA: MIT Press. Edwards, S. 1992. “Trade Orientation, Distor- Frankel, J. A., and D. Romer. 1999. “Does Trade tions and Growth in Developing Countries.” Cause Growth?” American Economic Review Journal of Development Economics 39 (1): 89 (3): 379–99. 31–57. Frederick, S., and G. Gereffi. 2011. “Upgrad- Escaith, H., N. Lindenberg, and S. Miroudot. ing and Restructuring in the Global Apparel 2010. “International Supply Chains and Trade Value Chain: Why China and Asia Are Out- Elasticity in Times of Global Crisis.” World performing Mexico and Central America.” Trade Organization Staff Working Paper Intern ation al Journ al of Technologic al ERSD-2010-08, Geneva. Learning, Innovation and Development 4 Ethier, W. 1982. “National and International (1–3): 67–95. Returns to Scale in the Modern Theory of Gereffi, G. 1999. “International Trade and Indus- International Trade.” American Economic trial Upgrading in the Apparel Commodity Review 72 (3): 389–405. Chain.” Journal of International Economics Eora MRIO (database). http://worldmrio.com/. 48 (1): 37–70. See Lenzen and others (2012, 2013). ———. 2014. “Global Value Chains in a Post– Evans, C. L., and J. Harrigan. 2005. “Distance, Washington Consensus World.” Review of Time, and Specialization: Lean Retailing in International Political Economy 21 (1): 9–37. General Equilibrium.” American Economic Gereffi, G., J. Humphrey, and T. Sturgeon. 2005. Review 95 (1): 292–313. “The Governance of Global Value Chains.” Fay, M., and M. Morrison. 2007. Infrastructure Review of International Political Economy 12 in Latin America and the Caribbean: Recent (1): 78–104. Developments and Key Challenges. Washing- Gibbon, P. 2008. “Governance, Entry Barriers, ton, DC: World Bank. Upgrading: A Re-Interpretation of Some GVC Feenstra, R. 1998. “Integration of Trade and Dis- Concepts from the Experience of African integration of Production in the Global Econ- Clothing Exports.” Competition and Change omy.” Journal of Economic Perspectives 12 12 (1): 29–48. (4): 31–50. Giovanni, J., and A. Levchenko. 2009. “Trade Feenstra, R., R. Lipsey, H. Deng, A. Ma, and H. Openness and Volatility.” Review of Econom- Mo. 2005. “World Trade Flows: 1962–2000.” ics and Statistics 91 (3): 558–85. Database, NBER w11040, National Bureau of Goldberg, P., A. K. Khandelwal, N. Pavcnik, and Economic Research, Cambridge, MA. P. Topalova. 2010. “Multiproduct Firms and Felbermayr, G. 2005. “Dynamic Panel Data Evi- Product Turnover in the Developing World: dence on the Trade-Income Relation.” Review Evidence from India.” Review of Economics of World Economics 141 (4): 583–611. and Statistics 92 (4): 1042–49. Fernandes, A., and C. Paunov. 2009. “Does Grossman, G., and E. Helpman. 1991a. Inno- Tougher Import Competition Foster Prod- vation and Growth in the Global Economy. uct Quality Upgrading?” World Bank Policy Cambridge: MIT Press. Research Working Paper 4894, Washington, ———. 1991b. “Quality Ladders in the Theory DC. of Growth.” Review of Economic Studies 58 Fiorentino, R., C. Touqueboeuf, and L. Verdeja. (1): 43–61. 2007. “The Changing Landscape of Regional ———. 1991c. “Trade, Knowledge Spillovers, Trade Agreements: 2006 Update.” WTO Dis- and Growth.” European Economic Review cussion Paper 12, World Trade Organization, 35: 517–26. Geneva. Grossman, G., and E. Rossi-Hansberg. 2008. Fisher, E., and F. Vega-Redondo. 2006. “The “Trading Tasks: A Simple Theory of Offshor- Linchpins of a Modern Economy.” Paper pre- ing.” American Economic Review 98 (5): sented at the annual meeting of the American 1978–97. Economic Association, Chicago. Grubel, H. G., and P. J. Lloyd. 1975. Intra- Flam, H., and E. Helpman. 1987. “Vertical Prod- Industry Trade: The Theory and Measurement uct Differentiation and North-South Trade.” of International Trade in Differentiated Prod- American Economic Review 77 (5): 810–22. ucts. New York, NY: Wiley. THE STRUCTURE OF TRADE LINKAGES AND ECONOMIC GROWTH 127 Haddad M., J. Lim, and C. Saborowski. 2013. Globalization.” Journal of Economic Perspec- “Trade Openness Reduces Growth Volatility tives 21: 131–54. When Countries Are Well Diversified.” Cana- Hummels, D., J. Ishii, and K. Yi. 2001. “The dian Journal of Economics 46 (2): 765–90. Nature and Growth of Vertical Specialization Hanson, G. H., R. J. Mataloni, and M. J. Slaugh- in World Trade.” Journal of International ter. 2005. “Vertical Production Networks in Economics 54 (1): 75–96. Multinational Firms.” Review of Economics Hummels, D., and P. Klenow. 2005. “The Variety and Statistics 87 (4): 664–78. and Quality of a Nation’s Exports.” American Harrison, A., and M. McMillan. 2011. “Offshor- Economic Review 95 (3): 704–23. ing Jobs? Multinationals and U.S. Manufac- Humphrey, J. 2008. “Private Standards, Small turing Employment.” Review of Economics Farmers and Donor Policy: EUREPGAP in and Statistics 93 (3): 857–75. Kenya.” Institute of Development Studies Harrison, A., and A. Rodríguez-Clare. 2010. Working Paper 308, University of Sussex, “Trade, Foreign Investment, and Industrial Brighton, United Kingdom. Policy.” Handbook of Development Econom- Humphrey, J., and H. Schmitz. 2002. “How ics, vol. 5, ed. D. Rodrik and M. Rosenzweig, Does Insertion in Global Value Chains Affect 4039–214. Amsterdam: North-Holland. Upgrading in Industrial Clusters?” Regional Hausmann, R., and C. Hidalgo. 2011. “The Net- Studies 36 (9): 1017–27. work Structure of Economic Output.” Journal Iacovone, L., and B. Javorcik. 2008. “Multi-Prod- of Economic Growth 16 (4): 309–42. uct Exporters: Diversification and Micro-Level Hausmann, R., J. Hwang, and D. Rodrik. 2007. Dynamics.” World Bank Policy Research “What You Export Matters.” Journal of Eco- Working Paper 4723, Washington, DC. nomic Growth 12 (1): 1–25. IATA (International Air Transport Association). Heckscher, E. 1919. “The Effect of Foreign Trade 2014. “Five-Year Forecast Shows Improved on the Distribution of Income.” Ekonomisk Outlook for Air Cargo.” Press Release No. 58. Tidskriff 21: 497–512. http://www.iata.org/pressroom/pr/Pages/2014 Helpman, E., and P. R. Krugman. 1985. Mar- -10-22-01.aspx. ket Structure and Foreign Trade: Increasing IMF (International Monetary Fund). 2013. Trade Returns, Imperfect Competition, and the Interconnectedness: The World with Global International Economy. Cambridge, MA: Value Chains. Washington, DC. MIT Press. Irwin, D. A., and M. Tervio. 2002. “Does Trade ———. 1989. Trade Policy and Market Structure. Raise Income? Evidence from the Twentieth Cambridge, MA: MIT Press. Century.” Journal of International Economics Hidalgo, C., and R. Hausmann. 2009. “The 58 (1): 1–18. Building Blocks of Economic Complexity.” Islam, N. 1995. “Growth Empirics: A Panel Data Proceedings of the National Academy of Sci- Approach.” Quarterly Journal of Economics ences 106 (26): 10570–75. 110 (1): 1127–70. Hidalgo, C. A., B. Klinger, A. L. Barabási, and Johnson, R., and G. Noguera. 2012. “Fragmen- R. Hausmann. 2007. “The Product Space Con- tation and Trade in Value Added over Four ditions the Development of Nations.” Science Decades.” American Economic Review 102 317 (5837): 482–87. (3): 407–11. Hinloopen, J., and C. van Marrewijk. 2001. Jones, C. 2000. “Comment on Rodríguez-Rodrik, “On the Empirical Distribution of the Balassa ‘Trade Policy and Economic Growth: A Skep- Index.” Weltwirtschaftliches Archiv 137 (1): tic’s Guide to the Cross-National Evidence.’” 1–35. NBER Macroeconomics Annual 2000, ed. Hoekman, B., and A. Nicita. 2011. “Trade Policy, Ben Bernanke and Kenneth Rogoff. Cam- Trade Costs, and Developing Country Trade.” bridge, MA: MIT Press. World Development 39 (12): 2069–79. Jones, R., and H. Kierzkowski. 1990. “The Role of Hu, Y., and D. Zhu. 2009. “Empirical Analysis of Services in Production and International Trade: the Worldwide Maritime Transportation Net- A Theoretical Framework.” In The Political work.” Physica A 388: 2061–71. Economy of International Trade: Festchrift in Hummels, D. 2007. “Transportation Costs and Honor of Robert A. Mundell, ed. R. Jones and International Trade in the Second Era of A. Krueger. Cambridge, MA: MIT Press. 128 LATIN AMERICA AND THE RISING SOUTH Kaplinsky, R., and M. Farooki. 2010. “What Lall, S., and M. Albaladejo. 2004. “China’s Com- Are the Implications for GVCs When the petitive Performance: A Threat to East Asian Market Shifts from the North to the South?” Manufactured Exports?” World Development Intern ation al Journ al of Technologic al 32 (9): 1441–66. Learning, Innovation and Development 4 Lancaster, K. 1990. “The Economics of Product (1): 13–38. Variety: A Survey.” Marketing Science 9 (3): Kaplinsky, R., M. Morris, and J. Readman. 2002. 189–206. “The Globalization of Product Markets and Lederman, D., and W. Maloney. 2012. Does What Immiserizing Growth: Lessons from the South You Export Matter? In Search of Empirical African Furniture Industry.” World Develop- Guidance for Industrial Policies. Washington, ment 30 (7): 1159–77. DC: World Bank. Kaplinsky, R., A. Terheggen, and J. Tijaja. 2011. Lee, H. Y., L. A. Ricci, and R. Rigobon. 2004. “China as a Final Market: The Gabon Tim- “Once Again, Is Openness Good for Growth?” ber and Thai Cassava Value Chains.” World Journal of Development Economics 75 (2): Development 39 (7): 1177–90. 451–72. Kawai, M., and G. Wignaraja. 2013. “Patterns of Lenzen, M., K. Kanemoto, D. Moran, and A. Free Trade Areas in Asia.” Asian Development Geschke. 2012. “Mapping the Structure of the Bank, Manila. World Economy.” Environmental Science and Keesing, D., and S. Lall. 1992. “Marketing Man- Technology 46 (15): 8374–81. ufactured Exports from Developing Countries: Lenzen, M., D. Moran, K. Kanemoto, and A. Learning Sequences and Public Support. In Geschke. 2013. “Building Eora: a Global Trade Policy, Industrialization and Develop- Multi-Regional Input-Output Database at ment, ed. G. Helleiner. Oxford: Oxford Uni- High Country and Sector Resolution.” Eco- versity Press. nomic Systems Research 25 (1): 20–49. Keller, W. 2004. “International Technology Dif- Limao, N., and A. Venables. 2001. “Infrastruc- fusion.” Journal of Economic Literature 42 ture, Geographical Disadvantage, Transport (3): 752–82. Costs, and Trade.” World Bank Economic Koopman, R., Z. Wang, and S. J. Wei. 2014. Review 15 (3): 451–79. “Tracing Value-Added and Double Count- Loayza, N., and P. Fajnzylber. 2005. Economic ing in Gross Exports.” American Economic Growth in Latin America and the Caribbean: Review 104 (2): 459–94. Stylized Facts, Explanations, and Forecasts. Kose, M. A., E. Prasad, K. S. Rogoff, and S. J. Washington, DC: World Bank. Wei. 2009. “Financial Globalization: A Reap- Loayza, N., and C. Raddatz. 2007. “The Struc- praisal.” IMF Staff Paper 56: 8–62. tural Determinants of External Vulnerability.” Kose, M., E. Prasad, and E. Terrones. 2004. “Vol- World Bank Economic Review 21 (3): 359–87. atility and Comovement in a Globalized World Loayza, N., R. Ranciere, L. Servén, and J. Ven- Economy: An Empirical Exploration.” In Mac- tura. 2007. “Macroeconomic Volatility and roeconomic Policies in the World Economy, Welfare in Developing Countries: An Intro- ed. H. Siebert. Berlin: Springer. duction.” World Bank Economic Review 21 Krugman, P. 1979. “Increasing Returns, Monop- (3): 343–57. olistic Competition, and International Trade.” Lopez-Gonzalez, J., and P. Holmes. 2011. “The Journal of International Economics 9 (4): Nature and Evolution of Vertical Specializa- 469–79. tion: What Role for Preferential Trade Agree- ———. 1980. “Scale Economies, Product Differ- ments.” National Center of Competence in entiation, and the Pattern of Trade.” American Research (NCCR) Working Paper 41, Berne. Economic Review 70 (5): 950–59. Lucas, R. 1988. “On the Mechanics of Economic ———. 1981. “Intra-Industry Specialization and Development.” Journal of Monetary Econom- the Gains from Trade.” Journal of Political ics 22 (1): 3–42. Economy 89 (5): 959–73. Lüthje, T. 2003. “Intra-Industry Trade in Inter- ———. 1991. Geography and Trade. Cambridge, mediate Goods and Final Goods in a Gen- MA: MIT Press. eral Equilibrium Setting.” Open Economies Lall, S. 2000. “The Technological Structure and Review 14 (2): 191–209. Performance of Developing Country Manufac- Lynn, B. C. 2005. End of the Line: The Rise and tured Exports. 1985–98.” Oxford Develop- Coming Fall of the Global Corporation. New ment Studies 28 (3): 337–69. York: Doubleday. THE STRUCTURE OF TRADE LINKAGES AND ECONOMIC GROWTH 129 Maertens, M., and J. Swinnen. 2009. “Trade, Notteboom, T. E. 2009. “Complementarity and Standards, and Poverty: Evidence from Sene- Substitutability among Adjacent Gateway gal.” World Development 37 (1): 161–78. Ports.” Environment and Planning A 41 (3): Maloney, W., and F. Valencia Caicedo. 2014. 743–62. “Engineers, Innovative Capacity, and Devel- Ohlin, B. 1933. “Interregional and International opment in the Americas.” World Bank Policy Trade.” In Vulnerable Places, Vulnerable Research Working Paper 6814, Washington, People: Trade Liberalization, Rural Pov- DC. erty and the Environment, ed. R. O’Ryan, Maritime International Secretariat Services. 2013. M. Niklitschek, E. Niklitschek, A. Ulloa, N. “Sustainable Development. IMO World Mar- Gligo, J. A. Cook, and P. Stedman-Edwards itime Day 2013.” http://www.ics-shipping (2010). Cambridge, MA: Harvard University .org/docs/default-source/resources/policy-tools Press. /sustainable-development-imo-world-mari Packenham, R. 1992. The Dependency Move- time-day-2013.pdf. ment: Scholarship and Politics in Develop- Marshall, A. 1879. “The Pure Theory of (Domes- ment Studies. Cambridge, MA: Harvard tic) Values.” In History of Economic Thought University Press. Books, ed. William J. Barber. Middletown, Palpacuer, F., P. Gibbon, and L. Thomsen. 2005. CT: Wesleyan University Press. “New Challenges for Developing Country ———. 1890. Principles of Economics. London: Suppliers in Global Clothing Chains: A Com- Macmillan. parative European Perspective.” World Devel- Matsubara, K. 2005. “Import Penetration and opment 33 (3): 409–30. Domestic Process Innovation.” Review of Penn World Table 7.1. https://pwt.sas.upenn.edu International Economics 13 (4): 646–58. /php_site/pwt_index.php. Matsuyama, K. 1992. “Agricultural Productiv- Piore, M., and C. Ruiz Durán. 1998. “Industrial ity, Comparative Advantage, and Economic Development as a Learning Process: Mexican Growth.” Journal of Economic Theory 58 (2): Manufacturing and the Opening to Trade.” 317–34. In Learning, Liberalization and Economic Miyagiwa, K., and Y. Ohno. 1997. “Strategic Adjustment, ed. M. Kagami, J. Humphrey, R&D Policy and Appropriability.” Journal of and M. Piore. Tokyo: Institute of Developing International Economics 42 (1): 125–48. Economies. Morales Sarriera, J., G. Araya, T. Serebrisky, C. Plummer, M. 2007. “Best Practices in Regional Briceño-Garmendía, and J. Schwartz. 2013. Trade Agreements: An Application to Asia.” Benchmarking Container Port Technical Effi- World Economy 30 (12): 1771–96. ciency in Latin America and the Caribbean: A Pritchett, L. 2000. “Understanding Patterns of Stochastic Frontier Analysis. Washington, DC: Economic Growth: Searching for Hills among World Bank. Plateaus, Mountains, and Plains.” World Morris, M., C. Staritz, and J. Barnes. 2011. Bank Economic Review 14 (2): 221–50. “Value Chain Dynamics, Local Embedded- Rajan, R., and A. Subramanian. 2008. “Aid and ness, and Upgrading in the Clothing Sectors of Growth: What Does the Cross-Country Evi- Lesotho and Swaziland.” International Jour- dence Really Show?” Review of Economics nal of Technological Learning, Innovation and Statistics 90 (4): 643–65. and Development 4 (1–3): 96–119. Ramey, G., and V. Ramey. 1995. “Cross-Country Narula, R., and J. Dunning. 2010. “Multinational Evidence on the Link between Volatility and Enterprises, Development and Globalization: Growth.” American Economic Review 85 (5): Some Clarifications and a Research Agenda.” 1138–50. Oxford Development Studies 38 (3): 263–87. Reyes, J., M. Garcia, and R. Lattimore. 2009. Newman, M. 2005. “A Measure of Betweenness “The International Economic Order and Trade Centrality Based on Random Walks.” Social Architecture.” Spatial Economic Analysis 4 Networks 27 (1): 39–54. (1): 73–102. Nickell, S. 1981. “Biases in Dynamic Models Ricardo, D. 1817. Principles of Political Economy with Fixed Effects.” Econometrica 49 (6): and Taxation. London: Bell and Sons. 1417–26. Rivera-Batiz, L. A., and P. M. Romer. 1991. Noguer, M., and M. Siscart. 2005. “Trade Raises “International Trade with Endogenous Tech- Income: A Precise and Robust Result.” Journal nological Change.” European Economic of International Economics 65 (2): 447– 60. Review 35 (4): 971–1001. 130 LATIN AMERICA AND THE RISING SOUTH Rodrigue, J. P., and C. Comtois. 2006. The Sinnott, E., J. Nash, and A. De la Torre. 2010. Geography of Transport Systems. New York: Natural Resources in Latin America and the Routledge. Caribbean: Beyond Booms and Busts? Wash- Rodríguez, F., and D. Rodrik. 2001. Trade Pol- ington, DC: World Bank. icy and Economic Growth: A Skeptic’s Guide Sturgeon, T., and O. Memevodic. 2010. “Map- to the Cross-National Evidence. Cambridge, ping Global Value Chains: Intermediate Goods MA: MIT Press. Trade and Structural Change in the World Rodríguez-Clare, A. 2010. “Offshoring in a Economy.” UNIDO Development Policy and Ricardian World.” American Economic Jour- Strategic Research Branch Working Paper nal: Macroeconomics 2 (2): 227–58. 05/2010, United Nations Industrial Develop- Rodrik, D. 2006. “What’s So Special about Chi- ment Organization, Vienna. na’s Exports?” China and World Economy 14 Sturgeon, T., and J. Van Biesebroeck. 2011. (5): 1–19. “Global Value Chains in the Automotive Rodrik, D., A. Subramanian, and F. Trebbi. 2004. Industry: An Enhanced Role for Developing “Institutions Rule: The Primacy of Institutions Countries?” International Journal of Tech- over Geography and Integration in Economic nological Learning, Innovation and Develop- Development.” Journal of Economic Growth ment 4 (1): 181–205. 9 (2): 131–65. Thoenig, M., and T. Verdier. 2003. “A Theory of Romer, P. 1990. “Endogenous Technological Defensive Skill-Biased Innovation and Global- Change.” Journal of Political Economy 98 (5): ization.” American Economic Review 93 (3): 71–102. 709–28. ———. 1993. “Ideas Gaps and Object Gaps in Timmer, M., A. Erumban, B. Los, R. Stehrer, and Economic Development.” Journal of Monetary G. De Vries. 2014. “Slicing Up Global Value Economics 32 (3): 543–74. Chains.” Journal of Economic Perspectives 28 RTA-IS (Regional Trade Agreements Information (2): 99–118. System) (database). World Trade Organization, UNCTAD (United Nations Conference on Trade Geneva. http://rtais.wto.org/. and Development). 2011. World Investment S achs , J., a nd A . Wa r ner. 1995. “Nat u- Report: Non-Equity Modes of International ral Resource Abundance and E conomic Production and Development. Geneva: Grow th.” N BE R Working Paper 5398, UNCTAD. National Bureau of Economic Research, ———. 2013. World Investment Report: Global Cambridge, MA. Value Chains: Investment and Trade for ———. 1997. “Sources of Slow Growth in Afri- Development. Geneva: UNCTAD. can Economies.” Journal of African Econo- Va n B i e s e b r o e c k , J . 2 011. “ D i s s e c t i n g mies 69 (3): 335–76. Intra-Industry Trade.” Economics Letters 110 Sanyal, K., and R. Jones. 1982. “The Theory of (2): 71–75. Trade in Middle Products.” American Eco- Venables, A. 2003. “Winners and Losers from nomic Review 72 (1): 16–31. Regional Integration Agreements.” Economic Schumpeter, J. A. 1942. Capitalism, Socialism Journal 113 (490): 747–61. and Democracy. New York: Harper & Row. Vernon, R. 1966. “International Investment and Schwab, K., and X. Sala-i-Martin, eds. 2013. The International Trade in the Product Cycle.” Global Competitiveness Report 2013–2014. Quarterly Journal of Economics 80 (1): Geneva: Word Economic Forum. 190–207. Segoura, I. 1998. “Return to Scale and External Wacziarg, R. 2001. “Measuring the Dynamic Economies: Empirical Evidence from Greek Gains from Trade.” World Bank Economic Two-Digit Manufacturing Industries.” Applied Review 15 (3): 393–429. Economics Letters 5 (8): 485–90. Wignaraja, G. 2014. “Anatomy and Emerging Servén, L. 2003. “Real-Exchange-Rate Uncer- Economic Issues in Asia–Latin American tainty and Private Investment in LDCs.” FTAs.” World Bank, background paper for Review of Economics and Statistics 85 (1): this report. 212–18. Wignaraja, G., and D. Lazaro. 2010. “North- Singh, T. 2010. “Does International Trade Cause South vs. South-South Asian FTAs: Trends, Economic Growth? A Survey.” World Econ- Compatibilities, and Ways forward.” UNU- omy 33 (11): 1517–64. CR IS Working Paper 3, United Nations THE STRUCTURE OF TRADE LINKAGES AND ECONOMIC GROWTH 131 University, Institute on Comparative Regional .org /about -the-industry/global-trade/top Integration Studies, Brugge, Belguim. -50-world-container-ports. Wilmsmeier, G. 2014. “Charging Practices in WTO (World Trade Organization). 2011. World European Ports: User Reaction on Differentia- Trade Report 2011. The W TO and Pref- tion of Charges.” Proceedings of the European erential Trade Agreements: From Coexis- Transport Conference, Frankfurt, September tence to Coherence. Geneva: World Trade 29 –October 1. Organization. Wilson, J., C. Mann, and T. Otsuki. 2003. WTO (World Trade Organization), and IDE “Trade Facilitation and Economic Develop- (Institute of Developing Economies)–JETRO ment: A New Approach to Quantifying the (Japan External Trade Organization). 2011. Impact.” World Bank Economic Review “Trade Patterns and Global Value Chains in 17 (3): 367–89. East Asia: From Trade in Goods to Trade in World Bank. 2009. World Development Report Tasks.” World Trade Organization and Insti- 20 09: Reshaping Economic Geography. tute of Developing Economies, Geneva and Washington, DC: World Bank. Tokyo. World Development Indicators (WDI) (database). Yi, M. 2003. “Can Vertical Specialization Explain World Bank, Washington, DC. http://data the Growth of World Trade?” Journal of Polit- .worldbank.org/data-catalog/world-development ical Economy 111 (1): 52–102. -indicators. Young, A. 1993. “Invention and Bounded Learn- World Shipping Council. n.d. “Top 50 World ing by Doing.” Journal of Political Economy Container Ports.” http://www.worldshipping 101 (3): 443–72. Big Emerging Markets, Big Labor Market Dislocations? 3 The evidence presented in this chapter indicates that the ongoing restructuring of the global economy has affected labor markets in Latin America and the Caribbean (LAC) in various ways. Labor market adjustments triggered by the rise of the South depend on differences in the composition of import and export baskets of new economic heavyweights versus tradi- tional ones (such as the United States, Japan, and European countries), and the similarity of trade structures between emerging market and LAC economies affects the adjustments. Of all emerging markets, China stands out as a particularly apparent force in the process of global restructuring, with asymmetric consequences across industries. Specifically, the rise of China embodies both supply and demand shocks. The former reduced the prices of manufactures, the latter raised commodity prices. Frictions in LAC labor markets probably resulted in large and long adjustments—but not necessarily in substantial long-term changes in wages and employ- ment. Economies in which manufacturing employed large shares of the workforce likely faced adjustments that resulted in lower economywide real wages or reductions in labor force par- ticipation even in the long run as a consequence of the rise of China. O ne of the main features of the rise of products. Among the large global economies, large developing economies in global Japan’s trade structure is perhaps most similar markets is that their economic struc- to China’s, although its share of global man- tures differ from Northern economies’ (chap- ufacturing is declining. The European Union ter 1 highlights trends in global market shares (with its 25 members) has trade structures that in agriculture, mining, and manufacturing). are similar to the United States. The biggest newcomer in the early 21st cen- From the viewpoint of Latin American tury is China, which has become the world’s and Caribbean (LAC) economies, what mat- largest exporter of manufactured products as ters most are changes in giants’ global market well as one of the world’s largest importers of shares across industries, as the size of eco- agricultural and mining products. In contrast, nomic “shocks” faced by a given industry in as this chapter shows, the United States tends a given (relatively small and open) economy to be a large net importer of manufactured will be proportional to the change in large goods, a relatively large exporter of agricul- economies’ global market share. Put another tural goods, and a modest importer of mining way, developing countries adjust to changes 133 134 LATIN AMERICA AND THE RISING SOUTH in the structure of the global economy rather given the similarity of trade in LAC and than to the levels of market shares. China. Indexes of the impact of China’s Given the above, we are interested in importance in global manufactures, agricul- studying how changes in the structure of the ture, and mining markets on a large sam- global economy have been associated with ple of LAC countries reveal heterogeneous asymmetric changes across markets for dif- impacts on exports within the region. It is ferent types of goods. Such an analysis is therefore likely that the impacts of the rise of relevant because the global demand for and China in global markets on domestic labor supply of various products likely changes the markets also differ within LAC. relative demand for labor across industries. If Third, the chapter examines trends in the this is the case, the labor market implications employment shares of formal and informal for LAC could be substantial, as some indus- manufacturing sectors in Argentina, Bra- tries will shrink while others expand. zil, and Mexico to provide insight into the Chapter 2 reviews potential growth effects employment structure of LAC economies. The of the structure (or quality) of exports asso- data suggest that in these economies the share ciated with South-South versus North-North of manufacturing employment—especially trade. It suggests that LAC’s bilateral trade formal employment—has declined since about with the South is less pro-growth than its 2000. The drop in the share of formal man- bilateral trade with the North, though there ufacturing employment was most apparent in is significant heterogeneity across countries Mexico, one of the countries most severely hit in the region. New research by Bown (2014) by the rise of China. The descriptive evidence reviewed in chapter 2 suggests that LAC gov- suggests that the impact of China was great- ernments have tended to impose temporary est in LAC labor markets in which the trade trade barriers (such as antidumping, counter- effects from China were largest. vailing, and safeguard duties) against China Fourth, the chapter presents the results of and other emerging markets, particularly an empirical analysis commissioned for this when rising foreign competition seems to report on the impact of the rise of China on have affected domestic labor markets. Real LAC labor markets since 2001, when China exchange rate appreciation, which may also surged onto the global stage after joining the have affected manufacturing industries, World Trade Organization (WTO). The focus could also have provided impetus for the is on China because it has been the most imposition of such trade restrictions. important South economy in the restructur- This chapter examines how changes in the ing of the global economy. This part of the structure of global markets affect domestic chapter studies the labor market adjustment labor markets in LAC in five steps. First, it paths of Argentina, Brazil, and Mexico.1 documents trends in global market shares The results, provided by Artuç, Lederman, (for imports and exports) of selected emerg- and Rojas (2015), indicate that the impact ing markets, including China as well as other of China was substantial in the short run major economies from the North and South, but modest in the long run; labor market in manufacturing, agriculture, and mining. frictions increased the short-run pain of the Second, it analyzes similarities and dif- adjustment for workers, but the opposing ferences between the structure of exports impacts of China through exports (of man- in LAC and the major global players from ufactures) and imports (of agriculture and both the North and the South, emphasizing mining) tended to cancel each other out in exports of manufactures. The evidence sug- the long run, at least in Argentina and Brazil. gests that some LAC countries have export Mexico probably fared a bit worse: the model structures that are similar to those of China estimates suggest that the negative effects on and other major global economies, whereas labor demand in manufacturing were too others are quite dissimilar. A key question large to be compensated for by the relatively is the size of the economic shocks emanat- small effects on labor demand in agriculture ing from the restructuring of global markets and mining. BIG EMERGING MARKE TS, BIG LABOR MARKE T DISLOCATIONS? 135 Last, the chapter speculates on the poten- China played the leading role in the pro- tial impact of labor market adjustments on the cess of global economic restructuring. Its income of LAC households in the bottom 40 share of global manufacturing exports rose percent of the income distribution. It does so from 7.6 percent in 2001 to 14.7 percent in by discussing the “intensity” of each sector’s 2011. This epic increase was accompanied by use of workers that belonged to households its increased appetite for commodities (that is, in the bottom 40 percent of the distribution. decreasing export shares) as well as declines Agriculture appears to employ a relatively in the export shares of agriculture and min- larger share of workers from the bottom 40 ing (including energy). China’s share of world percent than mining or manufacturing. Hence exports for all sectors rose from 6.9 percent to the extent that China’s rising demand for in 2001 to slightly more than 12 percent in agricultural commodities was strong, the 2011. Exports of manufactured products resulting adjustments may have worked in from Korea also grew, albeit on a smaller favor of the bottom 40 percent, particularly in scale than China’s, rising from 2.8 percent countries such as Argentina and Brazil. of global exports of manufactures in 2001 In addition to summarizing the main find- to 3.8 percent in 2011. Similar to the case ings, the chapter’s conclusion draws on Ribe, for China, its share of commodities exports Robalino, and Walker (2010) and Hollweg declined, and its share of total global exports and others (2014) to suggest that it may be rose, from 2.5 percent to 3.1 percent. worthwhile for policymakers to think about In contrast, Russia’s rise in global mar- how social protection policies can help reduce kets exhibited different patterns. Its natural labor market adjustment costs when econ- resource wealth enabled it to gain global omies face long-lasting structural changes market share in mining and energy, increas- emanating from the permanent reconfigura- ing its global share of exports in this sector tion of the global economy. from 7.4 percent in 2001 to 10.5 percent in 2011. Unlike in China and Korea, the share of manufacturing (and agricultural) exports The rise of the South and the remained stagnant during this period. Hence, restructuring of global markets the rise of the South appears to have affected in manufacturing, agriculture, different industries differently, depending on and mining the size and endowments of the emerging It cannot be overstated that the economic markets themselves. impact of the restructuring of the global The global trade structure of North (or economy through trade flows has differed high-income) economies greatly differs from across sectors. Chapter 1 documents such that of emerging markets. Indeed, in con- trends by examining changes in global export trast to China, Korea, and Russia, the United and import shares across groups of coun- States experienced a dramatic decline in its tries and China. This chapter takes a closer share of global manufacturing exports, from look at the role of large economies’ weight more than 13 percent in 2001 to less than 9 in global trade flows in the manufacturing, percent in 2011 (see figure 3.1). Japan’s share agriculture, and mining sectors. of global manufactures exports also declined, Figure 3.1 shows the evolution of global from 8.3 percent to 6.2 percent, during this export shares for China, the Republic of period, while the European Union’s share fell Korea, and the Russian Federation on the from 39 percent to less than 36 percent. As one hand and the United States, Japan, and their performance in commodity sectors did the European Union on the other. These not compensate for the drop in manufactured figures largely confirm the central tenet of product exports, all three economic power- this report—that since 2000 large emerging houses experienced declines in their shares of markets have risen in importance in global total global exports. markets while the weight of the North has On the demand side, as on the sup- declined.2 ply side, China appears to be a central 136 LATIN AMERICA AND THE RISING SOUTH FIGURE 3.1 Global export market shares of selected large economies, by sector, 2001, 2006, and 2011 a. China b. Korea, Rep. 2001 7.6 2001 2.8 Manufacturing 2006 12.1 Manufacturing 2006 3.5 2011 14.7 2011 3.8 2001 3.4 2001 0.5 Agriculture 2006 3.2 Agriculture 2006 0.2 2011 2.8 2011 0.2 2001 1.3 2001 0.0 Mining 2006 0.8 Mining 2006 0.0 and utilities and utilities 2011 0.4 2011 0.1 2001 6.9 2001 2.5 Total 2006 10.4 Total 2006 3.0 2011 12.0 2011 3.1 0 2 4 6 8 10 12 14 16 0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 Share of world exports (%) Share of world exports (%) c. Russian Federation d. Japan 2001 1.1 2001 8.3 Manufacturing 2006 1.4 Manufacturing 2006 7.1 2011 1.5 2011 6.2 2001 1.5 2001 0.3 Agriculture 2006 2.0 Agriculture 2006 0.2 2011 1.3 2011 0.4 2001 7.4 2001 0.1 Mining 2006 9.1 Mining 2006 0.0 and utilities and utilities 2011 10.5 2011 0.0 2001 1.6 2001 7.4 Total 2006 2.4 Total 2006 6.1 2011 2.9 2011 5.1 0 2 4 6 8 10 12 0 1 2 3 4 5 6 7 8 9 Share of world exports (%) Share of world exports (%) e. United States f. European Uniona 2001 13.1 2001 39.0 Manufacturing 2006 9.5 Manufacturing 2006 38.4 2011 8.8 2011 35.3 2001 17.0 2001 27.2 Agriculture 2006 14.3 Agriculture 2006 28.0 2011 15.1 2011 23.3 2001 1.5 2001 9.8 Mining 2006 Mining 2006 1.2 8.4 and utilities and utilities 2011 1.5 2011 6.4 2001 12.2 2001 36.2 Total 2006 8.7 Total 2006 34.4 2011 8.0 2011 30.7 0 2 4 6 8 10 12 14 16 18 0 5 10 15 20 25 30 35 40 45 Share of world exports (%) Share of world exports (%) Source: Calculations based on data from World Integrated Trade Solution (WITS)/Comtrade. Note: Sectoral classification of trade flows is based on the International Standard Industrial Classification (ISIC), Revision 3. The agriculture sector corresponds to ISIC codes 0111–0500, mining to ISIC codes 1010–1429, and manufacturing to ISIC codes 1511–3699. a. The European Union includes 25 member countries. BIG EMERGING MARKE TS, BIG LABOR MARKE T DISLOCATIONS? 137 player (figure 3.2). Its global share of agri- the rise of China. The export similarity index, cultural imports rose from 4.7 percent in first proposed by Finger and Kreinin (1979), 2001 to 13.2 percent in 2011. Its share of has been widely used in the international global imports of mining and energy rose trade literature.4 It measures the percentage of even more dramatically, from 3.4 percent to a country’s export basket of products that is 15.3 percent. Korea did not come close to also exported by another country. China in importance, as its global share of Argentina’s export structure is least simi- imports remained virtually unchanged across lar to that of China and most similar to that the three broad industries. For its part, Russia of the European Union (followed closely by increased its share of global manufacturing the world as a whole). Brazil is only slightly imports (from about 0.7 percent in 2001 to different from Argentina. Indeed, only more than 1.9 percent in 2011) and of agricul- roughly 30 percent of its manufacturing tural imports (from 1.3 percent to 2.9 percent exports coincided with exports from China over the same period). over the whole period (versus about 20 per- In the high-income North, the United cent in Argentina). Like Argentina, Brazil is States and Japan have been the mirror image most similar to the European Union, with of China: their agriculture and mining import about 60 percent of its exports also exported shares have fallen as China’s have risen. In by the European Union. South Africa and contrast, the European Union’s global import the United States are also similar to Brazil in shares have remained roughly constant across terms of export structure. the three broad industries. Mexico, in contrast, is quite differ- Overall, the story of the global restructur- ent: about 55 percent of its manufacturing ing across industries is clear. The increasing exports are also exported by China, a level weight of China in particular has had asym- only slightly below that of Japan (and the metric consequences for different sectors. world). The large commodity exporters from The rise of China—and to a smaller extent the developing world (namely, Russia and Korea—flooded global markets for manufac- South Africa) appear to be the most dissim- tured products and increased imports of agri- ilar to Mexico in terms of export structure. cultural and mining commodities. In sum, it appears that on the export side, It is likely that the effect on workers has Mexico has remained quite similar to, and also varied across sectors. Labor markets in thus a competitor of, China for most of the countries with manufacturing trade struc- 21st century. In contrast, Argentina and Bra- tures similar to that of China have probably zil have export structures that are different. been more severely affected, as China poses a They have much more overlap with advanced direct threat to the competitiveness of major economies, such as the United States and manufacturing exports. the European Union. Consequently, one can speculate that the rise of China proba- bly has presented more severe challenges for A closer look at manufactures economic adjustment for Mexico than for exports and the role of China Argentina or Brazil. through the lens of export Figure 3.4 presents findings by Artuç, similarity Lederman, and Rojas (2015) on the potential Figure 3.3 presents data on the evolution of gains and losses of export growth for a large similarity between manufactured product sample of LAC countries. Their export index exports of Latin American economies and is closely related to the indexes used by Lall various large economies (from North and and Weiss (2004), Hanson and Robertson South) as well as the world as a whole. To ease (2009), and Freund and Ozden (2009). exposition, the panels focus on Argentina, Lall and Weiss (2004) compare Latin Brazil, and Mexico. 3 These three countries American and Chinese exports at the four- cover the spectrum of trade structures in the digit International Standard Industrial Clas- region in terms of exposure, particularly to sification (ISIC) level to identify categories 138 LATIN AMERICA AND THE RISING SOUTH FIGURE 3.2 Global import market shares of selected large economies, by sector, 2001, 2006, and 2011 a. China b. Korea, Rep. 2001 3.9 2001 2.0 Manufacturing 2006 5.9 Manufacturing 2006 2.2 2011 7.8 2011 2.4 2001 4.7 2001 2.4 Agriculture 2006 8.4 Agriculture 2006 2.2 2011 13.2 2011 2.3 2001 3.4 2001 6.1 Mining 2006 Mining 2006 7.3 5.8 and utilities and utilities 2011 15.3 2011 6.6 2001 3.9 2001 2.4 Total 2006 6.2 Total 2006 2.7 2011 9.2 2011 3.1 0 2 4 6 8 10 12 14 16 18 0 1 2 3 4 5 6 7 Share of world imports (%) Share of world imports (%) c. Russian Federation d. Japan 2001 0.7 2001 4.8 Manufacturing 2006 1.2 Manufacturing 2006 4.0 2011 1.9 2011 3.9 2001 1.3 2001 9.7 Agriculture 2006 2.2 Agriculture 2006 6.8 2011 2.9 2011 5.6 2001 0.3 2001 13.2 Mining 2006 Mining 2006 11.2 0.2 and utilities and utilities 2011 0.1 2011 10.8 2001 0.6 2001 5.7 Total 2006 2.2 Total 2006 5.0 2011 1.6 2011 4.9 0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 0 2 4 6 8 10 12 14 Share of world imports (%) Share of world imports (%) e. United States f. European Uniona 2001 19.2 2001 39.1 Manufacturing 2006 16.0 Manufacturing 2006 39.0 2011 12.8 2011 34.8 2001 10.9 2001 41.9 Agriculture 2006 9.8 Agriculture 2006 41.6 2011 8.1 2011 34.0 2001 20.2 2001 33.5 Mining Mining 2006 18.9 2006 33.0 and utilities and utilities 2011 14.5 2011 28.7 2001 19.1 2001 38.7 Total 2006 16.1 Total 2006 38.3 2011 12.9 2011 33.9 0 5 10 15 20 25 0 5 10 15 20 25 30 35 40 45 Share of world imports (%) Share of world imports (%) Source: Calculations based on data from WITS/Comtrade. Note: Sectoral classification of trade flows is based on the International Standard Industrial Classification (ISIC) classification, Revision 3. The agriculture sector corresponds to ISIC codes 0111–0500, mining to ISIC codes 1010–1429, and manufacturing to ISIC codes 1511–3699. a. The European Union includes 25 member countries. BIG EMERGING MARKE TS, BIG LABOR MARKE T DISLOCATIONS? 139 FIGURE 3.3 Export similarity indexes in in which China gained market share at the manufacturing for Argentina, Brazil, and expense of Latin America between 1990 and Mexico, 1999–2011 2002. They fi nd that 30 percent of trade in a. Argentina 1990 was in industries in which Chinese 0.7 exports were increasing and Latin American Manufacturing exports similarity index exports were decreasing but that the threat 0.6 from China gradually decreased: by 2002 0.5 China seemed to be negatively affecting only 0.4 11 percent of Latin American exports. Lall and Weiss (2014) conclude that over time, 0.3 Latin American trade structures evolved 0.2 to complement those of China. These find- ings are somewhat different from the ones 0.1 presented in the previous section on export 0 similarity, which indicate that the similarity 1999 2003 2006 2009 2011 of exports between the three LAC countries b. Brazil and China remained relatively stable between 0.7 1999 and 2011. Manufacturing exports similarity index 0.6 Freund and Ozden (2009) fi nd that Chi- na’s export growth had only a small negative 0.5 effect on overall Latin American exports. 0.4 They show that the rise of China hurt Mexi- 0.3 co’s industrial exports but had no significant impact on the rest of Latin America. For 0.2 Mexico they fi nd that a 10 percent increase 0.1 in China’s industrial exports reduced Mexi- 0 co’s industrial export growth by 7.9 percent. 1999 2003 2006 2009 2011 However, they conclude that China’s con- tinuing export growth may be affecting the c. Mexico wage distribution, because export growth is 0.7 concentrated in high-wage industries. This Manufacturing exports similarity index 0.6 evidence is largely consistent with the export 0.5 similarity indexes presented above. Box 3.1 presents the index proposed by 0.4 Artuç, Lederman, and Rojas (2015), which 0.3 measures the trade impact of the changes 0.2 in China’s global market shares. The index is consistent with the assumption that the 0.1 growth of global markets during the 21st 0 century was exogenous to the performance of 1999 2003 2006 2009 2011 LAC economies—that is, the growth rates of China World United States China and the rest of the world were unaf- European Union Korea, Rep. Japan fected by the policies of LAC economies or Russian Federation South Africa their growth performance. By relying on this assumption, the index measures how the Sources: Calculations based on data from WITS/Comtrade; index proposed by Finger and Kreinin 1979. increase in China’s global share of manufac- Note: The higher the index, the greater the similarity between the manu- turing exports reduced “residual demand” for facturing export baskets of two economies. LAC exports of manufactures (of the products in which China gained market share). On the demand side, the index measures the increase 140   LATIN AMERIC A AND THE RISING SOUTH Box 3.1  Construction of the China effect index For each good g in group G, exports from China Summing across countries, ROW exports in t 2 and the world in t 1 are given. Defining the rest of can be expressed as follows: the world (ROW) as the world excluding China, this assumption implies that the ROW’s share in world’s XW g,t 5 1 1 1 rg 2 2 exports in t1 is also given. Let r ig be country i’s export X ROW g,t2 X ROW g,t XW 1 growth rate of good g between t 1 and t 2 such that g,t 1 As it is assumed that ROW exports are given, X ig,t 5 1 1 1 r ig 2 X ig,t (B3.1.1) 2 1 from this expression rg is defined as Total exports of goods G are obtained by sum- ming over all products g, as follows: X ROW g,t XW g,t 2 1 rg 5 ° ¢° ¢ 21 g g [ GX ig,t 5 g g[G 1 1 1 XW g,t X ROW g,t X iG,t 5 r ig 2 X ig,t 2 1 2 2 1 which can be rewritten as (B3.1.2) XC g,t XC g,t XW g,t 2 1 1 rg 5 2 ° 2 ¢° ¢ (B3.1.5) The percentage change in total exports of G XW g,t XW g,t X ROW g,t between t 1 and t 2 can be calculated by dividing 2 1 1 equation (B3.1.2) by total exports of G in t 1 (and Substituting equation (B3.1.5) into equation subtracting 1): (B3.1.3) yields the desired index (which represents the percentage change in country i ’s exports of G), Index 5 g g [ Gr ig X ig,t defined as follows: g g 9[ GX ig 9t 1 (B3.1.3) Index 5 2 a ° XC XC XW X ig,t g g 9[ GX ig 9,t 1 g,t 2 g,t 1 g,t 1 1 2 ¢° ¢ To distribute the growth of China among the g[G XW g,t XW g,t X g,tROW economies in the ROW while excluding other 2 1 1 1 sources of export growth, it is assumed that world The index for imports is obtained from this rela- exports do not change from t 1 to t 2 (that is, China’s tionship (by changing China’s exports by the addi- export growth perfectly crowds out the ROW’s tive inverse of imports and substituting the world’s exports). This assumption implies that X W g,t2 5 XW g,t1 . exports by the world’s imports). If it is also assumed that the export growth rate of good g is the same for all countries, equation (B3.1.1) can be written as follows: X ig,t i Xg ,t 5 1 1 1 rg 2 2 1 (B3.1.4) XW g,t XW g,t 2 1 in LAC agricultural and mining exports of least affected country in this sample would products that China increasingly demanded be Cuba. For LAC as a whole, manufactures as its share of global imports rose. exports would grow by 10 percent less than Panel a of figure 3.4 shows the estimation they would have if China’s global share had results of the China effect for manufactures not risen. exports over the period 2001–11. The index Given that the analyses to be discussed demonstrates that Haiti would be hit hard- throughout the remainder of the chapter est, losing 19 percent of its manufactures focus on Mexico, Brazil, and Argentina, exports (mainly textiles and apparel). The it is worth noting that Mexico (along with FIGURE 3.4 Effects of the rise of China on gross exports of selected countries in Latin America and the Caribbean, by sector, 2001–11 a. Manufactured exports Haiti Honduras El Salvador Mexico Dominican Republic Costa Rica Guatemala St. Lucia Panama Nicaragua LAC St. Kitts and Nevis St. Vincent and the Grenadines Dominica Grenada Colombia Brazil Peru Jamaica Belize Ecuador Uruguay Chile Argentina Bolivia Suriname Venezuela, RB Paraguay Guyana Cuba Ϫ20 Ϫ18 Ϫ16 Ϫ14 Ϫ12 Ϫ10 Ϫ8 Ϫ6 Ϫ4 Ϫ2 0 Manufacturing exports (percentage change) b. Agricultural exports c. Mining exports Paraguay Brazil Argentina Chile Guyana Honduras Brazil Peru Uruguay Cuba Bolivia Jamaica LAC Guyana Cuba LAC Nicaragua Bolivia Venezuela, RB Uruguay St. Kitts and Nevis Haiti Dominican Republic Grenada Suriname Dominica Jamaica Belize Mexico Paraguay Chile Colombia Guatemala Argentina Peru Dominican Republic Belize Venezuela, RB Costa Rica Mexico El Salvador Suriname St. Vincent and the Grenadines Guatemala Haiti St. Lucia Panama Ecuador Honduras Nicaragua Dominica Panama Grenada St. Vincent and the Grenadines Ecuador Costa Rica St. Lucia El Salvador Colombia St. Kitts and Nevis 0 2 4 6 8 10 12 14 16 0 5 10 15 20 25 30 35 Agricultural exports Mining exports (percentage change) (percentage change) Source: Artuç, Lederman, and Rojas 2015, based on data from World Integrated Trade Solution (WITS)/Comtrade. Note: LAC = Latin America and the Caribbean. Sectoral classification of trade flows is based on the International Standard Industrial Classification (ISIC), Revision 3. The agriculture sector corresponds to ISIC codes 0111–0500, mining to ISIC codes 1010–1429, and manufacturing to ISIC codes 1511–3699. See box 3.1 for details. 141 142 LATIN AMERICA AND THE RISING SOUTH Central America and the Caribbean) was on LAC’s exports of manufactured goods but among the most affected economies. It suf- a positive and often large effect on those in fered greatly because of a large overlap with agriculture and mining sectors. How much China on a broad set of exported manufac- these trade patterns affected LAC labor mar- tured products, including textile and apparel kets is an important economic and empirical as well as electronics. Because the export question. structures of Argentina (and Chile and Uru- guay) are less similar to that of China, the impact on their manufacturing exports was Recent trends in manufacturing milder. employment in Latin America These results are consistent with evidence and the Caribbean derived from different empirical methods. To One place to start the discussion of the labor analyze the effect of China’s export growth market implications of the rise of China is on LAC economies, Hanson and Robertson employment in manufacturing. The focus (2009) use a gravity model of trade in which is on Argentina, Brazil, and Mexico; survey exporters produce differentiated goods and employment data are also available for other compete with Chinese exporters under countries (results available upon request).5 monopolistic competition. They first estimate Figure 3.5 shows trends in the shares of the changes in exporter “fi xed effects” asso- formal and informal employment in manu- ciated with export growth. They then sim- facturing industries. The trends are divided ulate manufacturing export growth in LAC into periods just before or around 2000 after setting China’s export growth rate to and after 2000. (Several years of data were zero within a counterfactual scenario. Their pooled into time periods to avoid sampling results indicate that China impeded man- errors in the data, as the surveys were not ufacturing export growth by 1.1 percent in designed to be representative of workers at Argentina, 1.4 percent in Brazil, 2.3 percent the industry level). The objective is to deter- in Chile, and 3.1 percent in Mexico. mine whether the years after China’s inclu- China’s growth increased other countries’ sion in the global trading system (through agricultural and mining product exports, as its accession to the WTO) show evidence of its global share of imports of these commod- declines in the share of employment in man- ities ballooned after 2000. Panels b and c of ufacturing in countries where competition figure 3.4 show the effects on LAC countries’ from China was strongest. exports of these commodities. In Argentina both formal and informal Regarding agricultural exports, the esti- employment in manufacturing declined after mation predicts that Paraguay would benefit 2000 (relative to observed employment shares most, through its global exports of soy, with in the 1990s). The share of formal employ- Argentina a close second (panel b). Regarding ment in manufacturing industries fell from mining, Brazil would benefit the most among about 14.7 percent of the employed labor this sample of LAC economies, driven mainly force in 1991–99 to 8.8 percent in 2000–05 by China’s imports of iron ore. Chile is just and rose only slightly, to 9.9 percent, in 5 percentage points behind, given its high 2006–12. Informal employment declined dependence on copper exports, which China from 5.8 percent of the employed labor force imported heavily after 2000. Peru is also a in 1991–99 to 5.5 percent in 2000–05 and major exporter of copper and other mining 4.5 in 2006–12. commodities. Honduras falls between Chile The trends for Brazil are less stark. For- and Peru on this index; it benefitted mainly mal employment in manufacturing industries from China’s increase in imports of nonfer- fell from more than 10.1 percent in 1990–99 rous mining products, such as zinc. to 9.3 percent in 2001–05, followed by a Overall, the trade analysis presented thus slight increase to 9.9 percent in 2006–11. far suggests that China had a negative effect But the share of employment in informal BIG EMERGING MARKE TS, BIG LABOR MARKE T DISLOCATIONS? 143 manufacturing rose, from 3.6 percent in the FIGURE 3.5 Employment shares in the formal and informal 1990s to 4.7 percent in 2001–05 and 4.3 per- manufacturing sectors of Argentina, Brazil, and Mexico, before cent in 2006–11. Overall, the figures show and after 2000 stagnant, if not declining, trends in manufac- a. Argentina turing employment. Formal manufacturing employment 1991–99 14.7 Formal in Mexico fell from 13.4 percent of the 2000–05 8.8 employed workforce in 2000–04 to 9.8 per- 2006–12 9.9 cent in 2006–12. This 3.6 percentage point decline is dramatic, given the short amount 1991–99 5.8 Informal of time between the two periods. Informal 2000–05 5.5 employment in manufacturing was relatively 2006–12 4.5 stable during this time, hovering slightly above 8.2 percent in 2000–04 and 2006–12. 0 2 4 6 8 10 12 14 16 Share of employment in manufacturing sectors (%) Thus in the three LAC economies under study, manufacturing employment, especially b. Brazil in the formal sector, appears to have fallen or 1990–99 10.1 remained stagnant at best. Much caution is Formal needed, however, in identifying the causes of 2001–05 9.3 employment trends, as these economies expe- 2006–11 9.9 rienced numerous shocks during this time. 1990–99 3.6 The next section relies on research by Informal Artuç, Lederman, and Rojas (2015) that was 2001–05 4.7 commissioned for this study. It assesses the 2006–11 4.3 quantitative importance of the rise of China 0 2 4 6 8 10 12 in global markets as a determinant of labor Share of employment in manufacturing sectors (%) market outcomes. Their analysis combines empirical analyses with theoretical modeling c. Mexico to make inferences about the role of China in 2000–04 13.4 Formal shaping domestic labor markets. 2006–12 9.8 Labor market adjustment Informal 2000–04 8.2 paths in response to the rise 2006–12 8.3 of China 0 2 4 6 8 10 12 14 16 One approach to analyzing the effect of the Share of employment in manufacturing sectors (%) rise of China on foreign labor markets has been to estimate “reduced-form” econometric Sources: Calculations based on data from Encuesta Permanente de Hogares-Continua (EPHC) in Argentina, Pesquita Nacional por Amostra de Domicilios (PNAD) in Brazil, and Encuesta Nacional de models. Such specifications model the impact Ingresos y Gastos de los Hogares (ENIGH) in Mexico. of the rise of China in global markets on local Note: Informal workers are defined as workers without social security benefits. labor markets as proportional to the share of workers employed in industries in which China had substantial exports over time.6 by other high-income countries (used as an A good example of this approach is the instrumental variable) increased unemploy- article by Autor, Dorn, and Hanson (2013), ment, reduced labor force participation, and which studies the implications of the rise of reduced wages in local labor markets in the China on local labor markets (defined as United States that housed import-competing “commuting zones”) within the United States manufacturing industries. Preliminary results through imports of Chinese goods. The of similar research on Mexican labor markets authors argue that changes in Chinese imports underway at the Central Bank of Mexico, led 144 LATIN AMERICA AND THE RISING SOUTH FIGURE 3.6 Simulated short- and long-run impacts of the rise The study by Autor, Dorn, and Hanson of China on wages in Argentina, Brazil, and Mexico, by sector relies on clever econometrics to identify the a. Argentina impact of imports from China on local U.S. 12 labor markets. The study by Artuç, Leder- 10 man, and Rojas (2015) uses a combination of econometrics and theory. Their approach 8 can be summarized in two steps. First, the 6 authors estimate industry-specific labor Percent 4 mobility costs. They compute intersectoral 2 employment transitions from individual 0 worker panel datasets for Argentina, Bra- zil, and Mexico by following the methods Ϫ2 described in Arias and others (2014). Broadly Ϫ4 speaking, sectors with larger numbers of Ϫ1 0 1 2 3 4 5 6 7 8 9 10 11 12 13 Years incoming workers as a share of industry b. Brazil employment are identified as having lower 35 entry costs than sectors with smaller num- 30 bers of incoming workers. 25 S econd , t he aut hors use st a nda rd 20 industry-level data and a simple model of labor 15 demand across industries to trace the impacts Percent 10 of trade shocks emanating from China on 5 manufacturing industries, agriculture, and 0 mining. Figures 3.6–3.8 show the simulated Ϫ5 impact of China on industry wages,7 informal employment shares, and labor force participa- Ϫ10 Ϫ1 0 1 2 3 4 5 6 7 8 9 10 11 12 13 tion (called the “residual sector”). Years Across industries the authors report high c. Mexico overall interindustry labor mobility costs, 35 30 with estimates for the three countries ranging 25 from 0.5 times the average annual wage (for 20 entry into informal agriculture for workers 15 coming from formal agriculture) to roughly 8 times the average annual wage (for entry Percent 10 5 into the formal mining sector from any infor- 0 mal sector). These magnitudes are consis- Ϫ5 tent with estimates by Hollweg and others Ϫ10 (2014); Artuç, Lederman, and Porto (2015); Ϫ15 Artuç, Chaudhuri, and McLaren (2010); and Ϫ1 0 1 2 3 4 5 6 7 8 9 10 11 12 13 Dix-Carneiro (2014). Years As in Arias and others (2014), three fea- Economy average Agriculture Mining tures regarding labor mobility costs are com- Manufacturing Services mon to Argentina, Brazil, and Mexico. First, Source: Artuç, Lederman, and Rojas 2015. it is less costly to become formal if a worker Note: The model assumes that shocks to all sectors occurred simultaneously. See Artuç, Lederman stays in the same industry. Second, the highest and Rojas (2015) for technical details. entry costs involve moving from the informal sector in one industry to the formal sector by Daniel Chiquiar, are qualitatively similar in another. Third, the lowest entry costs are to those reported by Autor, Dorn, and Han- associated with movements from the formal to son for the United States. the informal sector within the same industry. BIG EMERGING MARKE TS, BIG LABOR MARKE T DISLOCATIONS? 145 In Argentina and Mexico, the lowest entry FIGURE 3.7 Simulated short- and long-run impacts of the rise of costs are incurred in moving from formal to China on informal employment in Argentina, Brazil, and Mexico informal employment within the restaurant a. Argentina and hotel sector. In Brazil the lowest entry 9 cost is associated with moving from formal 8 to informal employment within agriculture. 7 6 In all three countries, the highest cost is in 5 moving from informal employment in any Percent 4 sector besides agriculture or mining to for- 3 mal employment in these sectors. This cost 2 limited the movement of workers into these 1 0 sectors—precisely the ones positively affected Ϫ1 by the rise of China. Ϫ2 The sectors most affected by China were Ϫ2 Ϫ1 0 1 2 3 4 5 6 7 8 9 10 11 12 agriculture in Argentina, mining in Brazil, Years and manufacturing in Mexico. As expected, b. Brazil in Argentina employment in the agricultural 20 sector increased in both the formal sec- tor (16.0 percent) and the informal sector 15 (7.7 percent). Employment in the mining sec- tor also increased, with formal employment 10 Percent rising 5.6 percent and informal employment 5 rising 2.0 percent. The negative shock to the manufacturing sector reduced employment 0 in that sector, with formal employment fall- ing 2.3 percent and informal employment Ϫ5 1.0 percent. Ϫ2 Ϫ1 0 1 2 3 4 5 6 7 8 9 10 11 12 Although the percentage increases are Years larger for agriculture and mining than for c. Mexico manufacturing, the reduction in employ- 15 ment in manufacturing offset the increase in employment in the other sectors because 10 of its weight in the overall economy. Formal 5 Percent manufacturing employs about 9 times as many workers as formal mining and about 0 45 times as many workers as formal agricul- ture. Informal manufacturing employs about Ϫ5 41 times more workers than informal mining and 8 times more than informal agriculture.8 Ϫ10 The aggregate estimated effect of the rise of Ϫ2 Ϫ1 0 1 2 3 4 5 6 7 8 9 10 11 12 Years China was thus a reduction in employment— Economy wide Agriculture Mining Manufacturing that is, an increase in the residual sector—of about 0.3 percent (see figure 3.8). Source: Artuç, Lederman, and Rojas 2015. Note: The model assumes that shocks to all sectors occurred simultaneously. Panel a of figure 3.6 shows the adjust- ment path of real wages in Argentina. Right after the shock, wages increase in agricul- ture and mining and fall in manufacturing, across sectors, wages tend to move toward as the demand for labor increases in the two their initial level. In the new steady state, rising sectors and decreases in manufactur- there is an increase in the agriculture real ing. In the long run, as labor is reallocated wage (of about 3 percent); real wages in the 146 LATIN AMERICA AND THE RISING SOUTH FIGURE 3.8 Simulated short- and long-run impacts of the rise constant in the long run, it is reallocated of China on the residual sector in Argentina, Brazil, and Mexico from the manufacturing to the agriculture a. Argentina and mining sectors. 6 Artuç, Lederman, and Rojas (2015) fi nd 5 similar results for Brazil. Their simulations 4 show that mining employment rises about 3 40 percent in the formal sector and 16 per- cent in the informal sector. There is also an Percent 2 increase of about 10 percent in the number 1 of workers in formal agriculture. In the neg- 0 atively shocked sector (manufacturing), the Ϫ1 number of formal workers falls by about Ϫ2 6 percent and the number of informal work- Ϫ2 Ϫ1 0 1 2 3 4 5 6 7 8 9 10 11 12 ers falls by roughly 3 percent. Years Brazil’s manufacturing sector employs b. Brazil substantially more workers than does min- 6 ing or agriculture. The rise of China there- 5 fore leads to a reduction in the aggregate 4 level of employment, albeit a small one. In 3 the simulations, the residual sector increases Percent about 1 percent (figure 3.8, panel b). Thus, 2 as in Argentina, the positive shock on mining 1 and agriculture offsets the negative shock on 0 manufacturing, leaving the aggregate level of Ϫ1 employment similar to that before the shock. Ϫ2 Moreover, when the shocks from China hit Ϫ2 Ϫ1 0 1 2 3 4 5 6 7 8 9 10 11 12 Brazil, real wages in the mining sector rise Years about 31 percent. Nonetheless, as labor c. Mexico moves from manufacturing to mining, real 6 wages begin to decline, eventually reaching 5 roughly the same initial level in the new long- 4 run equilibrium, after the process of adjust- 3 ment has taken its course. In short, there is Percent 2 a reallocation of labor from manufacturing to mining and agriculture, leaving real wages 1 and the aggregate level of employment almost 0 unchanged in the long run. In the short run, Ϫ1 real wages in the mining sector rise. Ϫ2 Mexico experiences an increase in formal Ϫ2 Ϫ1 0 1 2 3 4 5 6 7 8 9 10 11 12 Years and informal employment in agriculture and mining and a decrease in formal and infor- Source: Artuç, Lederman, and Rojas 2015. mal manufacturing employment. Informal Note: The residual sector includes unemployed workers and workers who drop out of the labor force. The model assumes that shocks to all sectors occurred simultaneously. employment decreases 6 percent in manufac- turing and increases 10 percent in mining and 7 percent in agriculture. Formal employment other sectors remain close to their original in manufacturing falls (by about 14 percent) level. The negative shock on manufacturing and rises in mining (by 25 percent) and agri- may offset the positive shock on agriculture culture (by 6 percent). Total employment falls. and mining, keeping wages and the aggregate The residual sector increases 5 percent in the level of labor almost at their preshock levels. long run (see panel c of figure 3.8). In the Although the aggregate level of labor remains short run, there is a 28 percent increase in real BIG EMERGING MARKE TS, BIG LABOR MARKE T DISLOCATIONS? 147 wages in mining and an 11 percent decrease FIGURE 3.9 Evolution of relative wages in Brazil to Mexico, in real wages in manufacturing. In the long 2001–09 run, however, there is only a slight decrease 0.86 in real wages in manufacturing and services, 0.84 and a slight increase in real wages in mining. 0.82 In sum, for Argentina and Brazil, the pos- itive shocks on agriculture and mining offset 0.80 the negative shock on manufacturing, leaving 0.78 Ratio the total level of employment and real wages 0.76 almost at their initial level. In both countries, 0.74 a larger positive shock on mining and agri- 0.72 culture is needed to offset a smaller shock 0.70 on manufacturing, because manufacturing 0.68 employs a larger proportion of workers. In 0.66 Mexico the larger shock on manufacturing 2001 2002 2003 2004 2005 2006 2007 2008 2009 reduces employment in the long run and Sources: National average wages in local currency are from the International Labour Office. They lower wages. These simulation results are were converted to international purchasing power parity constant 2005 US dollars using the con- consistent with observed data on wages in version factor from World Development Indicators. Brazil and Mexico. As shown in Figure 3.9, the ratio of Brazil’s average wage relative to Mexico’s rose since the early 2000s. reflects the fact that the estimates of labor These simulation results are inextrica- mobility costs, which are high, drive the ble from the modeling of the labor markets sharp short-term adjustments relative to the in the three countries and should thus be milder long-term adjustments. interpreted with caution. The results could Third, the rise of China probably shaped be overstating the impact of China in both the trade structure of other large emerging the short and long run for three key reasons. markets, such as Korea, Russia, and South First, the index of the China effects for each Africa. A proper analysis would need to take of the selected broad industries focuses on the changing trade patterns of these (and the gross rather than net impacts on exports: other) countries into account. In particular— the authors used only Chinese exports of and as discussed previously—Russia con- manufactures, not Chinese imports of manu- trolled a rising share of global commodity factures as well. For most LAC countries that markets but imported a growing share of appear in figure 3.4, the net effect of China manufactures; Korea (like China) increased (after taking account of the fact that China its export share of manufactures. The United also imports some manufactured goods that States’ and Japan’s shares of global manu- LAC economies export) is somewhat smaller factured products exports fell while China’s than the gross effects.9 rose. Thus to the extent that China’s trade Second, the underlying assumptions of impacts were associated with some compen- the simulations posit that the rise of China sating effects from other large markets (rel- affected LAC labor markets instantly at ative to the size of LAC’s trade flows), the the beginning of the 21st century, but the simulated labor market implications of the changes in global market shares that drive rise of China may be overstated. the estimated effects of China through global These important caveats notwithstanding, trade shares occurred gradually after 2000 recent research also provides grounds to think (recall the trends depicted in figure 3.1). Con- that the long-run impacts of China reported sequently, the simulations may exaggerate here may be understated. New research sug- the magnitude of the China effects in the gests that labor mobility may be even more short run but not necessarily the relationship handicapped than Artuç, Lederman, and between the short-term dynamics and the Rojas (2015) estimate. Autor, Dorn, and long-term effects. This dichotomy of biases Hanson (2013) fi nd persistent effects across 148 LATIN AMERICA AND THE RISING SOUTH U.S. “commuting zones,” which implies that Costa, Garred, and Pessoa (2014) use labor may not be mobile enough across such Brazilian census data to show that between zones to equalize (or attenuate) the interin- 2000 and 2010, manufacturing wages and dustry wage differential when industries are in-migration rates grew more slowly and concentrated (or agglomerated) across phys- wage inequality widened more in local labor ical space. Ongoing research by the Central markets that were more affected by Chinese Bank of Mexico could corroborate the results import competition.10 They cannot discuss with Mexican data (Chiquiar 2014). adjustment dynamics because of data limita- New research on Brazil also seems to tions (namely, the census provides information suggest that a lack of labor mobility across on only two points in time, 2000 and 2010). geographic space can cause long-lasting Still, their fi ndings could imply long-lasting impacts of trade shocks that permanently effects that are at odds with the model-driven change relative prices across industries. results of high short-term displacements pro- A first piece of evidence comes from Dix- vided in the analysis presented above. Carneiro and Kovak (2014). Using an Another piece of evidence comes from approach comparable to that of Autor, Dorn, Morten and Oliveira (2014). They study how and Hanson (2013), they find long-lasting transport networks, mainly roads, affect the interindustry and interregional effects within extent of labor market integration across Brazil of the trade reforms implemented in territorial units within Brazil. Their paper the late 1980s and early 1990s. The authors is motivated by highly persistent wage dif- examine regions of Brazil that had manufac- ferentials between Brasilia and other regions turing industries that employed large num- even within categories of skilled labor. They bers of workers and were exposed to tariffs. fi nd wage differentials over time associated They find long-lasting declines in real wages with (the lack of) road transport linkages, in those industries and regions as well as dif- which affect labor mobility. The high costs of ferential wage effects that tended to grow physical mobility in Brazil may thus be driv- rather than decline over time. These results ing persistent wage differentials across geo- are in stark contrast to the findings of Artuç, graphic space. Lederman and Rojas (2015). This over-time magnification effect can reflect a variety of explanations that share one thing in com- Potential distributional mon: lack of interregional labor mobility. If implications of China-induced workers choose not to move from, say, São labor market adjustments Paulo to rural areas, then declines in the rela- The World Bank Group and other organiza- tive price of manufactured production in São tions, including the International Monetary Paulo because of trade reforms (or the rise Fund, have recently pushed to put distribu- of China) will result in permanent relative tional issues at the forefront of the develop- effects on manufacturing wages in São Paulo ment policy debate. In 2013 the World Bank relative to wages elsewhere. (In the case of Group and its shareholders set two new long- Brazil, these trends would lead to wage con- term objectives: eradicating extreme pov- vergence, as wages in São Paulo are higher erty and raising the incomes of the bottom than wages in rural Brazil.) The sluggish 40 percent of the income distribution within adjustment of sector-specific capital could countries. Although the analyses of the labor magnify such effects: as capital (or machines market adjustments in LAC brought about employed) in the manufacturing industry of by the rise of China were not designed to São Paulo (for instance) begins to depreciate, assess the distributional consequences of the real wages of workers still employed in these adjustments, the results can be used to the region’s manufacturing industry would speculate about how China’s demand and fall further, with a corresponding reduction supply shocks may have affected demand for in the marginal product of labor as the capi- workers from the bottom 40 percent of the tal workers have to work with declines. distribution. BIG EMERGING MARKE TS, BIG LABOR MARKE T DISLOCATIONS? 149 TABLE 3.1 Percentage of workers in bottom 40 percent of income distribution in Argentina, Brazil, and Mexico, by sector Sector Argentina Brazil Mexico Agriculture 53.8 65.2 75.9 Mining/utilities 11.0 15.6 10.8 Manufacturing 32.0 28.4 38.2 Services 41.9 40.1 36.5 Average (excluding respondents who did not report sector) 34.8 37.4 40.6 Respondents who did not report sector of employment 27.1 47.9 24.8 Sources: Calculations based on the following surveys: Argentina: EPHC (urban coverage only), 2006–12; Brazil: PNAD (urban and rural coverage), 2004–11; Mexico: ENIGH (urban and rural coverage), 2000–02. Note: Only individuals with positive income are included. All surveys are pooled by country; computations are based on sample expansion weights. Table 3.1 shows the share of employees in in Mexico) indicate that this industry tends total employment that fall in the bottom 40 to employ workers at rates that fall between percent of the distribution in agriculture, man- agriculture and mining. The services sec- ufacturing, mining, and services. This mea- tor employs a slightly larger share of poor sure can be interpreted as an indicator of the workers than manufacturing in Argentina intensity of the use of poor workers. Industries (41.9 percent) and Brazil (40.1) but a slightly with higher ratios can be seen as pro-poor in smaller share in Mexico (36.5 percent). the sense that when demand for output from Given these industry-specific intensities in these industries rises, demand for poor work- the use of poor workers, several points can be ers tends to rise more in industries that use made about the impact of the rise of China on poor labor relatively intensively. The ratio is the distribution of income within LAC. First, analogous to factor intensities in production it is likely that the positive demand shock on functions but has the advantage of not requir- agriculture resulted in a relative increase in ing the computation of factor intensities (that the demand for labor provided by poor house- is, the use of unskilled labor relative to skilled holds. Second, the positive shock on mining labor in the production of a unit of output in probably led to the opposite effect: an increase each industry), which would then need to be in the relative demand for labor provided by mapped to the “assets” of the poor (mainly households in the top 60 percent of the distri- unskilled labor). Examining the use of “poor” bution. Third, in the best of cases, manufac- labor is thus a shortcut that requires little data turing seems to be neutral in Argentina and manipulation while still providing a clear pic- Mexico (its ratios are close to 40 percent in ture of relative labor demands. these countries); manufacturing industries in The data tell a consistent story across the Mexico are more frequently tilted in favor of three economies: agriculture employs a larger demand for labor from the upper deciles of share of workers from the bottom 40 percent the distribution. Thus no simple conclusions of the income distribution than other sectors can be drawn regarding the distribution of (53.8 percent in Argentina, 65.2 in Brazil, income, although there are clear indications and 75.9 percent in Mexico) across countries. of the direction of the various effects. This relatively larger share implies that agri- In summary, the impact of the rise of China culture disproportionately employs workers is complex and not easily identified because belonging to the bottom 40 percent, as the it was characterized by multiple shocks shares for this industry exceed 40 percent. that pushed LAC labor markets in opposite In contrast, the shares for mining are directions. Ultimately, for the distribution 11.0 percent in Argentina, 15.6 percent in Bra- of income, what matters most is the size of zil, and 10.8 percent in Mexico. The figures the positive demand shock on agriculture. In for manufacturing (32.0 percent in Argen- countries where this shock was large, such as tina, 28.4 percent in Brazil, and 38.2 percent Argentina and to a lesser extent Brazil, the 150 LATIN AMERICA AND THE RISING SOUTH overall effect was probably more egalitarian were not large enough to compensate for the than in Mexico, although the negative impact decline in the demand for labor by manufac- on manufacturing in Mexico may have con- turing industries. Indeed, it is plausible that tributed to a more egalitarian distribution wage inflation was subdued in Mexico during of income. This analysis, however, does the period analyzed, at least relative to Argen- not yield a definitive conclusion regarding tina and Brazil. Wage data from employment income distribution within countries. surveys in Mexico and Brazil suggest that wage inflation was higher in Brazil than in Mexico, especially in 2003–09. Concluding remarks The positive shock on agriculture from Hollweg and others (2014) argue that the the rise of China was probably beneficial for costs of physical mobility, as well as other sharing prosperity with the bottom 40 per- factors, such as industry-specific skills that cent of the income distribution in LAC. How- limit the employment mobility of workers ever, the decline of prices of manufactured across industries, may explain high estimated goods may have been marginally favorable labor mobility costs. Such costs seem to be for the bottom 40 percent only in Mexico, much more binding than the usual suspects where manufacturing industries appear to of regulatory barriers to hiring and fir- have employed relatively few poor workers. ing workers. The evidence reviewed in this Given the multiple effects across industries, chapter suggests that policy makers in LAC however, tracing the distributional conse- should pay more attention to these types of quences of China’s rise for LAC economies costs, especially in the context of shifting remains a task for future research. global trade patterns that are neither static The technical literature from both the nor inconsequential for the well-being of World Bank Group and academia iden- workers in the region. tifi es two key policy areas to examine: (a) The ongoing restructuring of the global the sluggish adjustment of labor markets economy has affected LAC economies in because of labor mobility costs across both different ways and with different magni- industries and space, especially when indus- tudes. The effects depend on the extent to tries are spatially concentrated, and (b) skills which the newly emergent global economic mismatches and transport costs, which may heavyweights export and import goods that be slowing adjustments. Hollweg and others differ from the goods traded by the United (2014) argue that developing countries could States, Japan, and Europe. China has been deal with the fi rst issue with social protec- the dominant force in this process of global tion strategies that focus on displaced work- restructuring, with asymmetric conse- ers’ capacities to fi nd employment in other quences across industries. The rise of China industries. A complicating factor in design- can thus be seen as embodying both supply ing trade adjustment assistance programs and demand shocks, with supply shocks in LAC and elsewhere in the developing dominating in manufacturing industries and world is that a complex network of social demand shocks dominating in commodity assistance programs (such as various types markets. of conditional cash transfer programs often Of the three countries analyzed in detail in force simultaneously together) and vari- (Argentina, Brazil, and Mexico), Mexico ous types of worker training and retrain- seems to have experienced the most adverse ing programs already exists. It is not clear consequences. It was hardest hit because a whether or not it is wise to implement yet large share of its employment was in manu- another type of worker assistance program facturing and its export structure was most to deal with permanent (or long lasting) similar to China at the beginning of the 21st trade shocks. century. As a result, it lost jobs, as the increase In dealing with the root causes of labor in demand for labor in agriculture and mining mobility costs a consensus is beginning BIG EMERGING MARKE TS, BIG LABOR MARKE T DISLOCATIONS? 151 to emerge in the academic literature that on export patterns from country to country. It goes beyond regulations that raise the costs is defined as follows: of labor churning from the viewpoint of Similarityi, j 5 a min 1 xg, i , xg, j 2 g[G employers. In this light, taking a cold-headed look at the role of domestic transport infra- where i and j are countries; G is a group of structures might end up being a long-term products (for example, manufactured goods); and xg,i is the share of product g in total policy agenda well worth pursuing. Look- exports of goods G in country i. The index ing forward, the distinction between global varies from 0 to 1, with 1 indicating identi- supply and demand shocks is important for cal export composition among two countries understanding other sources of develop- (perfect similarity). ment challenges associated with the rise of 5. The data required for this analysis come from the South in global markets. Of particular household employment surveys with infor- relevance may be the nature of LAC’s net mation on industry of employment as well as connections with the global economy (where workers’ formality status, defined as eligibility “net” refers to the region’s persistent current for social security (retirement benefits). Results account deficits, the broadest measure of an are very similar when using access to health economy’s net trade balance with the rest of insurance. In addition to Argentina, Brazil, and the world). Persistent external deficits may Mexico, such data are available for Bolivia, be a symptom of persistently low domestic Chile, Costa Rica, Ecuador, El Salvador, Peru, savings (or put another way, persistently Paraguay, and Uruguay. Overt-time coverage varies across countries. The data for Argentina, high consumption relative to output). To Brazil, and Mexico include a panel component, the extent that the rise of the South—which which allows analyses of labor mobility across has been dominated by emerging markets industries and formality status. Panel data were such as China that have persistently high not available for other countries, which were external surpluses—has manifested itself in therefore excluded from the analysis. further persistent declines in LAC savings 6. Costa, Garred, and Pessoa (2014) character- rates (and thus external deficits), it is likely ize the approach of Autor, Dorn, and Hanson that engineering economic growth with low (2013) as “shift-share methodology,” because savings will continue to be an analytical it relies on employment shares as the indica- and policy priority for the region. Chapter 5 tor of exposure to changes in China’s global takes on this complex topic of growth with shares. Bartik (1991) developed this approach. Topalova (2007) applied it to study the impact low domestic savings in a changing global of trade reforms across territories within India economy. after 1991. 7. Figure 3.6 shows the average wage by indus- Notes try for both formal and informal workers. The results refer to the simulation of three simul- 1. These countries were selected because of the taneous shocks on agriculture, mining, and availability of panel data for workers employed manufacturing. in both the formal and informal sectors. 8. The size of manufacturing relative to other sec- 2. For the sake of brevity, this chapter analyzes tors was calculated using the panel component only a few countries. Results for other major of the household employment surveys, the developing economies are available upon sample used for the simulations of the labor request. None of the omitted emerging mar- market adjustment paths. kets seem to have had global industry shares 9. These results are available upon request. above 1 percent or increases of more than a 10. They define “local” labor markets as Brazilian fraction of a percentage point. “microregions,” a territorial unit defined by 3. Results are available for numerous other LAC the Brazilian statistics agency (IBGE) accord- countries as well (results are available upon ing to a criteria related to the level of eco- request). nomic integration. It is somewhat comparable 4. The export similarity index, proposed by Fin- to the U.S. “commuting zones” used by Autor, ger and Kreinin (1979), provides information Dorn, and Hanson (2013). 152 LATIN AMERICA AND THE RISING SOUTH References Dix-Carneiro, R., and B. Kovak. 2014. “Trade Refor m a nd Reg iona l D y na m ic s: Ev i- Arias, J., E. Artuç, D. Lederman, and D. Rojas. dence from 25 Years of Brazilian Matched 2014. “Trade, Informal Employment and Employer-Employee Data.” Paper presented Labor Adjustment Costs.” World Bank Policy at the Summer National Bureau of Economic Research Working Paper 6614, Washington, Research Workshop on Labor Economics. July. DC. https://sites.google.com/site/rafaeldixcarneiro Artuç, E., S. Chaudhuri, and J. McLaren. 2010. /research. “Trade Shocks and Labor Adjustment: A Finger, J. M., and M. Kreinin. 1979. “A Measure Structural Empirical Approach.” American of Export Similarity and Its Possible Uses.” Economic Review 100 (3): 1008–5. Economic Journal 89 (356): 905–12. Artuç, E., D. Lederman, and G. Porto. 2015. Freund, C., and C. Ozden. 2009. “The Effect “A Mapping of Labor Mobility Costs in the of China’s Exports on Latin American Trade Developing World.” Journal of International with the World.” In China’s and India’s Chal- Economics 95 (1): 28–41 lenge to Latin America , ed. D. Lederman, Artuç, E., D. Lederman, and D. Rojas. 2015. M. Olarreaga, and G. Perry. Latin American “The Rise of China and Labor Market Development Forum Series. Washington, DC: Adjustments in Latin America.” World Bank World Bank. Policy Research Working Paper 7155, Wash- Hanson, G., and R. Robertson. 2009. “China and ington, DC. the Recent Evolution of Latin America’s Man- Autor, D. H., D. Dorn, and G. H. Hanson. 2013. ufacturing Exports.” In China’s and India’s “The China Syndrome: Local Labor Market Challenge to Latin America, ed. D. Lederman, Effects of Import Competition in the United M. Olarreaga, and G. Perry. Latin American States.” American Economic Review 103 (6): Development Forum Series. Washington, DC: 2121–68. World Bank. Bartik, T. J. 1991. Who Benefits from State and Hollweg, C., D. Lederman, D. Rojas, and E. Local Economic Development Policies? W. E. Ruppert-Bulmer. 2014. Sticky Feet: How Upjohn Institute for Employment Research, Labor Market Frictions Shape the Impact of Kalamazoo, MI. International Trade on Jobs and Wages. Direc- Bown, C. 2014. “International Shocks and Inter- tions in Development Series, Trade. Washing- national Agreements: Latin America’s Trade ton, DC: World Bank. Policy in a Multipolar World.” Paper pre- Lall, S., and J. Weiss. 2004. “China’s Competi- sented at the Authors’ Workshop on LAC in tive Threat to Latin America: An Analysis for a Multipolar World, Office of the Chief Econ- 1990–2002.” QEH Working Paper, Oxford omist for Latin America and the Caribbean, University, Oxford. World Bank Group, Washington, DC, Febru- Morten, M., and J. Oliveira. 2014. “Migration, ary 27–28. Roads and Labor Market Integration: Evidence Chiquiar, D. 2014. Presentation at the conference from a Planned Capital City.” Departments on “The Global Insertion of Asian Economies,” of Economics, Stanford University, Stanford Center for Latin American Monetary Studies CA, and Clemson University, Clemson, SC. (CEMLA)–Central Bank of Mexico, June 5–6. https://sites.google.com/site/jaquemdeoliveira www.cemla.org/actividades/2014/2014-06 /research. -InsertionAsianEconomies.html Ribe, H., D. A. Robalino, and I. Walker. 2010. Costa, F. J. M., J. Garred, and J. P. Pessoa. 2014. Achieving Effective Social Protection for All in “Winners and Losers from a Commodi- Latin America and the Caribbean. Directions ties-for-Manufactures Trade Boom.” Centre in Development Series, Human Development. for Economic Performance Discussion Paper Washington, DC: World Bank. 1269, London School of Economics and Politi- Topalova, P. 2007. “Trade Liberalization, Pov- cal Science, London. erty, and Inequality: Evidence from Indian Dix-Carneiro, R. 2014. “Trade Liberalization Districts.” In Globalization and Poverty, ed. and Labor Market Dynamics.” Econometrica A. Harrison. Chicago: University of Chicago 82 (3): 825–85. Press. The Changing Patterns of Financial Integration in Latin 4 America and the Caribbean This chapter expands the stylized facts documented in chapter 1 to describe how Latin Amer- ica and the Caribbean (LAC) has been integrating financially with both the North and the South. It shows that the largest increases took place in LAC’s investments abroad, although the investments of the rest of the world in LAC (including investment by other LAC countries) also rose. Higher gross domestic product (GDP) growth alone does not explain these patterns: LAC countries have become more important in global financial transactions even relative to their GDP. The changes reflect significant increases in portfolio investments, syndicated loans, and merger and acquisition (M&A) flows (growth of greenfield investment, which was already high, was more subdued). Despite these increases, cross-border investments into LAC coun- tries far outweigh foreign investments by LAC countries, and LAC countries have been losing ground with respect to other South regions as receivers of North flows. Moreover, although M&A flows to LAC increased, there is no evidence that they have raised labor productivity, as North-North flows often do. C hapter 1 documents several important • What role did new connections play in facts about how the South has been the evolution of these flows? growing and integrating with the • Did LAC receive investments in the North in both trade and finance. This chap- sectors in which it has a comparative ter complements those facts by addressing advantage, or were inflows related to four questions about financial integration in the comparative advantage of the coun- Latin America and the Caribbean (LAC): try sending the capital? • To what extent were these flows associ- • How are LAC countries connecting ated with labor productivity growth? financially with countries in the North, countries in other regions of the South, To shed light on these questions, this chap- and other countries in LAC, and how ter analyzes how LAC has been integrating did these connections evolve during the financially with the rest of the world and 2000s? how important it is in international financial 153 154 LATIN AMERICA AND THE RISING SOUTH transactions. The chapter also examines the investment grew less than other flows in extent to which LAC’s financial integration is recent years, but these cross-border invest- related to its trade integration and the degree ments were already well established at the to which foreign direct investment (FDI) has beginning of the 2000s, especially between increased labor productivity. LAC and countries elsewhere in the South. Four main patterns emerge from the anal- The increase in flows from LAC has ysis. First, like the rest of the South, LAC occurred in both the primary and the heavy countries appear to be increasingly connected manufacturing sectors. In contrast, the main with the rest of the world in terms of both trends driving flows to LAC have been M&A cross-border portfolio holdings and capi- and syndicated loans in the primary sector. tal flows. The largest increases took place The different growth trajectories across types in LAC’s investments abroad, although the of investment may reflect the fact that as investments of the rest of the world in LAC LAC has become more developed, investors also increased across all types of financial have become more comfortable conducting flows. more arms’ length transactions and shift- Second, despite these increases, cross- ing to types of contracts that require less or border investments into LAC countries far no actual production in the target countries outweigh foreign investments by LAC coun- (providing loans and purchasing securi- tries. LAC is more important as a receiver ties rather than opening a foreign plant, for than as a sender of investments. example).1 Third, LAC’s connections with other Increases in both the number of new con- South countries grew more rapidly than with nections (extensive margin) and the intensity North countries, especially during the second of preexisting connections (intensive margin) half of the 2000s. This growth has increased improved LAC’s connections with the rest the participation of South countries as send- of the world. For portfolio investments, the ers of resources, particularly in merger and intensive margin explains almost all of the acquisition (M&A) flows, to LAC countries. growth in cross-border holdings. In contrast, In addition, North-LAC flows have been for syndicated loans, M&A, and greenfield increasing at a slower pace than North- flows, the extensive margin plays a more South flows. LAC countries have therefore important role, especially in connections been losing ground with respect to the South between LAC and countries in other South as receivers of North flows. Despite these regions and within LAC. North-LAC flows changes, the North remains by far the prin- were well established in the 1990s; the inten- cipal source (receiver) of the flows to (from) sive margin drove their growth. LAC countries. The dynamics of trade flows partly Fourth, within-LAC flows have increased explain these patterns. Greenfield investment substantially, in some cases more than flows and trade seem to be complements: countries to the North, reflecting a higher degree of in the North and South invest in the sectors connectivity among the countries of the in which they have a relative comparative region. advantage, not necessarily in the sectors in What is behind these patterns of inte- which LAC has a comparative advantage. gration? Although higher gross domestic This complementarity is also observed in product (GDP) growth explains much of the South-LAC flows of syndicated loans. It growth, the data indicate that LAC coun- is not observed in M&A flows or North- tries have become more important in global LAC syndicated loan flows. In these cases, financial transactions even relative to GDP. foreign investments have gone to sectors in The patterns reflect large increases in portfo- which the receiver country has a comparative lio investments, syndicated loans, and M&A advantage. flows, the types of investments that experi- The search by companies for improve- enced the highest growth rates. Greenfield ments in efficiency by dispersing production T H E C H A N G I N G PAT T E R N S O F F I N A N C I A L I N T E G R AT I O N 155 stages across countries could explain the trends might change at some point and that increase in flows, especially FDI flows, across the more rapid increase in LAC’s investments countries. Therefore, the rising participa- abroad might be evidence of this shift. In net tion of South economies in global financial terms, the patterns are the counterpart of flows, particularly FDI, can be perceived as the persistent current account deficits run by a potential driver of economic growth. FDI many countries in the region. To the extent not only directly eases financing constraints that these deficits are reduced, net capital in recipient economies, it can also be a major inflows to LAC will diminish. To the extent conduit of technology diffusion and learning that LAC will have to repay the money it has spillovers. borrowed, investments in LAC are likely to The findings in this chapter provide some stabilize. Furthermore, as LAC becomes evidence that North-North M&A flows in richer, it will invest more abroad, particu- the manufacturing sector generally have a larly in the North, with which the growth positive impact on labor productivity growth, differentials are more consistently positive in whereas North-South, South-North, and LAC’s favor. South-South flows do not. These findings Third, the recent expansion in capital suggest that the rise in South connections has flows within LAC and between LAC and not led to increases in labor productivity in other regions of the South reflects an increase LAC. Differences in the sectoral composition in the extensive margin. To the extent that of M&A flows or differences in the structure these new connections are stable and coun- of ownership of target companies by North tries learn to invest in one another, it is pos- and South countries do not explain these sible that growth in the intensive margin will patterns. Trade costs are also an unlikely accelerate, following growth in the extensive explanation. margin, as countries may invest more and What do the patterns documented in this more in the links that have already been chapter mean for policymakers, research- established, especially if there is dynamic ers, and practitioners interested in LAC? learning in these connections. Although inevitably speculative in nature, Fourth, under some plausible assumptions, the broad set of stylized facts presented here the analysis on M&A and growth provides leads to some conclusions and predictions. some evidence that North-North flows gen- They also raise several questions. erally have a positive impact on labor produc- First, the observed dynamics of financial tivity growth whereas South-South flows do flows shed some light on where future expan- not. This finding is surprising, as one might sion might be. The patterns suggest that LAC expect South-South flows to have a positive is gaining ground in the types of investments impact on growth in recipient countries for that are more arms’ length. To the extent that several reasons. For instance, technologies it makes North and South countries more from a South country may be closer to tech- willing to invest in LAC using new instru- nologies of the South recipient country and ments, improving the financial contracting thus more easily adopted. Although technol- environment can ease further expansion ogy and knowledge spillovers may still take of these investments. Expansion of LAC’s place, the effects of reallocations, economies financial transactions might take place even of scale, and increased competition may be when LAC is more connected to the rest of large enough to offset them. the world financially than on the real side, in Part of the explanation for the positive particular because its financial connections effect of North-North on labor productivity with the South and other LAC countries are may be the “absorptive capacity” of firms still small relative to investments from the operating in high-income environments. North. These firms are characterized by high-skilled Second, LAC has received more flows than labor forces, superior management practices, it has sent abroad. One could argue that these and higher rates of investment in innovation. 156 LATIN AMERICA AND THE RISING SOUTH Exploring these possibilities more for- from firm-level transaction data on M&A mally remains an important topic for future from Thomson Reuters’ SDC Platinum and research. In light of the increased importance on (announced) greenfield investment from of South flows for LAC countries, a more the Financial Times ’ fDi Markets. The in-depth analysis of the differentiated effect M&A data are for 1990–2011; they cover that source countries can have on the poten- 139 source and 162 recipient countries. The tial for growth-enhancing effects of FDI also greenfield investment data are for 2003–11; seems to be in order. The role of distance they cover 157 source and 193 recipient (including all of its aspects, such as trade countries. Data on syndicated loans come and financial barriers, cultural differences, from Thomson Reuters’ SDC Platinum trans- and the degree of information asymmetries) action-level database for 1996–2012; they between source and receiver countries merits cover 111 source and 183 recipient countries. further examination as well. South countries The analysis excludes offshore financial cen- still send and receive the majority of their ters. Box 4.1 compares this bilateral data cross-border financial investments to and with the balance of payments data. Because from North countries, but neighboring South the CPIS data are on stock holdings, the esti- countries come in second place as a share of mates on portfolio assets are much larger these investments: countries in LAC typically than the estimates on syndicated loans, invest in other LAC countries. The largest M&A, and greenfield investment, which non-North recipients of FDI from LAC coun- are based on annual transactions. There- tries are other LAC countries, which account fore, these different datasets cannot be com- for 32 percent of M&A flows and 60 percent pared in terms of size. The evolution of these of greenfield investment. transactions and the differences within these datasets are very informative, however. South countries have been growing more The role of Latin America and rapidly than North countries. As a result, the Caribbean in international they now capture an important share of these financial transactions flows. Across all types of transactions, LAC Chapter 1 describes some important facts has been gaining ground, as both a receiver about international financial transactions. It and a sender (table 4.1). provides evidence that South countries have LAC countries are increasingly connected been gaining space in the global economic with the rest of the world. Investments of the landscape (Set of Facts 1). The growth of the rest of the world to LAC have increased in South is manifested in increasing South-South almost all categories. The largest increases, connections, in addition to South-North and however, have been in LAC’s investments North-South ones (Set of Facts 2). A strong abroad. LAC countries’ portfolio holdings in degree of regional clustering is observed in the rest of the world (North and South coun- both trade and financial connections (Set of tries) rose from an average of $45.3 billion Facts 3). (in 2011 U.S. dollars) in 2001–05 to an aver- What is the role of LAC in these three sets age of $152.5 billion in 2006–11. Growth of facts and other related patterns discussed in cross-border syndicated loans and M&A in chapter 1? To address this question, this flows has been substantial as well. Between chapter presents the results of analyses using 2001–05 and 2006–11, the average annual bilateral data on portfolio investments, FDI, volume of syndicated loan from LAC to the and syndicated bank loans. Data on portfo- rest of the world jumped from $2.1 billion to lio assets come from the Coordinated Portfo- $4.0 billion and the volume of M&A flows lio Investment Surveys (CPIS) conducted by rose from $4.2 billion $12.9 billion. When the International Monetary Fund (IMF) for using a longer time span the growth is even 2001–11; they cover 75 source countries and more impressive. Greenfield growth has 207 recipient countries. 2 Data on FDI come been more subdued, but these cross-border T H E C H A N G I N G PAT T E R N S O F F I N A N C I A L I N T E G R AT I O N 157 TABLE 4.1 Cross-border investment, by pairs of regions and type of investment (annual average, millions of 2011 U.S. dollars) Financial flows from region A to region B Portfolio invest- ments of region A Greenfield in region B Syndicated loans Mergers and acquisitions investment Region Region 1996– 1996– A B 2001–05 2006–11 2000 2001–05 2006–12 1990–95 2000 2001–05 2006–11 2003–05 2006–11 LAC North 44,325 146,054 598 2,055 3,614 1,362 2,331 4,193 10,065 1,991 2,705 LAC South 928 6,442 9 49 357 0 342 56 2,791 3,858 4,051 North LAC 291,555 573,452 59,914 46,498 64,932 6,489 46,961 23,333 30,935 70,923 79,262 South LAC 1,847 10,527 968 1,591 5,623 517 876 405 7,579 15,348 14,894 LAC LAC 3,475 11,370 109 558 1,619 709 3,388 3,627 6,054 10,069 10,054 Sources: Calculations based on data from International Monetary Fund (IMF) Coordinated Portfolio Investment Survey (CPIS), SDC Platinum, and fDi Markets. Note: Portfolio data are stockholdings; data on syndicated loans, mergers and acquisitions, and greenfield investment are annual transactions. The North includes the G-7 members and 19 other European countries. The South includes all other economies (excluding countries in Latin America and the Caribbean [LAC]). Sample excludes offshore centers. investments were already better established and $0.4 billion for M&A during 2001–05. at the beginning of the 2000s (compared In 2006–11 they reached $5.6 billion (an with syndicated loans and M&A).3 increase of 253 percent) and $7.6 billion (an Although there has been an important increase of 1,771 percent), respectively. In rise in the role played by LAC as a sender contrast, average annual North-LAC flows region, cross-border flows to LAC countries of syndicated loans rose just 40 percent far outweigh flows from LAC countries. (from $46.5 billion to $64.9 billion), and For syndicated loans, M&A, and greenfield flows of M&A increased just 33 percent investment, total flows to LAC countries (from $23.3 billion to $30.9 billion). More- from North and South countries in 2006–11 over, because North-LAC flows increased at were almost nine times larger than flows from a slower pace than North-South flows, LAC LAC countries to North and South countries. countries lagged other South countries in These sharp differences may explain why this regard.4 LAC’s financial connections in all types of Flows within LAC countries have also financial transactions have increased more increased substantially, in some cases more as a sender than as a receiver, especially with than flows to the North, reflecting a higher respect to other South countries. degree of connectivity within the region. Port- Another notable feature of the growth folio holdings averaged $3.5 billion during in LAC’s financial integration with the rest 2001–05; for 2006–11 they reached $11.4 of the world is that (except for greenfield billion (an increase of 227 percent). Between investment) LAC’s connections with other 2001–05 and 2006–11, the average annual South countries have been growing faster volume of syndicated loans within LAC rose than its connections with North countries, from $0.6 billion to $1.6 billion (an increase especially during the second half of the of 190 percent), and M&A flows soared from 2000s. This growth has increased the par- $3.6 billion to $6.1 billion (an increase of 67 ticipation of South countries as financiers percent). In contrast, greenfield investment— of LAC countries, particularly in M&A. the level of which was already high in the first Annual flows from South to LAC countries half of the 2000s (compared with syndicated averaged $1.6 billion for syndicated loans loans and M&A)—remained stagnant. 158 LATIN AMERICA AND THE RISING SOUTH BOX 4.1 How do bilateral data compare with balance of payments data? How do the bilateral data used in this chapter For foreign direct investment (merger and acqui- compare with the flows reported by the financial sition and greenfield flows), the two databases can account of the balance of payments (BoP)? In partic- be compared directly (figure B4.1.1). ular, do bilateral flows systematically underestimate At the region-year-level, figure B4.1.1 shows or overestimate the flows reported by the BoP, or do similar values and a significant positive correla- they move in a manner that is consistent with the tion between the two datasets. However, for South flows derived from the BoP? countries (both inflows and outflows), the bilateral BoP data come from the International Monetary data seem to slightly overestimate the flows reported Fund, which provides annual country-level infor- in the BoP accounts, possibly because the bilateral mation for 1970–2012 on different types of capital data are gross inflows whereas the BoP data are net flows, measured in current U.S. dollars. The data are inflows (net of inflows and outflows of foreigners). divided into the current account and the financial In addition, the greenfield data used in this chapter account. The financial account is divided into four reflect announced investments; they may differ from subcategories: direct investments, portfolio invest- the actual flows recorded in the BoP data. Still, at ments, other investments, and international reserve the country-year-level, the correlation between the assets. The BoP data provide aggregate figures for bilateral data and the BoP data is still high (0.89 for each country with respect to the rest of the world. outflows and 0.86 for inflows). The bilateral data need to be aggregated before the For syndicated loans, direct comparison between two databases can be compared. the bilateral data and the BoP data is not possible, FIGURE B4.1.1 Comparison between bilateral and balance of payments account data on mergers and acquisitions and greenfield investment, 2003–11 a. Flows to North countries b. Flows from North countries 1600 2500 1400 2000 1200 US$ (billions) US$ (billions) 1000 1500 800 600 1000 400 500 200 0 0 2003 2005 2007 2009 2011 2003 2005 2007 2009 2011 c. Flows to South countries d. Flows from South countries 1600 700 1400 600 US$ (billions) US$ (billions) 1200 500 1000 400 800 300 600 400 200 200 100 0 0 2003 2005 2007 2009 2011 2003 2005 2007 2009 2011 BoP data Bilateral data Sources: Calculations based on data from SDC Platinum and fDi Markets (bilateral data) and IMF (balance of payments data). Note: BoP = balance of payments. The North includes the G-7 members and 19 other European countries. The South includes all other economies (including countries in Latin America and the Caribbean).Offshore centers are excluded from the sample. (continued) T H E C H A N G I N G PAT T E R N S O F F I N A N C I A L I N T E G R AT I O N 159 BOX 4.1 How do bilateral data compare with balance of payments data? (continued) because the “other investment” category in the BoP nized under three broad subcategories: “revaluations database covers not only syndicated loans but short- due to changes in exchange rates,” “revaluations due and long-term trade credits, loans, currency, and to price changes,” and “other changes in volume.” deposits (transferable and other, such as savings and The CPIS does not contain enough information to term deposits, savings and loan shares, and shares in distinguish between transactions and other flows. credit unions), as well as accounts receivables and pay- Cross-border securities transactions can therefore be ables (IMF 1993). Thus syndicated loans enter only as derived from the CPIS only with significant noise. part of the other investment category in the BoP. These caveats notwithstanding, the analysis For portfolio investments, the BoP database cov- computes a proxy for transactions using the CPIS ers transactions in equity and debt securities, whereas holdings and measures the correlation between this the bilateral database used in this chapter (the Coor- variable and the flows covered in the BoP database. dinated Portfolio Investment Survey [CPIS]) contains Because the CPIS database does not include infor- information about the holdings of portfolio invest- mation on revaluations caused by price changes, ment securities (that is, the stock of bilateral invest- the proxy variable simply computes the difference ments). In principle, the holdings information could between the holdings at the end of the period and be used to estimate the investment flows. However, the holdings at the beginning of the period. Despite according to the CPIS guide, flows reflect changes these shortcomings, the correlation between the two associated with both transactions and other flows variables is significant (0.69 for outflows and 0.82 (IMF 2002). “Other flows” covers changes recog- for inflows). Although flows between LAC and South cement company Cemex. This $14.2 billion countries have increased more rapidly, the transaction accounted for 85 percent of LAC- North is still by far the principal source South flows between 2006 and 2011. Exclud- (receiver) of the flows to (from) LAC coun- ing these two transactions, non-LAC South tries. Figure 4.1 shows that North coun- countries would have received just 4 percent tries are still the main destination of LAC of LAC M&A flows in 2006–11 and LAC cross-border flows in portfolio invest- countries would have received 44 percent. ments, syndicated loans, and M&A. During North countries are by far the main 2006–11, North countries accounted for source of cross-border flows to LAC coun- 89 percent of LAC’s portfolio investments tries, accounting for 96 percent of portfolio abroad, 65 percent of LAC’s syndicated investments, 90 percent of syndicated loans, loans, and 53 percent of LAC’s M&A flows. 69 percent of M&A, and 76 percent of green- Greenfield investment is the only type of flow field flows in 2006–11 (figure 4.2). Given the for which North countries are not the main faster growth of South connections, however, destination of LAC flows. there has been a gradual decline in the share The South increased its participation as of North countries, particularly in M&A. receivers of M&A flows during 2006–11 Because the patterns documented are (from 1 percent to 15 percent), whereas par- expressed in constant dollars, they could ticipation by LAC countries declined (from have been driven by the fact that real eco- 46 percent to 32 percent). Two large trans- nomic activity was growing relatively rapidly actions affected these results. The first was in LAC countries. However, even relative the 2006 acquisition of the Canadian mining to LAC’s GDP, cross-border portfolio hold- company Inco by the Brazilian company Vale. ings, syndicated loans, and M&A flows to This $17.2 billion deal accounted for 28 per- and from LAC rose (exceptions are North- cent of LAC-North flows between 2006 and LAC syndicated loans and M&A flows) 2011. The second was the 2007 acquisition of (table 4.2). In contrast, greenfield flows grew the Australian Rinker Group by the Mexican more slowly than LAC’s GDP.5 160 LATIN AMERICA AND THE RISING SOUTH FIGURE 4.1 Cross-border investment shares by Latin America and Caribbean (LAC) countries in North, South, and other LAC countries, by type of investment, selected years a. Portfolio investments 2001–05 2006–11 2% 7% 7% 4% 91% 89% b. Syndicated loans 2001–05 2006–12 21% 29% 2% 77% 65% 6% c. Mergers and acquisitions 2001–05 2006–11 32% 46% 53% 53% 15% d. Greenfield investment 2003–05 2006–11 13% 16% 24% 63% 60% 24% North South LAC Sources: Calculations based on data from CPIS, SDC Platinum, and fDi Markets. Note: The North includes the G-7 members and 19 other European countries. The South includes all other economies (excluding countries in Latin America and the Caribbean [LAC]). Offshore centers are excluded from the sample. T H E C H A N G I N G PAT T E R N S O F F I N A N C I A L I N T E G R AT I O N 161 FIGURE 4.2 Cross-border investment shares by North, South, and Latin America and Caribbean (LAC) countries, by type of investment, selected years a. Portfolio investments 2001–05 2006–11 2% 2% 98% 96% b. Syndicated loans 2001–05 2006–12 2% 3% 8% 96% 90% c. Mergers and acquisitions 2001–05 2006–11 2% 13% 14% 17% 69% 85% d. Greenfield investment 2003–05 2006–11 10% 10% 16% 14% 74% 76% North South LAC Sources: Calculations based on data from CPIS, SDC Platinum, and fDi Markets. Note: The North includes the G-7 members and 19 other European countries. The South includes all other economies (excluding countries in Latin America and the Caribbean[LAC]). Offshore centers are excluded from the sample. 162 LATIN AMERICA AND THE RISING SOUTH TABLE 4.2 Shares of cross-border investment by source and receiver region, normalized by GDP of Latin America and the Caribbean (annual average, percent) Financial flows from region A to region B Portfolio invest- ments of region A Greenfield in region B Syndicated loans Mergers and acquisitions investment Region Region 1996– 1996– A B 2001–05 2006–11 2000 2001–05 2006–12 1990–95 2000 2001–05 2006–11 2003–05 2006–11 LAC North 1.63 3.15 0.02 0.07 0.08 0.06 0.08 0.16 0.24 0.08 0.06 LAC South 0.03 0.14 0.00 0.00 0.01 0.00 0.01 0.00 0.07 0.14 0.09 North LAC 10.83 12.43 2.11 1.74 1.43 0.28 1.67 0.88 0.67 2.63 1.69 South LAC 0.07 0.22 0.03 0.06 0.12 0.02 0.03 0.02 0.15 0.55 0.33 LAC LAC 0.13 0.25 0.00 0.02 0.03 0.03 0.12 0.14 0.14 0.35 0.22 Sources: Calculations based on data from CPIS, SDC Platinum, and fDi Markets. Note: GDP = gross domestic product. Portfolio figures are based on stockholdings; figures on syndicated loans, mergers and acquisitions, and greenfield investment are based on annual transactions. The North includes the G-7 members and 19 other European countries. The South includes all other economies (excluding countries in Latin America and the Caribbean [LAC]). Sample excludes offshore centers. Figure 4.3 shows the annual evolution LAC’s connections with the rest of the world of different types of LAC flows. It indicates reflects increases in both margins. that the integration of LAC countries with Figures 4.4–4.7 show the evolution of the the rest of the world has not been smooth. extensive margin for each type of investment Cross-border flows to and from LAC have and the total value of these connections. To been characterized by boom and bust pat- measure the active connections, the analysis terns. Moreover, the growth periods for computes the number of country pairs that different types of investments seem to be cor- have a positive flow in each year as a share of related, particularly in syndicated loans and the number of country pairs with a positive M&A (the data for which the sample periods or zero flow. Annex figure 4A.1 shows the are longer) and for LAC as a receiver. In both number of active connections.6 When com- cases there was an increase in flows to LAC paring the level of the extensive margin of countries during the mid-1990s, a decrease at portfolio investments with the level of syndi- the beginning of the 2000s, and a rise since cated loans, M&A, and greenfield flows, one then and until the 2008–09 global financial needs to recall that portfolio investments are crisis. The global financial crisis seems to stocks and the other measures are flows. The have had a different effect on these two types extensive margins for these types of transac- of investments (box 4.2). tions are thus expected to be very different.7 In general, the evidence shows that for all types of investment, LAC countries are con- Growth in the intensive and nected with a higher percentage of North extensive margins countries than South ones. Nevertheless, How much of the growth in LAC’s connec- since the beginning of the 2000s, LAC coun- tions reflects growth in the intensive mar- tries have broadened their connections with gin (which captures increases in the value of the South, including other LAC countries. transactions for existing connections) and Moreover, for syndicated loans and M&A how much reflects growth in the extensive flows, there is a noticeable downward trend margin (which captures increases in the pro- in the percentage of North-LAC active con- portion of active connections)? The analy- nections. The evidence also shows that the sis in this section shows that the growth of percentage of active connections is greater T H E C H A N G I N G PAT T E R N S O F F I N A N C I A L I N T E G R AT I O N 163 FIGURE 4.3 Cross-border investment to and from countries in Latin America and the Caribbean a. Portfolio investments 800 700 600 2011 constant US$ (billions) 500 400 300 200 100 0 90 91 92 93 94 95 96 97 98 99 00 01 02 03 04 05 06 07 08 09 10 11 19 19 19 19 19 19 19 19 19 19 20 20 20 20 20 20 20 20 20 20 20 20 b. Syndicated loans 150 125 2011 constant US$ (billions) 100 75 50 25 0 90 91 92 93 94 95 96 97 98 99 00 01 02 03 04 05 06 07 08 09 10 11 19 19 19 19 19 19 19 19 19 19 20 20 20 20 20 20 20 20 20 20 20 20 LAC as sender LAC as receiver (continued) 164 LATIN AMERICA AND THE RISING SOUTH FIGURE 4.3 Cross-border investment to and from countries in Latin America and the Caribbean (continued) c. Mergers and acquisitions 150 125 2011 constant US$ (billions) 100 75 50 25 0 90 91 92 93 94 95 96 97 98 99 00 01 02 03 04 05 06 07 08 09 10 11 19 19 19 19 19 19 19 19 19 19 20 20 20 20 20 20 20 20 20 20 20 20 d. Greenfield investment 150 125 2011 constant US$ (billions) 100 75 50 25 0 90 91 92 93 94 95 96 97 98 99 00 01 02 03 04 05 06 07 08 09 10 11 19 19 19 19 19 19 19 19 19 19 20 20 20 20 20 20 20 20 20 20 20 20 LAC as sender LAC as receiver Sources: Calculations based on data from CPIS, SDC Platinum, and fDi Markets. Note: Offshore centers are excluded from the sample. LAC = Latin America and the Caribbean. T H E C H A N G I N G PAT T E R N S O F F I N A N C I A L I N T E G R AT I O N 165 BOX 4.2 How did the global financial crisis affect investment in and by the region? The effects of the 2008–09 global crisis varied across other developed countries’ sovereign debt), whose types of investment, as figure 4.3 indicates. Portfolio prices fell by much less. holdings and syndicated loan flows to Latin America The crisis did not affect flows of syndicated loans and the Caribbean (LAC) countries declined more from or to LAC countries until 2009, when it sharply than flows from LAC countries. This finding is con- hit flows to and to a lesser extent from LAC. In both sistent with evidence showing that foreign investors cases the effects have persisted: even in 2011 flows to pulled out sharply from emerging economies when and from LAC were smaller than they were in 2007. the crisis hit (Broner and others 2013). In contrast, M&A seems to be the only case in which flows merger and acquisition (M&A) flows from LAC fell from LAC were affected significantly more than more than M&A flows to LAC. The crisis did not flows to LAC. M&A flows from LAC decreased appear to affect greenfield flows to or from LAC. 77 percent in 2008. Part of this decrease reflects the This finding is consistent with a large body of litera- fact that the 2007 flow was very large because of the ture showing that foreign direct investment tends to $14.2 billion acquisition of the Australian Rinker be more stable than other types of flows (Sarno and Group by the Mexican cement company Cemex, Taylor 1999; Levchenko and Mauro 2007). which represented 41 percent of 2007 LAC M&A Several examples illustrate the behavior of dif- outflows. Even excluding this deal, however, the ferent types of flows and the magnitude of their decline was significant (61 percent) and much larger collapse during the global financial crisis. Portfolio than the decrease in flows to LAC (26 percent). In investments by LAC countries abroad decreased by addition, the contractionary effects of the crisis 18 percent in 2008. This decline was much more lasted longer in the case of flows from LAC. moderate than the 44 percent drop of foreign port- The crisis also affected the extensive margin of folio investments in LAC. However, both reductions the cross-border investments, especially for North- are significant given that these values are stocks LAC connections. Following the crisis, the percent- (not flows). These effects were temporary, however: age of active North-LAC connections decreased for by 2009 both holdings were very close to their 2007 portfolio holdings, syndicated loans, and M&A values. flows. Although the downward trend for syndicated The behavior of asset prices during the crisis may loans and M&A flows started at the beginning of explain the size of these fluctuations. As De la Torre the 2000s, the decrease was steeper after the cri- and others (2010, 2012) note, foreign investors held sis. In contrast, for portfolio holdings the extensive equity positions in LAC, whose value dropped sub- margin of North-LAC connections had been increas- stantially during the crisis, whereas LAC investors ing steadily until 2007; it has decreased every year held debt abroad (including U.S. Treasury bonds and since the crisis. when LAC is a receiver: the percentage of LAC to North countries was around seven active North-LAC links is larger than the times greater than the percentage of active percentage of active LAC-North links (except connections between LAC and South coun- for portfolio investments), and the percentage tries. Since the beginning of the 2000s, of active South-LAC links exceeds the per- however, LAC countries have broadened centage of active LAC-South links. their connections with the South, including Regarding portfolio investments, figure other LAC countries. The share of active 4.4 shows that LAC countries are connected LAC-South connections increased from (as both receivers and senders) with a much 4 percent in 2001 to almost 11 percent in higher percentage of North countries than 2011, for example, and the share of active South ones. Over the entire sample period, LAC-LAC connections rose from 24 percent the percentage of active connections from to 42 percent. 166 LATIN AMERICA AND THE RISING SOUTH FIGURE 4.4 Cross-border holdings of and extensive margin for portfolio investments, 2001–11 a. Latin America and the b. Latin America and the Caribbean to the North Caribbean to the South 200 80 12 14 180 70 2011 constant US$ (billions) 2011 constant US$ (billions) 10 12 160 Active connections (%) Active connections (%) 60 140 10 8 120 50 8 100 40 6 80 6 30 60 4 4 20 40 2 2 20 10 0 0 0 0 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 c. North to Latin America d. South to Latin America and the Caribbean and the Caribbean 800 60 16 16 2011 constant US$ (billions) 700 50 2011 constant US$ (billions) 14 14 Active connections (%) Active connections (%) 600 12 12 40 500 10 10 400 30 8 8 300 6 6 20 200 4 4 10 100 2 2 0 0 0 0 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 e. Latin America and the Caribbean to Latin America and the Caribbean 14 45 Active connections (right axis) 40 Total holdings 2011 constant US$ (billions) 12 Active connections (%) 35 10 30 8 25 6 20 15 4 10 2 5 0 0 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 Source: Calculations based on data from CPIS. Note: The extensive margin is the percentage of active connections—the number of country pairs that have a positive investment in each year divided by the number of country pairs with a positive or zero value in the last year of the sample. The North includes the G-7 members and 19 other European coun- tries. The South includes all other economies (excluding countries in Latin America and the Caribbean). Offshore centers are excluded from the sample. Figure 4.5 shows that for syndicated loans, was 11 percent for North-LAC connections LAC countries as receivers are much more and less than 2 percent for South-LAC and connected with North countries than with LAC-LAC connections. The larger number South countries, including other countries in of banks in the North that have tradition- LAC. In 2011 the share of active connections ally engaged in syndicated loans may explain T H E C H A N G I N G PAT T E R N S O F F I N A N C I A L I N T E G R AT I O N 167 FIGURE 4.5 Cross-border flows of and extensive margin for syndicated loans, 1996–2012 a. Latin America and the b. Latin America and the Caribbean to the North Caribbean to the South 1.4 0.30 7 4.5 2011 constant US$ (billions) 2011 constant US$ (billions) 4.0 1.2 Active connections (%) Active connections (%) 6 0.25 3.5 5 1.0 3.0 0.20 4 0.8 2.5 0.15 3 2.0 0.6 1.5 0.10 2 0.4 1.0 1 0.2 0.05 0.5 0 0.0 0.0 0.00 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 c. North to Latin America and the d. South to Latin America and the Caribbean 10 Caribbean 1.2 2011 constant US$ (billions) 2011 constant US$ (billions) 120 20 Active connections (%) Active connections (%) 9 18 1.0 100 8 16 7 14 0.8 80 12 6 60 10 5 0.6 8 4 40 0.4 6 3 4 2 20 0.2 2 1 0 0 0 0.0 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 e. Latin America and the Caribbean to 2012 Latin America and the Caribbean Active connections (right axis) 4 3.0 Total flows 2011 constant US$ (billions) Active connections (%) 2.5 3 2.0 2 1.5 1.0 1 0.5 0 0.0 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 Source: Calculations based on data from SDC Platinum. Note: The extensive margin is the percentage of active connections—the number of country pairs that have a positive flow in each year divided by the num- ber of country pairs with a positive or zero flow in the last year of the sample. The North includes the G-7 members and 19 other European countries. The South includes all other economies (excluding countries in Latin America and the Caribbean). Offshore centers are excluded from the sample. these figures. However, there is a clear down- within LAC. The percentage of active sender ward trend in both the extensive margin and connections and the total amount financed by total flows of North-LAC connections (espe- LAC in other regions is very low, suggesting cially during the crisis years) and an upward that banks in LAC are still not engaging in trend in connections from South to LAC and this type of transaction across borders. For 168 LATIN AMERICA AND THE RISING SOUTH example, flows from LAC to other countries In line with the previous findings, fig- reached $5.3 billion in 2011, about 8 percent ure 4.6 shows that during the 1990s, LAC of North-LAC flows. Possibly because of the countries as receivers of M&A were much small share of the extensive margin, syndi- more connected with North countries than cated loans display the greatest degree of vol- with South countries, including countries in atility in the percentage of active connections. LAC. In 1999 the share of active connections FIGURE 4.6 Cross-border flows of and extensive margin for mergers and acquisitions, 1990–2011 a. Latin America and the b. Latin America and the Caribbean to the North Caribbean to the South 30 4.0 18 0.25 2011 constant US$ (billions) 2011 constant US$ (billions) 3.5 16 25 Active connections (%) Active connections (%) 14 0.20 3.0 20 12 2.5 0.15 10 15 2.0 8 1.5 0.10 10 6 1.0 4 0.05 5 0.5 2 0 0.0 0 0.00 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 c. North to Latin America d. South to Latin America and the Caribbean and the Caribbean 80 16 25 1.4 2011 constant US$ (billions) 2011 constant US$ (billions) Active connections (%) Active connections (%) 70 14 1.2 20 60 12 1.0 50 10 15 0.8 40 8 0.6 30 6 10 20 4 0.4 5 10 2 0.2 0 0 0 0.0 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 e. Latin America and the Caribbean to Latin America and the Caribbean Active connections (right axis) 10 5.0 Total flows 2011 constant US$ (billions) 9 4.5 Active connections (%) 8 4.0 7 3.5 6 3.0 5 2.5 4 2.0 3 1.5 2 1.0 1 0.5 0 0.0 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 Source: Calculations based on data from SDC Platinum. Note: The extensive margin is the percentage of active connections—the number of country pairs that have a positive flow in each year divided by the num- ber of country pairs with a positive or zero flow in the last year of the sample. The North includes the G-7 members and 19 other European countries. The South includes all other economies (excluding countries in Latin America and the Caribbean). Offshore centers are excluded from the sample. T H E C H A N G I N G PAT T E R N S O F F I N A N C I A L I N T E G R AT I O N 169 was almost 10 percent for North-LAC, stock holdings during 2006–11, few new 0.1 percent for South-LAC, and 3 percent for connections were created; the increase LAC-LAC connections. However, since the reflects a deepening of the intensive margin 2000s there has been a downward trend in (see table 4.3). Average LAC-North holdings the extensive margin of North-LAC connec- increased from $44.3 billion in 2001–05 to tions and an upward trend in the extensive $146.1 billion in 2006–11, but only 0.08 margins of LAC-LAC and South-LAC con- percent of the increase was attributable to nections. Although these developments have new connections. The increase in North- narrowed the gap in the extensive margin LAC investments attributable to new con- across regions, the North-LAC percentage nections was only 0.1 percent. This pattern of active connections still remains larger. In reflects the fact that even at the beginning 2011, 7 percent of North-LAC, 4 percent of of the 2000s, the extensive margin (or the LAC-LAC, and 1 percent of South-LAC links portfolio links) between LAC and North were active. As senders, LAC countries are countries had already been well established. equally connected with North countries and For South-LAC and LAC-LAC links, there other LAC countries: in 2011 the share of was some increase in investments as a result both types of active connections was about of new connections, but this increase rep- 4 percent. In contrast, the share of active con- resents less than 10 percent of the expansion nections (and the total amount financed) by in cross-border holdings. LAC in South countries is very low and dis- For syndicated loans and M&A, new plays significant volatility. connections played a more important role Figure 4.7 shows that greenfield flows to in augmenting cross-border flows, espe- and from LAC countries share three char- cially LAC-South and LAC-LAC flows. For acteristics that are not observed in the other example, within LAC, 92 percent of syn- types of investments. First, the share of active dicated loan flows during 2006–11 were connections within LAC (6 percent) is larger attributable to connections established since than it is for LAC-North links (4 percent). 1996–2000. New connections represented Second, the average value of LAC-South and 57 percent of 2006–11 M&A flows. Even LAC-LAC links is higher than the value of between 2001–05 and 2006–11 there was a LAC-North connections. Third, there is an significant increase in flows between South upward trend in the percentage of active and LAC (and within LAC) as a result of North-LAC connections, reflecting the new connections, suggesting that syndicated increasing number of North countries invest- loan and M&A links are still expanding. ing in LAC. New connections represented a much smaller To explicitly capture the growth of the fraction of North-LAC flows in 2006–11 (2 intensive margin, table 4.3 shows both the percent for syndicated loans and 18 percent evolution of the flows for different regions for M&A). This result suggests that North- with respect to LAC and the share of the LAC links were already well established in increase in these flows that is driven by new the 1990s. For greenfield investment, a large connections relative to the initial period (for fraction of the 2006–11 flows was attribut- each type of flow) and the previous period. able to new connections, mostly LAC-South Overall, the intensive margin accounts and South-LAC links. for almost all of the growth of cross-border To capture the growth in the extensive and portfolio investments. It also explains North- intensive margins more formally, tables 4.4 LAC flows. For other types of investments, and 4.5 show the results of regressions that the extensive margin plays a more import- include source and receiver fixed effects and ant role, especially in LAC-South and within gravity controls. The extensive margin regres- LAC links. sions (table 4.4) are probit regressions in For portfolio investments, although there which the dependent variable is an indicator was an important increase in the value of variable that takes the value of 1 when there 170 LATIN AMERICA AND THE RISING SOUTH FIGURE 4.7 Cross-border flows of and extensive margin for greenfield investment, 2003–11 a. Latin America and the b. Latin America and the Caribbean to the North Caribbean to the South 6 6 12 0.8 2011 constant US$ (billions) 2011 constant US$ (billions) 5 5 10 Active connections (%) Active connections (%) 0.6 4 4 8 3 3 6 0.4 2 2 4 0.2 1 1 2 0 0 0 0.0 2003 2004 2005 2006 2007 2008 2009 2010 2011 2003 2004 2005 2006 2007 2008 2009 2010 2011 c. North to Latin America and the d. South to Latin America and the Caribbean Caribbean 120 25 18 2.5 16 2011 constant US$ (billions) 100 2011 constant US$ (billions) 20 2.0 Active connections (%) Active connections (%) 14 80 12 15 1.5 10 60 8 10 1.0 40 6 5 4 0.5 20 2 0 0 0 0.0 2003 2004 2005 2006 2007 2008 2009 2010 2011 2003 2004 2005 2006 2007 2008 2009 2010 2011 e. Latin America and the Caribbean to Latin America and the Caribbean 16 9 14 8 2011 constant US$ (billions) Active connections (right axis) Active connections (%) 12 7 Total flows 6 10 5 8 4 6 3 4 2 2 1 0 0 2003 2004 2005 2006 2007 2008 2009 2010 2011 Source: Calculations based on data from fDi Markets. Note: The extensive margin is the percentage of active connections—the number of country pairs that have a positive flow in each year divided by the number of country pairs with a positive or zero flow in the last year of the sample. The North includes the G-7 members and 19 other European countries. The South includes all other economies (excluding countries in Latin America and the Caribbean). Offshore centers are excluded from the sample. is a positive bilateral flow between the two differences in latitude and longitude, differ- countries involved and 0 otherwise. These ences in time zones, whether they share a regressions include gravity control variables, common language, whether they have a com- which help explain different levels of financial mon legal origin, and whether the receiver flows between each country pair based on the (sender) country is (or was) a colony of the geographic distance between the countries, sender (receiver). The regressions also control TABLE 4.3 Intensive margin of financial connections across regions Financial flows from region A to region B Portfolio invest- ments of region Greenfield A in region B Syndicated loans Mergers and acquisitions investment Region Region 2001– 2006– 1996– 2001– 2006– 1990– 1996– 2001– 2006– 2003– 2006– A B Item 05 11 2000 05 12 95 2000 05 11 05 11 LAC North Value of investment (millions of 2011 dollars) 44,325 146,054 598 2,055 3,614 1,362 2,331 4,193 10,065 1,991 2,705 Investment due to nonexisting links during 0.08 35 62 13 5 11 11 initial period (percent) Investment due to nonexisting links during 0.08 35 16 13 59 13 11 previous period (percent) LAC South Value of investment (millions of 2011 dollars) 928 6,442 9 49 357 0 342 56 2,791 3,858 4,051 Investment due to nonexisting links during 10 100 89 100 100 100 71 initial period (percent) Investment due to nonexisting links during 10 100 96 100 100 100 71 previous period (percent) North LAC Value of investment (millions of 2011 dollars) 291,555 573,452 59,914 46,498 64,932 6,489 46,961 23,333 30,935 70,923 79,262 Investment due to nonexisting links during 0.10 2 2 5 7 18 5 initial period (percent) Investment due to nonexisting links during 0.10 2 2 5 3 10 5 previous period (percent) South LAC Value of investment (millions of 2011 dollars) 1,847 10,527 968 1,591 5,623 517 876 405 7,579 15,348 14,894 Investment due to nonexisting links during 6 9 25 79 80 85 42 initial period (percent) Investment due to nonexisting links during 6 9 29 79 66 56 42 previous period (percent) LAC LAC Value of investment (millions of 2011 dollars) 3,475 11,370 109 558 1,619 709 3,388 3,627 6,054 10,069 10,054 Investment due to nonexisting links during 6 72 92 55 26 57 16 initial period (percent) Investment due to nonexisting links during 6 72 28 55 3 33 16 previous period (percent) Sources: Calculations based on data from CPIS, SDC Platinum, and fDi Markets. Note: This table provides statistics on portfolio investments, syndicated loans, mergers and acquisitions, and greenfield cross-border flows and shows how much of the increase was driven by new connections relative to the initial period and relative to the previous period. The North includes the G-7 members and 19 other European countries. The South includes all other economies (excluding countries in Latin America and the Caribbean). Sample excludes offshore centers. LAC = Latin America and the Caribbean. T H E C H A N G I N G PAT T E R N S O F F I N A N C I A L I N T E G R AT I O N 171 172 LATIN AMERICA AND THE RISING SOUTH for source- and target-country dummies and in financial connections across regions. The region-pair dummies (North-North, North- reported results correspond to the differences South, North-LAC, South-North, South- between these trend coefficients. South, South-LAC, LAC-North, LAC-South, The regressions in table 4.5 are ordinary and LAC-LAC). Having controlled for these least squares regressions with the log of the factors, the regressions measure the trends bilateral flows (the value of the connections) TABLE 4.4 Extensive margin of cross-border financial flows Dependent variable: 1 if investment > 0, 0 otherwise Portfolio Mergers and Greenfield investments Syndicated loans acquisitions investment Differences between trend coefficients (LAC-North) trend – (South-North) trend 0.014 0.008 –0.002 –0.021 (LAC-South) trend – (South-South) trend –0.051*** 0.013 –0.016 –0.009 (LAC-LAC) trend – (LAC-South) trend –0.012 –0.005 –0.002 0.019 (North-LAC) trend – (North-South) trend –0.025*** –0.022*** –0.017*** 0.024** (South-LAC) trend – (South-South) trend –0.042*** –0.011 0.008 0.027** Number of observations 120,078 264,401 386,584 217,350 Sources: Calculations based on data from CPIS, SDC Platinum, and fDi Markets. Note: Regressions are probit models in which the dependent variable is an indicator variable that takes the value of 1 when there is a bilateral positive flow (or investment holding) between two countries and 0 otherwise. Data are aggregated at the country-country level. The regressions include gravity control variables that help explain the levels of financial flows between each country pair based on the geographic distance between the countries, differences in latitude and longitude, differences in time zones, whether they share a common language, whether they have a common legal origin, and whether the receiver (sender) country is (or was) a colony of the sender (receiver). They also control for source- and target-country dummies and region-pair dummies. The North includes the G-7 members and 19 other European countries. The South includes all other economies (excluding countries in Latin America and the Caribbean [LAC]). Sample excludes offshore centers. Standard errors are clustered by country pairs. Significance level: * = 10 percent, ** = 5 percent, *** = 1 percent. TABLE 4.5 Region-to-region financial flows Mergers and Portfolio investments Syndicated loans acquisitions Greenfield investment Log(hold- Log(flows Log(flows Log(flows ings scaled scaled by scaled by scaled by Log(holdings) by GDP) Log(flows) GDP) Log(flows) GDP) Log(flows) GDP) Differences between trend coefficients (LAC-North) trend – 0.049** 0.060*** -0.006 0.007 –0.056** –0.046* 0.021 0.025 (South-North) trend (LAC-South) trend – –0.018 –0.008 0.065 0.063 –0.095 –0.092 –0.082 –0.075 (South-South) trend (LAC-LAC) trend – –0.066 –0.056 –0.092* –0.067 0.004 0.028 0.115 0.115 (LAC-South) trend (North-LAC) trend – –0.091*** –0.081*** –0.067*** 0.051*** –0.056*** –0.044*** 0.024 0.028 (North-South) trend (South-LAC) trend – –0.116*** –0.107*** –0.051*** –0.036* –0.068** –0.057** –0.085* –0.079* (South-South) trend Number of observations 6,012 6,012 5,089 5,089 6,160 6,160 4,601 4,601 R–squared 0.968 0.987 0.982 0.992 0.933 0.980 0.969 0.988 Sources: Calculations based on data from CPIS, SDC Platinum, and fDi Markets. Note: This table includes both country-region (outflows) and region-country (inflows) observations. When indicated, flows (or holdings) are scaled using the geometric mean between the sender’s and receiver’s GDPs. Control variables are country-region dummies. The North includes the G-7 members and 19 other European countries. The South includes all other economies (excluding countries in Latin America and the Caribbean [LAC]). Sample excludes offshore centers. Significance level: * = 10 percent, ** = 5 percent, *** = 1 percent. T H E C H A N G I N G PAT T E R N S O F F I N A N C I A L I N T E G R AT I O N 173 as the dependent variable. Unlike the regres- the initial period for greenfield is 2003–05 sions in table 4.4, these regressions use instead of 2001–05). regional (not country-level) data. There is Portfolio holdings and greenfield flows therefore no need to use gravity controls. exhibit a higher degree of connectivity, even Because the regressions drop all observations in the first years of the 2000s. This pattern in which the bilateral flows are equal to zero is expected in the case of portfolio hold- (because of the use of logs), they isolate the ings, given that they are stock measures. For effect of the extensive margin and thus cap- greenfield investment, the result may indicate ture changes in the intensive margin. How- that this type of investment is the preferred ever, this measure of the intensive margin mode by LAC participants investing within differs from the more precise one used in the region. There were 105 portfolio holding table 4.3, which accounts only for the growth connections during 2001–05, and 59 green- in the intensity of the connections that were field flows during 2003–05. Greenfield links previously established. The reported results grew significantly in the years that followed, correspond to the differences between the with the number of connections reaching 122 trend coefficients. during 2006–11; the number of portfolio The main conclusion from tables 4.4 holding links increased less rapidly, to 124. and 4.5 is that North-LAC connections are During 2001–05 there were far fewer syndi- increasing more slowly than North-South cated loan links (66) and M&A connections connections in both the intensive and exten- (38). The number of M&A links did not sive margins, except for greenfield invest- increase significantly over the 1996–2000 ment. LAC is therefore losing ground to other period, when there were 48 links within regions of the South in terms of flows from LAC. In contrast, the number of links for the North. In addition, the regressions for syndicated loans rose from just 10. the intensive margin show that South-LAC Figure 4.9 describes LAC-South con- connections are increasing more slowly than nections. As in LAC-LAC links, there is South-South connections. In contrast, for increasing connectivity in all four types of LAC as a sender, there is no clear evidence investments. Except in the case of portfolio that LAC-North (LAC-South) connections investments, however, the number of links is are increasing more slowly than South-North much lower than it is within LAC. Moreover, (South-South) connections. just two countries, Brazil and Mexico, seem The maps in chapter 1 show how LAC to be driving the flows. Brazil accounted has become more connected with the rest of for 10 of the 17 connections between LAC the world. This chapter uses similar maps and South countries for syndicated loans in to show the connections between LAC and 2006–11; these links represented 91 percent countries in other South regions and coun- of the value of these flows. Mexico accounted tries in LAC. This analysis highlights the role for 7 of the 15 LAC-South links for M&A that large countries (in particular, Brazil and (93 percent of the value of these flows).8 For Mexico) have played. It shows that Brazil and greenfield flows, Brazil and Mexico together Mexico seem to drive LAC-South connec- accounted for 43 of the 69 links (83 percent tions but that their role in LAC-LAC links is of the value of these flows). more subdued. In contrast, Brazil and Mexico play a much Figure 4.8 depicts every connection less critical role as senders within LAC (see within LAC with flows greater than $1 mil- figure 4.8). Other countries, such as Chile lion (measured at 2011 prices). It shows that or Colombia, are also important. Of the 38 between 2001–05 and 2006–11, the number syndicated loan connections within LAC in of connections rose 18 percent for portfolio 2006–11, Brazil and Mexico accounted for links, 46 percent for syndicated loans, 94 just 7, representing 47 percent of the value percent for M&A, and 107 percent for green- of these flows. They accounted for 18 of field investment (because of data restrictions, the 62 M&A links (46 percent of the value 174 LATIN AMERICA AND THE RISING SOUTH FIGURE 4.8 Extensive margin of cross-border financial flows within Latin America and the Caribbean, by type of investment, selected years a. Portfolio investments 2001–05 2006–11 b. Syndicated loans 1996–2000 2001–05 2006–12 c. Mergers and acquisitions 1996–2000 2001–05 2006–11 d. Greenfield investment 2003–05 2006–11 Sources: Calculations based on data from CPIS, SDC Platinum, and fDi Markets. Note: Each line represents a flow or stock greater than $1 million (measured at 2011 prices) between two countries in Latin America and the Caribbean. Offshore centers are excluded from the sample. T H E C H A N G I N G PAT T E R N S O F F I N A N C I A L I N T E G R AT I O N 175 FIGURE 4.9 Extensive margin of cross-border financial flows from Latin America and the Caribbean to countries in other regions of the South, by type of investment, selected years a. Portfolio investments 2001–05 2006–11 b. Syndicated loans 1996–2000 2001–05 2006–12 c. Mergers and acquisitions 1996–2000 2001–05 2006–11 d. Greenfield investment 2003–05 2006–11 Sources: Calculations based on data from CPIS, SDC Platinum, and fDi Markets. Note: Each line represents a flow or stock greater than $1 million (measured at 2011 prices) between a country in Latin America or the Caribbean (LAC) and a country in another South region. The South includes all countries outside LAC that are not in the North (G-7 members and 19 other European countries). Offshore centers are excluded from the sample. 176 LATIN AMERICA AND THE RISING SOUTH of these flows). In greenfield investment, the and apparel (including leather); and wood two countries accounted for only 30 of the and paper-related products. The heavy man- 122 links but accounted for 62 percent of the ufacturing sector includes the following value of the flows. subsectors: refined petroleum and related products, chemicals and plastics, nonmetal- lic minerals, metals, machinery and equip- Financial flows and trade flows ment, and transport equipment.10 The globalization of LAC, which started in Figure 4.10 shows the average flows to and the late 1980s and continued strongly during from LAC countries by receiver and sender the 1990s, accelerated and intensified in the region, as well as the sectoral composition 2000s. A growing body of evidence suggests of financial flows for different sample peri- that the patterns of financial globalization ods. The patterns for LAC as a sender show changed during the 2000s.9 Chapter 1 docu- that no single sector explains the increase ments some important facts about the nature in financial flows from LAC countries. For of these changes. In particular, it shows that LAC-South and LAC-LAC flows, the pri- LAC is increasingly connected with other mary sector drove the growth in syndicated South countries in both trade and finance loans, and the heavy manufacturing sector (Set of Facts 2). largely accounted for the increase in M&A In addition to size, one aspect of both trade and greenfield flows. In contrast, for LAC- and financial flows that has been changing North flows, the heavy manufacturing sector significantly for LAC (as well as other South accounted for most of the growth in syndi- countries) is their composition. The sectoral cated loans, and the primary sector powered composition of LAC’s connections with other the increase in M&A and greenfield flows. South countries is generally different from the For LAC as a receiver, the primary sector composition of its connections with North drove the growth in syndicated loans and countries, in both trade and finance (Set of M&A flows. The value of North-LAC syn- Facts 3). An important question is the extent dicated loans to the primary sector grew 175 to which financial flows reflect the dynamics percent between 2001–05 and 2006–11, and of trade connections. This section sheds light the value of M&A flows rose 360 percent. on the links between these two types of flows Flows to the heavy manufacturing sector and the importance of the link for LAC. also increased during this period, although Here the analysis studies the role played growth was more subdued. In contrast, there by the different sectors in the growth in was a small decrease in both North-LAC and financial flows to and from LAC countries. South-LAC greenfield flows. The reduction It also examines the links between trade and in flows to the primary sector accounts for financial flows in LAC. Sector-level data on the decline in North-LAC flows, whereas the foreign investments (M&A and greenfield) decrease in both the primary and heavy man- and syndicated loans are matched with sec- ufacturing sectors accounts for the drop in tor-level trade data from Comtrade covering South-LAC flows. 14 sectors in 215 countries during 1990– There has been growing interest in 2012. For ease of exposition of the broad understanding the link between interna- trends, the analysis groups these sectors into tional trade and financial flows. The clas- three broad categories: primary, light manu- sical Heckscher-Ohlin-Mundell paradigm facturing, and heavy manufacturing sectors. predicts that trade is an important factor The primary sector includes the following in international capital flows. It argues that subsectors: agriculture, hunting, forestry, exports are based on endowments, the North and fishing; mining; and crude petroleum exports capital, and trade and capital flows and natural gas. The light manufacturing are substitutes. Countries invest in countries sector includes the following subsectors: to which they cannot export their goods, food, beverages, and tobacco; textiles thereby gaining access to domestic markets. T H E C H A N G I N G PAT T E R N S O F F I N A N C I A L I N T E G R AT I O N 177 FIGURE 4.10 Sectoral composition of cross-border financial flows to and from Latin America and the Caribbean, by type of investment, selected years a. Latin America and the b. Latin America and the Caribbean as sender Caribbean as receiver Syndicated loans 1996–2000 LAC–North 1996–2000 North–LAC 2001–05 2006–12 2001–05 1996–2000 2006–12 LAC–South 2001–05 2006–12 1996–2000 South–LAC 1996–2000 2001–05 LAC–LAC 2001–05 2006–12 2006–12 0 0.2 0.4 0.6 0.8 1.0 1.2 0 5 10 15 20 25 30 35 40 45 50 Average flow, 2011 constant US$ (billions) Average flow, 2011 constant US$ (billions) Mergers and acquisitions 1990–95 LAC–North 1990–95 1996–2000 North–LAC 2001–05 1996–2000 2006–11 2001–05 1990–95 LAC–South 2006–11 1996–2000 2001–05 1990–95 2006–11 South–LAC 1990–95 1996–2000 LAC–LAC 1996–2000 2001–05 2001–05 2006–11 2006–11 0 2 4 6 8 10 12 0 2 4 6 8 10 12 14 16 18 Average flow, 2011 constant US$ (billions) Average flow, 2011 constant US$ (billions) Greenfield investment LAC–LAC LAC–South LAC–North 2003–05 North–LAC 2003–05 2006–11 2006–11 2003–05 2006–11 South–LAC 2003–05 2003–05 2006–11 2006–11 0 1 2 3 4 5 6 7 8 0 10 20 30 40 50 60 Average flow, 2011 constant US$ (billions) Average flow, 2011 constant US$ (billions) Primary Light manufacturing Heavy manufacturing Sources: Calculations based on data from SDC Platinum and fDi Markets. Note: Primary sector includes the following subsectors: agriculture, hunting, forestry, and fishing; mining; and crude petroleum and natural gas. Light manufacturing sector includes the following subsectors: food, beverages, and tobacco; textiles and apparel (including leather); and wood and paper-related products. Heavy manufacturing sector includes the following subsectors: refined petroleum and related products, chemicals and plastics, nonmetallic minerals, metals, machinery and equipment, and transport equipment. The North includes the G-7 members and 19 other European countries. The South includes all other economies (excluding countries in Latin America and the Caribbean [LAC]). Offshore cen- ters are excluded from the sample. 178 LATIN AMERICA AND THE RISING SOUTH As a consequence, trade integration reduces gravity models—in which aggregate trade incentives for capital to flow to capital-scarce is one of the key variables capturing dis- countries. tance and transaction costs—explain cap- Recent theoretical work on international ital flows.11 The most disaggregated level at investments argues that trade and capital which the links between financial and trade flows can be complements rather than sub- flows have been studied is the country pair stitutes and that the South exports capital to level, generally using pooled data on both the North (Antràs and Caballero 2009; Ju exports and imports. and Wei 2011; Jin 2012). Part of these effects The empirical relevance of the interac- may be rooted in firm-level motives to export tion between trade and capital flows is not and invest abroad (Greenaway and Kneller yet fully understood. In particular, little is 2007; Alfaro and Charlton 2009). known about the cross-country sectoral allo- Some empirical papers use data from cation of capital and how it is related to the the early 2000s to understand whether sectoral composition of exports. The analysis FIGURE 4.11 Sectoral composition of cross-border financial flows to and from Latin America and the Caribbean, by type of investment, 2003–11 average a. Syndicated loansa b. Mergers and acquisitions LAC-North 13 17 70 LAC-North 42 30 28 LAC-South 91 9 LAC-South 8 23 69 North-LAC 55 9 35 North-LAC 48 28 24 South-LAC 63 4 33 South-LAC 90 6 3 LAC-LAC 59 10 31 LAC-LAC 27 39 34 0 10 20 30 40 50 60 70 80 90 100 0 10 20 30 40 50 60 70 80 90 100 Percent Percent c. Greenfield investment d. Trade LAC-North 28 8 64 LAC-North 29 17 54 LAC-South 62 4 34 LAC-South 43 33 25 North-LAC 31 8 61 North-LAC 6 14 80 South-LAC 38 4 58 South-LAC 12 17 71 LAC-LAC 42 5 53 LAC-LAC 19 23 58 0 10 20 30 40 50 60 70 80 90 100 0 10 20 30 40 50 60 70 80 90 100 Percent Percent Primary Light manufacturing Heavy manufacturing Sources: Calculations based on data from SDC Platinum, fDi Markets, and Comtrade. Note: Primary sector includes the following subsectors: agriculture, hunting, forestry, and fishing; mining; and crude petroleum and natural gas. Light manufacturing sector includes the following subsectors: food, beverages, and tobacco; textiles and apparel (including leather); and wood and paper-related products. Heavy manufacturing sector includes the following subsectors: refined petroleum and related products, chemicals and plastics, nonmetallic minerals, metals, machinery and equipment, and transport equipment. The North includes the G-7 members and 19 other European countries. The South includes all other economies (excluding countries in Latin America and the Caribbean). Offshore centers are excluded from the sample. a. The average for syndicated loans is for the 2003–12 period. T he C hanging P atterns o F Financial I ntegration   179 conducted here expands on the literature by For LAC-LAC links, greenfield and exploring sectoral data in both trade and trade flows are also very similar. Both are financial connections. tilted toward heavy manufacturing, which Figure 4.11 shows the sectoral composi- accounts for 53 percent of greenfield flows tion for 2003–11, the period for which data and 58 percent of trade flows within LAC. are available for all types of investments (for The patterns for syndicated loans and M&A a breakout by subregion within LAC, see are quite different from the patterns for annex figure 4A.2). Unlike figure 4.10, figure trade, with primary industry accounting for 4.11 shows only the percentage share of each the bulk of syndicated loans (59 percent of sector; it does not display the volumes of the flows) and light manufacturing dominating flows. It indicates that the sectoral composi- M&A flows (39 percent of flows). tion of greenfield and trade flows is similar The regressions in table 4.6 explore in (for LAC as both a sender and a receiver). In more detail the relation between financial and contrast, the sectoral composition of syndi- trade flows using country pair–level infor- cated loans and M&A flows tends to differ mation at the sectoral level, covering all 14 from the composition of trade flows. sectors. In particular, they link financial For foreign investments by LAC countries, flows with the comparative advantages of the the sectoral composition of syndicated loans source and receiver country. The relative com- and (especially) greenfield flows is similar to parative advantage (RCA) for both the source the composition of trade flows. Heavy man- and receiver country is constructed following ufacturing captures the largest share of LAC- Vollrath (1991), as shown in equation (4.1): Xi,j,t / 1 g 4j Xi,j,t 2 Xi,j,t 2 1 g 4i Xi,j,t 2 Xi,j,t 2 / 3 1 g 4i,j Xi,j,t 2 g 4j Xi,j,t 2 2 1 g 4i Xi,j,t 2 Xi,j,t 2 4 RCAi,j,t 5 ln e f (4.1) North flows (70 percent of syndicated loans, where Xi,j,t refers to the exports of country i 64 percent of greenfield flows, and 54 percent in industry j in period t. of trade flows), and the primary sector cap- The dependent variable is specified as tures the largest share of LAC-South flows log(1 + flows), in order to explicitly account (91 percent of syndicated loans, 62 percent for the large number of observations equal to of greenfield flows, and 43 percent of trade zero. All regressions control for both fixed- flows). The patterns for M&A are different: source and host-country effects. The regres- LAC countries finance the primary sector in sions also include sector dummies and gravity North countries (42 percent of LAC-North controls. flows) and the heavy manufacturing sector The first pattern that emerges from table 4.6 in South countries (69 percent of LAC-South is that even after controlling with gravity vari- flows).12 ables for common factors that can jointly drive For LAC as a receiver, greenfield flows trade and lending decisions, countries in both are similar to trade flows, in the sense that the North and South invest more in countries both are substantially tilted toward heavy with which they have larger trade flows (mea- manufacturing from both the North and the sured as the sum of exports and imports). This South. In contrast, the patterns for syndi- positive relation appears in all three types of cated loans and M&A differ from those of investments considered (syndicated loans, trade given the fact that North and South M&A, and greenfield investment). countries finance relatively more the primary In general, there is a positive relation sector. For example, in M&A the primary between the RCA of the source country and sector represents 48 percent of North-LAC financial flows. In syndicated loans from flows and 90 percent of South-LAC flows. South and LAC countries, in M&A flows 180 TABLE 4.6 Global financial and trade flows Dependent variable: Log(flows+1) Syndicated loans Mergers and acquisitions Greenfield investment North source South and LAC North source South and LAC North source South and LAC countries source countries countries source countries countries source countries (1) (2) (3) (4) (5) (6) Log(total trade+1) 0.0657*** 0.0064*** 0.0244*** 0.0033*** 0.0652*** 0.0096*** (0.0050) (0.0009) (0.0023) (0.0004) (0.0038) (0.0006) RCA of source country –0.0031 0.0005* 0.0050*** 0.0001 0.0180*** 0.0012*** (0.0022) (0.0003) (0.0009) (0.0001) (0.0016) (0.0001) RCA of receiver country 0.0113*** –0.0010** 0.0050*** 0.0000 –0.0013 –0.0006** (0.0024) (0.0004) (0.0010) (0.0002) (0.0018) (0.0003) RCA of source country ∗ LAC target dummy –0.0004 0.0000 –0.0034*** –0.0001 –0.0089*** 0.0001 (0.0041) (0.0003) (0.0012) (0.0001) (0.0029) (0.0002) RCA of receiver country ∗ LAC target dummy –0.0013 0.0014** –0.0012 0.0005* 0.0035 0.0004 (0.0050) (0.0006) (0.0017) (0.0003) (0.0034) (0.0005) Number of observations 540,707 1,743,205 498,248 2,127,160 408,610 1,994,079 R–squared 0.208 0.043 0.089 0.019 0.172 0.031 Sum of LAC coefficients RCA of source country + RCA of source coun- –0.0035 0.0005*** 0.0017 0.0000 0.0091*** 0.0013*** try ∗ LAC target dummy RCA of receiver country + RCA of receiver 0.0100** 0.0004 0.0038** 0.0005* 0.0022 -0.0002 country ∗ LAC target dummy Sources: Calculations based on data from SDC Platinum, fDi Markets, and Comtrade. Note: This table explores the relation between financial and trade flows using sector-level data. The dependent variable is the financial flow between two countries. Total trade is measured as the sum of exports and imports. Relative comparative advantage (RCA) is based on Vollrath (1991). The Latin America and the Caribbean (LAC) target dummy equals 1 if the receiver is a LAC country. All regressions include gravity control variables that help explain levels of financial flows between each country pair based on the geographic distance between the countries involved, differences in latitude and longitude, differences in time zones, whether they share a common language, whether they have a common legal origin, and whether the receiver (sender) country is (or was) a colony of the sender (receiver). The regressions also control for source- and target-country dummies and sector dummies. The North includes the G-7 members and 19 other European countries. The South includes all other economies (excluding LAC countries). Sample excludes offshore centers. Standard errors are clustered by country pairs. Significance level: * = 10 percent, ** = 5 percent, *** = 1 percent. T H E C H A N G I N G PAT T E R N S O F F I N A N C I A L I N T E G R AT I O N 181 from North countries, and in greenfield flows flows to LAC countries from both the North from both North and South and LAC coun- and the South are substantially tilted toward tries, foreign investments have gone to sectors heavy manufacturing, a sector in which in which the source country has a compara- (overall) LAC countries do not have a com- tive advantage. parative advantage. The evidence on the relation between the RCA of the receiver country and financial flows is mixed. In general, North countries Foreign direct investment tend to invest more in sectors in which the and GDP growth receiver country has a comparative advan- As documented above and in chapter 1, the tage, whereas South countries, including period between 1990 and 2010 was charac- countries in LAC, invest more in sectors in terized not only by a sharp increase in finan- which the receiver has a comparative dis- cial flows across the world but also by the advantage. In flows from North countries, rise of South economies as important players there is a positive relation between the RCA in the global landscape of financial flows. of the receiver country and financial flows Many observers view the rising participation in syndicated loans and M&A flows but no of South economies in global financial flows significant relation for greenfield flows. In broadly and in FDI in particular as a poten- contrast, in flows from the South, including tial driver of economic growth. FDI flows can LAC, there is a negative relation for syn- not only directly ease financing constraints in dicated loans and greenfield flows and no recipient economies, they can also be a major statistically significant relation for M&A. conduit of technology diffusion and learn- Using interaction variables for the cases ing spillovers.13 Indeed, policymakers in the in which LAC is a receiver, table 4.6 breaks South, including LAC, place attracting FDI down the relation between trade and finan- and multinational corporations (MNCs) high cial flows in LAC countries. Most of the on their agendas. They use incentives such as interactions variables are insignificant, sug- income tax holidays, tariff exemptions, and gesting that the relation between capital subsidies to infrastructure to attract foreign flows and the RCA is not significantly differ- firms. According to a census of investment ent for LAC. promotion agencies carried out by the World In comparing the LAC-specific results with Bank in 2004, 78 of the 110 countries sur- the aggregate results, two main differences veyed were offering fiscal or financial conces- emerge. First, when LAC is a receiver, the sions to foreign companies that decided to set comparative advantage of North countries is up production or other facilities within their less related to financial flows of M&A and borders (Harding and Javorcik 2011, 2012). greenfield investment (in fact, for M&A the Behind these efforts is the belief that foreign RCA does not play any role). Second, regard- presence benefits the host country by poten- ing M&A flows to LAC, South and LAC tially raising aggregate productivity in the countries tend to invest more in industries in economy; by introducing advanced technolo- which the receiver country has a comparative gies (both hard technologies, such as machin- advantage. This pattern can be observed in ery and blueprints, and soft technologies, figure 4.11, which shows that almost 90 per- such as management techniques and informa- cent of the M&A flows from South countries tion); and by fostering positive externalities to LAC countries are related to the primary to local firms through technological diffu- sector, the sector in which LAC has a compar- sion and knowledge spillovers.14 This section ative advantage based on natural resources. goes deeper into this issue by examining the Summing up, the evidence suggests that nature of financial linkages and growth. LAC’s comparative advantage seems to have FDI in general and the activities of MNCs helped attract syndicated loans and M&A in particular may prompt productivity- but not greenfield investment. Greenfield enhancing reallocations within the host 182 LATIN AMERICA AND THE RISING SOUTH economy even in the absence of (productiv- Horizontal spillovers can take place when ity) spillover effects on local firms.15 Labor local firms learn about new technologies, and production may get reallocated toward production processes, and marketing or more productive sectors and to relatively management techniques by observing foreign more productive firms within sectors. The- affiliates operating in their industry or by oretical work exploring firm heterogeneity hiring workers trained by foreign affiliates. highlights that firms self-select into becom- MNCs may have incentives to prevent hor- ing MNCs and that only the most produc- izontal spillovers, which could benefit local tive establishments within a country can competitors. afford the extra cost of setting up production In contrast, they may have an incentive facilities abroad. MNCs are thus likely to to facilitate vertical spillovers, especially belong to the upper tier of the productivity through backward linkages.16 Transfer of distribution of firms in their home country knowledge and technology to local firms (Helpman, Melitz, and Yeaple 2004). They in upstream sectors (such as their suppli- are also more likely to invest in local firms ers of intermediate inputs) may lead to an with higher productivity and greater growth improvement in the performance of inter- potential. Fiercer competition in the host mediate input suppliers and to lower input country market that could arise from the prices. Similar effects can be achieved by presence of MNCs may force local firms to subjecting local suppliers to more stringent use their existing resources more efficiently requirements for product quality and on-time or to search for new technologies (Blom- delivery, which provide them with incentives strom and Kokko 1998). The least compet- to upgrade their production management and itive local firms may exit the market as a technology. result of more intense competition for factors In addition, the increased presence of of production or market share (“business MNCs may boost demand for intermediate stealing” effects). products produced domestically, which may There are also important theoretical rea- allow local suppliers to reap the benefits of sons why MNCs can bring advanced technol- scale economies. The forward linkage chan- ogy and know-how to South host countries nel may also be at play. Domestic firms could and, consequently, be closely linked to greater become more productive through improved technological diffusion and knowledge spill- access to new, better-quality, or less expen- overs. The theory of MNCs argues that sive intermediate inputs produced by MNCs these firms rely heavily on intangible assets, in more upstream sectors. such as firm-specific technologies and man- Ample evidence documenting the techno- agement techniques and/or well-established logical edge of MNCs over non-MNC firms brand names, which gives them an “own- backs these ideas. In 2002, for example, ership advantage” over other organizations MNCs accounted for almost half of total (Dunning 1988). Subsidiaries operating in global research and development (R&D) South economies could therefore potentially expenditure and almost 70 percent of global benefit from aggregate technological advan- business R&D (Javorcik 2013). Patenting tages from MNCs through direct transfers is another area where MNCs have a clear (Ethier 1986; Markusen 2004). advantage. Across regions the headquar- In addition, there can be spillover effects ters of MNCs hold more patents than local from MNC subsidiaries to local firms. These firms in the country where the headquarters spillovers can be horizontal (affecting firms is located (Lederman and others 2014). Sub- in the same industry) or vertical (affecting sidiaries of MNCs also have productivity firms in different industries or along the pro- and managerial advantages over local firms duction chain). They can take place through in host countries. Lipsey (2002) reviews the demonstration effects, labor turnover, imita- empirical evidence on productivity differ- tion, and reverse engineering. ences between foreign-owned and local firms. T H E C H A N G I N G PAT T E R N S O F F I N A N C I A L I N T E G R AT I O N 183 He reports that most studies find a posi- performance of target firms in particular and tive and significant productivity advantage of host countries more broadly. of foreign-owned over local firms. Bloom Javorcik and Spatareanu (2011) are among and others (2012) show that foreign-owned the few researchers to tackle this issue. They firms use better management practices than use firm-level panel data from Romania to local firms.17 In LAC countries, the authors examine whether the origin of foreign inves- argue, the quality of management practices tors affects the degree of vertical spillovers by foreign-owned firms is much closer to best from FDI. They find that MNCs from the practices than to local practices, giving sup- United States have a positive effect on Roma- port to the idea that multinational affiliates nian firms, whereas MNCs from the Euro- “import” knowledge from headquarters. pean Union (EU) have no effect. According The evidence on the technological and to the authors, trade costs (particularly dis- managerial advantage of MNCs is over- tance and preferential trade agreements) can whelming. In contrast, the literature finds explain this finding. Compared with EU mixed results regarding the effect of MNC MNCs, U.S. firms find it more expensive to activity on local firms, especially firms oper- import inputs from the home country, lead- ating in the same industry as the MNC affil- ing to greater incentives to create backward iate.18 Many researchers argue that these linkages and more potential for technology mixed results reflect the fact that spillovers and knowledge spillovers. from MNC presence are not automatic; they Other country characteristics could also hinge on a range of factors, from the level affect the degree of spillovers from MNC of education and financial development of activity. One is the level of development of the the host country to the initial level of pro- source and host countries. FDI from highly ductivity of local firms.19 Others provide developed countries may bring more modern, some evidence that the negative competition state-of-the-art technologies and manage- effect outweighs the positive effect of knowl- ment practices and therefore lead to greater edge spillovers, especially in developing growth effects. However, these technologies countries.20 may be too sophisticated for less developed An important aspect of the relation target countries; the difference in the level between FDI inflows and growth and spill- of economic development of the source and over effects that has received much less atten- target countries may be what matters. MNCs tion in the literature is the identity of the host from North countries operating in other and home countries. Inherent to this issue are North countries potentially give rise to larger the motives for FDI. Often-cited theoretical spillovers to the host country because more reasons for FDI by North countries in South developed countries are typically closer to countries are differences in relative input the world technological frontier and may be costs (such as lower labor costs) and mar- better able to absorb the technology, know- ket access (Yeaple 2003). The rationale for how, and intangible assets brought by foreign cross-border FDI may be different for South firms. Technologies from South countries countries, which are typically endowed with may be more suitable, cheaper, and easier to larger and cheaper labor forces than North adapt in other South countries. countries. It is possible that South acquirers Is there evidence of differential growth- may relocate manufacturing activity to their enhancing effects of FDI inflows depending home country while keeping the existing on the level of development of the home and distribution networks in the host country host country? This question is particularly (Chari, Wenjie, and Dominguez 2012). Tech- relevant given the increasing prominence of nology transfer is less likely in South-North South countries as both senders and receivers than in North-South transactions. For these of FDI flows. reasons, the origin of foreign acquirers may Drawing on Didier, Nguyen, and Pienkna- have consequences for the postacquisition gura (2015), the rest of this section examines 184 LATIN AMERICA AND THE RISING SOUTH whether the impact of FDI from the North that financial FDI (for example, investments on productivity in the host country is differ- by foreign financial institutions) only follows ent from the impact of FDI from the South. growth whereas industrial FDI (for example, The outcome could depend on whether the investments by foreign manufacturing firms) recipient is a North or a South country. This both follows and alters growth. Put another issue is particularly relevant for LAC coun- way, growth prospects drive both industrial tries given the significant increase in LAC’s and financial FDI, but only industrial FDI connections with other South countries, espe- potentially leads to growth. Hence the anal- cially during the second half of the 2000s. ysis focuses on the impact on productivity Four types of FDI are studied: from North growth of the component of industrial FDI to North, from North to South, from South that is orthogonal to financial FDI. This to North, and from South to South. Because component is not driven by the growth-fol- of data limitations, the analysis examines lowing motive and can thus be considered only M&A flows. The exercise is conducted exogenous to productivity growth. Box 4.3 at the sectoral level, covering 23 host man- provides details on the model setup and iden- ufacturing sectors. The sample includes tification strategy. 52 host countries (18 North countries and 34 Table 4.7 shows the estimates of the South countries, including 6 from LAC) and impact of FDI on labor productivity growth 98 source countries (22 North countries and in manufacturing sectors around the world. 76 South countries, including 16 from LAC). It reports only the second-stage regressions, The data cover 1993–2010. Unlike the previ- which estimate the impact of this exogenous ous part of this chapter, this part of the chap- component of industrial FDI on productivity ter includes LAC in the South. growth. The results pooling all host coun- Endogeneity is an important issue when tries show that FDI from the North improves addressing this question. Most of the empir- labor productivity growth of the recipient ical evidence on FDI and growth is about sector in the host country, whereas FDI from association, not causation. 21 Theoretically, the South has no effect on labor productivity the relationship could go either way (or both (column 1). The increase in labor productiv- ways): MNCs could invest in local firms (or ity growth following FDI inflows reflects the countries) that are better performing (“cherry net effects of reallocations, technology trans- picking”), or MNC presence could lead to fer and knowledge diffusion, and economies improvements in performance through the of scale as well as the effects of increased channels mentioned earlier. competition resulting from foreign entry into It is key to identify the exogenous com- the sector. This positive effect of FDI from ponents of FDI that are not caused by but the North on productivity takes place with a rather lead to growth-enhancing effects. 22 one-year lag. The identification strategy adopted by Didier, The positive impact of FDI from the Nguyen, and Pienknagura (2015) is similar North on labor productivity growth is to that of Fons-Rosen and others (2013). observed only in North recipient countries The idea is that financial investors do not (columns 2 and 3 of table 4.7): North-North actively manage their targets, at least in part FDI flows affect labor productivity growth because of their limited expertise on ways in recipient sectors, but North-South flows to improve their day-to-day operations. In do not. Moreover, this positive effect of contrast, industrial investors typically have North-North flows is larger than the effect the relevant expertise; they attempt to inter- obtained for the pooled sample. These find- vene and improve the target firm’s operations ings are consistent with the results of Chen and management in order to raise their prof- (2011), who provides empirical evidence that itability and productivity. The underlying public U.S. target firms exhibit greater gains assumption of the identification strategy is in labor productivity when acquired by firms T he C hanging P atterns o F Financial I ntegration   185 Box 4.3  Model setup and identification strategy Drawing on Didier, Nguyen, and Pienknagura growth-following and growth-enhancing motives. (2015), this box examines whether the impact on The growth-following motive does not drive the the productivity of the host country of FDI from the component of industrial FDI orthogonal to finan- North is different from the impact of FDI from the cial FDI, which can thus be considered exogenous South. Equation B4.3.1 gives the baseline regression to productivity growth. specification: Intuitively, this exogenous instrument can be thought of along the lines of a portfolio choice prob- FDIN FDIS Dlog 1 Prod 2 c,s,t 5 a 1 bN a b 1 bS a b lem. Suppose a foreign bank expects host sector s Inv c,s,t Inv c,s,t in host country c to grow at an average annual rate 1 gc,t 1 gs,t 1 uc,s,t (B4.3.1) of 5 percent over the next five years. Based on this expectation, the bank decides to invest $1 million where Dlog 1 Prod 2 c,s,t is the growth of rate in in the sector. Now suppose a foreign manufacturing labor productivity, measured as the ratio of real firm has the same information set as the foreign bank value added to total employment in host country c, does: it also expects host sector s in host country c to host sector s, and time t; 1 FDIN/Inv 2 c,s,t is the M&A grow 5 percent a year over the next five years. How- component of FDI from North countries to host ever, this firm also believes that if it invests in this sector s in host country c at time t divided by total sector, it will boost growth to an average annual rate investment in host sector s in host country c at time of 7 percent over the next five years. Hence the firm t; 1 FDIS /Inv 2 c,s,t is the M&A component of FDI from decides to invest more than $1 million in the sector. South countries to host sector s in host country c at The growth-enhancing component associated with time t divided by total investment in host sector s in the firm’s investment arguably drives this difference host country c at time t ; gc,t are country-time fixed between the two entities’ investments. effects; gs,t are sector-time fixed effects; and uc,s,t is A two-step procedure is adopted in order to deal the error term. The regressions also include lagged with this endogeneity issue. First, the exogenous values of the FDI to investment ratios. component of industrial FDI is constructed. Indus- This baseline regression specification suffers trial FDI from the North to a given country-sector from endogeneity biases. MNCs can invest in well- host in a given year is regressed against total finan- performing sectors (“cherry picking”), or they can cial FDI from the world (financial FDI from both target low-performing sectors (with high growth the North and South) to that country-sector-year potential) and then improve firm performance. To (equation B.4.3.2a). The assumption is that indus- account for sector- and country-specific selection trial firms are informed about investments by finan- issues, the regressions include fixed effects. The cial institutions around the world, not just in their country-time and sector-time fixed effects aim at own countries. The error term from this regression capturing all time-varying changes within a country (PNc,s,t ) is the exogenous component of industrial FDI and within sectors that may attract MNC activity. from the North to host country c in sector s at time These changes include the macroeconomic environ- t. An analogous regression setup is used to construct ment, a better location, policy reforms, technolog- the exogenous component of industrial FDI from the ical improvements, and relative price changes. The South to host country c, in sector s, at time t (equa- identity of the investor is used to construct an tion B4.3.2b). 23 exogenous instrument in order to deal with unob- Ind FDIN Fin FDIW served heterogeneity at the country-sector-time a b 5 aN 1 f N a b Inv Inv level. More specifically, the underlying assump- c,s,t c,s,t tion of the identification strategy is that only the 1 gc,t 1 gs,t 1 eN c,s,t (B4.3.2a) growth-following component drives financial FDI Ind FDIS Fin FDI W (for example, investment by foreign financial insti- a b 5 aS 1 f S a b Inv Inv tutions) whereas industrial FDI (for example, invest- c,s,t c,s,t ment by foreign manufacturing firms) reflects both 1 gc,t 1 gs,t 1 eS c,s,t (B4.3.2b) (continued) 186   LATIN AMERIC A AND THE RISING SOUTH Box 4.3  Model setup and identification strategy (continued) In the second step, the error terms from the two One important concern with this setup is that, for equations replace the FDI to investment ratios in a variety of reasons, many South countries do not equation B4.3.1. Equation B4.3.3 gives the regres- allow financial inflows. These inflow restrictions sion specification of productivity growth on these may restrict the level of financial FDI. The two-step residuals: setup described above is not able to fully eliminate the growth-following component in industrial FDI, Dlog 1 Prod 2 c,s,t 5 a 1 bNeN c,s,t 1 bSeS c,s,t because financial FDI is constrained at lower than expected levels. The residuals in equations (B4.3.2a) 1 gc,t 1 gs,t 1 uc,s,t (B4.3.3) and (B4.3.2b) may thus be larger than expected and no longer fully exogenous to productivity growth. The results of the estimations of this two-step pro- Exclusion of countries in the sample with no cedure are reported and discussed in the main text. observed financial FDI mitigates this issue. from developed countries than they do when foreign parent companies’ investment in they are acquired by developing country staff training suggests that a significant firms. The results in table 4.7 also show that increase in foreign ownership is likely to lead FDI from the South typically has no impact to improvements in the subsidiary’s perfor- on productivity growth, in either North or mance. MNCs also typically transfer more South countries (columns 2 and 3). sophisticated technologies and management It is possible that FDI to or from the South techniques to their wholly owned affiliates occurs largely in sectors where positive spill- than to their partially owned affiliates, 24 and overs are more limited, such as sectors with they may be more likely to transfer technol- relatively short quality ladders, for example. ogy to local suppliers, in order to increase Financial flows to LAC countries from the their productivity and reduce input prices. South are indeed biased toward the primary Therefore, productivity improvements may sector, as discussed in chapter 1. Particularly be more marked when M&A investments striking is the share of FDI flows from other lead to greater control of local firms.25 South regions to LAC: on average, 90 percent To examine whether the extent of foreign of all M&A and 38 percent of all green- ownership affects the results presented so far, field investment went to the primary sector the analysis considers only cross-border M&A during the 2000s. The sectoral composition transactions that lead to ownership of at least of the M&A component of FDI inflows 50 percent of the target firm (columns 4–6 of within manufacturing sectors does not seem table 4.7). The results are generally robust to to explain the patterns documented so far, this restriction on the magnitude of FDI trans- however. Increases in productivity growth actions. They reveal a positive impact of FDI in either light or heavy manufacturing sec- from the North on labor productivity growth tors are typically observed in the aftermath in North recipient countries but not in South of North-North flows but not North-South, recipient countries. FDI from the South does South-North, or South-South flows. not lead to systematic productivity effects in The ownership structure of subsidiar- recipients in the North or the South. ies of MNCs is another important factor. One possible explanation for the lack of The larger the stakes MNCs have in local consistent positive effects of FDI from the subsidiaries, the more they control the pro- North to the South is the difference in the level duction processes, operations, and manage- of development. South countries may be too ment and the greater are their incentives to far from the technology level of the North and improve them (the “pushy parent” analogy thus not able to efficiently absorb the North’s in Arnold and Javorcik 2009). For instance, technology in its production processes. TABLE 4.7 Foreign direct investment and labor productivity growth in the host country Dependent variable: Growth in labor productivity in host country c sector s Transactions in which posttransaction ownership is at least All transactions 50 percent of shares All target North target South target All target North target South target countries countries countries countries countries countries Source countries (1) (2) (3) (4) (5) (6) North FDI / Inv t 0.005 0.008 –0.017 0.001 0.005 –0.025 (0.009) (0.011) (0.016) (0.009) (0.011) (0.017) North FDI / Inv t–1 0.024** 0.028** –0.000 0.024** 0.028** –0.004 (0.010) (0.013) (0.014) (0.011) (0.013) (0.014) North FDI / Inv t–2 0.009 0.007 0.020 0.006 0.005 0.020 (0.007) (0.008) (0.018) (0.008) (0.009) (0.022) North FDI / Inv t–3 –0.010 0.000 –0.027 –0.008 –0.002 –0.022 (0.008) (0.010) (0.018) (0.009) (0.011) (0.020) South FDI / Inv t –0.045 –0.002 –0.078 –0.055 –0.038 –0.075 (0.031) (0.025) (0.052) (0.037) (0.046) (0.057) South FDI / Inv t–1 –0.025 –0.008 –0.037 –0.024 0.015 –0.052 (0.023) (0.024) (0.037) (0.026) (0.032) (0.040) South FDI / Inv t–2 0.022 –0.017 0.048 0.023 –0.017 0.050 (0.042) (0.024) (0.068) (0.045) (0.025) (0.075) South FDI / Inv t–3 –0.012 –0.038 0.029 –0.029 –0.046 0.009 (0.026) (0.028) (0.040) (0.027) (0.029) (0.046) Number of observations 8,885 4,030 4,855 8,414 4,032 4,382 R-squared 0.354 0.394 0.373 0.363 0.395 0.389 Source: Didier, Nguyen, and Pienknagura 2015. Note: This table shows the impact of FDI (foreign direct investment) on labor productivity growth in manufacturing sectors around the world. The estimation method is panel fixed effects. The regressions include country-year and sector-year fixed effects. Box 4.3 in the text describes the procedure followed and the identification assumption made. FDI is measured by the exogenous component of industrial FDI to investment ratio into the host country c sector s. The North includes the G-7 members and 19 other European countries. The South includes all other economies (including countries in Latin America and the Caribbean). Sample excludes offshore centers. Significance level: * = 10 percent, ** = 5 percent, *** = 1 percent. 187 188 LATIN AMERICA AND THE RISING SOUTH Another possible explanation is related to and thus more easily adopted. Although tech- trade costs. The share of intermediate inputs nology and knowledge spillovers may still sourced locally by MNCs (which may be take place, the effects of reallocations, econo- an important factor determining the poten- mies of scale, and increased competition may tial for technology and knowledge spillover be large enough to offset them. effects) is likely to increase with the distance Exploring these possibilities more formally between the host and the source economy. is an important topic for future research. In However, the distance between North coun- light of the rise of South-South FDI flows over tries is on average smaller than the distance the past decade, a more in-depth analysis of between North and South countries, making the differentiated effects that source countries this explanation unlikely. may have on the potential for growth-enhanc- Although the no-effect of South-North ing effects of FDI seems to be in order. The FDI flows finding is not entirely surprising, role of distance between source and receiver one might have expected South-South flows countries (including all aspects it may cap- to have a positive impact on growth outcomes ture, such as trade and financial barriers, cul- of the recipient country. For instance, technol- tural differences, the degree of information ogies from a South country may be closer to asymmetries, and so on) also merits further the technologies of the South recipient country examination. Number of active connections Number of active connections Number of active connections Number of active connections 0 10 20 30 40 50 60 0 10 20 30 40 50 60 0 50 100 150 200 250 300 350 0 10 20 30 40 50 60 1990 1990 1990 1990 1991 1991 1991 1991 1992 1992 1992 1992 Annex 4A 1993 1993 1993 1993 1994 1994 1994 1994 1995 1995 1995 1995 1996 1996 1996 1996 1997 1997 1997 1997 1998 1998 1998 1998 1999 1999 1999 1999 2000 2000 2000 2000 2001 2001 2001 2001 2002 2002 2002 2002 2003 2003 2003 2003 2004 2004 2004 2004 2005 2005 2005 2005 2006 2006 2006 2006 2007 2007 2007 2007 2008 a. Latin America and the Caribbean as sender 2008 2008 2008 2009 2009 2009 2009 2010 2010 2010 2010 North 2011 2011 2011 2011 ica and the Caribbean [LAC]). Offshore centers are excluded from the sample. Sources: Calculations based on data from CPIS, SDC Platinum, and fDi Markets. Number of active connections Number of active connections Number of active connections Number of active connections South Syndicated loans 0 20 40 60 80 100 120 140 160 0 20 40 60 80 100 120 140 160 0 20 40 60 80 100 120 140 160 0 50 100 150 200 250 300 350 Portfolio investments Greenfield investment Mergers and acquisitions 1990 1990 1990 1990 LAC 1991 1991 1991 1991 1992 1992 1992 1992 1993 1993 1993 1993 1994 1994 1994 1994 1995 1995 1995 1995 1996 1996 1996 1996 1997 1997 1997 1997 1998 1998 1998 1998 1999 1999 1999 1999 2000 2000 2000 2000 2001 2001 2001 2001 2002 2002 2002 2002 2003 2003 2003 2003 2004 2004 2004 2004 2005 2005 2005 2005 2006 2006 2006 2006 2007 2007 2007 2007 2008 2008 2008 2008 b. Latin America and the Caribbean as receiver 2009 2009 2009 2009 2010 2010 2010 2010 2011 2011 2011 2011 Note: The North includes the G-7 members and 19 other European countries. The South includes all other economies (excluding countries in Latin Amer- ANNEX FIGURE 4A.1 Number of active cross-border connections, by type of investment and region T H E C H A N G I N G PAT T E R N S O F F I N A N C I A L I N T E G R AT I O N 189 190 LATIN AMERICA AND THE RISING SOUTH ANNEX FIGURE 4A.2 Sectoral composition of cross-border financial flows to and from Latin America and the Caribbean, by type of investment and subregion, 2003–11 average a. By sender, syndicated loansa b. By sender, mergers and acquisitions The Caribbean 91 3 6 The Caribbean 42 35 24 Central America 74 14 13 Central America 12 28 59 Pacific South America 58 11 32 Pacific South America 6 51 43 Other South America 34 17 49 Other South America 44 28 29 0 10 20 30 40 50 60 70 80 90 100 0 10 20 30 40 50 60 70 80 90 100 Percent Percent c. By sender, greenfield investment d. By sender, trade The Caribbean 57 17 25 The Caribbean 41 16 43 Central America 22 12 65 Central America 18 14 68 Pacific South America 15 11 74 Pacific South America 44 19 36 Other South America 51 3 46 Other South America 36 28 36 0 10 20 30 40 50 60 70 80 90 100 0 10 20 30 40 50 60 70 80 90 100 Percent Percent e. By receiver, syndicated loansa f. By receiver, mergers and acquisitions The Caribbean 50 22 27 The Caribbean 75 17 8 Central America 34 9 57 Central America 17 45 38 Pacific South America 76 13 11 Pacific South America 63 23 14 Other South America 68 7 25 Other South America 61 18 21 0 10 20 30 40 50 60 70 80 90 100 0 10 20 30 40 50 60 70 80 90 100 Percent Percent g. By receiver, greenfield investment h. By receiver, trade The Caribbean 26 1 74 The Caribbean 21 19 60 Central America 19 8 73 Central America 6 19 75 Pacific South America 68 3 29 Pacific South America 13 17 70 Other South America 25 7 68 Other South America 11 12 77 0 10 20 30 40 50 60 70 80 90 100 0 10 20 30 40 50 60 70 80 90 100 Percent Percent Primary Light manufacturing Heavy manufacturing Sources: Calculations based on data from SDC Platinum, fDi Markets, and Comtrade. Note: Caribbean: Antigua and Barbuda, Aruba, The Bahamas, Barbados, Belize, Bermuda, Cayman, Cuba, Dominica, Grenada, Guyana, Haiti, Jamaica, Puerto Rico, St. Kitts and Nevis, St. Lucia, St. Vincent and the Grenadines, Suriname, Trinidad and Tobago, Turks and Caicos Islands, the Virgin Islands. Central America: Costa Rica, the Dominican Republic, El Salvador, Gua- temala, Honduras, Mexico, Nicaragua, and Panama. Pacific South America: Chile, Colombia, Ecuador, and Peru. Other South America: Argentina, Bolivia, Brazil, French Guiana, Paraguay, Uruguay, and República Bolivariana de Venezuela. The primary sector includes agriculture, hunting, forestry, and fishing; mining; and crude petroleum and natural gas. The light man- ufacturing sector includes manufacturing of food, beverage, and tobacco; textiles and apparel (including leather); and wood and paper-related products. The heavy manufacturing sector includes manufacturing of refined petroleum and related products, chemicals and plastics, nonmetallic minerals, metals, machinery and equipment, and transport equipment. a. The average for syndicated loans is for the 2003–12 period. T H E C H A N G I N G PAT T E R N S O F F I N A N C I A L I N T E G R AT I O N 191 Notes information regarding the number of active connections. Given that the South category 1. For example, the average share of FDI flows includes many more countries, the extensive (M&A and greenfield) from North and South margin computed in these figures could be a to LAC countries decreased between 2003–05 misleading indicator of the number of active and 2006–11 (the share of North-LAC links connections between two regions. Annex fig- fell from 66 percent to 62 percent and the share ure 4A.1 tries to account for this. of South-LAC links from 89 percent to 79 per- 7. For portfolio investments, the extensive cent). However, this trend does not mean that margin may be underestimated if investors LAC is receiving less equity investments. The in a country hold internationally diversified results show a significant increase in the share mutual funds that invest in many other coun- of equity instruments for both North-LAC and tries. However, international mutual funds South-LAC connections. For North-LAC links, are not very well diversified (Didier, Rigobon, the equity share increased from 43 percent to and Schmukler 2013). Therefore, even if some 59 percent between 2001–05 and 2006–11. For portfolio investments are in mutual funds, the South-LAC links, the equity share grew even degree of diversification or the extensive mar- more, rising from 22 percent to 67 percent. gin may not be significantly larger. 2. The CPIS covers portfolio investment secu- 8. A large share of M&A flows is explained by rities held by monetary authorities but not the 2007 acquisition of the Australian Rinker their reserve assets. The central banks of Group by the Mexican cement company. Even many LAC countries (such as Brazil, Chile, after excluding this observation, however, Colombia, and Costa Rica) classify all their Mexico represented 58 percent of total flows. foreign securities as reserves assets. In these 9. Given the relatively short time span of the data cases, the CPIS database does not cover the on gross capital flows explored in this chapter, investments made by the central banks. Cen- it was not possible to disentangle the extent to tral banks from other LAC countries (such as which changes in the nature of financial inte- Mexico, Panama, and República Bolivariana gration of LAC countries are inherent to its de Venezuela) do not classify all their hold- globalization process or driven by changes in ings as reserves assets. For these countries, the the global landscape, such as changes associ- CPIS survey covers all their holdings that are ated with the rise of the South. not reserves assets. As a consequence of these 10. Transactions classified as “other manufactur- differences, the figures presented in the chap- ing” and “utilities and infrastructure” were ter may be lower than LAC countries’ actual dropped, as they do not clearly fit into either holdings in the rest of the world. light or heavy manufacturing. 3. The dataset for syndicated loans also covers 11. See, for example, Aviat and Coeurdacier 2012; the later period is thus 2006–12. For (2007); Stein and Daude (2007); Daude and simplicity, this period is referred to as 2006– Fratzscher (2008); Lane and Milesi-Ferretti 11 throughout this chapter. (2008); Dailami, Kurlat, and Lim (2012); and 4. The volume of syndicated loans from North Okawa and van Wincoop (2012). to South countries increased 86 percent and 12. The two large transactions described earlier M&A flows 94 percent over this period. (the 2006 acquisition of the Canadian com- North-South portfolio investments increased pany Inco by the Brazilian mining company 135 percent between 2001–05 and 2006–11, Vale and the 2007 acquisition of the Austra- and North-LAC portfolio investments rose 97 lian Rinker Group by the Mexican cement percent. See tables 4.4 and 4.5 for details. company Cemex) partly explain these dif- 5. Another way to account for the expansion in ferences. Excluding these two cases, heavy the real economy is to use the average GDP manufacturing accounts for a larger share of of the two regions as a benchmark. In addi- LAC-North (36 percent) than the primary tion, the flows between two regions can be sector (26 percent). The sectoral composition expressed as a fraction of total cross-border of LAC-North M&A flows therefore appears flows of each type of investment. In both more similar to that of trade. However, even cases, trends similar to the ones reported in after excluding these observations, heavy table 4.2 are evident (results not reported). manufacturing still accounts for a larger share 6. Figures 4.4–4.7 show the percentage of of LAC-South flows (25 percent) than the pri- active connections but they do not provide mary sector (20 percent). 192 LATIN AMERICA AND THE RISING SOUTH 13. For example, for aggregate (rather than bilat- increase with financial development. Kokko, eral) flows, average FDI inflows in the South Zejan, and Tansini (2001) provide evidence increased from about 1 percent of GDP in the that technological spillovers require a min- 1980s to about 5 percent in the 2000s. Aver- imum initial level of technology in the local age FDI outflows from South countries rose firms. Using data for Uruguay, they show that from 0.15 percent of GDP to 1.8 percent, a firms with higher initial levels of productiv- staggering 12-fold increase in the span of three ity experience larger positive spillovers from decades. Although the levels are different, the MNC activity. Blomstrom, Lipsey, and Zejan trends are similar to the ones reported in table (1994) find similar results using cross-country 4.1. See Broner and others (2013) for an anal- data. They show that growth spillovers from ysis of the dynamics of capital flows during inward FDI are positively correlated with a tranquil and crisis times. country’s wealth. 14. Romer (1993), for instance, argues that the 20. See, for example, Aitken and Harrison (1999), presence of MNCs can narrow both the Djankov and Hoekman (2000), and Konings “object gap” (the shortage of physical goods, (2001). such as factories and roads) and the “ideas 21. See Barba Navaretti, Galeotti, and Mattozzi gap” (the shortage of knowledge used to cre- (2004) for a survey of the literature. ate value added) in South economies. 22. A few studies focus on the growth-enhancing 15. As in the literature in international economics, effect in country-specific contexts. Arnold and the country that receives the MNC (or FDI) is Javorcik (2009) estimate the productivity labeled the “host country,” and the country effects of FDI for Indonesian firms, for exam- of origin of the capital is labeled the “home ple, and Guadalupe, Kuzmina, and Thomas country.” (2012) study Spanish firms. Fons-Rosen and 16. See, for example, Rodriguez-Clare (1996), others (2013) focus on the causal effect of for- Markusen and Venables (1999), Pack and eign investment on productivity using a global Saggi (2001), and Lin and Saggi (2007). firm-level database. 17. Bloom and Van Reenen (2007) find a positive 23. These first-stage regressions are estimated correlation between productivity and manage- with a tobit setup, as the dependent variable is ment practices. a nonnegative variable with a large number of 18. Many firm-level studies cast doubt on the observations at zero. existence of intraindustry spillovers from 24. This argument is in line with Grossman and FDI in developing countries and transition Hart (1986) and Hart and Moore (1990). economies (see Haddad and Harrison 1993 For some empirical evidence, see, for exam- on Morocco; Aitken and Harrison 1999 on ple, Mansfield and Romeo (1980), Ramach- República Bolivariana de Venezuela; Djankov andaram (1993), and Javorcik and Saggi and Hoekman 2000 on the Czech Repub- (2010). lic; Konings 2001 on Bulgaria, Poland, and 25. One could also argue that spillovers could be Romania; Javorcik 2004 on Lithuania; and more limited for full ownership of subsidiar- Javorcik and Spatareanu 2008 on Romania). ies than for partial ownership. One reason Nonetheless, there is some evidence of positive often cited for the practice of transferring spillovers from MNC affiliates to local firms less sophisticated technologies and manage- through backward linkages. For empirical ment techniques to partially owned affiliates evidence on vertical spillovers more broadly, than to wholly owned affiliates is the desire see Lopez-Cordova (2003); Javorcik (2004); of MNCs to minimize the potential for tech- Lopez-Cordova and Mesquita Moreira nology and knowledge leakages to competi- (2004); Kugler (2006); Blalock and Gertler tors in the host country. A local partner might (2008); Barrios, Görg, and Strob (2011); and use the knowledge acquired from a foreign Javorcik and Spatareanu (2011). See Harrison investor in other operations not involving the and Rodriguez-Clare (2010) for a comprehen- foreign shareholders, for example. This prac- sive review of the literature. tice may backfire for MNCs. Local compet- 19. Borenzstein, de Gregorio, and Lee (1998) find itors may be more able to absorb these less that the effect of FDI on growth depends on sophisticated technologies, which, combined the host country’s human capital. Alfaro and with better access to knowledge through the others (2004) find that spillovers from FDI actions of the local shareholders, may lead T H E C H A N G I N G PAT T E R N S O F F I N A N C I A L I N T E G R AT I O N 193 to greater technology and knowledge spill- Blomstrom, M., and A. Kokko. 1998. “Multina- overs. Moreover, firms with joint domestic tional Corporations and Spillovers.” Journal of and foreign ownership may face lower costs Economic Surveys 12 (3): 247–77. of finding local suppliers of intermediate Blomstrom, M., R. Lipsey, and M. Zejan. goods and may thus be more likely to engage 1994. “What Explains Developing Country in local sourcing than wholly owned foreign Growth?” In Convergence of Productivity: subsidiaries. Shared domestic and foreign Cross-National Studies and Historical Evi- ownership may thus lead to higher produc- dence, 9th ed., ed. W. Baumol. New York, NY: tivity spillovers to local producers in the sup- Oxford University Press. plying sectors. For empirical analysis of this Bloom, N., C. Genakos, R. Sadun, and J. Van issue, see, for example, Haddad and Harrison Reenen. 2012. “Management Practices across (1993) for Morocco, Aitken and Harrison Firms and Countries.” N BER Working (1999) for República Bolivariana de Venezu- Paper 17850, National Bureau of Economic ela, and Javorcik and Spatareanu (2008) for Research, Cambridge, MA. Indonesia. Bloom, N., and J. Van Reenen. 2007. “Measuring and Explaining Management Practices across Firms and Countries.” Quarterly Journal of References Economics 122 (4): 1351–408. Aitken, B., and A. Harrison. 1999. “Do Domestic Borenzstein, E., J. de Gregorio, and J. W. Lee. Firms Benefit from Direct Foreign Investment? 1998. “How Does Foreign Direct Investment Evidence from Venezuela.” American Eco- Affect Economic Growth?” Journal of Inter- nomic Review 89 (3): 605–18. national Economics 45 (1): 115–35. Alfaro, L., A. Chanda, S. Kalemli-Ozcan, and Broner, F., T. Didier, A. Erce, and S. Schmukler. S. Sayek. 2004. “FDI and Economic Growth: 2013. “Gross Capital Flows: Dynamics and The Role of Local Financial Markets.” Journal Crises.” Journal of Monetary Economics 60 of International Economics 64 (1): 89–112. (1): 113–33. Alfaro, L., and A. Charlton. 2009. “Intra- Chari, A., C. Wenjie, and K. Dominguez. 2012. Industry Foreign Direct Investment.” Ameri- “Foreign Ownership and Firm Performance: can Economic Review 99 (5): 2096–119. Emerging Market Acquisitions in the United Antràs, P., and R. Caballero. 2009. “Trade and States.” IMF Economic Review 60 (1): 1–42. Capital Flows: A Financial Frictions Perspec- Chen, W. 2011. “The Effect of Investor Origin tive.” Journal of Political Economy 117 (4): on Firm Performance: Domestic and Foreign 701–44. Direct Investment in the United States.” Jour- Arnold, M., and B. Javorcik. 2009. “Gifted Kids nal of International Economics 83 (2): 219–28. or Pushy Parents? Foreign Direct Investment Dailami, M., S. Kurlat, and J. Lim. 2012. “Bilat- and Plant Productivity in Indonesia.” Journal eral M&A Activity from the Global South.” of International Economics 79 (1): 42–53. World Bank Policy Research Working Paper Aviat, A., and N. Coeurdacier. 2007. “The Geog- 5953, Washington, DC. raphy of Trade in Goods and Asset Holdings.” Daude, C., and M. Fratzscher. 2008. “The Peck- Journal of International Economics 71(1): ing Order of Cross-Border Investment.” Jour- 22–51. nal of International Economics 74 (1): 94–119. Barba Navaretti, G., M. Galeotti, and A. Mat- De la Torre, A., C. Calderon, T. Didier, E. Levy tozzi. 2004. “Moving Skills from Hands to Yeyati, and S. Schmukler. 2010. Globalized, Heads: Does Importing Technology Affect Resilient, Dynamic: The New Face of Latin Export Performance in Textiles?” Research America and the Caribbean. World Bank LAC Policy 33 (6–7): 879–95. Semi-Annual Report, October, Washington, Barrios, S., H. Görg, and E. Strob. 2011. “Spill- DC. overs through Backward Linkages from Mul- De la Torre, A., T. Didier, C. Hevia, S. Pienkna- tinationals: Measurement Matters.” European gura, and S. Schmukler. 2012. Latin America Economic Review 55 (6): 862–75. Copes with Volatility: The Dark Side of Glo- Blalock, G., and P. Gertler. 2008. “Welfare Gains balization. World Bank LAC Semi-Annual from Foreign Direct Investment through Tech- Report, April, Washington, DC. nology Transfer to Local Suppliers.” Journal of Didier, T., H. Nguyen, and S. Pienknagura. 2015. International Economics 74 (2): 402–21. “FDI and Growth: Does the Source Country 194 LATIN AMERICA AND THE RISING SOUTH Matter?” World Bank, background paper for Hart, O., and J. Moore. 1990. “Property Rights this report. and the Nature of the Firm.” Journal of Politi- Didier, T., R. Rigobon, and S. Schmukler. 2013. cal Economy 98 (6): 1119–58. “Unexploited Gains from International Diver- Helpman E., M. Melitz, and S. Yeaple. 2004. sification: Patterns of Portfolio Holdings “Export versus FDI with Heterogeneous Firms.” around the World.” Review of Economics and American Economic Review 95 (1): 300–16. Statistics 95 (5): 1562–83. IMF (International Monetary Fund). 1993. Bal- Djankov, S., and B. Hoekman. 2000. “Foreign ance of Payments Manual . Washington, Investment and Productivity Growth in Czech DC. https://www.imf.org/external/pubs/ft Enterprises.” World Bank Economic Review /bopman/bopman.pdf. 14 (1): 49–64. ———. 2002. Coordinated Portfolio Investment Dunning, J. 1988. “The Eclectic Paradigm of Survey Guide. http://www.imf.org/external/ International Production: A Restatement and pubs/ft/cpis/2002/pdf/cpis_index.pdf. Some Possible Extensions.” Journal of Inter- Javorcik, B. 2004. “Does Foreign Direct Invest- national Business Studies 19 (1): 1–31. ment Increase the Productivity of Domestic Ethier, W. J. 1986. “The Multinational Firm.” Firms? In Search of Spillovers through Back- Quarterly Journal of Economics 101 (4): ward Linkages.” American Economic Review 805–33. 94 (3): 605–27. Fons-Rosen, C., S. Kalemli-Ozcan, B. Sørensen, ———. 2013. “International Technology Trans- C. Villegas-Sanchez, and V. Volosovych. 2013. fer.” In The Evidence and Impact of Finan- “Quantifying Productivity Gains from Foreign cial Globalization , ed. G. Caprio, T. Beck, Investment.” CEPR Discussion Paper 9434, S. Claessens, and S. Schmukler. Amsterdam: Center for Economic Policy Research, Wash- Elsevier. ington, DC. Javorcik, B., and K. Saggi. 2010. “Technologi- Greenaway, D., and R. Kneller. 2007. “Firm cal Asymmetry among Foreign Investors and Heterogeneity, Exporting, and Foreign Direct Mode of Entry.” Economic Inquiry 48 (2): Investment.” Economic Journal 117 (517): 415–33. 134–61. Javorcik, B., and M. Spatareanu. 2008. “To Grossman, S., and O. Hart. 1986. “The Costs Share or Not to Share: Does Local Participa- and Benefits of Ownership: A Theory of Verti- tion Matter for Spillovers from Foreign Direct cal and Lateral Integration.” Journal of Politi- Investment?” Journal of Development Eco- cal Economy 94 (4): 691–719. nomics 85 (1): 194–217. Guadalupe, M., O. Kuzmina, and C. Thomas. ———. 2011. “Does It Matter Where You Come 2012. “Innovation and Foreign Ownership.” From? Vertical Spillovers from Foreign Direct American Economic Review 102 (7): 3594–627. Investment and the Origin of Investors.” Haddad, M., and A. Harrison. 1993. “Are Journal of Development Economics 96 (1): There Positive Spillovers from Direct Foreign 126–38. Investment? Evidence from Panel Data for Jin, K. 2012. “Industrial Structure and Capital Morocco.” Journal of Development Econom- Flows.” American Economic Review 102 (5): ics 42 (1): 51–74. 2111–46. Harding, T., and B. Javorcik. 2011. “Roll out the Ju, J., and S. J. Wei. 2011. “When Is Quality of Red Carpet and They Will Come: Investment Financial System a Source of Comparative Promotion and FDI Inflows.” Economic Jour- Advantage?” Journal of International Eco- nal 121 (557): 1445–76. nomics 84 (2): 178–87. ———. 2012. “Foreign Direct Investment and Kokko, A., M. Zejan, and R. Tansini. 2001. Export Upgrading.” Review of Economics “Trade Regimes and Spillover Effects of FDI: and Statistics 94 (4): 964–80. Evidence from Uruguay.” Weltwirtschaftliches Harrison, A., and A. Rodriguez-Clare. 2010. Archiv 137 (1): 124–49. “Trade, Foreign Investment, and Industrial Konings, J. 2001. “The Effects of Foreign Direct Policy for Developing Countries.” In Hand- Investment on Domestic Firms.” Economics of book of Development Economics, vol. 5, ed. Transition 9 (3): 619–33. D. Rodrik and M. Rosenzweig. Amsterdam: Kugler, M. 2006. “Spillovers from Foreign Direct North-Holland. Investment: Within or between Industries?” T H E C H A N G I N G PAT T E R N S O F F I N A N C I A L I N T E G R AT I O N 195 Journal of Development Economics 80 (2): Markusen, J., and A. Venables. 1999. “Foreign 444–77. Direct Investment as a Catalyst for Industrial Lane, P., and G. M. Milesi-Ferretti. 2008. “Inter- Development.” European Economic Review national Investment Patterns.” Review of Eco- 43 (2): 335–56. nomics and Statistics 90 (3): 538–49. Okawa, Y., and E. van Wincoop. 2012. “Gravity Lederman, D., J. Messina, S. Pienknagura, and J. in International Finance.” Journal of Interna- Rigolini. 2014. “Latin American Entrepreneurs: tional Economics 87 (2): 205–15. Many Firms but Little Innovation.” World Bank Pack, H., and K. Saggi. 2001. “Vertical Technol- Publication 16457, Washington, DC. ogy Transfer via International Outsourcing.” Levchenko, A., and P. Mauro. 2007. “Do Some Journal of Development Economics 65 (2): Forms of Financial Flows Protect from Sudden 389–415. Stops?” World Bank Economic Review 21 (3): Ramachandaram, V. 1993. “Technology Transfer, 389–411. Firm Ownership, and Investment in Human Lin, P., and K. Saggi. 2007. “Multinational Firms, Capital.” Review of Economics and Statistics Exclusivity, and Backward Linkages.” Journal 75 (4): 664–70. of International Economics 71(1): 206–20. Rodriguez-Clare, A. 1996. “Multinationals, Link- Lipsey, R. E. 2002. “Home and Host Country ages, and Economic Development.” American Effects of FDI.” NBER Working Paper 9293, Economic Review 86 (4): 852–73. National Bureau of Economic Research, Cam- Romer, P. 1993. “Idea Gaps and Object Gaps in bridge, MA. Economic Development.” Journal of Monetary Lopez-Cordova, J. E. 2003. “NAFTA and Manu- Economics 32 (3): 543–73. facturing Productivity in Mexico.” Journal of Sarno, L., and M. Taylor. 1999. “Hot Money, the Latin American and Caribbean Econom- Accounting Labels and the Permanence of ics Association (LACEA) 4 (1): 55–98. Capital Flows to Developing Countries: An Lopez-Cordova, J. E., and M. Mesquita Moreira. Empirical Investigation.” Journal of Develop- 2004. “Regional Integration and Productivity: ment Economics 59 (2): 337–64. The Experiences of Brazil and Mexico.” In Stein, E., and C. Daude. 2007. “Longitude Mat- Integrating the Americas: FTAA and Beyond, ters: Time Zones and the Location of Foreign ed. A. Estevadeordal, D. Rodrick, A. Tay- Direct Investment.” Journal of International lor, and A. Velasco. Cambridge, MA: David Economics 71(1): 96–112. Rockefeller Center Series on Latin American Vollrath, T. 1991. “A Theoretical Evaluation Studies, Harvard University. of Alternative Trade Intensity Measures of Mansfield, E., and A. Romeo. 1980. “Technol- Revealed Comparative Advantage.” Review of ogy Transfers to Overseas Subsidiaries by World Economics 127 (2): 265–80. US–Based Firms.” Quarterly Journal of Eco- Yeaple, S. 2003. “The Role of Skill Endowments nomics 95 (4): 737–70. in the Structure of U.S. Outward Foreign Markusen, J. 2004. Multinational Firms and the Direct Investment.” Review of Economics and Theory of International Trade. Cambridge, Statistics 85 (3): 726–34. MA: MIT Press. Ascending with the South Winds: Will Low Saving in Latin America 5 and the Caribbean Be a Drag? This chapter analyzes the evolving connectivity between Latin America and the Caribbean (LAC) and the rising South based on net saving (domestic saving minus investment) and, hence, the relative importance of domestic versus external demand. It explores whether the low domestic saving rates in LAC impaired the region’s growth potential in the past and may continue to do so in the future given changes in the world environment (particularly the rise of China and, more generally, the South). The chapter identifies three channels through which domestic saving can affect growth: the real exchange rate, the interest rate (by way of the country’s risk rating), and the endogenous response of saving to growth, which exerts a mul- tiplier effect on growth through the first two channels. The analysis finds that the interest rate channel hindered growth in the 1980s and 1990s and boosted it in the 2000s. However, given the region’s significantly improved macro-financial policies and a more robust (equity-based rather than debt-based) form of international financial integration, LAC’s low saving is more likely to hinder future growth through the real exchange rate channel. This effect is stronger for countries relying more on domestic demand-oriented growth strategies and hence incur- ring recurrent current account deficits. Although a good case for saving-enhancing policies that promote competitiveness can thus be made for such countries, external factors (weak world demand and ample availability of finance) as well as domestic factors (social policy pressures) are likely to create difficult policy tradeoffs and tensions. T he long-run growth performance of A quick look at the data suggests that there Latin America and the Caribbean may be some link between LAC’s mediocre (LAC) has been unimpressive relative growth performance and its generally low to the United States: the comparison reveals saving rates. Except for República Bolivariana a history of convergence failure (figure 5.1). de Venezuela, all major LAC countries stand Given the slowdown that follows a decade of on the low side of the domestic saving dis- apparent improvement, concerns about low tribution (after controlling for gross domes- growth have risen to the forefront of the pol- tic product [GDP] per capita), and all Asian icy debate (figure 5.2). economies sit on the other side (figure 5.3). 197 198 THE RISE OF THE SOUTH FIGURE 5.1 Growth paths of Latin America and the Southeast Low saving rates seem to be connected Asian Tigers, 1950–2014 with appreciated real exchange rates. As mea- 100 sured by the Big Mac index, after controlling for GDP per capita, most LAC countries sit 90 on the appreciated side of the distribution 80 and most Asian countries stand on the other 70 side (figure 5.4).1 Yet while LAC’s saving Percentage points 60 rates have been persistently low on average, 50 its real exchange rates have not always been 40 so appreciated. In fact, as this chapter shows, exchange rates have appreciated strongly and 30 rapidly from the very depreciated levels that 20 prevailed in the 1980s and 1990s, when sov- 10 ereign risk ratings for LAC countries were 0 substantially lower than the ratings of their East Asian peers. 50 55 60 65 70 75 80 85 90 95 00 05 10 14 19 19 19 19 19 19 19 19 19 19 20 20 20 20 Latin America Southeast Asian Tigers These trends raise three important ques- tions. The first is whether the roots of the Sources: Based on data from the Maddison Project (Bolt and van Zanden 2013) and World Develop- ment Indicators. region’s low growth can at least in part be Note: Figure shows the weighted-average of per capita gross domestic product (GDP) for each traced back to its low saving. The second group of countries as a percentage of per capita GDP in the United States. Southeast Asian Tigers include Hong Kong SAR, China; the Republic of Korea; Singapore; and Taiwan, China. concerns the role the exchange rate and sovereign risk ratings may have played in channeling the impact of saving on growth. The third is how changes in the world envi- FIGURE 5.2 Growth rates in selected emerging economies, 2003–14 ronment and LAC’s macro-financial policy frameworks affected the saving-to-growth 12 connection in the past and are likely to do 11 so in the future. Key factors considered in 10 analyzing these questions are the rise of the South (a main focus of this report) and LAC’s 9 much-improved macro-financial “immune 8 system.”2 Chapters 1– 4 focus on the nature and Percentage points 7 implications of the evolving connectivity 6 between LAC and the emerging South from 5 a strictly microeconomic vantage point. They emphasize LAC’s arguably insufficiently 4 developed global value chains, excessive reli- 3 ance on primary products, relatively undiver- sified trade structures, and underexploited 2 trade and foreign direct investment (FDI) 1 spillovers. 0 This chapter emphasizes a new, seldom LAC Southeast Asia Eastern Europe China explored connectivity dimension, which 2003–11 (excluding2009) 2012 2013 2014 is based on the composition of aggregate demand—that is, the relative importance of Sources: Based on data from World Development Indicators (for 2003–11) and latest consensus esti- mates and projections by the World Economic Outlook (for 2012–14). domestic versus external demand (see Set of Note: LAC = Latin America and the Caribbean. Growth rates are weighted averages for each region. Facts 1 in chapter 1). An external demand– Emerging economies in Eastern Europe include Croatia, Estonia, Hungary, Lithuania, Poland, Roma- nia, the Slovak Republic, and Turkey. Emerging economies in Southeast Asia include Indonesia, the driven model of integration is one in which Republic of Korea, Malaysia, the Philippines, and Thailand. domestic saving is sufficient to ensure that ASCENDING WITH THE SOUTH WINDS 199 national income exceeds absorption (the sum FIGURE 5.3 Domestic saving rates in selected economies, adjusted of consumption and investment spending), for per capita GDP, 2012 giving rise to current account surpluses. In contrast, a domestic demand–driven model Ukraine South Africa of integration features systematic current Uruguay account deficits or, equivalently, a systematic Greece United Kingdom excess of investment over domestic saving. Brazil The fact that persistent current account defi- Portugal Colombia cits can undermine growth by being accom- New Zealand panied by external debt viability problems Egypt, Arab Rep. Chile or overvalued currencies naturally puts the France spotlight on the links between domestic sav- United States Costa Rica ing and growth. Ireland The debate as to whether saving matters Italy Peru for or is just a corollary of growth is an old Canada and familiar one.3 Sri Lanka Turkey This chapter puts this old debate under a Hungary new light. It examines the medium-term rela- Argentina Finland tionship between domestic saving and trend Russian Federation (as opposed to cyclical) growth from the per- Mexico Spain spective of three possible channels, two of Denmark which go from saving to growth and one of Australia Japan which goes from growth to saving. Pakistan The first “saving-causes-growth” channel Belgium Czech Republic is the real exchange rate (ER) channel. It is Estonia associated with the current account of the bal- Sweden Austria ance of payments, external competitiveness, Lithuania and the imperfect substitutability between Germany Indonesia tradable and nontradable goods. Because Thailand prices are likely to become more responsive Venezuela, RB Hong Kong SAR, China to aggregate demand as economies approach India their production frontier, the strength of the Norway Switzerland ER channel would be expected to rise in Malaysia times of high current account deficits. The Korea, Rep. Philippines second “saving-causes-growth” channel is China the interest rate (IR) channel. It is associated –30 –20 –10 0 10 20 30 with the capital account of the balance of Share of GDP (%) payments, the sovereign risk rating, and the LAC Asia Others imperfect substitutability between domestic and foreign financing. Given that the country Sources: Based on data from United Nations and World Development Indicators. risk rating is inherently a truncated variable Note: GDP = gross domestic product; LAC = Latin America and the Caribbean. The economies shown are the ones that appear in the Big Mac index. Figures shown are the residuals of a (bounded by a zero probability of default for regression. a AAA rating), one would expect the effects of low saving through the IR channel to also be nonlinear (gathering strength on the with a “growth-causes-saving” channel, the downside, as the risk of default and a balance endogenous saving (ES) channel, which is of payments crisis rises, while tapering off associated with the endogenous response of on the upside, as balance of payments via- domestic saving to growth. bility and capacity and willingness to pay is These conceptual distinctions translate ensured). The two channels in turn interact into distinct macroeconomic patterns linking 200 THE RISE OF THE SOUTH FIGURE 5.4 External competitiveness (Big Mac index), adjusted for that undersave should grow at a slower per capita GDP, 2012 rate despite having undervalued curren- cies. Where the ES channel dominates (the Hong Kong SAR, China Japan ES-on-steroids case), domestic saving should Malaysia cease to constrain growth, as an autonomous Russian Federation South Africa marginal increase in investment should gen- Korea, Rep. erate a marginal increase in domestic saving Australia Mexico of an equal or larger magnitude. Czech Republic In a background paper for this report, De la Argentina Ukraine Torre and Ize (2015) develop a medium-term Lithuania equilibrium model that assumes away all United States China transient dynamics, including the short-term Austria fluctuations around potential output and Indonesia India the exchange rate and interest rate dynamics Chile around the risk-adjusted interest rate parity Ireland condition. This model lays the ground for a Estonia New Zealand benchmarking framework, estimated with United Kingdom three-year data averages, that seeks to iden- Greece Portugal tify the long-run equilibrium footprints that Thailand set LAC apart from other South regions and Egypt, Arab Rep. Canada differentiate countries within LAC. Denmark The ER channel is found to dominate the Hungary Germany data for the sample as a whole (that is, at the Turkey world level). In contrast, the IR channel has Sri Lanka France played a uniquely important role in LAC’s Peru recent history. During the 1980s and 1990s, Spain Belgium it worked at full steam against growth, as a Italy result of both domestic policy factors (poor Finland Philippines macro-financial policies that led to capital Costa Rica flight and debt crises) and external factors Colombia Uruguay (high world interest rates) that kept LAC sover- Sweden eign ratings substantially below the ratings of Pakistan Brazil the Southeast Asian middle-income countries. Switzerland The adverse effects on growth of low country Norway Venezuela, RB ratings and frequent crises more than offset –60 –40 –20 0 20 40 60 80 the expansionary effects of deeply depreciated Big Mac Index real exchange rates. Instead, the IR channel worked in favor of growth for LAC during LAC Asia Others 2003–211, reflecting a mix of domestic policy Source: Based on data from World Development Indicators. factors (sounder macro management leading Note: GDP = gross domestic product, LAC = Latin America and the Caribbean. Figures shown are to much improved country risk ratings) and the residuals of a regression. external factors (very low world interest rates). These favorable effects more than offset the saving, the exchange rate, the sovereign risk contractionary effects on growth of strong rating, and growth, thereby giving rise to real exchange rate appreciation. several testable hypotheses. In particular, Looking forward, the potential conse- where the ER channel dominates, countries quences of low saving operating through the that undersave should have persistently over- IR channel are likely to be mitigated for most valued currencies and grow at a slower rate. major LAC countries—on the upside because Where the IR channel dominates, countries of the IR channel’s nonlinear nature (country ASCENDING WITH THE SOUTH WINDS 201 risk ratings have much less room to rise than examining separately the impulse responses in the past), on the downside because of the for pre- and post-2003 data, the SVAR finds improved macro-financial policies and altered that the traditional (pre-2003) tendency of composition of external financing in favor of LAC was to experience macro responses equity rather than debt.4 In contrast, the ER (especially in consumption and the current channel could become a significant hindrance account) that were much more pronounced to growth, as the region’s low saving, in tan- and persistent than in non-LAC emerging dem with much lower sovereign risk premi- economies. As a result, domestic saving in ums and the ample availability of external LAC declined strongly and persistently. How- finance, could result in higher current account ever, thanks to improved monetary and fiscal deficits and persistent pressures toward real policy frameworks, such amplification and exchange rate appreciations. The economet- persistence effects in consumption (saving), ric estimates reported in De la Torre and Ize investment, and the current account responses (2015) indeed suggest that, through the joint appear to have considerably dimmed after operation of the ER and ES channels, the 2003, counterbalanced by larger changes in impact on growth of a boost in saving could the real exchange rate (relative to LAC’s past). be substantial, particularly for countries with Looking forward, LAC’s penchant for persistent current account deficits. Thus LAC low saving, combined with weakness in countries that have consistently undersaved world demand and the region’s social policy relative to their benchmark, thereby incur- priorities, could exert further appreciation ring recurrent current account deficits, would pressures on real exchange rates, by calling most likely benefit from broad-based saving for more stimulative spending policies in a mobilization efforts. context of easy access to foreign finance. If Given the importance of shocks in the macro and debt management in LAC suc- macro-dynamics of the region—particularly ceeds in keeping default risk from rising global shocks, such as the rise of the South— (thereby keeping the IR channel under con- the analysis complements the medium-term, trol), these factors could boost the adverse three-year average equilibrium analysis with ER effects of low saving on long-run growth a structural vector autoregression (SVAR) through larger current account deficits and model that emphasizes fluctuations and more appreciated real exchange rates. The dynamics over time based on quarterly data. region’s policy makers may therefore face This approach, developed by Hevia and difficult trade-offs between short-run and Servén (2014) in another background paper longer-run objectives. for this report, explores the dynamics of sav- The rest of this chapter is structured as ing (consumption), investment, and the real follows. The next section provides a concep- exchange rate resulting from both domestic tual discussion of the three channels linking and global shocks. domestic saving and growth and relating The rise of the South, with China at its them to the literature. The following sec- epicenter, has given rise to three distinct tion looks at LAC’s macro history from the expansionary global shocks: a supply shock perspective of these channels, first from a (reducing the prices of LAC manufacturing broad-brush perspective, then by establish- imports), a demand shock (raising the prices ing a typology of countries based on their of LAC primary exports) and a monetary macroeconomic footprints, and finally by shock (maintaining low interest rates and easy breaking down the analysis into shorter sub- access to foreign finance). Based on data for periods affected by different global or domes- 1990–2012, the SVAR model finds that the tic shocks. The last section looks ahead, first mix of these three shocks would have natu- by arguing that low saving is likely to affect rally boosted the region’s growth—as in fact growth in the future more through the ER it did. However, it also boosted consump- than the IR channel, then by assessing the tion and appreciated real exchange rates. By likely growth-enhancing impact of policies 202 THE RISE OF THE SOUTH aimed at boosting saving, and finally by saving. Because imperfect substitutability can briefly reviewing the tensions and challenges originate from the real side of the balance associated with such a policy agenda. of payments (the current account) as well as the financial side (the capital account), abandoning this assumption gives rise to Concepts and literature review: two possible channels through which saving When does saving matter for may affect growth. The first is the real ER trend growth? channel, which involves the current account This section develops a conceptual frame- and operates through the imperfect substi- work on the links between domestic saving tutability of tradable and nontradable goods. and trend (as opposed to cyclical) growth. The second is the IR channel, which involves This issue has been debated in the theoret- the capital account and operates through ical and empirical literature as far back as the imperfect substitutability of foreign and the 1950s. The analysis presented here adds domestic assets. In either case the imperfect value to this debate by identifying three substitutability of goods or assets is itself well-defined channels through which saving associated with frictions that prevent the may affect growth. internalization of externalities. Thus private Domestic saving ceases to be a matter of saving decisions are not socially optimal, and interest for growth-oriented policy once any government policies can potentially improve of the three following conditions is met. First, the equilibrium.7 if private saving decisions are socially opti- Abandon next the assumption of perfect mal given the constraints, the private sector growth elasticity of domestic saving. Doing is doing the best it can and the government, so opens up a third channel, the ES channel, faced with the same constraints, has no com- through which saving may affect growth, but parative advantage over the private sector in this time it is through a multiplier effect on improving the outcome. Hence there is no the action of the first two channels. justification for saving-promoting policies. 5 The workings of all three channels are Second, if the supply of domestic saving is illustrated in figure 5.5, which links the dif- perfectly growth elastic (any increase in the ference between domestic saving (S D) and demand for saving needed to accommodate investment (I ) to growth ( g), through the higher growth is fully met by an increase in real exchange rate (e) in the case of the ER the supply of domestic saving), the saving channel and through the country risk rat- needed for growth automatically emerges ing (r) in the case of the IR channel. It also as soon as required. Third, if foreign and shows the ES channel, which links back domestic saving are perfect substitutes, for growth ( g) to domestic saving (S D). The any given level of desired investment, for- links across the channels are summarized eign saving compensates for any shortfall in in eight channel-specific elasticities (be, bI, domestic saving. Any change in the compo- b g , ge , gr , gI , gg, and a) and one general, sition of saving (for example, from domestic productivity-related elasticity, d, which links to foreign) then alters only the distribution of growth with investment. These elasticities set growth dividends across beneficiaries (from the basis for the medium-term macro model local residents to foreigners), not growth and benchmarking structure presented in itself.6 detail in De la Torre and Ize (2015) and sum- Hence if domestic saving were to be of marized in annex 5A, which is used as a basis consequence for growth, it would have to be for the analysis presented in later sections because at least one of the above conditions of this chapter. Because it focuses on medi- is not met, which in turn connects with three um-term structural relationships, the model possible channels linking saving and growth. assumes away all transient dynamics around Start by dropping the assumption of per- the equilibrium, including the short-term fect substitutability of domestic and foreign fluctuations around potential output and ASCENDING WITH THE SOUTH WINDS 203 the exchange rate and interest rate dynamics FIGURE 5.5 The three channels linking saving and growth around the sovereign risk–adjusted interest rate parity condition. Positive-learning externalities Consider first the ER channel (the blue arrows in figure 5.5), which involves imper- βg e fect good substitution, the current account, βs βi and the real exchange rate. Excess demand βe βe δ SD γe g γr γr for domestic saving, which implies an excess γi of absorption over output, raises the demand r γg for both tradables and nontradables. For a Negative crises price-taking small economy, excess demand externalities for tradables is resolved solely through quan- α tities (a widening of the current account defi- Multiplier effect cit as imports increase) at given world prices. In contrast, excess demand for nontradables Note: Blue arrows indicate the exchange rate (ER) channel, red the interest rate (IR) channel, and green (teal) the endogenous saving (ES) channel. raises their price relative to the price of trad- ables, thereby appreciating the real exchange rate. The extent of the appreciation depends employment; tradable and nontradables on the elasticity of the real exchange rate with are imperfect substitutes; and the external respect to the excess demand for saving, be, a debt viability condition is met, then a rise in key parameter to estimate. As noted above, domestic demand relative to national income this elasticity is likely to be nonlinear, rising (an increase in investment relative to domes- in times of current account deficits, when the tic saving) leads to an appreciation of the economy is more likely to be overheating. To equilibrium real exchange rate. the extent that tradables and nontradables In contrast, the links between domestic are not perfect substitutes—tradables gener- saving, the real exchange rate, and growth ating more positive growth spillovers than have received limited attention in the long- nontradables—less external competitiveness term growth literature. Several studies find depresses growth. This effect is captured by that the countries that have relied the most on bg, the responsiveness of growth to changes foreign saving are the ones that have grown in the real exchange rate. In addition, total the least (Aizenman, Pinto, and Radziwill investment may rise with a more undervalued 2004; Prasad, Rajan, and Subramanian exchange rate, either because the productiv- 2007). Yet the empirical growth literature ity spillovers of a larger tradable sector are (Eichengreen 2008; Haddad and Pancaro partly internalized or because the tradable 2010) has been generally skeptical or outright sector is more capital intensive. The elasticity critical of the feasibility of using the exchange of investment to changes in the real exchange rate as a robust, durable lever for growth. By rate, bI , captures these effects. For its part, and large, this literature reflects a disequilib- investment affects growth both indirectly, via rium view of the exchange rate rather than the exchange rate and learning-by-investing one in which the real exchange rate is firmly externalities (through be and bg), and directly, grounded in fundamentals (that is, saving). via factor accumulation (through d).8 There is, however, an emerging strand of What does the literature report about literature connecting saving, the exchange the ER channel? The causal link from net rate, and growth, albeit perhaps not yet as domestic saving (the current account) to directly as it could. This literature arguably the equilibrium real exchange rate has been originated with Rodrik (2008), who focuses amply studied in the context of short-term, mainly on the second leg of the link. He finds two-sector (tradables and nontradables) that countries with more depreciated real dependent-economy models.9 These mod- exchange rates grow faster, leading him to els establish that if the economy is in full posit that tradables are somehow special, in 204 THE RISE OF THE SOUTH that they produce more growth-enhancing and the sovereign risk premium (hence the positive externalities than nontradables.10 He local interest rate). Excess demand for saving therefore argues that maintaining a competi- widens the current account deficit, increas- tive real exchange rate is equivalent to a pol- ing the external debt. To the extent that this icy of across-the-board protection in favor of buildup raises the risk of default, it reduces externalities-rich tradable activities.11 Rodrik the country risk rating and raises the risk (2008) also provides empirical evidence on premium and the cost of capital, hindering the other leg of the ER channel, the saving investment and undermining growth.13 The to real exchange rate link. He shows econo- strength of these effects depends on the size metrically that countries that save more have of gr (the elasticity of the country rating with more depreciated real exchange rates. But he respect to the current account) and gg (the does not attempt to tease out the direction of elasticity of growth to changes in the rat- causality or elaborate on the rationale or pol- ing). Because a balance of payments crisis is icy implications of this link.12 an extreme event (it occurs only at the tail In contrast, Korinek and Servén (2010) of the distribution) and the risk premium is formalize both legs of the ER channel. They bounded below by zero (just as the risk rating develop a model in which the relative pro- is bounded by a zero probability of default), ductivities of the tradable and nontradable one would expect gr to be nonlinear. Improve- sectors—a key determinant of long-run equi- ment of the current account balance from a librium exchange rates (see for example Ricci, position of strength should have little or no Milesi-Ferretti and Lee 2008)—become a impact on the rating; instead, improvement in function of aggregate demand (hence saving). the current account balance from a position At the same time, the positive growth exter- of weakness should have a large impact. But nalities of the tradable sector derive from country ratings also affect growth indirectly, Romer-type learning-by-investing spillovers both through investment (with an elasticity in a setting in which the tradable sector is gI) and through the exchange rate (with an more capital intensive (and hence generates elasticity ge). Thus, while lower saving should more growth externalities) than the nontrad- appreciate the real exchange rate on account able sector (Romer 1986). Because external- of the ER channel, by worsening the coun- ities are not internalized, private agents save try’s risk premium (hence the rating) it should and invest too little, the tradable sector is depreciate the real exchange rate on account too small, and the economy grows less than of the IR channel.14 optimally. What does the literature say about the Itskhoki and Moll (2014) also explore IR channel? There is ample evidence that both legs of the ER channel, but this time the world is very close to full capital market based on agency frictions (collateral con- integration. Properly computed, marginal straints) as in Aghion and others (2009). rates of return to capital are largely equalized They justify the need for public intervention across countries (Caselli and Feyrer 2006). based on uninternalized pecuniary external- Moreover, increases in world saving (say, as ities that give rise to constrained-inefficient a result of a fiscal improvement in the United equilibria. Boosting saving raises the prof- States) have a one-for-one impact on invest- its of financially constrained firms, thereby ment across the world (Feyrer and Sham- promoting investment and growth. As in baugh 2009). However, tightly integrated Korinek and Servén (2010), the link with international financial markets do not neces- the exchange rate derives from the fact that sarily imply that foreign and domestic saving the tradable sector is more capital-intensive are perfect substitutes. Indeed, a large body than the nontradable sector. of literature links foreign debt accumulation Consider now the IR channel (the red to balance of payments crises (Eaton and arrows in figure 5.5), which involves imper- Gersovitz 1981; Corsetti and others 2012). fect asset substitution, the capital account, The adverse consequences on output and ASCENDING WITH THE SOUTH WINDS 205 growth of such crises has been analyzed from (Loayza, Schmidt-Hebbel, and Servén 2000). various angles, including from a theoretical In Granger causality studies, growth gener- perspective in which pecuniary externalities ally causes saving (Carroll and Weil 1993). are formally modeled (Jeanne and Korinek Countries undergoing growth transitions 2010) and a broad-based historical perspec- end up with permanently higher saving rates tive (Reinhart and Rogoff 2011).15 (Rodrik 2000). A number of studies (Guari- Finally, consider the ES channel (the green glia, Liu, and Song 2008; Yang, Zhang, and arrows in figure 5.5), in which domestic sav- Zhou 2011) find that the Chinese growth ing follows growth. In this case a reduction in acceleration of the past quarter of a century the rate of growth leads to a decline in saving, was largely a result of endogenous increases of a magnitude determined by the growth in corporate saving. However, the critical elasticity of saving (a). This process sets in condition for self-propelling growth does not motion a reinforcing process that further appear to have been tested or adequately dis- reduces growth, through either a less com- cussed. Yet this condition provides another petitive real exchange rate (the ER channel) testable hypothesis—namely, that where the or a higher sovereign risk premium (the IR ES channel dominates, the critical condition channel). However, should the ES channel be ad . 1 should hold. sufficiently strong (the case of ES on steroids, where ad . 1), a rise in investment could ignite a self-propelling increase in growth Looking back: Latin America and by boosting saving in excess of investment, the Caribbean under the spell of thereby raising net saving and depreciating the interest rate channel the exchange rate. In this case domestic sav- This section views LAC’s recent macro his- ing no longer matters for growth, even if pri- tory from the perspective of the saving and vate saving decisions are not socially optimal growth channels. The analysis suggests that or foreign and domestic saving are imperfect substitutes. The literature on the ES channel explains the positive growth elasticity of saving, FIGURE 5.6 Saving rates of higher-income countries in Latin which underlies the ES channel, in various America and the Caribbean and middle-income countries in Southeast Asia ways. On the household side, growth raises the income of middle-aged people, who Domestic saving (in 2005 dollars) as percent of GDP save more than both the young and the old 40 (Modigliani 1986). Moreover, consumption lags income growth as a result of habit for- 35 mation (Campbell and Cochrane 1999). On the firm side, as income and profits expand, 30 corporate saving can rise, as firms limit 25 dividend distribution to mobilize internal finance (Fazzari, Hubbard, and Petersen 20 1988) or increase output prices relative to wages (Lewis 1954; Kaldor 1958). A more 15 controversial strand of literature (Rowthorn 1982) supports the ES on steroids view by 10 1980 1982 1984 1986 1988 1990 1992 1994 1996 1998 2000 2002 2004 2006 2008 2010 2012 extending Keynesian concepts and con- structs beyond the short run. LAC1 Southeast Asian middle-income countries The empirical evidence generally cor- roborates that saving does follow growth. Sources: Based on data from United Nations and World Development Indicators. Note: LAC1 includes countries in Latin America and the Caribbean with annual per capita gross domestic Panel regressions find output growth to be product (GDP) of more than $5,000 (see annex table 5A.1 for list of countries). Middle-income countries a significant determinant of private saving in Southeast Asia include Indonesia, the Republic of Korea, Malaysia, the Philippines, and Thailand. 206 THE RISE OF THE SOUTH IR-type dynamics dominated in the past model on LAC’s macro responses to global quarter of a century. shocks based on data for 1990–2011. The section zooms in progressively from the general to the specific. It begins by tak- Latin America and the Caribbean ing a broad-brush view of the region in com- and the world parison with the world. It separates LAC into two groups based on their GDP per Consider first the contrasting evolution of capita—LAC1 (higher-income countries) domestic saving rates in the higher-income and LAC2 (lower-income countries)—and LAC countries (LAC1) and the Southeast examines the structural relations between Asian middle-income countries (figure 5.6). saving, the real exchange rate, the sovereign After dipping sharply to about 15 percent rating, and growth.16 The section then fine- of GDP during the early 1980s, the ratio tunes the analysis by breaking it down, first of domestic saving to GDP in LAC1 under- synchronically (by looking at the structural went a sustained recovery during the 1990s. macro patterns for subgroups of countries It stabilized around 22 percent by the early within LAC1) and then diachronically (by 2000s before jumping to just under 25 per- contrasting the broad macro features of the cent at the height of the commodity super- crisis-and-stabilization period [1980–2002] cycle (2005–08). This high rate of saving with the features of the growth recovery proved to be temporary, however: LAC’s period [2003–11]). In this last context, the domestic saving rate started to decline after analysis interprets the results of the SVAR 2009. Throughout the entire period, saving rates in LAC remained below the rates of the fast-growing Asian tigers by about 10 per- centage points of GDP. FIGURE 5.7 Domestic saving and real exchange rate gaps The region’s unimpressive saving perfor- mance stands out in the structural bench- 2 marking exercise of De la Torre and Ize (2015). To focus on medium-term equilibrium relationships, they use three-year averages 1 over 1981–2012 to estimate country-specific benchmarks and gaps. Benchmarks indicate National saving EAP MENA HI where an individual country’s main macro 0 ECA SSA variables are expected to lie, given the coun- LAC2 LAC1 try’s level of economic development (GDP per capita), structural (nonpolicy-related) charac- –1 teristics, exposure to global shocks (partic- ularly as they affect its terms of trade), and –2 the “typical” or average (policy-dependent) –1 –0.5 0 0.5 1 institutional features of its peers.17 Gaps Real exchange rate reflect the distance between where a country LAC1 countries per period is and where it is expected to be.18 Given the LAC1 countries 1990–2012 average controls, they provide a rough measure of the LAC2 countries per period LAC2 countries 1990–2012 average country’s policy-related shortfall or excess Other countries per period relative to its peers. Both benchmarks and Other groups of countries 1990–2012 average gaps are obtained based on a two-stage pro- cess. Simple benchmarks and gaps are first Sources: Based on data from United Nations and World Development Indicators. derived from the ordinary least squares (OLS) Note: Each period is a three-year average. EAP = East Asia and Pacific, ECA = Europe and Central estimates of the structural model displayed Asia, HI = high income, MENA = Middle East and North Africa, SSA = Sub-Saharan Africa. See table 5A.1 for list of countries in each group and annex 5A for details on how the benchmarks are in figure 5.5. By linearly combining these calculated. simple benchmarks and gaps, equilibrium ASCENDING WITH THE SOUTH WINDS 207 benchmarks and gaps are then obtained that FIGURE 5.8 Domestic saving and sovereign risk rating gaps are solutions of the structural model and, 2 therefore, take into account the cross-equa- tion linkages between endogenous variables.19 In this way the correlations across the gaps 1 of different macroeconomic variables reflect National saving the combined effect of the key elasticities EAP of the model (see annex 5A for methodolog- ECA MENA 0 ical details). LAC1 SSA HI LAC2 In particular, the saving and real exchange rate gaps should be negatively correlated –1 if the elasticity of the exchange rate with respect to changes in the current account deficit is positive and significant.20 Figure 5.7 –2 confirms this relationship. It plots the saving –1 –0.5 0 0.5 1 and real exchange rate gaps (that is, the dis- Sovereign risk rating tances from the horizontal and vertical axes) LAC1 countries per period LAC1 countries 1990–2012 average for the full sample, per country per period, LAC2 countries per period where each period is a three-year average. As LAC2 countries 1990–2012 average the fitted line indicates, on average countries Other countries per period that saved more (less) than their benchmark Other groups of countries 1990–2012 average throughout 1981–2012 had more (less) com- petitive real exchange rates. Sources: Based on data from United Nations, World Development Indicators, and Institutional This pattern, which clearly bears the signa- Investor. Note: Each period is a three-year average. EAP = East Asia and Pacific, ECA = Europe and Central ture of the ER channel, applies to every region Asia, HI = high income, MENA = Middle East and North Africa, SSA = Sub-Saharan Africa. See in the world except LAC. For the 1990–2012 table 5A.1 for list of countries in each group and annex 5A for details on how the benchmarks are calculated. subperiod, both LAC1 and LAC2 are located in the lower-left-hand quadrant of figure 5.7: although they undersaved, their real exchange rates were undervalued. Such association points to a unique historical influence of the gaps. This figure confirms Rodrik’s (2008) IR rather than the ER channel in LAC.21 result by neatly illustrating the second leg of Figure 5.8, which shows the correlation the ER channel: countries with more (less) between the saving and sovereign rating gaps, depreciated exchange rates grew more rapidly confirms that both LAC groups had a rating (slowly) than their benchmarks. The two LAC problem. It shows that the IR of low (high) groups are the only ones in the bottom-left- saving is associated with low (high) sover- hand quadrant: they grew more slowly despite eign ratings, all relative to the benchmark. their more depreciated real exchange rates, a Although all regions of the world reveal this finding that is again consistent with their hav- pattern, the two LAC groups again stand out, ing been under the spell of the IR channel. lying in the bottom-left-hand quadrant. Their undersaving is associated with large under- ratings, which in turn are consistent with the Low savers and high savers large real exchange rate undervaluations of Not all LAC countries followed the same figure 5.7. structural patterns, however. In particular, That the adverse impact of low saving on there are clear differences within the LAC1 growth in LAC occurred mainly through the group when countries are divided accord- IR channel can also be gleaned, indirectly ing to their 1990–2012 average saving and from figure 5.9, which shows the correlation real exchange rate positions relative to the between the growth and real exchange rate benchmarks. 208 THE RISE OF THE SOUTH FIGURE 5.9 Real exchange rate and growth gaps in selected group of countries (the Bahamas, Barbados, country groups Brazil, Costa Rica, and Uruguay), that also conform to the ER channel but on the low 10 saving side. Their exchange rates were over- valued on average, and they saved less than their benchmarks. 5 As illustrated in figure 5.11 (which comple- GDP per capita growth EAP MENA EAC ments figure 5.7 by showing where the four LAC1 HI LAC1 high savers and five LAC1 low savers 0 were located during 1990–2012 in terms of LAC2 SSA their average saving and real exchange rate gaps), these two groups fit the ER pattern. –5 The high savers are located above the fitted line, in the top-left-hand quadrant (over- saving and undervalued), while the low sav- –10 –1 –0.5 0 0.5 1 ers are located below the fitted line, in the Real exchange rate bottom-right-hand quadrant (undersaving and overvalued).22 LAC1 countries per period The five remaining LAC1 countries can LAC1 countries 1990–2012 average be assembled into two additional subgroups. LAC2 countries per period The first, composed of Colombia, Ecuador, LAC2 countries 1990–2012 average and Trinidad and Tobago, sits in the bottom- Other countries per period left-hand quadrant, with low domestic saving Other groups of countries 1990–2012 average rates but undervalued exchange rates. This pattern is consistent with the IR channel, in Sources: Based on data from United Nations, World Development Indicators, and Institutional Investor. which undersaving is associated with under- Note: Each period is a three-year average. EAP = East Asia and Pacific, ECA = Europe and Central valued exchange rates caused by low ratings. Asia, GDP = gross domestic product, HI = high income, MENA = Middle East and North Africa, SSA = Sub-Saharan Africa. See table 5A.1 for list of countries in each group and annex 5A for details on Remarkably, as shown in figure 5.10, panel how the benchmarks are calculated. b, this group later migrated to the right, as its real exchange rates appreciated significantly, with Colombia joining the group of ER low Figure 5.10 (together with the underlying savers. This massive rightward shift (a large regression lines for the sample as a whole) real appreciation) reflects the gradual easing shows the average saving and real exchange of the IR channel caused by steadily improv- rate gaps (relative to benchmarks) of all 14 ing country ratings. LAC1 countries, first for the period 1990– The last subgroup of LAC1 countries, 2012 as a whole (panel a) and then broken Argentina and República Bolivariana de down by two subperiods, the 1990s and the Venezuela, appears above the fitted line in 2000s (panel b). panel a of figure 5.10, saving more than Four well-differentiated groups of coun- their benchmarks throughout the 1990 – tries stand out in panel a. The first group, 2012 period. They shifted positions between comprising Chile, Mexico, Panama, and 1990 –99 and 2000 –12, from undervalu- Peru, occupies the top-left-hand quadrant. ation to overvaluation in República Bolivar- These countries saved more than their bench- iana de Venezuela and from overvaluation mark and had undervalued exchange rates. to undervaluation in Argentina (see panel b). This pattern conforms to what one would Although in principle such a pattern could expect for high savers under the ER channel. be consistent with the polar opposite side On the polar opposite side (the bottom- of the IR channel (that is, countries becom- right-hand quadrant of panel a) lies another ing overvalued as a result of stellar country ASCENDING WITH THE SOUTH WINDS 209 ratings), this explanation is not relevant FIGURE 5.10 Saving and real exchange rate gaps for higher- here, as these countries had sovereign rat- income countries in Latin America and the Caribbean ings well below those of the LAC1 countries a. 1990–2012 as a group (figure 5.12). A more plausible 0.5 explanation is the predominance of exten- ER high savers IR high savers CHL VEN sive exchange controls in both countries in PAN ARG the post–World War II period. Exchange PER MEX National saving controls reflected severe macroeconomic 0 ECU disequilibria with acute financial repression COL BHS BRA and chronic capital flight—hence excess URY TTO saving and current account surpluses. At the –0.5 same time, multiple exchange rate systems BRB tended to show up in the reported data as IR low savers overvaluations, given that the (more appre- CRI ciated) official exchange rate was typically –1 ER low savers used to measure the purchasing power par- –0.4 –0.2 0 0.2 ity index. Real exchange rate Even within the nine LAC1 ER countries, b. 1990–2001 and 2002–12 the macro-dynamics differed, depending on whether they were in the low-saver subgroup VEN CHL PAN VEN (the Bahamas, Barbados, Brazil, Costa Rica, ARG CHL MEX PER PAN ARG and Uruguay) or the high-saver subgroup 0 PER MEX ECU TTO CRI (Chile, Mexico, Panama, and Peru). Figure BRA BHS National saving COL ECU COL URY BHS BRA URY 5.13 shows the evolution of each of the main BRB structural gaps for these two subgroups. BRB TTO Consider first the saving gaps (panel a). –1 High savers exceeded their benchmarks throughout most of 1981–2012, except for CRI the most recent period, when their saving rates dipped somewhat below benchmark. –2 The low savers, by contrast, fell short of –0.6 –0.4 –0.2 0 0.2 0.4 their benchmark saving rates by a large mar- Real exchange rate gin throughout the entire period, especially 1990 – 2001 2002 – 2012 during the 1990s. Consider next the sovereign rating gaps Sources: Based on data from United Nations and World Development Indicators. Note: ER = exchange rate, IR = interest rate. The linear fit was calculated for the per-period version (panel b). Following an initial dip, ratings of the complete country sample for 1990–2012. Higher-income countries in Latin America and the rose steadily for both subgroups. In fact, Caribbean (LAC1) are countries with annual per capita gross domestic product of more than $5,000 (see annex table 5A.1 for list of countries). See annex 5A for details on how the benchmarks were after underperforming substantially relative calculated. Three-letter country codes correspond to ISO 3166 standard. to benchmark during the first part of the period, ratings ended up overperforming in the second part of the period, particularly among high savers. currencies became significantly overvalued Consider finally the real exchange rate by the end of the period. In contrast, high gaps (panel c). Both high and low savers savers were able to retain somewhat under- had substantially undervalued currencies in valued currencies by the end of the period. the 1980s and 1990s, and both experienced Although the experience of both subgroups substantial appreciation during the 2000s. is, of course, also consistent with the rising However, the real appreciation was much rating trends under the IR channel, only the more pronounced among low savers, whose very substantial differences in the size and 210 THE RISE OF THE SOUTH FIGURE 5.11 Saving and real exchange rate gaps in selected sign of their saving gaps can explain the stark country groups differences between them by the end of the 2 period. Low savers ended up paying a much heavier price in terms of exchange rate over- valuations, with significantly lower average 1 investment and growth rates than high savers (panels d and e). National saving EAP LAC1-HS HI 0 ECA MENA SSA Shocks, crises, and recoveries LAC1-LS Sizable external and domestic shocks—the –1 effects of which were not independent of LAC’s fundamental macro structure and policy framework—heavily inf luenced –2 macroeconomic developments in LAC over –1 –0.5 0 0.5 1 the past three decades. To help isolate the Real exchange rate dynamic impact of these shocks, the rest of this section combines the benchmark- LAC1 high-saver countries per period ing analysis with the SVAR methodology LAC1 high-saver countries 1990–2012 average of Hevia and Servén (2014), which is sum- LAC1 low-saver countries per period marized in annex 5A and used to interpret LAC1 low-saver countries 1990–2012 average Other countries per period the region’s responses to shocks during the Other groups of countries 1990–2012 average following three subperiods: the crisis period of the 1980s (observations 1–3), the macro Sources: Based on data from United Nations and World Development Indicators. stabilization period of the 1990s and early Note: LAC1 countries are countries in Latin America and the Caribbean (LAC) with annual per capita 2000s (observations 4–8), and the growth gross domestic product of more than $5,000 (see annex table 5A.1 for list of countries in all groups). recovery period that started in 2003 (obser- LAC1-HS (high savers) includes Chile, Mexico, Panama, and Peru. LAC1-LS (low savers) includes the Bahamas, Barbados, Brazil, Costa Rica, and Uruguay. EAP = East Asia and Pacific; ECA = Europe and vations 9–11). Central Asia; HI = high income; MENA = Middle East and North Africa; SSA = Sub-Saharan Africa. See Consider first the crisis decade of the annex 5A for details on how the benchmarks are calculated. 1980s. Major negative global shocks— including receding world demand in the wake of the second oil crisis and the U.S. Federal FIGURE 5.12 Country ratings for selected country groups Reserve’s 1981–82 disinflation efforts, which 80 brought world interest rates to historical 75 highs—hit the region’s weak macroeconomic 70 structures during this period. Low saving Sovereign risk rating index 65 rates in LAC, together with high real inter- 60 55 est rates in the United States (figure 5.14) 50 and widespread capital flight (that is, sav- 45 ing invested abroad rather than at home), 40 set the grounds for a perfect IR-style storm. 35 30 Rising external debt rapidly unfolded into 25 balance of payments and debt crises, under- 20 cutting growth. Indeed, this period recorded 1980 1984 1988 1992 1996 2000 2004 2008 2012 a peak number of crisis events, with LAC LAC1 Argentina and Venezuela, RB Southeast Asian experiencing many more crises than did the middle-income countries middle-income countries of Southeast Asia Source: Based on data from Institutional Investor database. (figure 5.15). The generally depressed terms Note: Middle-income countries in Southeast Asia include Indonesia, the Republic of Korea, Malaysia, the Philippines, and Thailand. LAC1 countries are countries in Latin America and the Caribbean (LAC) with of trade exerted further depreciating pres- annual per capita gross domestic product of more than $5,000 (see annex table 5A.1 for list of countries). sures on real exchange rates. But despite ASCENDING WITH THE SOUTH WINDS 211 FIGURE 5.13 Policy-adjusted gaps for high-saver and low-saver higher-income countries in Latin America and the Caribbean, 1981–2012 a. National saving gaps b. Sovereign risk rating gaps 0.6 0.6 0.4 0.4 0.2 0.2 0 0 –0.2 –0.2 –0.4 –0.4 –0.6 –0.6 –0.8 –0.8 0 1 2 3 4 5 6 7 8 9 10 11 12 0 1 2 3 4 5 6 7 8 9 10 11 12 c. Real exchange rate gaps d. Investment gaps 0.6 0.6 0.4 0.4 0.2 0.2 0 0 –0.2 –0.2 –0.4 –0.4 –0.6 –0.6 –0.8 0 1 2 3 4 5 6 7 8 9 10 11 12 0 1 2 3 4 5 6 7 8 9 10 11 12 e. GDP per capita growth gaps f. Current account gaps 6 0.6 4 0.4 2 0.2 0 0 –2 –0.2 –4 –0.4 –6 –0.6 –8 –0.8 0 1 2 3 4 5 6 7 8 9 10 11 12 0 1 2 3 4 5 6 7 8 9 10 11 12 LAC1 low savers LAC1 high savers Sources: Based on data from United Nations, World Development Indicators, and Institutional Investor. Note: Higher-income countries in Latin America and the Caribbean (LAC1) are countries with annual per capita gross domestic product of more than $5,000 (see annex table 5A.1 for list of countries). Each period is a three-year average. See annex 5A for details on how the benchmarks are calculated. frequent and significant currency devalu- instability, as reflected in the sharp dips in ations (indeed, fear of depreciation was the sovereign ratings (see figure 5.12). order of the day), the heavily discounted Although the SVAR exercise uses data that exchange rates were not effective in promot- start only in 1990, it can be loosely extrap- ing exports and growth, because of the drag olated to this earlier period. The data sug- exerted by macroeconomic imbalances and gest that shocks, particularly the negative 212 THE RISE OF THE SOUTH FIGURE 5.14 Real U.S. interest rate still low sovereign ratings (see figure 5.12), 12 high world real interest rates (see figure 5.14), and falling but still high incidence of 10 crisis events (see figure 5.15). Despite visible 8 progress in the fight against inflation, IR Percentage points 6 dynamics fed policy makers’ intense fears 4 of depreciation and concerns about sudden 2 stops and reversals in capital flows.24 0 Instead, the post-2002 growth recovery –2 period saw a 180-degree shift in the direc- –4 tion of the IR winds. Real exchange rates –6 in many LAC countries appreciated rapidly after 2002, wiping out much of the region’s 80 82 84 86 88 90 92 94 96 98 00 02 04 06 08 10 12 14 19 19 19 19 19 19 19 19 19 19 20 20 20 20 20 20 20 20 traditional undervaluation relative to bench- Sources: Based on data from the Board of Governors of the Federal Reserve System and the Federal Reserve Bank of Cleveland databases. marks and leading to significant currency Note: Series was constructed by deflating the (effective) monthly federal funds rate by the inflation overvaluation for the region’s ER low savers rate for the previous 12 months. (see figure 5.13, panel c). Fear of apprecia- tion became predominant. The substantial improvements in sovereign ratings largely FIGURE 5.15 Incidence of crises in Latin America and Southeast contributed to these appreciations. Indeed, Asia, 1980–2010 LAC country ratings converged to the level 70 of the middle-income countries of Southeast 60 Asia (see figure 5.12). The change largely reflected the region’s success in regain- Composite regional 50 ing macro stability thanks to significantly crises index 40 improved macroeconomic policy frame- works, particularly in countries that made 30 an early shift toward robust inflation tar- 20 geting. The adverse growth effects of these currency appreciations, however, were more 10 than offset by the favorable IR winds. As a 0 result, on average growth picked up strongly 80 82 84 86 88 90 92 94 96 98 00 02 04 06 08 10 between 2003 and 2012, during which time 19 19 19 19 19 19 19 19 19 19 20 20 20 20 20 20 LAC1 Southeast Asia LAC experienced a nontrivial process of con- vergence in GDP per capita (see figure 5.1).25 Source: Based on data from Reinhart and Rogoff 2011. Note: The variable shown in the figure was constructed by summing all the dummy variables for Two key external factors contributed sig- the different kinds of crises (currency, inflation, domestic debt, external debt and banking) across nificantly to this post-2002 outcome: the rise the countries within each region, and then dividing the resulting sum by the number of countries in the region times the number of kinds of crises (5). Hence, if the variable were to take the value of of China and the sharp decline in world inter- 100 in a certain year, it should be read as “all the countries within the region experienced every kind est rates, to historical lows. The rise of China of crisis that year.” Southeast Asia includes China; Indonesia; the Republic of Korea; Malaysia; the Philippines; Singapore; Taiwan, China; and Thailand. LAC1 includes countries in Latin America and entailed a global supply shock (a rise in world the Caribbean with annual per capita gross domestic product of more than $5,000 (see annex table output associated with declining prices for 5A.1 for list of countries). manufactured goods and hence lower global inflation) together with a global demand world demand shock of the 1980s, may have shock (which pushed commodity prices up). depressed saving in LAC more deeply and Taken jointly these two shocks unequivo- permanently than in other emerging econo- cally led to major gains for LAC’s terms of mies (figure 5.16).23 trade. However, because the supply shock During the macro stabilization period of was accompanied by large current account the 1990s, IR dynamics continued to under- surpluses (that is, excess saving) at the epi- mine growth, as evidenced by the rising but center of the shock (China and other East ASCENDING WITH THE SOUTH WINDS 213 FIGURE 5.16 Impulse responses in Latin America and the Caribbean and other emerging market economies to positive global demand shocks a. Investment b. Real exchange rate 2 0.010 0.005 Percentage points Percentage points 1 0.000 –0.005 0 –0.010 –1 –0.015 1 11 21 31 41 1 11 21 31 41 Quarters Quarters c. Saving rate d. Current account 0.3 0.15 0.2 0.05 Percentage points Percentage points 0.1 –0.05 0 –0.15 –0.1 –0.2 –0.25 1 11 21 31 41 1 11 21 31 41 Quarters Quarters LAC 70% of LAC model Non-LAC emerging market economies 70% of Non-LAC emerging market economies model Source: Hevia and Servén 2014. Note: Solid lines represent accepted model median deviation from the trend from a global demand shock, in terms of the sign restrictions defined by the authors for the model and the shock in Hevia and Servén (2014). Dotted bands encompass 70 percent of the accepted models. See table 5A.4 for details on the sign restrictions. Non-LAC (Latin America and the Caribbean) emerging market economies include Hungary, India, Indonesia, the Republic of Korea, the Philippines, Poland, the Russian Federation, South Africa, Thailand, and Turkey. Asian economies), its effects are likely to have in world interest rates led to an exchange dominated the effects of the global demand rate appreciation and a sharp decline in shock. The historically low world interest the region’s saving rates (figure 5.17). The rates that have generally prevailed since 2002 increases in global demand and global sup- represented, in turn, a major positive global ply resulting from the rise of the South also monetary shock that reflected the recycling in boosted LAC’s growth and appreciated its the North of current account surpluses orig- real exchange rates (see figures 5.16 and inating in the South, particularly East Asia, 5.18). The two shocks had opposite effects on as well as accommodative monetary policies saving, but with the supply shock dominat- in the North, particularly the United States. ing the demand shock, the combined effect Using the entire 1990–2011 period (that was a further reduction in LAC’s saving rates is, assuming unchanged economic structures and a deterioration of its current accounts. and institutional setups), the SVAR exercise For LAC’s commodity-exporting countries, shows that LAC’s responses were more pro- a prolonged upswing phase of the com- nounced and persistent than the responses of modity price supercycle further boosted the other South regions. In particular, the decline appreciation. 214 THE RISE OF THE SOUTH FIGURE 5.17 Impulse responses in Latin America and the Caribbean and other emerging market economies to global monetary easing a. Investment b. Real exchange rate 2.0 0.006 0.004 1.5 0.002 Percentage points Percentage points 1.0 0.000 –0.002 0.5 –0.004 0 –0.006 –0.008 –0.5 –0.010 –1.0 –0.012 1 11 21 31 41 1 11 21 31 41 Quarters Quarters c. Saving rate d. Current account 0.2 0.10 0.1 0.05 Percentage points Percentage points 0 0 –0.1 –0.05 –0.2 –0.10 –0.3 –0.15 –0.4 –0.20 1 11 21 31 41 1 11 21 31 41 Quarters Quarters LAC 70% of LAC model Non-LAC emerging market economies 70% of Non-LAC emerging market economies model Source: Hevia and Servén 2014. Note: Solid lines represent accepted model median deviation from the trend from a global monetary easing shock, in terms of the sign restrictions defined by the authors for the model and the shock in Hevia and Servén (2014). Dotted bands encompass 70 percent of the accepted models. See table 5A.4 for details on the sign restrictions. Non-LAC (Latin America and the Caribbean) emerging market economies include Hungary, India, Indonesia, the Republic of Korea, the Philippines, Poland, the Russian Federation, South Africa, Thailand, and Turkey. The regional pattern of accelerated both high savers and low savers, only the low growth, low and falling domestic saving, savers continued to appreciate into overvalu- appreciated real exchange rates, and deteri- ation (see figure 5.13, panel c). orating current accounts that prevailed in many LAC countries over the period 2003– 11 is therefore consistent with the domestic Looking ahead: Growth- macro-policy improvements captured by impairing effects of low saving the benchmarking exercise as well as the through the exchange rate response to global shocks captured by the channel SVAR exercise. However, the De la Torre This section builds on lessons from the recent and Ize (2015) benchmarking exercise goes past to look at the nature of the drag that low farther in explaining the differential patterns saving may exert on LAC’s future growth of appreciation across LAC countries. Struc- potential. It first assesses the relative roles tural saving differences made a huge differ- and importance of the ER and IR channels ence in the extent of the appreciation across in light of current changes in world demand countries. Although saving rates declined for and improvements in LAC’s macro-financial ASCENDING WITH THE SOUTH WINDS 215 FIGURE 5.18 Impulse responses in Latin America and the Caribbean and other emerging market economies to positive global supply shocks a. Investment b. Real exchange rate 2.0 0.006 1.5 0.002 Percentage points Percentage points 1.0 0 –0.002 0.5 –0.006 0 –0.010 –0.5 –1.0 –0.014 1 11 21 31 41 1 11 21 31 41 Quarters Quarters c. Saving rate d. Current account 0.30 0.10 0.25 0.20 0.05 0.15 Percentage points Percentage points 0.10 0 0.05 0 –0.5 –0.05 –0.10 –0.10 –0.15 –0.20 –0.15 –0.25 –0.30 –0.20 1 11 21 31 41 1 11 21 31 41 Quarters Quarters LAC 70% of LAC model Non-LAC emerging market economies 70% of Non-LAC emerging market economies model Source: Hevia and Servén 2014. Note: Solid lines represent accepted model median deviation from the trend from a global supply shock, in terms of the sign restrictions defined by the authors for the model and the shock in Hevia and Servén (2014). Dotted bands encompass 70 percent of the accepted models. See table 5A.4 for details on the sign restrictions. Non-LAC (Latin America and the Caribbean) emerging market economies include Hungary, India, Indonesia, the Republic of Korea, the Philippines, Poland, the Russian Federation, South Africa, Thailand, and Turkey. immune system. It then discusses the desir- closer to the regression line and neatly aligned ability of boosting saving, based on prelim- along an ER pattern such that the lower the inary results in De la Torre and Ize (2015) saving the more overvalued the exchange regarding the growth impact of increases in rates (figure 5.19). saving rates. It concludes with a brief review Instead, the IR channel now looks more of policy tensions and challenges. subdued. As long as the current world environment of low interest rates and con- strained demand lasts, it should facilitate The coming threat of the exchange access to foreign finance. At the same time, rate channel thanks to public sector deleveraging and With the end of the commodity price bonanza international reserve accumulation, LAC and low international interest rates, the prob- rebalanced its portfolio of net external lia- lem of low saving is likely to come back to bilities from debt to equity, thereby achiev- haunt most of the region, this time mainly ing a more resilient form of international through the ER channel. Indeed, except for financial integration. LAC actually became a couple of outliers, the region is now much a net creditor with respect to the rest of the 216 THE RISE OF THE SOUTH FIGURE 5.19 Saving and exchange rate gaps for higher-income sovereign ratings at high levels for much of countries in Latin America and the Caribbean, 2011–12 averages the region. They will likely continue to do so in the future, further muting the down- 0.5 PAN* sides of the IR channel.26 Indeed, the SVAR ARG* tests performed by Hevia and Servén (2014) VEN CHL find that the adoption of inflation targeting MEX PER 0 and prudent fiscal rules in the region have National saving TTO* ECU BHS* CRI resulted in significantly smoother responses COL of output, consumption (hence saving), and URY BRA investment to global supply and demand –0.5 shocks, counterbalanced (at least in the case of the inflation targeters) by larger responses of the real exchange rate. These outcomes BRB* –1 are consistent with the argument that LAC’s –0.4 –0.2 0 0.2 0.4 low saving rates are more likely to constrain Real exchange rate future growth through the ER channel than Sources: Based on data from United Nations and World Development Indicators data. through the IR channel. LAC has been tran- Note: The linear fit was calculated for the complete country sample for 1990–2012. The gaps were sitioning from an IR world dominated by fear calculated as the difference between the actual and benchmark values for the variables. See annex of depreciation to an ER world dominated by 5A for details on how the benchmarks are calculated. Higher-income countries in Latin America and the Caribbean are countries with annual per capita GDP of more than $5,000 (see annex table fear of appreciation.27 5A.1 for list of countries). Three-letter country labels correspond to the ISO 3166 standard. * = due to However, any mitigation of the IR prob- missing data for 2011–12 period, the latest available period was used. lem will only exacerbate the ER problem. By enhancing access to foreign finance and world in debt contracts (reflecting external allowing for higher current account deficits, debt deleveraging and international reserve it will surely generate further real exchange accumulation). In addition, it became a rate appreciation pressures. Indeed, the more active user of foreign equity finance, already worsening trend of the region’s cur- which led to a rising net debtor position in rent account balances, as evidenced in figure risk-sharing equity contracts, particularly 5.13 (panel f), is a reminder that policy action FDI, with respect to the rest of the world may become required in the future not just (figure 5.20). To be sure, the associated pay- on macro stability–related grounds but also ment of dividends will still be a challenge on growth-related grounds.28 for balance of payments viability. More- over, some FDI (particularly FDI related to The scope for policy response the commodity cycle) may actually reflect retained earnings from multinational cor- The scope for a policy response to such a porations held in the form of liquid assets threat ultimately depends on two key empiri- and hence prone to sudden stops. Overall, cal findings. The first is the direction of cau- however, the change in composition should sality, which needs to be carefully ascertained help shield the region’s external liabili- to make sure that the correlation between sav- ties from rollover and currency risks and ing and growth reflects mainly the workings dampen the impact on external financing of the ER and IR channels rather than the ES costs once world interest rates start to rise. channel (that is, reverse causality from growth Significant improvements in monetary to saving). The second is the range of possible policy (the shift in several of the major LAC policy impacts, which also needs to be estab- countries to inflation-targeting-cum-exchange- lished in order to justify the pain associated rate-flexibility regimes) and fiscal policy (the with policies designed to boost saving. introduction of sounder government debt De la Torre and Ize (2015) report pre- management and fiscal responsibility rules in liminary findings on both issues. They first several LAC countries) have helped stabilize conduct OLS–based estimates of the five ASCENDING WITH THE SOUTH WINDS 217 structural equations (one for each of the FIGURE 5.20 Composition of foreign assets and liabilities in five endogenous variables) that underpin the selected countries in Latin America and the Caribbean, 1990–2011 model. (These equations are depicted in figure 10 5.5 and formalized in equations 5A.5–5A.9 5 in annex 5A.) These estimates produce a set of elasticities that are consistent with all 0 three channels linking saving and growth. 29 –5 Percent of GDP Moreover, they also support the nonlinearity –10 of the ER and IR channels (as expected, the –15 responses of the real exchange and country rating to changes in the current account are –20 higher in economies with current account –25 deficits). Yet the OLS estimates are affected –30 by endogeneity problems and, as they do –35 not take the cross-equation correlations into 19 0 19 1 19 2 19 3 94 19 5 19 6 19 7 98 20 9 20 0 20 1 20 2 20 3 20 4 20 5 20 6 20 7 20 8 09 20 0 11 9 9 9 9 9 9 9 9 0 0 0 0 0 0 0 0 0 1 19 19 19 20 account, yield very limited impacts.30 To overcome these limitations, De la Torre Net equity position Net debt position and Ize (2015) conduct alternative estimates Source: Based on data from Lane and Milesi-Ferretti 2007. based on instrumented reduced forms (where Note: GDP = gross domestic product, LAC = Latin America and the Caribbean. Figures are for LAC7 (Argentina, Brazil, Chile, Colombia, Mexico, Peru, and Uruguay). Ratios are calculated at the country the endogenous variables are regressed level and then averaged across countries. against all exogenous variables) instead of structural forms. These reduced forms use as instruments the exogenous variables that most strongly explain each of the endoge- nous variables when running the structural (OLS) equations. The structural elasticities for the instrumented variables are then cal- pressures that might otherwise result from culated backward, based on the mathemat- large real appreciations.34 It would also help ical restrictions imposed by the model. 31 By reduce the potential IR drag on growth aris- fully capturing the cross-equation linkages, ing from unsustainable balance of payment this approach leads to considerably higher trajectories. elasticities than the ones obtained through What can governments do to increase the structural form estimates. 32 Remark- aggregate saving? Although economists ably, while the ES channel also comes out often throw in the towel when pressed to much stronger, the threshold condition for think about saving as a policy variable, a self-propelling growth (ad . 1) is never saving-boosting reform agenda is not beyond verified.33 reach. It could involve actions on the fiscal, financial sector, and social safety net fronts. On the fiscal side, public sector saving can be Policy tensions and challenges directly increased by raising revenues, reduc- These results should be taken with some cau- ing public consumption, or both, and tax and tion, as they are still preliminary and subject subsidy policy can be used to foster private to confirmation. However, they suggest that saving at the household and corporate levels. for many LAC countries, particularly coun- Actions affecting the financial services sector tries incurring recurrent current account might involve regulations to promote saving deficits, a sustained effort to raise domestic and investment rather than consumption (by, saving would most likely yield substantial for example, facilitating the channeling of long-run growth benefits. Higher aggre- saving into long-term finance, expanding gate saving would promote growth through financial inclusion from the deposit-taking the ER channel and limit the protectionist and payments side rather than the lending side, 218 THE RISE OF THE SOUTH and preventing credit-fueled consumption counting on strong demand from the rest booms). On the social safety net side, fos- of the world to help pull them out of their tering saving might require redesigns of the current slump. At the same time, low-cost health, pensions, and unemployment safety external financing remains readily available, nets that promote self-reliance (private sav- especially for highly rated LAC countries, ing) rather than excessive reliance on the many of which have significant room to state (public saving). However, efforts to raise increase indebtedness. Prevailing world con- aggregate saving could run into potentially ditions can thus induce countries to maintain acute policy conflicts, both macroeconomic or expand domestic demand while externally (between short- and long-term growth objec- financing larger current account deficits. tives) and distributional (across generations Doing so would help sustain world demand and within the current generation). but possibly at the expense of LAC countries’ On the macro side, LAC policy mak- external competitiveness and hence long-run ers would first need to find the right policy growth potential. In the extreme, it could balance and timing from a purely long-run weaken the balance of payments, thus resus- trade-off between present and future con- citating the IR channel. Conversely, current sumption. Further complications are likely weaknesses in world demand exacerbate the to arise, however, because policy makers risk for countries of falling into a slump when also need to address the potential conflicts raising their saving rate. between short-term growth objectives (which On the distributional side, curtailing con- rely on strong countercyclical aggregate sumption today could benefit future gener- demand management to close output gaps ations, but it would do so on the shoulders while keeping inflation low and stable) and of the current generation. This effort could long-term growth objectives (which require run up against a brick wall, politically and strong aggregate saving as a complement to socially, especially if the cuts in consump- supply-side productivity-enhancing reforms). tion fall on the poorer segments of the popu- Navigating such treacherous waters clearly lation, something that would be particularly requires good timing and proper use of off- explosive in LAC’s unequal societies. Man- setting policies, particularly monetary pol- aging these conflicts would be facilitated by icies. A policy shift toward easier monetary policies designed to encourage asset building and tighter fiscal, as well as further progress among the poor by, for instance, investing in building up the region’s countercyclical fis- in health, education, and housing. Cutting cal and monetary policy capacity, should help public spending, particularly public invest- in this regard.35 ment, without affecting the quality of the These country-specific macro manage- business-enabling environment could also ment difficulties are further complicated be a major challenge. Double-duty work by the current world environment, as weak on productivity-enhancing supply-side world demand puts a premium on spend- reforms would help ease these conundrums ing, rather than saving. High-income coun- and increase the government’s maneuvering tries, European countries in particular, are room. A S C E N D I N G W I t H t H E S O U t H W I N D S    219 annex 5a  The benchmarking The a, b’s, and g’s are the elasticities asso- approach ciated with the ES, ER, and IR channels, respectively; d is a structural elasticity link- The growth model ing growth to investment, which reflects pro- On the demand side, consider the follow- ductivity; and the e’s are residuals. The main ing IS-LM (investment-saving /liquidity features of the model are highlighted in the preference–money supply) and interest rate following differential equations: parity equilibrium conditions, where e is the de be 1 gegr real equilibrium exchange rate (an apprecia- 52 nS (5A.10) dpS D tion raises e), r the domestic real interest rate, r* the real world interest rate, r the country’s dg 1 5 2 3 be 1 bg 1 dbi 2 1 gr 1 gg 1 dgi 2 rating, s(.) the risk premium, and the p’s are dpS D exogenous variables (controls) affecting each macro aggregate or price that may reflect 1 gegr 1 bg 1 dbi 2 4 nS (5A.11) the country’s level of development, struc- tural characteristics, external shocks, policy de d 1 I 2 SD 2 5 1 be 1 gegr 2 choices, or catastrophic policy outcomes: dpI dpI 2 2 1 1 2 I 1 e , r , pI 2 5 SD 1 g, pS 2 1 SF 1 e , r , pS 2 (5A.1) 1 be 1 gegr 2 1 1 2 ad 2 D F 5 nI (5A.12) 2 D r 5 r* 1 s 1 r , ps 2 (5A.2) where Plugging (5A.2) into (5A.1) and sub- D 5 1 1 a A 2 bI be 2 gI ge 2 bIgegr (5A.13) suming the world interest rate and country premium-specific factors into an expanded set A 5 1 be 1 gegr 2 1 bg 1 d bI 2 1 gr 1 gg 1 dgI 2 of investment factors p9 I , yields the following reduced-form IS-LM equilibrium condition: (5A.14) 2 1 1 1 2 I 1 e , r , pI92 5 SD 1 g , pS 2 1 SF 1 e , r , pS 2 (5A.3) Given D . 0, equation (5A.10) indicates D F that the exchange rate depreciates in response The model is completed on the supply side to a positive saving innovation if be 1 gegr . with a simple reduced-form growth equation: 0. Given gr , 0, this condition is satisfied if 1 _ 1 the be term (the ER channel) dominates the g 5 g 1 I , e, r , pg 2 (5A.4) gegr term (the IR channel). Equation (5A.11) indicates that the impact For estimation purposes, the above model can on growth of a saving innovation can be bro- be linearized and the exchange rate and country ken down as a sum of three terms, which pro- ratings expressed as a function of net saving: vide a convenient means to size up the relative e 5 be 1 I 2 SD 2 1 g e r 1 a ne n 1 j 1 pj 1 ee (5A.5) strengths of the ER and IR channels as well j51 as the multiplier effect of the ES channel. The first term, be 1 bg 1 dbi 2 , sums up the growth r 5 g r 1 I 2 SD 2 1 a nrj pj 1 er (5A.6) 2 n effect, both direct (through growth) and indi- j51 rect (through investment), of an increase in I 5 bI e 1 g I r 1 a nIj pj 1 eI (5A.7) 2 1 n saving, as carried through the ER channel. Similarly, the second term, gr 1 gg 1 dgi 2 , sums j51 up the direct and indirect growth effects of SD 5 a g 1 a nS 1 n an increase in saving, as carried through the j pj 1 eS (5A.8) j51 IR channel. The third term, gegr 1 bg 1 dbi 2 , picks up the interaction of the two chan- g 5 d I 1 b g e 1 g g r 1 a ng n 1 2 1 p 1 eg (5A.9) j j nels coming from the impact of the country j51 rating on the exchange rate. Finally, given 220   THE RISE OF tHE SOUtH A , 0, the aA term in (5A.13) indicates that level of economic development (as proxied by any growth stimulus carried by the ER or IR its per capita GDP) and for country-specific channels is multiplied, in proportion to a, by structural (nonpolicy-related) features and the ES channel. external shocks. It then controls for devia- Equation (5A.12) indicates that the tions from the policy benchmarks set by the exchange rate appreciates in response to a country’s peers. Gaps reflect country-specific boost in investment if net saving declines policy choices associated with either devi- d 1I 2 SD2 . 0, which will be the case as long ations from benchmarks for all identified as bg 1 gegr . 0 (the ER channel dominates policy-influenced variables or any latent the IR channel) and ad , 1 (the ES channel is policy difference remaining embedded in not on steroids). the residuals. 36 Finally, gaps are expressed as solutions of the underlying macro model. benchmarking As shown below, doing so ensures full model Benchmarks help make countries compa- consistency and incorporates into the gaps rable. They provide an indication of where the correlations across variables derived from a country should be given its stage of eco- the basic model elasticities. nomic development, its structural (nonpolicy- Equations (5.A5)–(5.A9) can be expressed related) characteristics, and the level of in vector form: policy-related variables that is “typical” (albeit not necessarily optimal) for countries Xk t 5 AX k t 1 BYk t 1 CPk t 1 Nk 1 Mt 1 Lkt at similar stages of economic development. (5A.15) The benchmarking framework in De la Torre and Ize (2015) follows a three-step Pk t 5 B9 Y k t 1 Dk t (5A.16) procedure. It first controls for the country’s annEx Table 5a.1  Country group composition Region Countries Higher-income countries in Latin America Argentina, the Bahamas, Barbados, Brazil, Chile, Colombia, Costa Rica, Ecua- and the Caribbean (LAC1) dor, Mexico, Panama, Peru, Trinidad, Uruguay, Venezuela, RB Lower-income countries in Latin America Belize, Bolivia, Dominican Republic, El Salvador, Guatemala, Guyana, Hondu- and the Caribbean (LAC2) ras, Nicaragua, Paraguay East Asia and Pacific (EAP) Bangladesh, Bhutan, Cambodia, China, Fiji, Hong Kong, SAR, China, India, Indonesia, the Republic of Korea, Malaysia, Pakistan, Papua New Guinea, the Philippines, Sri Lanka, Thailand, Tonga, Vietnam Europe and Central Asia (ECA) Albania, Armenia, Azerbaijan, Belarus, Bosnia and Herzegovina, Bulgaria, Cro- atia, the Czech Republic, Estonia, Georgia, Greece, Hungary, Kazakhstan, the Kyrgyz Republic, Latvia, Lithuania, Macedonia, Moldova, Mongolia, Romania, Slovenia, Tajikistan, Turkmenistan, Ukraine High income Australia, Belgium, Canada, Cyprus, Denmark, Finland, France, Germany, Iceland, Ireland, Israel, Italy, Japan, New Zealand, Norway, Portugal, Spain, Sweden, Switzerland, the United Kingdom, the United States Middle East and North Africa (MENA) Algeria, the Islamic Republic of Iran, Jordan, Lebanon, Morocco, Syria, Tunisia, Turkey Sub-Saharan Africa (SSA) Angola, Benin, Botswana, Burkina Faso, Burundi, Cameroon, Chad, Côte d’Ivoire, Equatorial Guinea, Ethiopia, Gabon, The Gambia, Ghana, Guinea, Guinea-Bissau, Kenya, Lesotho, Madagascar, Malawi, Mali, Mauritania, Mau- ritius, Mozambique, Namibia, Niger, Rwanda, Senegal, South Africa, Sudan, Swaziland, Togo, Uganda, Zambia Note: The dividing line between LAC1 and LAC2 countries is per capita income of $5,000 a year. A S C E N D I N G W I t H t H E S O U t H W I N D S    221 where Xtk is a vector of endogenous macro controls for identifiable structural differences variables for country k at time t; Y tk is a vector across countries include trade and capital of identifiable country-specific fundamentals openness, demographics, and dependence or external shocks; Pk t is a vector of identifi- on natural resource extraction. Controls for able policy choices; Nk are country-specific differential country exposure to external effects, which may reflect unidentified shocks include terms of trade and safe haven country-specific endowments, preferences, effects. Policy controls include fiscal policy or policy choices; Mt are worldwide dynamic (the fiscal balance and public consumption) disturbances; and L k t are normally distrib- and the country’s macroeconomic record, uted country-specific dynamic disturbances. as determined by its exposure to inflation- The policy controls contain a universal com- ary or debt crises. The sample covers 119 ponent, B 9Y kt , which is a predictable func- countries with annual data over the period tion of the country’s fundamentals, and 1981–2011 (not all countries’ data cover the a country-specific component, D k t , which whole period; see annex table 5A.2 for data reflects the country’s choice (good or bad) to definitions and sources). To better capture deviate from that predictable level. medium-term relations, three-year averages Replacing Pk t from equation (5A.16) in are used throughout instead of annual data. equation (5A.15) yields The United Nations (UN) database, which provides national accounts in real terms Xk t 5 AX k t 1 1 B 1 CB 9 2 Y k t 1 1 CDk t 1 Nk 2 (that is, where each component of aggregate demand is deflated by its own price deflator) is used to limit the possible biases that would 1 Mt 1 Lk (5A.17) t otherwise result from terms of trade change. Consistent with other studies of saving (see For policies that match the policies adopted Loayza, Schmidt-Hebbel, and Servén 2000), on average by other countries with similar the income measure used to calculate domes- fundamentals (that is, for which CDk 1 Nk t tic saving is gross national disposable income 5 0), the solution of equation (5A.17) yields ^ k, and (GNDI), equal to GDP plus net factor income a set of policy-neutral benchmarks, X ~ t ^ k}, (that is, GNP) and net unrequited transfers; policy-neutral average gaps, Xk 5 E {X k 2X t t t total domestic saving is then calculated as such that GNDI minus consumption expenditure. To ^ k 5 AX X ^ k 1 1 B 1 CB9 2 Y k 1 M (5A.18) facilitate comparison of exchange rates across t t t t countries (rather than across time for the ~ ~ X k 5 AX k 1 E 5 CD k t 1 Nk 6 (5A.19) same country), the World Bank’s purchasing power parity (PPP) conversion factor divided The gaps thus reflect country specifici- by the country’s nominal exchange rate with ties, which can be unidentified endowments, respect to the U.S. dollar (the national price preferences, or policy choices embedded index [NPI]) is defined as the exchange rate.37 in the N term, or identified policy devia- The real exchange rate and the country rat- tions from peer choices embedded in the ing are both regressed against net domestic CD term. Because they are linearly related saving where the latter equals the current through the A matrix, the cross-correlations account and is expressed as the difference of the gaps reflect the elasticities embedded between investment and gross domestic sav- in the model. ing (all as shares of GDP) with the coefficient of investment constrained to be the opposite Model specification of that of saving. Time clustering is used to All equations are systematically controlled for control for serial correlation of errors, and the country’s level of economic development, time fixed effects are used to control for as measured by GDP per capita. Additional worldwide shocks. 222 THE RISE OF THE SOUTH ANNEX TABLE 5A.2 Data description and sources Variable Description Source Domestic saving Domestic saving as a share of gross domestic product (GDP), expressed in UN data and WDI logs. Domestic saving is gross national disposable income (GDP plus net factor income and net unrequited transfers). Gross saving/GDP is from United Nations (UN); net factor payments/GDP and net unrequited transfers/GDP are from World Development Indicators (WDI). Investment Investment as a share of GDP, expressed in logs UN data Sovereign risk rating Country risk rating, expressed in logs Institutional Investor database GDP per capita growth Per capita income growth rate WDI Real exchange rate Ratio of purchasing power parity conversion factor to nominal exchange rate WDI with respect to the U.S. dollar, expressed in logs Current account Calculated as difference between investment and domestic saving UN data and WDI GDP per capita Per capita income, expressed in logs WDI Old-age dependency ratio Ratio of old people in the working population to the total work population WDI Population Total population WDI Population growth Rate of population growth WDI Fuel exports Oil exports as a share of GDP. Fuel exports are from WDI, with missing data WDI and Wealth of Nations filled through linear prediction using World Bank Wealth of Nations data. data Trade openness Ratio of imports plus exports to GDP, expressed in logs WDI Capital openness Capital openness index Chinn and Ito (2006) Net unrequited transfers Net unrequited transfers as a share of GDP WDI Share of foreign direct invest- Ratio of FDI to capital stock. FDI is from WDI; capital stock is from UNCTAD and Penn World ment (FDI) in total capital Penn World Table. Table 7.1 Policy-determined spending Public consumption as a share of GDP, expressed in logs WDI on nontradables Fiscal balance Fiscal balance as a share of GDP The Economist Countries Profiles Quality of institutional World Governance environment Simple average of corruption and rule of law indexes Indicators Inflation crisis Inflation crisis dummy Reinhart and Rogoff (2011) External debt crisis External debt crisis dummy Reinhart and Rogoff (2011) Risk appetite–safe haven Calculated as the VIX index times a safe haven dummy that is equal to 1 VIX and S&P 500 for the United States, Japan, and Switzerland and 0 for the rest of the world. VIX data are extrapolated backward using the S&P 500 index. Terms of trade changes Terms of trades expressed in logs. Data missing from WDI are completed WDI and International through smooth pasting with data from International Financial Statistics. Financial Statistics ASCENDING WITH THE SOUTH WINDS 223 ANNEX TABLE 5A.3 Data definitions and sources Variable Definition and proxy Source Domestic Gross domestic product (GDP) Real GDP (deviation from log-linear trend) National sources, International Financial Statistics (IFS), Organisation for Economic Co-operation and Development (OECD) Consumption Aggregate private plus public consumption National sources, IFS, OECD Investment Real aggregate investment National sources, IFS, OECD Current account Current account as a share of GDP National sources, IFS, OECD Inflation Consumer price index (deviation from log-linear trend) National sources, IFS, OECD Exchange rate Logarithm of real effective exchange rate National sources, IFS, OECD Global Global economic activity Proxied by real U.S. GDP (deviation from log-linear trend) OECD Global inflation Proxied by U.S. consumer price index (deviation from log- OECD linear trend) Global short-term interest rate Proxied by three-month U.S. Treasury rate St. Louis Federal Reserve Bank Global long-term interest rate Proxied by slope of U.S. yield curve (defined as spread St. Louis Federal Reserve Bank between 10-year and 3-month U.S. Treasury rates) Spreads on emerging economies’ Barclays’ corporate high-yield spread Bloomberg sovereign debt Real global commodity index (deviation from log-linear International Monetary Fund Commodity prices trend) ANNEX TABLE 5A.4 Signs and length restrictions on global and domestic shocks Sign restriction for domestic shock (for equation 5B.1) Current account Real effective Type of shock Output Consumption Investment Inflation as share of GDP exchange rate Supply +/3 +/3 ? –/1 ? +/1 Demand +/3 +/3 ? +/1 ? –/1a Monetary +/3 +/3 ? +/1 ? +/1 Sign restriction for global shock (for equation 5B.2) Commodity Short-term World output World inflation Term premium Credit spread prices interest rate Supply +/3 –/1 ? ? +/1 ? Demand +/1 +/1 ? ? +/1 +/1 b Monetary +/1 +/1 ? ? +/1 –/1 c Commodity +/3 –/1 ? ? –/1 ? Note: Plus (minus) signs indicate that a positive (negative) restriction is imposed on the sign of the response to the shock; ? means no restrictions are imposed. Figures represent the number of quarters the shock lasted, including the quarter in which it occurred. The first row of the table should be read as “A domestic supply (noncommodity) shock is assumed to increase global output and consumption on impact and for the next two quarters, to reduce inflation on impact, and to raise the real effective exchange rate on impact.” a. The rationale for the drop in the real exchange rate is the following: an increase in domestic demand leads to an increase in consumption of both tradable and nontradable goods. The increased demand puts pressure on the nominal prices of both tradable and nontradable goods. Yet the price of tradable goods is fixed in international markets. Therefore, the demand shock leads to an increase in the relative price of nontradable goods (that is, to a real appreciation). b. A global monetary shock reduces the short-term interest rate on impact. It is precisely this sign that allows one to disentangle demand from monetary shocks. c. Note how the impact on commodity prices allows one to disentangle a commodity price shock from a supply (noncommodity) global shock. 224   THE RISE OF tHE SOUtH annex 5b  The SVaR where country i 5 1, 2,…, I ; time t 5 methodology 1, 2,…, Ti (thus allowing for unbalanced panel); Aj is a 6 3 6 matrix on lagged val- Hevia and Servén (2014) assess the contri- ues for j 5 1, 2,…, p; X t is a k 3 1 vector bution of domestic and foreign shocks to with exogenous global variables; B h is a k macroeconomic dynamics across LAC coun- 3 k matrix capturing the impact of current tries. The contributions of these shocks are (h 5 0) and lagged (h . 0) exogenous vari- estimated using structural vector autore- ables on the variables of interest; and Pit is a gressions (SVARs) with exogenous variables. 6 3 1 vector of independent and identically SVARs impose restrictions on a reduced form distributed (i.i.d.) residuals with mean zero of vector autoregression (VAR) to identify, and covariance matrix V. Because of the rel- or recover, structural shocks or policy shifts atively short span of the time series, Aj and with clear economic meaning. Bh do not depend on the particular country i but capture instead generic properties for the Macro variables and shocks “representative” country. The analysis focuses on the impacts of shocks The vector of exogenous global variables X t on the evolution of six variables: GDP, aggre- follows an independent VAR given by the follow- gate consumption, aggregate investment, ing equation: inflation, the current account, and the real X t 5 b 1 a C j X t 2 j 1 vt r exchange rate. Many of these variables dis- (5B.2) play persistence and nonstationarity. The j51 analysis is conducted at levels, extracting where b is a k 3 1 vector of i.i.d. reduced- log-linear trends from GDP, investment, and form shocks with covariance matrix S orthogo- consumption. Data limitations and the risks nal to Pit for all i and t. of overparametrization limit the number of Identification endogenous domestic variables to six. Domestic and external shocks are not directly Shocks, identified based on sign restric- interpretable by tracing the impact of Pit and tions, include four external shocks (global vt on the macroeconomic variables of inter- supply, demand, commodity, and monetary est, as these shocks are contemporaneously shocks) and three domestic shocks (domestic correlated and do not have any structural or supply, demand, and monetary shocks). The economic interpretation. However, assum- external shocks are modeled as separate VARs ing those shocks are a linear combination of independent of the domestic variables.38 The structural shocks and imposing sign restric- assumption is that domestic variables do not tions on the impulse responses of the endog- affect the evolution of the global variables— enous variables at different horizons allow that is, that the developing countries consid- a correct identification of the econometric ered are small enough relative to the world specification (see table 5A.4).41 that their actions do not affect global quanti- ties and prices.39 External factors are proxied Empirical Implementation by their U.S. counterparts.40 Data availability, which varies by country, Setup dictates the sample period used to estimate the panel. The period ranges from first quar- The vector Y it , which collects the macro ter 1987 to fourth quarter 2012. Given the variables of interest for country i at time t, exogeneity of the global block of the model, evolves according to a panel VAR with com- equations (5B.1) and (5B.2) are estimated mon slope coefficients but individual country independently. Based on the Hannan-Quinn fixed effects ai given the following economet- criterion, three lags are selected for the ric model: endogenous variables of panel VAR (1), Yit 5 a i 1 a AjYit 2 j 1 a Bh X t 2 h 1 Pit (5B.1) p q only the contemporaneous response (no lag) j51 h50 is selected for the exogenous variables, and ASCENDING WITH THE SOUTH WINDS 225 two lags are selected for the global block in 5. In this case the government could improve the panel VAR (2). Once the parameters of the outcome only by easing the constraints faced reduced-form models (1) and (2) are esti- by private agents, typically through policies mated, structural shocks are identified by aimed at enhancing the enabling environment. 6. Saying that domestic saving is immaterial to imposing the sign restrictions, following growth because foreign and domestic saving Rubio-Ramírez, Waggoner, and Zha’s (2010) are perfect substitutes is tantamount to saying procedure.42 Because all structural shocks are that current and capital account imbalances mutually orthogonal, it is possible to decom- are immaterial to growth for the same reason. pose the variance of the forecast errors of 7. The first-best approach—to internalize such each variable into the portions attributable to externalities through Pigouvian taxes or sub- each of the identified structural shocks. It is sidies—may not be feasible. The state may also possible to recover the realized history of thus have to use second-best macro-oriented identified shocks, domestic or foreign. instruments (for example, fiscal policy, social security reforms) to directly raise the domestic saving rate. Public policy may itself contribute to the collective action failure underpinning Notes the suboptimality of private saving. For exam- 1. The Big Mac index is published by the Econo- ple, social safety nets can lead to undersaving mist as an informal way of measuring the rel- if private agents unduly rely on the state to ative purchasing power of two currencies. It is support them in old age or unforeseen contin- obtained by dividing the price of a Big Mac in gencies but the state fails to mobilize the fiscal one country (in its currency) by the price of a saving required to uphold its promises. Big Mac in another country (in its currency). 8. Thus, while consumption and investment This value is then compared with the actual affect the exchange rate in the same way, their exchange rate. impact on growth is clearly different. 2. Various issues of the semiannual reports 9. See, for example, Dornbusch (1980) and Vegh issued by the World Bank’s Chief Economist (2013). Office for LAC (http://go.worldbank.org 10. Berg and Miao (2010) find some evidence /WTVI133GT0) address the improvement in in support of tradable sector externalities. LAC’s macro-financial policy management, However, the identification of greater posi- starting with the April 2008 issue, entitled tive externalities in tradables and their role “Latin America’s New Immune System: How in the link between real exchange rates and Is it Coping with the Changing External growth remains elusive. See, for instance, Environment?” Giles and Williams (2000) and Harrison and 3. This debate has been particularly hot in LAC. Rodriguez-Clare (2009). On one side are economists who emphasize 11. In a somewhat similar vein, Levy Yeyati, the recessionary impact of fiscal frugality and Sturzenegger, and Gluzmann (2013) find that argue in favor of a possible virtuous growth countries that pursue exchange intervention circle in which domestic demand (particu- policies geared at keeping or enhancing exter- larly public investment) induces growth and, nal competitiveness grow faster, although the through it, raises saving. On the other side transmission channel between the exchange are economists who note that past domestic rate and growth is via higher investment demand–oriented policies in the region ended rather than increased exports. in a collapse of growth during the 1980s and 12. This neglect arguably reflects the discomfort to sluggish growth during the 1990s. They economists tend to have in viewing aggregate point out that the domestic demand–led high domestic saving as a policy-relevant variable. growth of the last decade (fueled as it was by Indeed, in an earlier contribution (Rodrik favorable terms of trade and international 2000), Rodrik himself concluded that “the liquidity conditions) is now losing steam. evidence provides no support for the view that 4. See, for example, De la Torre and others domestic saving is the binding constraint to (2010, 2012) for a discussion of the switch economic growth .… Policies geared towards of external net liability positions from debt to raising domestic saving do not deserve equity across LAC. priority.” 226 THE RISE OF THE SOUTH 13. In a short-run dynamic setting, the risk pre- consumption, and imports in periods in which mium would also be expected to affect the the terms of trade are rising (falling). There- exchange rate dynamically through the inter- fore, the national accounting data used in De est rate parity condition. This reverse effect is la Torre and Ize (2015) essentially eliminate ignored in the model presented here, because terms of trade–related valuation effects. in an equilibrium setting all transient dynam- 19. This procedure amounts to replacing, when ics are turned off. Thus the domestic interest calculating benchmarks, the observed values rate is simply given by the world interest rate of the macro variables on the right-hand side plus the country risk premium. of the regressions by their equilibrium values. 14. The ER channel dominates the IR channel 20. The real exchange rate is measured in units of when be 1 gegr . 0 (see annex 5A). foreign currency; hence a depreciation reduces 15. Another potentially important negative its value. growth externality of balance of payments cri- 21. Shifts along the regression line conform with ses is that by correlating exchange rate depre- the ER channel and fit the population as a ciations with downturns, they raise the risk whole. Instead, shifts away from the regres- premium on foreign-currency-denominated sion line conform with the IR channel and financial instruments in financially dollarized only fit the outlying LAC countries. economies. The higher premium effectively 22. The fact that the low saving ER countries lie “taxes” domestic saving, thereby inhibiting below the regression line in figure 5.11 sug- investment (particularly long-term investment gests that had it not been for their low sover- such as infrastructure) and promoting capital eign ratings (that is, the IR channel), their real flight (Ize 2013). exchange rates would have been much more 16. The dividing line between lower- and high- overvalued. er-income countries is per capita income of 23. Figure 5.16 shows a rise in LAC saving rates $5,000 a year. See annex table 5A.1 for the (rather than a decline) because it measures list of LAC1 and LAC2 countries. the impact of a positive (rather than negative) 17. Controls include demographics, natural global demand shock. resources, the terms of trade, remittances, 24. Remarkably, low and high savers in the region commercial and capital openness, fiscal policy, exhibited very distinct saving behaviors (see and several other factors (see a full description figure 5.13, panel a). Although saving rates in annex table 5A.2). Moreover, the impact of collapsed in the low-saving ER countries, they global shocks is largely neutralized through rose slightly and stabilized in the high-saving the use of time fixed effects. These controls ER countries, suggesting that the high-saving eliminate short-run fluctuations, thereby ER countries managed to avoid or mitigate allowing a tighter focus on the medium-term the (exchange rate–anchored) stabilization-in- equilibrium linkages and interactions between duced consumption booms of the low-saving these variables. At the same time, by making ER countries. As a result, the ER high savers countries comparable, they place countries (or avoided the large fluctuations in real exchange regions) on the same map in a way that is both rates and overvaluation tendency that affected revealing and meaningful. Benchmarks are the ER low savers (see figure 5.13, panel c). typical of what comparable countries do but 25. Most of the major LAC countries avoided an are not necessarily optimal. Moreover, causal- economic contraction in 2009, even as the ity cannot be ascertained when the underlying advanced economies of the world were caught elasticities are derived through simple (nonin- up in a great recession. strumented) ordinary least squares estimates. 26. Chile, Colombia, Mexico, Peru, Trinidad and 18. To avoid mixing price and quantity effects, Tobago, and Uruguay are currently in the elite the benchmarks and gaps, as well as regres- group of “investment-grade” countries. sion results from De la Torre and Ize (2015), 27. LAC exchange rates did depreciate between presented throughout this chapter are based 2012 and 2014. However, such depreci- on the real ratios of aggregate demand com- ations—which under current macro and ponents—consumption (hence saving), invest- financial conditions in LAC play a helpful ment, and exports and imports—to GDP, role in absorbing shocks and dampening the with each component deflated by its own amplitude of the cyclical downturn—can deflator. Nominal ratios to GDP underesti- be explained largely as short-term fluctu- mate (overestimate) the volume of investment, ations derived from changes in the world ASCENDING WITH THE SOUTH WINDS 227 environment (the U.S. Federal Reserve’s expected returns. Experience shows, however expected tightening of monetary policy) and (and the economic literature emphasizes), that the worsening in the region’s terms of trade. investment or production subsidies targeted They are unlikely to be a durable way to avoid to specific sectors can be distortionary and the spell of the ER channel. wasteful, especially if the main obstacles to 28. In effect, the tendency for LAC to generate investment lie elsewhere (in poorly defined current account deficits—as a result of its and deficiently enforced contract rights, for low-saving, domestic demand–based macro- instance, or in skills constraints). In contrast, economic structure—remained strong even in policy actions aimed at boosting aggregate the midst of the recent commodity supercycle. saving can be more neutral and efficient. The current account surpluses it generated Korinek and Servén (2010) present a similar were all too brief: already by the end of 2007 argument. LAC taken as a whole was back in current 35. Over time higher saving would facilitate coun- account deficit territory. tercyclical management by freeing monetary 29. However, the key correlations are generally policy. It would reduce both the fear of appre- more significant for middle-income countries ciation that constrains the central bank’s abil- than for low- or high-income countries. ity to raise the interest rate when the economy 30. Under the OLS estimates, for example, an overheats and the fear of depreciation (and the increase of 10 percentage points of GDP associated pass-through effects) that constrains in domestic saving (which would put LAC the central bank’s ability to lower the interest broadly on par with the middle-income coun- rate when the economy goes into a slump. tries of Southeast Asia) would raise annual per 36. Admittedly, the residuals could also reflect capita income growth through the ER channel unaccounted fundamentals rather than policy by no more than about 0.2 percentage points differences. of GDP. For the ES channel, from each dollar 37. Rodrik (2008) uses a similar index, albeit of additional investment, only about 9 cents from a different database (the Penn World would be self-financed by the induced increase Table instead of the World Bank’s World in saving caused by the higher growth. Development Indicators). 31. Because the system is overdetermined (it has 38. This modeling follows a large body of empiri- more equations than unknowns), the presence cal literature, including Raddatz (2007, 2008) of some of the instruments in more than one and Canova (2005). structural equation can be taken into account 39. Global variables taken into account include when needed. overall global economic activity, world short- 32. An increase of 10 percentage points of GDP and long-term interest rates, the cost to emerg- in domestic saving would boost annual per ing economies of issuing debt, the level of capita growth by up to 1.8 percentage points global commodity prices, and a variable of on account of the ER channel, 0.8 percentage inflation (see table 5A.3 for data definitions points on account of the IR channel, and one and sources). full percentage point on account of the ES 40. Tests of robustness give very similar results channel. when considering Group of Seven aggregates. 33. Although only a range (instead of a point 41. Fry and Pagan (2011) provide a critical estimate) can be inferred for a , an addi- review of structural VARs identified by sign tional dollar of investment would induce an restrictions. increase in saving of about 40 cents at the 42. This approach consists of using an arbitrarily middle of the range and 80 cents at the top of identified VAR—in this case a Cholesky the range. decomposition of the covariance matrix of 34. The underlying market failure calling for the reduced-form residuals—and randomly government intervention under the ER chan- rotating this identification matrix until the nel (the lack of internalization of the positive required sign restrictions are satisfied. The learning spillovers associated with investment random rotation is performed 1,000 times by in and producing tradables) could in principle postmultiplying the Cholesky identification be addressed through investment subsidies or matrix by an orthonormal matrix obtained by other promotion policies focused on specific applying the QR decomposition to a random exporting sectors. In an ideal world, such pol- 6 ϫ 6 matrix whose elements are drawn from icies would boost domestic saving by raising a standard normal. Medians are reported. 228 THE RISE OF THE SOUTH References De la Torre, A., T. Didier, C. Hevia, S. Pienkna- gura, and S. Schmukler. 2012. Latin America Aghion, P., D. A. Comin, P. Howitt, and I. Tecu. Copes with Volatility: The Dark Side of Glo- 2009. “When Does Domestic Saving Matter for balization . World Bank LAC Semi-Annual Economic Growth?” Harvard Business School Report, April, Washington, DC. Working Paper 09-080, Cambridge, MA. De la Torre, A., and A. Ize. 2015. “Does Latin Aizenman, J., B. Pinto, and A. Radziwill. 2004. America Have a Saving Problem?” World “Sources for Financing Domestic Capital: Is Bank, background paper for this report. Foreign Saving a Viable Option for Emerging Dornbusch, R. 1980. Open Economy Macroeco- Countries?” NBER Working Paper 10624, nomics. New York: Basic Books. National Bureau of Economic Research, Cam- Eaton, J., and M. Gersovitz. 1981. “Debt with bridge, MA. Potential Repudiation: Theoretical and Empir- Berg, A., and Y. Miao. 2010. “The Real Exchange ical Analysis.” Review of Economic Studies Rate and Growth Revisited: The Washington 48: 289–309. Consensus Strikes Back.” IMF Working Paper Eichengreen, B. 2008, “The Real Exchange Rate 10/58, International Monetary Fund, Wash- and Economic Growth.” Working Paper 4, ington, DC. Commission on Growth and Development, Bolt, J., and J. L. van Zanden. 2013. The First World Bank, Washington, DC. Update of the Maddison Project: Re-Estimating Fazzari, S., G. Hubbard, and B. Petersen. 1988. Growth before 1820. Maddison Project “Financing Constraints and Corporate Invest- Working Paper 4. The Maddison Project. ment.” Brooking Papers on Economic Activity http://www.ggdc.net/maddison /maddison I: 141–95. -project/publications/wp4.pdf. Feyrer J., and J. Shambaugh. 2009. “Global Sav- Campbell, J., and J. Cochrane. 1999. “Force of ing and Global Investment: The Transmission Habit: A Consumption-Based Explanation of of Identified Fiscal Shocks.” NBER Working Aggregate Stock Market Behavior.” Journal of Paper 15113, National Bureau of Economic Political Economy 107 (2): 207–51. Research, Cambridge, MA. Canova, F. 2005. “The Transmission of U.S. Fry, R., and A. Pagan. 2011. “Sign Restrictions in Shocks to Latin America.” Journal of Applied Structural Vector Autoregressions: A Critical Econometrics 20: 229–51. Review.” Journal of Economic Literature 49 Carroll, C., and D. Weil. 1993. “Saving and (4): 938–60. Growth: A Reinterpretation.” NBER Work- Giles, J., and C. Williams. 2000. “Export-Led ing Paper 4470, National Bureau of Economic Growth: A Survey of the Empirical Literature Research, Cambridge, MA. and Some Non-Causality Results.” Depart- Caselli, F., and J. Feyrer. 2006. “The Marginal ment of Economics, University of Victoria, Product of Capital.” Quarterly Journal of Greater Victoria, British Columbia, Canada. Economics 122 (2): 535–68. Guariglia, A., X. Liu, and L. Song. 2011. “Inter- Chinn, Menzie D., and H. Ito. 2006. “What Mat- nal Finance and Growth: Microeconometric ters for Financial Development? Capital Con- Evidence on Chinese Firms.” University of trols, Institutions, and Interactions.” Journal Nottingham Discussion Paper 09/11, Notting- of Development Economics 81(1): 163–92. ham, United Kingdom. Corsetti, G., K. Kuester, A. Meier, and G. Muel- Haddad, M., and C. Pancaro. 2010. “Can Real ler. 2012. “Sovereign Risk, Fiscal Policy and Exchange Rate Undervaluation Boost Exports Macroeconomic Stability.” IMF Working and Growth in Developing Economies? Yes, Paper 12/33, International Monetary Fund, but Not for Long.” Economic Premise 20, Washington, DC. World Bank, Washington, DC. De la Torre, A., C. Calderon, T. Didier, E. Levy Harrison, A., and A. Rodriguez-Clare. 2009. Yeyati, and S. Schmukler. 2010. Globalized, “Trade, Foreign Investment and Industrial Resilient, Dynamic: The New Face of Latin Policy.” MPRA Paper 15561, Munich Personal America and the Caribbean. World Bank RePEc Archive. LAC Semi-Annual Report, October, Wash- Hevia, C., and L. Servén. 2014. “The Macroeco- ington, DC. nomic Impact of Global and Domestic Shocks ASCENDING WITH THE SOUTH WINDS 229 on Latin America.” Background paper for this Raddatz, C. 2007. “Are External Shocks Respon- report. sible for the Instability of Output in Low- International Financial Statistics (database). Inter- Income Countries?” Journal of Development national Monetary Fund, Washington, DC. Economics 84 (1): 155–87. http://www.imf.org/external/data.htm. ———. 2008. “External Shocks and Macroeco- Itskhoki O. and B. Moll, 2014. “Optimal Devel- nomic Volatility in Latin America.” World opment Policies with Financial Frictions.” Bank, Development Economics Vice Presi- NBER Working Paper 19994, National Bureau dency (DEC), Washington, DC. of Economic Research, Cambridge, MA. Reinhart, C. M., and K. S. Rogoff. 2011. “From Ize, A. 2013. “Financial Dollarization,” in The Financial Crash to Debt Crisis.” American Evidence and Impact of Financial Globaliza- Economic Review 101 (5): 1676–706. tion, J. Caprio ed., London: Elsevier Publishing. ———. 2013. “Financial and Sovereign Debt Jeanne, O., and A. Korinek. 2010. “Excessive Vol- Crises: Some Lessons Learned and Those For- atility in Capital Flows: A Pigouvian Taxation gotten.” IMF Working Paper 13/266, Interna- Approach.” American Economic Review 100 tional Monetary Fund, Washington, DC. (2): 403–07. Ricci, L. A., G. M. Milesi-Ferretti, and J. Lee. Kaldor, N. 1958. “Alternative Theories of Distri- 2008. “Real Exchange Rates and Funda- bution.” Review of Economic Studies 23 (2): mentals: A Cross-Country Perspective.” IMF 83–100. Working Paper WP/08/13, International Mon- Korinek, A., and L. Servén. 2010. “Undervalu- etary Fund, Washington, DC. ation through Foreign Reserve Accumulation.” Rodrik, D. 2000. “Saving Transitions.” World World Bank Policy Research Working Paper Bank Economic Review 14 (3): 481–507. 5250, Washington, DC. ———. 2008. “The Real Exchange Rate and Eco- Lane, P. and G.M. Milesi-Ferretti. 2007. “The nomic Growth.” Brookings Papers on Eco- External Wealth of Nations Mark II: Revised nomic Activity 39 (2): 365–439. and Extended Estimates of Foreign Assets and Romer, P. 1986. “Increasing Returns and Long– Liabilities, 1970–2004.” Journal of Interna- Run Growth.” Journal of Political Economy tional Economics 73: 223–50. 94 (5): 1002–37. Levy Yeyati, E., F. Sturzenegger, and P. A. Gluz- Rowthorn, R. 1982. “Demand, Real Wages and mann. 2013. “Fear of Appreciation.” Journal Economic Growth.” Studi Economici 18: of Development Economics 101: 233–47. 3–54. Lewis, W. A. 1954. “Economic Development with Rubio-Ramírez, J. F., D. F. Waggoner, and T. Zha. Unlimited Supplies of Labor.” Manchester 2010. “Structural Vector Autoregressions: School of Economic and Social Studies 22 (2): Theory of Identification and Algorithms for 139–91. Inference.” Review of Economic Studies 77 Loayza, N., K. Schmidt-Hebbel, and L. Servén. (2): 665–96. 2000. “What Drives Saving across the World?” World Development Indicators. World Bank, Review of Economics and Statistics 82 (2): Washington, DC. http://data.worldbank.org 165–81. /data-catalog/world-development-indicators. Lucas, R. 1988. “On the Mechanics of Economic World Economic Outlook (database). Interna- Development.” Journal of Monetary Econom- tional Monetary Fund, Washington, DC. ics 22 (1): 3–42. http://www.imf.org/external/data.htm. Modigliani, F. 1986. “Life Cycle, Individual Vegh, C. 2013. Open Economy Macroeconom- Thrift and the Wealth of Nations.” American ics in Developing Countries. Cambridge, MA: Economic Review 76 (3): 297–313. MIT Press. Penn World Table (database). http://www.rug.nl Yang, D. T., J. Zhang, and S. Zhou. 2011. “Why /research/ggdc/data/pwt/. Are Saving Rates So High in China?” NBER Prasad, E., R. Rajan, and A. Subramanian. 2007. Working Paper 16771, National Bureau of “Foreign Capital and Economic Growth.” Economic Research, Cambridge, MA. Department of Economics, Cornell University, Ithaca, NY. Environmental Benefits Statement The World Bank Group is committed to reducing its environmental footprint. In sup- port of this commitment, the Publishing and Knowledge Division leverages 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 ship- ping distances decreased, resulting in reduced paper consumption, chemical use, greenhouse gas emissions, and waste. The Publishing and Knowledge Division follows the recommended standards for paper use set by the Green Press Initiative. The majority of our books are printed on 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 Chlo- rine Free (TCF), Processed Chlorine Free (PCF), or Enhanced Elemental Chlorine Free (EECF) processes. More information about the Bank’s environmental philosophy can be found at http://crinfo.worldbank.org/wbcrinfo/node/4.