Market Liberalism, Growth, and Economic Development in Latin America

Routledge studies in development economics

1 Economic Development in the Middle East Rodney Wilson 2 Monetary and Financial Policies in Developing Countries Growth and stabilization Akhtar Hossain and Anis Chowdhury 3 New Directions in Development Economics Growth, environmental concerns and government in the 1990s Edited by Mats Lundahl and Benno J. Ndulu 4 Financial Liberalization and Investment Kanhaya L. Gupta and Robert Lensink 5 Liberalization in the Developing World Institutional and economic changes in Latin America, Africa and Asia Edited by Alex E. Fernández Jilberto and André Mommen

6 Financial Development and Economic Growth Theory and experiences from developing countries Edited by Niels Hermes and Robert Lensink 7 The South African Economy Macroeconomic prospects for the medium term Finn Tarp and Peter Brixen 8 Public Sector Pay and Adjustment Lessons from five countries Edited by Christopher Colclough 9 Europe and Economic Reform in Africa Structural adjustment and economic diplomacy Obed O. Mailafia 10 Post-apartheid Southern Africa Economic challenges and policies for the future Edited by Lennart Petersson 11 Financial Integration and Development Liberalization and reform in sub-Saharan Africa Ernest Aryeetey and Machiko Nissanke

12 Regionalization and Globalization in the Modern World Economy Perspectives on the Third World and transitional economies Edited by Alex E. Fernández Jilberto and André Mommen 13 The African Economy Policy, institutions and the future Steve Kayizzi-Mugerwa 14 Recovery from Armed Conflict in Developing Countries Edited by Geoff Harris 15 Small Enterprises and Economic Development The dynamics of micro and small enterprises Carl Liedholm and Donald C. Mead 16 The World Bank New agendas in a changing world Michelle Miller-Adams 17 Development Policy in the Twenty-First Century Beyond the post-Washington consensus Edited by Ben Fine, Costas Lapavitsas and Jonathan Pincus 18 State-Owned Enterprises in the Middle East and North Africa Privatization, performance and reform Edited by Merih Celasun 19 Finance and Competitiveness in Developing Countries Edited by José María Fanelli and Rohinton Medhora

20 Contemporary Issues in Development Economics Edited by B.N. Ghosh 21 Mexico Beyond NAFTA Edited by Martín Puchet Anyul and Lionello F. Punzo 22 Economies in Transition A guide to China, Cuba, Mongolia, North Korea and Vietnam at the turn of the twenty­first century Ian Jeffries 23 Population, Economic Growth and Agriculture in Less Developed Countries Nadia Cuffaro 24 From Crisis to Growth in Africa? Edited by Mats Lundal 25 The Macroeconomics of Monetary Union An analysis of the CFA franc zone David Fielding 26 Endogenous Development Networking, innovation, institutions and cities Antonio Vasquez-Barquero 27 Labour Relations in Development Edited by Alex E. Fernández Jilberto and Marieke Riethof 28 Globalization, Marginalization and Development Edited by S. Mansoob Murshed

29 Programme Aid and Development Beyond conditionality Howard White and Geske Dijkstra 30 Competitiveness Strategy in Developing Countries A manual for policy analysis Edited by Ganeshan Wignaraja 31 The African Manufacturing Firm An analysis based on firm surveys in sub-Saharan Africa Dipak Mazumdar and Ata Mazaheri 32 Trade Policy, Growth and Poverty in Asian Developing Countries Edited by Kishor Sharma 33 International Competitiveness, Investment and Finance A case study of India Edited by A. Ganesh Kumar, Kunal Sen and Rajendra R. Vaidya 34 The Pattern of Aid Giving The impact of good governance on development assistance Eric Neumayer 35 New International Poverty Reduction Strategies Edited by Jean-Pierre Cling, Mireille Razafindrakoto and François Roubaud 36 Targeting Development Critical perspectives on the millennium development goals Edited by Richard Black and Howard White

37 Essays on Balance of Payments Constrained Growth Theory and evidence Edited by J.S.L. McCombie and A.P. Thirlwall 38 The Private Sector after Communism New entrepreneurial firms in transition economies Jan Winiecki, Vladimir Benacek and Mihaly Laki 39 Information Technology and Development A new paradigm for delivering the internet to rural areas in developing countries Jeffrey James 40 The Economics of Palestine Economic policy and institutional reform for a viable Palestine state Edited by David Cobham and Nu’man Kanafani 41 Development Dilemmas The methods and political ethics of growth policy Melvin Ayogu and Don Ross 42 Rural Livelihoods and Poverty Reduction Policies Edited by Frank Ellis and H. Ade Freeman 43 Beyond Market-Driven Development Drawing on the experience of Asia and Latin America Edited by Makoto Noguchi and Costas Lapavitsas

44 The Political Economy of Reform Failure Edited by Mats Lundahl and Michael L. Wyzan

52 The New Famines Why famines exist in an era of globalization Edited by Stephen Devereux

45 Overcoming Inequality in Latin America Issues and challenges for the twenty­first century Edited by Ricardo Gottschalk and Patricia Justino

53 Development Ethics at work Explorations – 1960–2002 Denis Goulet

46 Trade, Growth and Inequality in the Era of Globalization Edited by Kishor Sharma and Oliver Morrissey 47 Microfinance Perils and prospects Edited by Jude L. Fernando 48 The IMF, World Bank and Policy Reform Edited by Alberto Paloni and Maurizio Zanardi 49 Managing Development Globalization, economic restructuring and social policy Edited by Junji Nakagawa 50 Who Gains from Free Trade? Export­led growth, inequality and poverty in Latin America Edited by Rob Vos, Enrique Ganuza, Samuel Morley and Sherman Robinson 51 Evolution of Markets and Institutions A study of an emerging economy Murali Patibandla

54 Law Reform in Developing and Transitional States Edited by Tim Lindsey 55 The Asymmetries of Globalization Edited by Pan A. Yotopoulos and Donato Romano 56 Ideas, Policies and Economic Development in the Americas Edited by Esteban Pérez-Caldentey and Matias Vernengo 57 European Union Trade Politics and Development Everything but arms unravelled Edited by Gerrit Faber and Jan Orbie 58 Membership Based Organizations of the Poor Edited by Martha Chen, Renana Jhabvala, Ravi Kanbur and Carol Richards 59 The Politics of Aid Selectivity Good governance criteria in World Bank, U.S. and Dutch development assistance Wil Hout

60 Economic Development, Education and Transnational Corporations Mark Hanson 61 Achieving Economic Development in the Era of Globalization Shalendra Sharma 62 Sustainable Development and Free Trade Shawkat Alam 63 The Impact of International Debt Relief Geske Dijkstra 64 Europe’s Troubled Region Economic development, institutional reform and social welfare in the Western Balkans William Bartlett 65 Work, Female Empowerment and Economic Development Sara Horrell, Hazel Johnson and Paul Mosley 66 The Chronically Poor in Rural Bangladesh Livelihood constraints and capabilities Pk. Md. Motiur Rahman, Noriatsu Matsui and Yukio Ikemoto 67 Public–Private Partnerships in Health Care in India Lessons for developing countries A. Venkat Raman and James Warner Björkman

68 Rural Poverty and Income Dynamics in Asia and Africa Edited by Keijiro Otsuka, Jonna P. Estudillo and Yasuyuki Sawada 69 Microfinance A reader David Hulme and Thankom Arun 70 Aid and International NGOs Dirk-Jan Koch 71 Development Macroeconomics Essays in memory of Anita Ghatak Edited by Subrata Ghatak and Paul Levine 72 Taxation in a Low Income Economy The case of Mozambique Channing Arndt and Finn Tarp 73 Labour Markets and Economic Development Edited by Ravi Kanbur and Jan Svejnar 74 Economic Transitions to Neoliberalism in Middle-Income Countries Policy dilemmas, crises, mass resistance Edited by Alfedo Saad-Filho and Galip L. Yalman 75 Latecomer Development Innovation and knowledge for economic growth Banji Oyelaran-Oyeyinka and Padmashree Gehl Sampath

76 Trade Relations between the EU and Africa Development, challenges and options beyond the Cotonou Agreement Edited by Yenkong Ngangjoh-Hodu and Francis A.S.T. Matambalya 77 The Comparative Political Economy of Development Africa and South Asia Edited by Barbara Harriss-White and Judith Heyer 78 Credit Cooperatives in India Past, present and future Biswa Swarup Misra 79 Development Economics in Action (2nd edition) A study of economic policies in Ghana Tony Killick 80 The Multinational Enterprise in Developing Countries Local versus global logic Edited by Rick Molz, Cătălin Ratiu and Ali Taleb 81 Monetary and Financial Integration in West Africa Temitope W. Oshikoya 82 Reform and Development in China What can China offer the developing world Edited by Ho-Mou Wu and Yang L. Yao

83 Towards New Developmentalism Market as means rather than master Edited by Shahrukh Rafi Khan and Jens Christiansen 84 Culture, Institutions, and Development New insights into an old debate Edited by Jean-Philippe Platteau and Robert Peccoud 85 Assessing Prospective Trade Policy Methods applied to EU–ACP economic partnership agreements Edited by Oliver Morrissey 86 Social Protection for Africa’s Children Edited by Sudhanshu Handa, Stephen Devereux and Douglas Webb 87 Energy, Bio Fuels and Development Comparing Brazil and the United States Edited by Edmund Amann, Werner Baer and Don Coes 88 Value Chains, Social Inclusion and Economic Development Contrasting theories and realities Edited by A.H.J. (Bert) Helmsing and Sietze Vellema 89 Market Liberalism, Growth, and Economic Development in Latin America Edited by Gerardo Angeles-Castro, Ignacio Perrotini-Hernández, and Humberto Ríos-Bolívar

Market Liberalism, Growth, and Economic Development in Latin America

Edited by Gerardo Angeles-Castro, Ignacio Perrotini-Hernández, and Humberto Ríos-Bolívar

First published 2011 by Routledge 2 Park Square, Milton Park, Abingdon, Oxon OX14 4RN Simultaneously published in the USA and Canada by Routledge 711 Third Avenue, New York, NY 10017 This edition published in the Taylor & Francis e-Library, 2011.

To purchase your own copy of this or any of Taylor & Francis or Routledge’s collection of thousands of eBooks please go to www.eBookstore.tandf.co.uk. Routledge is an imprint of the Taylor & Francis Group, an informa business © 2011 Selection and editorial matter, Gerardo Angeles­Castro, Ignacio Perrotini­Hernández and Humberto Ríos­Bolívar; individual chapters, the contributors The right of Gerardo Angeles­Castro, Ignacio Perrotini­Hernández and Humberto Ríos­Bolívar to be identified as authors of the editorial material and of the authors for their individual chapters has been asserted by them in accordance with the Copyright, Designs and Patent Act 1988. All rights reserved. No part of this book may be reprinted or reproduced or utilized in any form or by any electronic, mechanical, or other means, now known or hereafter invented, including photocopying and recording, or in any information storage or retrieval system, without permission in writing from the publishers. Trademark notice: Product or corporate names may be trademarks or registered trademarks, and are used only for identification and explanation without intent to infringe. British Library Cataloguing in Publication Data A catalogue record for this book is available from the British Library Library of Congress Cataloging in Publication Data Market liberalism, growth, and economic development in Latin America / edited by Gerardo Angeles­Castro, Ignacio Perrotini­Hernández and Humberto Ríos-Bolivar. p. cm. Includes bibliographical references and index. 1. Latin America–Economic policy. 2. Latin America–Commercial policy. 3. Free trade–Latin America. 4. Economic development­­Latin America. I. Angeles­Castro, Gerardo. II. Perrotini­Hernández, Ignacio. III. Ríos­Bolivar, Humberto. HC125.M325185 2010 330.98–dc22 2010047759 ISBN 0-203-81612-9 Master e-book ISBN

ISBN: 978­0­415­57374­0 (hbk) ISBN: 978­0­203­81612­7 (ebk)

Contents

List of figures List of tables List of contributors

xiv xvii xx

Foreword

xxii

JAIME ROS

Acknowledgements List of acronyms Editors’ introduction

xxv xxvi 1

GERARDO ANGELES­CASTRO, IGNACIO PERROTINI­HERNáNDEz AND HUMBERTO RíOS-BOLíVAR

PART I

Trade liberalization, development and regional integration 1 Has trade liberalisation in poor countries delivered the promises expected?

5

7

PENéLOPE PACHECO­LóPEz AND A.P. THIRLWALL

2 Beyond the Washington Consensus: the quest for an alternative development paradigm for Latin America IGNACIO PERROTINI­HERNáNDEz, JUAN ALBERTO VázqUEz­MUñOz AND BLANCA L. AVENDAñO­VARGAS

26

xii

Contents

3 Foreign trade and per capita income: new evidence for Latin America and the Caribbean

59

HUMBERTO RíOS-BOLíVAR AND OMAR NEME-CASTILLO

4 Regional integration and its effects on inward FDI in developing countries: a comparison between North–South (Mexico) and South–South (Brazil) integration

81

THOMAS GODA

5 Trade blocs as determinants of trade flows in South American countries: an augmented gravity approach

110

CLEMENTE HERNáNDEz­RODRíGUEz

PART II

Trade reforms and development experience: case studies in Latin America

131

6 Downhill or the long agony of Argentinian development

133

ALCINO FERREIRA CâMARA-NETO AND MATíAS VERNENGO

7 The determinants of FDI in Chile: a gravity model approach

149

MATTEO GRAzzI

8 Assessment of the distributive impact of trade reforms in Uruguay

168

FERNANDO BORRAz, DANIEL FERRéS AND MáXIMO ROSSI

PART III

Economic liberalization, development and growth in Mexico 9 Economic liberalisation and income distribution: theory and evidence in Mexico

193

195

GERARDO ANGELES-CASTRO

10 How risk factors affect growth in Mexico: a free-market liberalism approach FRANCISCO VENEGAS­MARTíNEz

220

Contents 11 Anti- inflationary policy and financial fragility: a microeconomic analysis case study of Mexico, 1990–2004

xiii 233

IGNACIO PERROTINI­HERNáNDEz, BLANCA L. AVENDAñO­VARGAS AND JUAN ALBERTO VázqUEz­MUñOz

12 Technological innovation and sectoral productivity in the Mexican economy: regional evidence

250

JOSé CARLOS TREJO­GARCíA, HUMBERTO RíOS­BOLíVAR AND ANA LILIA VALDERRAMA­SANTIBáñEz

13 The robustness of Okun’s law – evidence from Mexico: a quarterly validation, 1985.1–2006.4

264

EDUARDO LORíA AND LEOBARDO DE JESúS

Index

277

Figures

1.1 2.1a 2.1b 2.1c 2.1d 2.1e 2.2a 2.2b 2.2c 2.2d 2.2e 2.3a 2.3b 2.3c 2.3d 2.3e 2.4a 2.4b 2.4c 2.4d 2.4e 2.5a 2.5b 2.5c 2.5d 2.5e 2.6a 2.6b 2.6c 2.6d 2.6e 4.1

The trade-off between growth and the balance of payments Argentina’s GDP (annual growth rate), 1961–2009 Brazil’s GDP (annual growth rate), 1961–2009 Chile’s GDP (annual growth rate), 1961–2009 Mexico’s GDP (annual growth rate), 1961–2009 China’s GDP (annual growth rate), 1961–2009 Argentina’s annual rate of inflation, 1961–2009 Brazil’s annual rate of inflation, 1970–2009 Chile’s annual rate of inflation, 1961–2009 Mexico’s annual rate of inflation, 1961–2009 China’s annual rate of inflation, 1987–2009 Argentina: trade balance–GDP ratio, 1960–2009 Brazil: trade balance–GDP ratio, 1960–2009 Chile: trade balance–GDP ratio, 1960–2009 Mexico: trade balance–GDP ratio, 1960–2009 China: trade balance–GDP ratio, 1978–2008 Argentina’s development gap, 1960–2008 Brazil’s development gap, 1960–2008 Chile’s development gap, 1960–2008 Mexico’s development gap, 1960–2008 China’s development gap, 1960–2008 Argentina’s gross investment and gross savings, 1960–2009 Brazil’s gross investment and gross savings, 1960–2009 Chile’s gross investment and gross savings, 1960–2009 Mexico’s gross investment and gross savings, 1960–2009 China’s gross investment and gross savings, 1960–2008 Argentina’s income elasticity of imports, 1961–2007 Brazil’s income elasticity of imports, 1961–2007 Chile’s income elasticity of imports, 1961–2007 Mexico’s income elasticity of imports, 1961–2007 China’s income elasticity of imports, 1979–2007 Foreign direct investment, net inflows, comparison with ROW

18 32 32 33 33 34 35 35 36 36 37 39 39 40 40 41 42 42 43 43 44 45 46 46 47 47 48 48 49 49 49 90

Figures 4.2 4.3 4.4 4.5 4.6 4.7 4.8 4.9 4.10 4.11 4.12 4.13 4.14 5.1 5.2 5.3 6.1 6.2 6.3 6.4 6.5 6.6 7.1 7.2 8.1 8.2 8.3 8.4 8.5 10.1 10.2 10.3 11.1 11.2 11.3

xv

Foreign direct investment, net inflows, comparison with other middle-income countries 91 Foreign direct investment as percentage of GDP 92 Mexico inward FDI stock, 1990–2000 95 Mexico inward FDI flows, 2001–2006 96 Brazil inward FDI stock, 1990–2000 97 Brazil inward FDI flows, 2001–2006 97 Mexico inward FDI stock by regions, 1990–2000 98 Mexico inward FDI flows by region, 2001–2006 98 Brazil inward FDI stock by regions, 1990–2000 99 Brazil inward FDI inflows by regions, 2001–2006 100 Dow Jones Index composite average and FDI inflows in Brazil and Mexico, 1990–2006 103 Brazil FDI inflows and exchange rate 104 Mexico FDI inflows and exchange rate 104 Graphic evolution of exports by country member of Mercosur 114–15 Graphic evolution of exports by country member of the CAN 117–19 Graphic evolution of gravity equation fixed effects for Mercosur and CAN 125 Income per capita 133 Business cycle 136 Output decomposition 139 Real exchange rate 140 Gini coefficients 144 Wage share 145 FDI flows into Chile 1985–2006 151 FDI inflows into Chile by geographical origin 1974–2006 154 Consumption effect: compensating variation as per cent of income by income distribution 177 Compensating variation as per cent of income by income distribution: traded good 180 Compensating variation as per cent of income by income distribution: non-tradable goods effect 182 Compensating variation as per cent of income by income distribution: labour income effect 183 Compensating variation as per cent of income by income distribution: total effect 183 Optimal w* as a function m and h 225 229 Consumption as a function of x0 and t0 Observed growth rate from 1930 to 2002 230 Growth, returns and interest rate: aggregate levels, 1990:02–2004:04 238 Growth and firms’ indebtness: aggregate levels, 1990:02–2004:04 240 Net wealth, growth and debt: aggregate levels, 1990:02–2004:04 240

xvi

Figures

11.4 11.5 11.6 11.7 13.1 13.2 13.a.1 13.a.2 13.a.3 13.a.4 13.a.5 13.a.6 13.a.7 13.a.8 13.a.9

Composition of firms according to financial structure, 1990:02–2004:04 Composition of firms’ assets according to financial structure, 1990:02–2004:04 Probability of hedge, speculative and Ponzi finance among firms, 1990:02–1994:04 Probability of hedge, speculative and Ponzi finance among firms, 1995:01–2004:04 Mexico: macroeconomic rate of unemployment, 1970–2004 GDP, unemployment, output gap and employment rate, 1985.1–2006.4 Model 1 first differences Model 1: first differences Diagnostic tests: correlogram, density, qqplot, cusum residual Model 2: output gap Model 2: output gap Diagnostic tests: correlogram, density, qqplot, cusum residual Model 3: fitted trend and elasticity Model 3: fitted trend and elasticity Diagnostic tests: correlogram, density, qqplot, cusum residual

241 242 247 248 266 267 270 271 271 272 272 273 273 273 274

Tables

1.1 3.1 3.2 3.3 4.1 4.2 4.3 5.1 5.2 5.3 5.4 5.5 6.1 6.2 6.3 6.4 6.5 7.1 7.2 7.3 7.4 7.5 7.6 8.1 8.2 8.3 8.4 8.5 8.6

A comparison of Gini ratios Latin America and the Caribbean countries considered in the sample Panel data regressions of per capita income Panel data regressions of per capita income Inward FDI stock as a percentage of GDP, 1990–2006 Transaction values of cross­border M&As of privatised firms Share of cross­border M&A sales in total FDI inflows in Mexico and Brazil Mercosur and the CAN Gravity equation for the panels of Mercosur and CAN: random effects Gravity equation for the panels of Mercosur and CAN: fixed effects Exogeneity Hausman Test for distance in the gravity equation Distance as the only regressor in the gravity equation Growth and productivity Fiscal policy Investment composition Debt sustainability indicators Manufacturing exports FDI inflows into Chile by sector, 1974–2006 FDI inflows into Chile by geographical origin, 1974–2006 Variable description Summary statistics Random effects estimation of the baseline equation Random effects estimation of the augmented equation Trade openness coefficient Intra and extra Mercosur trade flows: simple average Tariff structure: Uruguay Price changes from Mercosur Unit­root test: tradable and non­tradable prices Prices cointegration

14 68 71 76–7 94 101 102 112 123 124 126 127 135 140 142 143 143 153 155 160 161 162 164 170 170 178 179 181 182

xviii Tables 8.7a 8.7b 8.8 8.9 8.10 8.11

Poverty: before and after trade reform Poverty: before and after trade reform Change in income inequality Unit-root test: ADF Engle-Granger: cointegration test Poverty and inequality effects of liberalisation in the beef international trade 8.12 Change in the probability of being employed after a Free Trade Agreement with USA 9.1 Average real hourly income per level of education 9.2 Changes in average hourly income and share of educational levels 9.3 Average real hourly income and educational attainment per sector 9.4 Income bill share per sector and level of education 9.5 Performance of sectors (labour income) 9.6 Returns to education (labour income) 9.7 Average real monthly income, Gini, and composition of income receivers 9.8 Decomposition of household Gini by income source 9.9 The effect of market openness on income (traded sector) 10.1 Optimal consumption shares, parameters, and estimates 11.1 Typology of the growth rate of firms 11.2 Average values for the sample of firms as a whole 11.3 Average percentage composition of the companies according to their financial structure 11.4 Dependent variable: bit(t subscript directly beneath superscript i) 11.5 Dependent variable: wit(t subscript directly beneath superscript i) 11.6 Dependent variable: Fit(t subscript directly beneath superscript i) 11.7 Average of the estímate probability of financial structure of firms 12.1 Variables 12.2 Estimate for municipalities with high density of population (urban) and weighted by the product 12.3 Estimate for municipalities with low density of population (rural) and weighted by the product 12.4 Estimate for municipalities with high density of population (urban) and weighted by the number of registered units 12.5 Estimate for municipalities with low density of population (urban) and weighted by the number of registered units 13.1 Okun models 13.2 Mexico: Okun estimations

184 185 185 186 187 188 188 201 202 203 204 206–7 209 211 213 214 231 236 239 242 243 245 246 248 255 258 259 260 260 265 268

Tables 13.a.1 Basic statistics and unit roots, 1985.1–2006.4 13.a.2 Granger Causality Test, 1985.1–2006.4 for an unrestricted VAR(5) 13.a.3 Mexico: Okun’s law, 1985.1–2006.4 13.a.4 Model 2 13.a.5 Model 3

xix 269 269 270 271 273

Contributors

Gerardo Angeles-Castro is Research Economist and Head of Research and Graduate Studies at the School of Economics in the Instituto Politécnico Nacional, Mexico. Blanca L. Avendaño-Vargas is Research Economist at the Faculty of Economics in the Benemérita Universidad Autónoma de Puebla, Mexico. Fernando Borraz is Senior Researcher at the Central Bank of Uruguay and also Head of the Department of Economics and full-time Professor at Montevideo University, Uruguay. Leobardo de Jesús is Research Economist at the Universidad Nacional Autónoma de México (UNAM), Mexico. Alcino Ferreira Câmara-Neto is Dean of the Centre for Economic and Legal Sciences (CCJE) and Chairman of the Development Council in the Federal University of Rio de Janeiro (UFRJ); he is also Chief Editor of the Journal of Applied Social Science of the CCJE­UFRJ, Brazil. Daniel Ferrés is Assistant Professor at Montevideo University, Uruguay. Thomas Goda is PhD student in Economics at the London Metropolitan University, UK. Matteo Grazzi is Economist at the Science and Technology Division of the Inter­American Development Bank Washington, DC, USA. He is also an external researcher at the Centre for Research on Latin American Studies and Transition Economies Studies (ISLA) of the Bocconi University, Milan, Italy. Clemente Hernández-Rodríguez is Coordinator of Research and Consultancy Centre in Administration and Management, and Associate Director of Research in the Asia Pacific Institute at Tecnológico de Monterrey, Campus Guadalajara, Mexico. Eduardo Loría is Research Economist at the Universidad Nacional Autónoma de México (UNAM), Mexico. Omar Neme-Castillo is Research Economist at the School of Economics in the Instituto Politécnico Nacional, Mexico.

Contributors

xxi

Penélope Pacheco-López is consultant for UNIDO­Mexico’s Office in the area of the competitiveness of the Mexican manufacturing sector. Ignacio Perrotini-Hernández is Research Economist and Head of the Graduate Faculty of Economics at the Universidad Nacional Autónoma de México (UNAM), Mexico; he is also the Editor of Investigación Económica (Economic Research), a bilingual leading journal in Latin America. Humberto Ríos-Bolívar is Research Economist at the School of Economics in the Instituto Politécnico Nacional, Mexico; he is also the Editor of the journal Panorama Económico (Economic Panorama). Jaime Ros is currently Professor of Economics at the Universidad Nacional Autónoma de México (UNAM). He also taught Economics at the University of Notre Dame and is a Faculty Fellow at the Helen Kellogg Institute of International Studies. Máximo Rossi is Professor of Economics at De la República University, Uruguay. A.P. Thirlwall is Professor of Applied Economics at the University of Kent, UK. José Carlos Trejo-García is PhD student in Economics at the School of Economics in the Instituto Politécnico Nacional, Mexico. Ana Lilia Valderrama-Santibáñez is Research Economist at the School of Economics in the Instituto Politécnico Nacional, Mexico. Juan Alberto Vázquez-Muñoz is Research Economist at the Faculty of Economics in the Benemérita Universidad Autónoma de Puebla, Mexico. Francisco Venegas-Martínez is Research Economist at the School of Economics in the Instituto Politécnico Nacional, Mexico. Matías Vernengo is Associate Professor at the Department of Economics, University of Utah, USA.

Foreword Jaime Ros

The impact of economic liberalization on growth and development in developing countries is by now the subject of a vast literature that includes a variety of perspectives. This book takes up this subject and makes a substantial contribution to this literature, concentrating largely, although not exclusively, on Latin America. Have the processes of trade liberalization since the 1980s lived up to their promises? What went wrong with the policy package known as the Washington Consensus? What is the case for a developmental industrial policy in order to address the shortcomings of economic liberalization? What are the links between international trade and the level of economic development? What have been the effects of regional trade agreements on trade flows and foreign direct investment (FDI)? Do North–South trade agreements really attract more FDI than South– South agreements? What about the role in FDI of bilateral investment treaties? These are some of the key questions addressed by this book through an array of statistical and econometric methods applied to samples of developing and, in particular, Latin American countries, before and after and cross country comparisons, as well as case studies of Argentina, Chile, Mexico and Uruguay. The first part of the book focuses on the effects of trade liberalization and the processes of regional integration in Latin America. Trade liberalization has not lived up to expectations regarding growth performance, poverty reduction and income distribution as argued by Pacheco-López and Thirlwall in their review of the evidence for developing countries (a conclusion shared by country studies in the second and third parts of this volume). This, at least in part, is due to the fact that trade liberalization, especially in poor countries, has had deleterious effects on the balance of payments constraint on growth and also because, contrary to the orthodox predictions of the Stolper–Samuelson theorem, income inequalities have tended to increase in many developing countries as unskilled labour has not benefited from the expansion of trade. More generally, the policy package known as the Washington Consensus has not improved economic performance and closed the development gap in the large Latin American economies. This is attributed by Perrotini­Hernández, Vázquez­Muñoz, and Avendaño­Vargas to the fall in investment rates (with the exception of Chile among the large Latin American economies) and the increase in the income elasticity of imports which contrasts sharply with the Chinese experience. The policy implications derived

Foreword

xxiii

by the authors include an alternative development strategy centred on the implementation of an industrial policy for development (Perrotini­Hernández, Vázquez­Muñoz, and Avendaño­Vargas) as well as a trade strategy for development centred on acquiring dynamic comparative advantages and a call for a new world trade order that works for development in poor countries (Pacheco­López and Thirlwall). All this doesn’t mean that trade liberalization and regional trade agreements such as Mercosur and the Andean Community have not led to an increase in trade flows among trading partners (without reducing extra­regional trade flows). The increase in trade flows is shown by Hernández­Rodríguez using a gravity model which relates trade to income, distance between trading partners and membership in a trading bloc. Trade agreements have also been accompanied by increasing flows of FDI as shown by Goda in his interesting comparison of the experiences of Brazil under Mercosur and Mexico under NAFTA. Also, these trends in trade and FDI have generally had positive static effects on per capita income according to Ríos-Bolívar and Neme-Castillo in a study of a set of 21 Latin American and Caribbean countries. The second part of the book deals with trade reforms and development experience presenting three case studies of Latin American countries (Argentina, Chile, and Uruguay). Taking a long view of Argentina’s development, Ferreira and Vernengo argue that the recent growth and equity performance (except for the period since 2003) under economic liberalization compare rather unfavourably to that of the import substitution industrialization period. Grazzi examines the links between trade and investment liberalization (following the establishment of bilateral investment treaties and free trade agreements) and the flows of FDI in Chile, the third largest recipient of foreign investment in Latin America (after Brazil and Mexico) in absolute terms and the second in per capita terms after Trinidad and Tobago. Using an augmented gravity model, the author finds that bilateral investment treaties have a positive and significant impact on FDI inflows, while there is little or no evidence of a significant effect of double taxation treaties and free trade agreements. The experience of Uruguay under Mercosur is the subject of the chapter by Borraz, Ferrés and Rossi. They conclude that while specific groups of the population (those with higher and lower incomes) reaped more of the mild gains from trade than the middle income groups, they could not find any evidence about absolute losers resulting from Mercosur. The third part focuses on Mexico and contains a heterogeneous collection of chapterss dealing with different aspects of its economic performance. A first trend that characterizes the recent past is the rise and fall of inequality during the economic liberalization period. This is the subject of the chapter by AngelesCastro who finds that income inequality worsens after liberalization, mainly due to an increase in the skilled labour premium, a reduction of agricultural incomes relative to services and a negative impact of market openness on wages in the traded sector. This worsening of inequality is followed after 1998 by a decrease in inequality associated to the decrease in the returns to skilled labour and a weaker effect of trade on wages in the traded sector. Venegas-Martínez focuses

xxiv

Foreword

on growth performance from 1930 to 2002. The author brings currency, market, debt and fiscal risk factors in an intertemporal optimizing model of endogenous growth showing that risk factors may lead to qualitative changes in the determinants of growth in contrast with the deterministic setting. Perrotini­Hernández, Avendaño­Vargas, and Vázquez­Muñoz use a Minskyan framework to evaluate Mexico’s inflation targeting monetary policy, concluding that the management of the interest rate under this regime may encourage Ponzi financing. Using a production function approach, Ríos­Bolívar and Valderrama­Santibáñez study the relationship between the labour productivity growth and investment in research and development (R&D) for the manufacturing sector, trade and services, finding that investment in R&D has a positive effect on productivity growth in the three sectors. Finally, Loría and de Jesús focus on the relationship between unemployment and output growth, confirming for Mexico the validity of Okun’s law that relates the change in unemployment to the rate of GDP growth. Overall, this is a very valuable volume of essays that contributes significantly to the literature on its subject. It must be read by all those interested in the effects of trade and investment liberalization in developing countries, who will find in it not only new evidence but also fresh, intelligent and generally heterodox perspectives on the subject matter.

Acknowledgements

The editors wish to thank Gabriela Cruz­González, Estefania Molerés­Regalado, Jonathan Ortiz­Galindo, and Paulina Salazar­Rivera, for assisting in the editing of this book. Usual disclaimers apply.

Acronyms

ACP ADF AFTz ALADI

African, Caribbean and Pacific Countries Augmented Dickey-Fuller Andean Free Trade zone Asociación Latinoamericana de Integración (Latin American Association of Integration) APEC Asian­Pacific Economic Cooperation BADECEL Banco de Datos Estadísticos de Comercio Exterior (Statical Data Bank for Foreing Trade) BADEINSO Social Statistics and Indicators BCIA Banco de Crédito Industrial Argentino (Argentinian Industrial Bank of Credit) BCU Central Bank of Uruguay BIS Bank for International Settlements BIT Bilateral Investment Treaties BM Banco de México CAN Andean Community CB Central Bank CEMPE Center of Economic Modeling and Forecasting CEPII Centre d’Etudes Prospectives et d’Informations Internationales (Centre for Prospective Studies and International Information) CES Constant Elasticity of Substitution CET Common External Tariff CETES Certificados de Tesorería (Mexico’s Treasury bonds) CISEA Centro de Investigaciones Sociales sobre el Estado y la Administración (Centre for Social Research on the State and Administration) CONAPO Consejo Nacional de Población (National Council of Population) CPI Consumer Price Index DTTs Double Taxation Treaties ECH Encuesta Continua de Hogares ECLAC Economic Commission for Latin America and the Caribbean ENIGH Encuesta Nacional de Ingresos y Gastos de los Hogares (National Survey of Households’ Expenditures an Incomes)

Acronyms EPAs ESE EU FDI FE FEM FGLS FL FTAs FTAA GDP GLS GNP IADB IFAD IHS IMF INE INEGI IPN ISI IT IV LM Mercosur MF MNCs MSM MUR M&As NAFTA OECD OLS PCSE PCY PPP RE REM RIAs RoO ROW R&D SBTC SETH

xxvii

Economic Partnership Agreements Escuela Superior de Economía (Superior School of Economics) European Union Foreign Direct Investment Fixed-Effects Fixed Effects Model Feasible Generalised Least Squares Financial Liberalisation Free Trade Agreements Free Trade Area of the Americas Gross Domestic Product Generalised Least Squares Gross National Product Inter American Development Bank International Fund for Agricultural Development Inverse Hyperbolic Sine International Monetary Fund Instituto Nacional de Estadística (National Institute of Statistics) Instituto Nacional de Estadística, Geografía e Informática (National Institute for Statistics, Geography and Informatics) Instituto Politécnico Nacional (National Polytechnic Institute) Import Substituting Industrialization Inflation Targeting Instrumental Variables Lagrange Multiplayer Test Mercado Común del Sur (South American Common Market) Ministry of Finance Multinational Companies Mexican Stock Market Macroeconomic Unemployment Rate Mergers and Acquisitions North American Free Trade Agreement Organisation for Economic Co-operation and Development Ordinary Least Squares Panel Corrected Standard Errors Method Per Capita Income Purchasing Power Parity Random-Effects Random Effects Model Regional Integration Agreements Rules of Origin Rest of the World Research and Development Skill-Biased Technological Change Skill Enhancing Trade Hypothesis

xxviii Acronyms SIMBAD SITC SM SST TL TNCs TOBIT UK UN UNCTAD US USA WC WDI WIDER WTO YPF

Municipal Information System Database Standard International Trade Classification Secretaría del Mercosur (Mercosur’s Secretary) Stolper-Samuelson theorem Trade Liberalization Transnational Companies Tobit Model (it was first proposed by James Tobin, and involves aspects of Probit analysis) United Kingdom United Nations United Nations Conference on Trade and Development United States United States of America Washington Consensus World Development Indicators World Institute for Development Economics Research World Trade Organization Yacimientos Petrolíferos Fiscales (Fiscal Oilfields)

Editors’ introduction Gerardo Angeles-Castro, Ignacio Perrotini-Hernández and Humberto Ríos-Bolívar

The principal themes pursued in this book emerge from the great transformation that the Latin American and the Caribbean economies experienced in the aftermath both of the foreign debt crisis of 1982 and of the macroeconomic stabilisation policies that vividly and painfully produced the so-called ‘lost decade’ of the 1980s. Latin America implemented an economic liberalisation process during the late 1980s and the 1990s. The main policy reforms involved in that course can be summarised as privatisation of state owned firms, trade openness, deregulation of the foreign direct investment (FDI) regime and fiscal discipline. Latin American countries have also embarked in regional trade agreements, the most important ones being Mercosur and the North American Free Trade Agreement (NAFTA). The book compares results from the experience of North–South and South–South moulds of integration. Thus, the impacts of these policies on growth, development, technological progress, poverty and inequality are analysed. Orthodox and heterodox economic policies and theories are discussed along with relevant empirical evidence with a view to assess the relative merits of the various policy reforms applied by different countries in the region, on the one hand, and the experience of integration into the global economy, on the other. There are 13 chapters in this collection linked in varying ways to the series of economic reforms introduced in the region in the last decades. In this introduction we do not intend to provide a comprehensive résumé of the chapters of the present book or to exhaust the great richness of the analyses and debates contained therein. We just want to give a general appraisal of the editorial structure of the volume. There are many authors in this volume, and undoubtedly they disagree with each other on many important economic issues and policy questions. Yet, we have chosen to bring these chapters together because they indisputably share a common concern, namely the search for an alternative economic policy model that best suits Latin America’s future economic development. The book is organised in three parts.

Part I The first part comprises five chapters with the theme of trade liberalisation, development and regional integration. It is very appropriate to begin the book

2

G. Angeles-Castro et al.

with a contribution written by Pacheco-López and Thirlwall, who examine the impact of trade liberalisation on the trade-off between development and the balance of payments. The authors conclude that a new world trade order is needed where poor countries are allowed the acquisition of dynamic comparative advantage. Perrotini-Hernández, Vázquez-Muñoz and Avendaño-Vargas take up the issue of the impressive development gap that separates the world’s most advanced economy, the United States, and Latin America. They discuss the relative merits of the Washington Consensus strategy within a context where ‘classic conditions for market failure’ are ubiquitous. They make the case for a developmental industrial policy which calls for a fiscal policy regime aimed at stabilising investment, growth and employment. Ríos-Bolívar and Neme-Castillo study the relationship between international trade and per capita income in 21 countries of Latin America and the Caribbean in the period 1977–2007. They estimate the relationship between imports of capital goods, physical capital, per capita income and human capital. Among other things, they find the existence of a ‛literacy trap’. Thomas Goda wrestles with the relationship between regional integration and FDI in the Mercosur and the NAFTA areas. He compares the impacts of FDI in North–South and South–South integration environments and finds empirical results which may be seen as conflicting with received theory. Therefore, he states, regional integration as an effective means of attracting FDI should not be taken for granted. Hernández-Rodríguez, in turn, builds a gravity model to explore the factors that influence trade flows within trade blocs and outside such blocs. He concludes that geographic factors are most important, followed by income size.

Part II In the second part of the volume a set of three case studies is presented. CâmaraNeto and Vernengo offer a critical assessment of the process of economic development of Argentina from the mid-1940s to the present time. Their historical approach leads them to identify three stages of economic growth (the state-led growth model from 1946 to 1976, the liberal stage from 1976 to 2001 and the commodity boom period from 2002 to 2008) which have been unsustainable altogether. In addition, Grazzi examines in some detail the determinants of FDI inflows into Chile through the estimation of a gravity model from 1990 to 2005. His econometric results lead him to conclude that investments in Chile are mostly of the resource-seeking type; that FDI is negatively affected by distance and positively affected by the source country’s GDP and GDP per capita. This second part ends with an analysis of the impact of trade liberalisation on income distribution in Uruguay. Borraz, Ferrés and Rossi study the long-term effects on poverty and inequality resulting from Uruguay integrating into Mercosur. Their empirical results show that even if gains from trade liberalisation can be observed, those benefits are unevenly distributed. Therefore, trade liberalisation is not necessarily and always a pro-poor growth strategy. The ultimate effect of economic liberalisation and market integration depends on the impact of price

Editors’ introduction 3 changes of both tradable and non-tradable goods on the income share of the various social classes participating in the production process.

Part III The book ends with Part III which contains five chapters, though mainly theoretically heterogeneous in their inspiration, solidly anchored in a thorough empirical analysis of various features of the Mexican economy. This closing part should discourage any scrutiny of Mexico’s recent development experience through the narrow prism of a particularly raw version of economic religiosity. Some of the chapters here view the economic reform with the greatest of suspicion, whereas others entertain a more congenial approach. Angeles-Castro deals with factors driving changes in income distribution in post-reform Mexico. He argues that the main factors that contributed to increase inequality following economic liberalisation are the relative expansion of the average income in the service sector in relation to the agricultural and manufacturing sectors, a negative relationship between wages and market openness, and the increase in skill premium. On the other hand, between 1998 and 2006, income dispersion fell gradually and some of the factors driving this trend are the decrease in returns to skill and a weaker effect of market openness in wages. Venegas-Martínez works out a stochastic endogenous growth model that accounts for how risk factors (currency, market, debt and fiscal risk factors) affect economic growth. His model, in which a Brownian motion and a Poisson process drive exchange-rate expectations and a geometric Brownian motion guides a tax rate on wealth, sheds new lights to carry out simulation experiments and empirical research. He argues that his model explains the average and variance of Mexico’s economic growth over the period of time under analysis. In addition, Perrotini-Hernández, Avendaño-Vargas and Vázquez-Muñoz work with data for 47 non-financial firms quoted in the Mexican stock exchange market from 1990 to 2004, and conduct a microeconomic analysis of financial instability in Mexico. They conclude that financial fragility can occur when the central bank targets low inflation and appreciates the exchange rate due to the presence of high pass-through effects from currency volatility on to the price level. Trejo-García, Ríos-Bolívar and Valderrama-Santibáñez examine in some detail the relationship between human capital productivity and economic growth under market openness conditions and, focusing on the behaviour of three economic sectors – manufacturing, commerce and services – they show that investment on research and development has a positive effect and in most cases the coefficients are statistically significant. Lastly, Loría and de Jesús assess the validity of Okun’s Law in Mexico. They estimate Okun’s Law for the Mexican economy from 1985 to 2006 and found robust evidence in favour of bidirectional causality between output and unemployment. Finally, the analyses comprising this volume provide rich diagnoses and empirical analysis which, we hope, may contribute to the debate for the enhancement of economic development of Latin America. May this book inspire further research on the topics dealt with.

Part I

Trade liberalization, development and regional integration

1

Has trade liberalisation in poor countries delivered the promises expected? Penélope Pacheco-López and A.P. Thirlwall

Trade liberalisation has not lived up to its promises. But the basic logic of trade – its potential to make most, if not all, better off – remains. Trade is not a zero-sum game in which those who win do so at the cost of others; it is, or at least can be, a positive-sum game, in which everybody is a winner. If that potential is to be realised, first we must reject two of the long-standing premises of trade liberalisation: that trade liberalisation automatically leads to more trade and growth, and that growth will automatically ‘trickle down’ to benefit all. Neither is consistent with economic theory or historical experience. (Stiglitz, 2006)

1.1 Introduction The last decades have witnessed tremendous pressure on poor developing countries to liberalise their trade. The free trade mantra preached by developed countries and major international development organisations has become like a religion, holding out the promise that if poor countries adopt the faith, they will somehow be ‘saved’. The broad purpose of this chapter is to challenge this simplistic view. The chapter is based on a review of the vast literature of theory and case studies (including research of our own) on the relation between trade liberalisation and economic performance across the world (see Thirlwall and Pacheco-López, 2008), which leads us to four general, but important, conclusions. The first is that while there can be static gains from trade (if certain crucial assumptions are met) there is nothing in the theory of trade per se which demonstrates conclusively that trade liberalisation will launch a country on a higher sustainable growth path. Even Jagdish Bhagwati (2001), the high priest of free trade, is honest about that (see below). Second, the impact of trade liberalisation on reducing world poverty has been minimal, and may have increased it. Third, trade liberalisation has almost certainly worsened the distribution of income between rich and poor countries, and between unskilled wage-earners and other workers within countries, contrary to the predictions of orthodox theory. Finally, the evidence is fragile that the economic growth performance of countries that have liberalised extensively is in any way superior to countries that have not. The timing, sequencing and context of liberalisation are of prime importance in

8

P. Pacheco-López and A.P. Thirlwall

determining the impact of liberalisation. What really matters for growth performance is domestic economic policy and growth-supportive institutions. This will lead us at the end of the chapter to a brief discussion of trade strategy for development.

1.2 What is wrong with orthodox trade theory? Orthodox trade theory is based on Ricardo’s (1817) law of comparative advantage, and the Heckscher–Ohlin theorem which argues that countries will gain by specialising in the production of goods which use their most abundant factor of production (Heckscher, 1919; Ohlin, 1933). Paul Samuelson (1962) cites Ricardo’s theory of comparative advantage as one of the few laws in economics ‘that is both true and non-trivial’. There are, indeed, static welfare gains to be had by countries specialising in goods in which they have the greatest comparative advantage (or lowest opportunity cost), but two crucial, often-forgotten, assumptions need to be met. The first is that in the process of resources reallocation, full employment is preserved, but this is not guaranteed. If unemployment arises, the welfare gains from greater specialisation may be offset by the welfare losses of unemployment. As Keynes (1930) rightly says ‘free trade assumes that if you throw men out of work in one direction you re-employ them in another. As soon as this link in the chain is broken the whole of the free trade argument breaks down.’ The second crucial assumption is that in the process of freeing trade, balance of payments equilibrium is preserved, which is also not guaranteed. In orthodox theory, the balance of payments is assumed to look after itself without affecting output and employment. This was the implicit assumption of the gold standard adjustment mechanism, and is also implicit in the theory of flexible exchange rates. But if trade liberalisation leads to a faster growth of imports than exports and the nominal exchange rate is not an efficient balance of payments adjustment weapon, then output will need to contract to reduce imports, leading to welfare losses. As we shall see later, this has been the experience of many developing countries forced to liberalise prematurely. In fact, the existence of unemployment provides one of the major economic arguments for protection, as outlined in Johnson’s (1964) classic paper on tariffs and economic development. Unemployment means that the social cost of labour is less than the private cost so that a welfare gain is possible by encouraging more domestic employment until the social cost of production is equal to the world price of goods. A subsidy to labour, however, is the first best policy because an equivalent tariff would reduce consumer surplus. Johnson also outlines some of the other classic economic arguments for protection such as the infant industry argument, the externalities argument and the optimal tariff argument. But Rodrik (1988) is correct that despite the body of trade theory which legitimises protection, the arguments have still not penetrated the vast literature on trade policy in developing countries even though the market imperfections that the arguments reflect are more serious in developing countries than in developed countries.

Trade liberalisation in poor countries

9

As well as potential static gains from trade (although not guaranteed, and in any case small (see Dowrick, 1997)) there are also possible dynamic gains which arise through the greater flow of ideas, new knowledge, investment and economies of scale if the domestic market for output is small. The dynamic effects of trade, however, depend primarily on what countries specialise in; whether natural resource activities or manufacturing. John Stuart Mill (1848) pointed this out in the nineteenth century, and Stiglitz (2006) today makes the same enduring point: Without protection, a country whose static comparative advantage lies in, say agriculture, risks stagnation; its comparative advantage will remain in agriculture, with limited growth prospects. Broad-based industrial protection can lead to an increase in the size of the industrial sector which is, almost everywhere, the source of innovation; many of these advances spill over into the rest of the economy as do the benefits from the development of institutions, like financial markets, that accompany the growth of an industrial sector. Moreover, a large and growing industrial sector (and the tariffs on manufactured goods) provides revenues with which the government can fund education, infrastructure, and other ingredients for broad-based growth. In other words, if trade is to be an engine of growth, poor countries need to acquire new comparative advantage in goods that have favourable production and demand characteristics. Structure matters for economic growth. This is recognised in ‘new’ trade theory pioneered by Krugman (1984, 1986) in the 1980s, who shows there is a case for protecting industries with spillovers and externalities, and for using import substitution for export promotion. In most standard growth models, however, the effect of trade on growth is ambiguous. For example, in the canonical neoclassical Solow model (1956), trade cannot affect the steady-state growth rate, because it is treated as an exogenous constant. Only in the ‘new’ growth theories of, for example, Grossman and Helpman (1991a, 1991b) does trade have the potential to raise the growth rate permanently through learning and spillover effects, but they have to be continuous. Bhagwati (2001), the most ardent advocate of free trade, even for poor developing countries, frankly admits: Those who assert that free trade will lead necessarily to greater growth either are ignorant of the fine nuances of the theory and the vast quantity of literature to the contrary on the subject at hand or are nonetheless basing their argument on a different premise; that is, that the preponderant evidence on the issue (in the post-war period) suggests that free trade tends to lead to greater growth after all. In fact, where theory includes several models that can lead in different directions, the policy economist is challenged to choose the model that is most appropriate to the reality she confronts. And I would argue that, in the present instance, we must choose the approach that generates favourable outcomes for growth when trade is liberalised.

10

P. Pacheco-López and A.P. Thirlwall

The issue is empirical, but certainly history is not on the side of the free-traders. None of the now-developed countries transformed their economies on the basis of laissez-faire, laissez-passer. Great Britain started to protect and foster industries as early as the late fifteenth century under Henry VII, and did not start dismantling the structure of protection until the repeal of the Corn Laws in 1848. From then on Great Britain preached free trade, but it had already attained technological superiority in the world economy, and such preaching, as List (1885) remarked, was like ‘kicking away the ladder’. The United States followed Great Britain’s protectionist route at the end of the eighteenth century under the influence of the Treasury Secretary, Alexander Hamilton, who, in 1791, first coined the term ‘infant industry’. Adam Smith’s advice to the United States in his Wealth of Nations (1776) was to pursue free trade: Were the Americans, either by combination or by any other sort of violence, to stop the importation to European manufactures, and, by thus giving a monopoly to such of their own countrymen as could manufacture the like goods, divert any considerable part of their capital into this employment, they would retard instead of accelerating the further increase in the value of their annual produce, and would obstruct instead of promoting the progress of their country towards real wealth and greatness. If the United States had followed Smith’s advice, it would have remained an economic backwater instead of becoming the richest country in the world based on high productivity in industry. The same can be said of modern-day economic giants, such as Japan and South Korea, whose comparative advantage once lay in rice, but who, through selective protection, import substitution, export promotion and directed credit, transformed themselves into industrial power-houses (see Chang, 2005). The newly industrialising countries of South-East Asia, and particularly China, are pursuing the same route to development – transforming their industrial structure through deliberate policy intervention – and are growing fast as a consequence. Stiglitz (2006) is right when he says that ‘economists who promise that trade liberalisation will make everybody better off are being disingenuous. Economic theory (and historical experience) suggests the contrary.’ All we know is that as countries get richer they dismantle trade restrictions, not that they get richer because they liberalise trade. The issue for poor developing countries today is not whether to protect, but how to protect in order to ensure the dynamic efficiency of their nascent industrial activities.

1.3 Poverty and income inequality within countries At this moment in time, nearly one billion of the world’s population live on less than $1 a day, and 2.7 billion live on less than $2 a day (Chen and Ravallion, 2004). In other words, over one-third of the world’s population lives in absolute poverty. Advocates of trade liberalisation promise that the freeing of trade will lift people out of poverty. The former European Trade Commissioner, Peter

Trade liberalisation in poor countries

11

Mandelson, wrote in the Guardian newspaper (3 October 2008) that globalisation is the greatest engine of poverty reduction the world has ever seen. If he had looked at the facts, however, he would see that since 1980 the absolute number of people in poverty has not decreased. The number on less than $2 a day has increased from 2.4 billion to 2.7 billion, and the number on less than $1 a day (excluding China) has increased from 848 million to 870 million. The reduction in the total number living on less than $1 a day is because of a fall in China in the early 1980s, but this was due to agricultural reforms not to trade liberalisation. Winters et al. (2004), in their survey of trade liberalisation and poverty, claim that ‘theory provides a strong presumption that trade liberalisation will be poverty alleviating in the long run and on average’. This is simply not true, because, as we have seen, the theory of trade liberalisation says nothing definite about economic growth. The impact of trade liberalisation on poverty depends on its effects on employment and prices. Trade liberalisation can easily cause poverty by throwing people out of work. For example, since the NAFTA agreement was signed between the US, Canada and Mexico in 1994, two million Mexican maize farmers have lost their jobs because they cannot compete with subsidised maize from the US. Trade liberalisation can provide new opportunities in the export sector, but only if the sector is prepared. The statistical research on the relation between trade liberalisation and poverty is very inconclusive. The most comprehensive study is by Ravallion (2006) who takes 75 countries where there have been at least two household surveys on poverty, and runs a simple regression of the percentage change in the poverty rate on the percentage change in the ratio of trade to GDP (as a proxy for liberalisation). There is a statistically significant negative coefficient of 0.84, but the correlation is very fragile. For example, controlling for initial conditions makes the relation insignificant, and adding other control variables makes no difference. Ravallion concludes ‘it remains clear that there is considerable variation in the rates of poverty reduction at a given rate of expansion of trade volume’. Equally, however, ‘based on the data available from cross-country comparisons, it is hard to maintain the view that expanding trade, in general, is a powerful force for poverty reduction in developing countries’. At the same time as the absolute numbers in poverty have been increasing, the distribution of income within poor countries has also been widening, contrary to the orthodox predictions of the Hecksher–Ohlin (and Stolper–Samuelson, 1941) theorems. Goldberg and Pavcnik (2007), in their survey of the distributional effects of globalisation in developing countries, say: ‘while inequality has many different dimensions, all existing measures for inequality in developing countries seem to point to an increase in inequality which in some cases is severe’. The major cause of income inequality is wage inequality between skilled and unskilled workers. Orthodox trade theory predicts a narrowing of wage inequality in poor countries because their comparative advantage should lie in the production and export of goods using abundant unskilled labour. This narrowing has not happened for four main reasons: first, trade-related, skillbiased technical change; second, competition between poor countries; third,

12

P. Pacheco-López and A.P. Thirlwall

flows of FDI adding to the demand for skilled labour; and finally, in some cases, depressed demand for unskilled labour where trade liberalisation has caused balance of payments difficulties (see Arbache et al. 2004 for a case study of Brazil). By far the most detailed study of the impact of trade liberalisation on the distribution of income is that by Milanovic (2005a). First, in his introductory survey of the existing literature, he remarks: The conclusions run nearly the full gamut, from openness reducing the real income of the poor to openness raising the income of the poor proportionately less than the income of the rich to raising both the same in relative terms. Note, however, that there are no results that show openness reducing inequality; that is, raising the income of the poor more than the income of the rich – let alone raising the absolute income of the poor by more. Milanovic’s own research takes 321 household surveys from 95 countries in 1988, and 113 countries in 1993 and 1998 covering 90 per cent of the world’s population. Income inequality is measured, not by a summary statistic such as the Gini ratio or Theil index, but by the income of the ith decile of the population relative to the mean level of income of the whole population. For each decile, income inequality is then related to trade openness measured by the ratio of total trade to GDP, and also to openness interacted with the level of income to test whether the effect of openness on inequality varies with the level of income. A regression is run for each of the ten deciles using the same independent variables. Two striking results emerge. First, increased openness reduces the income share of the bottom six deciles. Second, the adverse effect of openness on inequality is less the higher a country’s per capita income. The turning point for the poor to benefit from increased trade is approximately US$7,500 at 1990 prices. Barro (2000) and Spilimbergo et al. (1999) also find openness worsens income inequality up to a certain point and then the effect diminishes. Milanovic concludes: ‘openness would therefore seem to have a particularly negative impact on poor and middle income groups in poor countries – directly opposite to what would be expected from the standard Hecksher–Ohlin–Samuelson framework’. The contrary conclusion to the above studies of Dollar and Kraay (2002, 2004) that ‘growth is good for the poor’ arises from their unusual procedure of measuring trade in nominal US dollars, but measuring GDP at purchasing power parity (PPP). Since GDP at PPP is much higher than in nominal dollars, this considerably understates the ratio of trade to GDP in poor countries. For example, China’s exports as a share of GDP measured at PPP are only 7 per cent compared to 26 per cent if both trade and GDP are measured in nominal dollars. It is this latter ratio that affects the income distribution, and which should be used in studies of trade and income distribution.

Trade liberalisation in poor countries

13

1.4 International and global inequality Not only has the distribution of income within poor countries been increasing over time, but also the distribution of income between poor and rich countries. Again, this contradicts the prediction of orthodox neoclassical theory (Solow, 1956) which argues that because the productivity of capital should be higher in poor capital-scarce countries than in rich capital-saturated countries, poor countries should grow faster than the rich leading to a convergence of living standards across the world. There are many non-orthodox models to explain divergence, associated with the names of Myrdal (1957), Hirschman (1958), Kaldor (1970) and various Marxist writers, but nonetheless the orthodoxy prevails despite the evidence. The measurement of international inequality takes each country’s average per capita income as a single unit, regardless of the distribution of income within countries, and a Gini ratio can be calculated either unweighted or weighted by population. Global inequality, by contrast, not only measures inequality between countries but also within countries as well, giving a higher figure. The most recent calculations of the Gini ratio of international inequality by Milanovic (2005b), and of global inequality by Milanovic (2005b), Bourguignon and Morrisson (2002) and Sala-i-Martin (2002) are shown in Table 1.1. The unweighted Gini ratio for international inequality shows a steady historical rise from 1820, and also in the post-war period of trade liberalisation from 1952. The populationweighted Gini ratio of international inequality shows a slight decline in recent years due to the fast growth of populous countries such as China and India. If China is taken from the sample, the population-weighted Gini ratio is also shown to rise. The Gini ratio for global inequality has increased over time but has been relatively static in recent years because while between-country inequality (population weighted) has fallen slightly, income inequality within countries has increased, particularly in China between the rural and urban sectors. The question is: how much of this rising and persistent inequality across the world is due to trade liberalisation? It is not easy to answer this question, but attempts can be made. One methodological approach is to interact a measure of trade openness with the level of per capita income (PCY) to test whether the impact of openness varies with the level of development. This is what Dowrick and Golley (2004) do, taking over 100 countries for two separate time periods, 1960–80 and 1980–2000, and regressing the growth of PCY on (1) trade as a per cent of GDP, (2) an interaction term of trade openness and a country’s level of PCY, (3) a dummy variable for specialisation in primary products, measured as more than 50 per cent of exports and (4) a number of control variables. Separate regressions are also run for developed and less developed countries. For the first period 1960–80, a higher trade share of one percentage point is associated with 0.11 per cent faster growth, and the poorer the country, the slightly greater the benefit from openness, meaning that trade was a force for convergence. But for the second period, 1980–2000, this result is reversed. The impact of the trade share on the growth of PCY is now negative (–0.072) and the interaction term

0.2 0.29 0.31 0.37 0.35 0.35 0.45 0.46 0.47 0.5 0.53 – 0.54

0.12 0.26 0.3 0.37 0.4 0.4 0.57 0.55 0.54 0.53 0.52 – 0.5

– – – – – – – – – 0.62 0.65 0.64 –

0.5 0.56 0.59 0.61 0.62 – 0.64 0.64 0.66 – 0.66 (1992) – –

Bourginon and Morrisson (2002)

Milanovic (2005b)

Unweighted

Population weighted

Global (or world) inequality

International inequality1

Note 1 Adapted from Milanovic (2005b), Table 11.1.

1820 1870 1890 1913 1929 1938 1952 1960 1978 1988 1993 1998 2000

Year

Table 1.1 A comparison of Gini ratios

– – – – – – – – 0.66 (1970) 0.65 0.64 0.63 0.63

Sala-i-Martin (2002)

Trade liberalisation in poor countries

15

with the level of PCY is positive (+0.009) indicating that poor countries suffered from trade openness more than rich countries, leading to divergence. Dividing the 1980–2000 sample of countries into 33 poorest countries and the rest shows no significant effect of the trade share on growth in the poorest countries, but the richer countries gained about 0.012 per cent growth for a one percentage point increase in the trade share. Specialisation in primary products had a strong negative effect on growth in the 1980–2000 period, reducing it on average by nearly 1 per cent; and the impact was even stronger in the poor country group – a difference of –1.76 per cent. Dowrick and Golley’s conclusion is that ‘trade has promoted strong divergence in productivity [between countries] since 1980’.

1.5 Trade liberalisation and trade performance The main purpose of trade liberalisation is to promote, or allow, the most efficient allocation of a country’s resources to maximise its welfare. We have already criticised the static nature of orthodox trade theory, and outlined some of its limiting assumptions, but what has been the effect of trade liberalisation in practice on countries’ trade performance, and ultimately on the growth of living standards? This requires detailed statistical analysis. Research on export performance before and after liberalisation gives mixed results depending on the context in which trade liberalisation takes place, particularly the domestic economic policy being pursued and world economic conditions. Also, in econometric studies, results differ according to the methodology used and how liberalisation is measured. The most comprehensive recent study is that of Santos-Paulino and Thirlwall (2004) who take a panel of 22 developing countries from the four ‘regions’ of Africa, Latin America, East Asia and South Asia that undertook significant trade liberalisation during the period 1972–97. Trade liberalisation is measured by two indicators: first by duties on exports, and second by a dummy variable taking the value of one in the year when trade liberalisation took place (and continued) and zero otherwise. Panel data and time series/cross-section estimation techniques are then applied to the determination of export growth using a conventional export growth equation of the form: xt = ao + a1(rert) + a2(zt) + a3(dxt) = a4(libt)

(1.1)

where x is the growth of export volume, rer is the rate of change of the real exchange rate, z is the growth of world income, dx is the duty on exports, lib is the liberalisation dummy and t is time. Depending on the estimation technique used, the central estimate is that trade liberalisation has raised export growth by approximately two percentage points, or by one-quarter compared to the pre-liberalisation export growth rate. The estimated coefficient on the export duty variable is negative, but small (roughly –0.2). The coefficient on the liberalisation dummy variable is consistently in the range one to two taking the full sample of 22 countries, but the quantitative effect (shown in parentheses) differs between the four regions: Africa (3.58), South Asia (2.54), East Asia (2.42) and Latin America (1.66).

16

P. Pacheco-López and A.P. Thirlwall

For a country’s overall economic performance to improve, however, it is not enough for export growth to accelerate. Export growth must be shown to outpace import growth, otherwise balance of payments difficulties will arise. In the literature on trade liberalisation, very little attention has been paid to import growth, or to the balance between export growth and import growth. This is a serious weakness of trade liberalisation studies, but is a reflection of the fact that in orthodox trade and growth theory the balance of payments is either assumed to look after itself, or deficits as regarded as a form of consumption smoothing and have no long run effect on real variables. Country studies by Melo and Vogt (1984) for Venezuela, Mah (1999) for Thailand, and Bertola and Faini (1991) for Morocco all show a significant impact of trade liberalisation on import growth and on the sensitivity of imports to domestic income growth, but the most detailed study is that by Santos-Paulino and Thirlwall (2004) who take the same 22 countries as for export growth and test three hypotheses: (1) trade liberalisation, measured by a shift dummy variable (lib), significantly increases import growth; (2) reductions in tariffs (dm) raise import growth and (3) a more liberal trade regime increases the income and price elasticities of demand of imports (measured by interacting the liberalisation dummy with the growth of income and real exchange rate variables, liby and librer, respectively). The import growth equation specified to capture these effects is: mt = bo + b1(rert) + b2(yt) + b3(dmt) + b4(libt) + b5(libyt) + b6(librert)

(1.2)

The results may be summarised as follows. Trade liberalisation itself, controlling for all other factors, has increased the growth of imports by between five and six percentage points, which represents a near doubling of the preliberalisation import growth. The independent effect of tariff cuts has been to raise the growth of imports by between 0.2 and 0.5 percentage points for a one percentage point cut in tariff rates. Liberalisation has increased the elasticity of imports to both domestic income and exchange rate changes by between 0.2 and 0.5 percentage points. We have examined ourselves (Pacheco-López and Thirlwall, 2006) the direct effect of trade liberalisation on the income elasticity of demand for imports in 17 Latin American countries over the period 1977 to 2002 using a simplified version of equation (1.2): mt = co + c1(rert) + π1(yt) + π2(libyt)

(1.3)

where p1 is the income elasticity of demand for imports in the pre-liberalisation period and (p1 + p2) is the income elasticity in the post-liberalisation period. We find an increase of 0.55 from 2.08 to 2.63, which more or less offsets the increase in export growth post-liberalisation, leaving the GDP growth rate of countries consistent with balance of payments equilibrium virtually unchanged. This increase in the income elasticity of demand for imports in Latin America as a result of trade liberalisation is confirmed using the technique of rolling regressions taking 13

Trade liberalisation in poor countries

17

overlapping periods starting from 1977–90 and ending in 1989–2002. The estimated income-elasticity starts at 2.04 and ends at 2.82, giving an annual trend rate of increase of approximately 0.04 percentage points. If trade liberalisation raises the growth of imports by more than exports, or raises the income elasticity of demand for imports by more than in proportion to the growth of exports, the balance of trade (or payments) will worsen at a given growth of output, unless the currency can be manipulated to raise the value of exports relative to imports. Surprisingly, very little research has been done on the balance of payments effect of trade liberalisation. The first major studies were by Parikh for UNCTAD (1999) and for WIDER (Parikh, 2002). The first study examined 16 countries over the period 1970–95, with the main result that trade liberalisation seems to have worsened the trade balance by 2.7 per cent of GDP (which is substantial). The second study extends the analysis to 64 countries, with the general conclusion: The exports of most of the liberalising countries have not grown fast enough after trade liberalisation to compensate for the rapid growth of imports during the years immediately following trade liberalisation. The evidence suggests that trade liberalisation in developing countries has tended to lead to a deterioration in the trade account. Santos-Paulino and Thirlwall (2004) take the same sample of 22 developing countries as for the impact of trade liberalisation on export and import growth previously discussed, and estimate the following equation:

(1.4) where TB/GDP is the trade balance to GDP ratio, BP/GDP is the balance of payments to GDP ratio, tt is the terms of trade and the other variables are as defined in equations (1.1) and (1.2). The equations are estimated using panel data techniques over the period 1976–98. The most important result is that the switch to a more liberal trading regime has worsened, on average, the trade balance by 2 per cent of GDP (which is similar to the Parikh estimate), and the balance of payments by 1 per cent of GDP. For a group of 17 Least Developed Countries over the period 1970–2001, Santos-Paulino (2007) finds a deterioration in the trade balance ratio of 4 per cent of GDP. In our own research (Pacheco-López and Thirlwall, 2007) on 17 Latin American countries over the period 1977–2002 we find a deterioration in the trade balance of between 1.3 and 2.3 per cent of GDP depending on the method of estimation used (whether a panel, or time/series/cross-section, estimator, using as control variables the first three variables in equation 1.4). All these results show that trade liberalisation has impacted unfavourably on the trade balance

18

P. Pacheco-López and A.P. Thirlwall

and current account balance of liberalising countries. Such a deterioration, if it cannot be financed by sustainable capital inflows, may either trigger a currency crisis or necessitate a severe deflation of domestic demand (and therefore growth) to control imports. As UNCTAD (2004) says in its Least Developed Countries Report 2004: ‘this critical [balance of payments] constraint on development and sustained poverty reduction is conspicuously absent in the current debate on trade and poverty’; and also, it may be added, in the debate on the wisdom of excessive trade liberalisation in poor vulnerable countries. Indeed, the ultimate test of successful trade liberalisation, at least at the macro-level without regard to distributional effects, is whether it lifts a country on to a higher growth path consistent with a sustainable balance of payments; or, in other words, whether it improves the trade-off between growth and the balance of payments, as illustrated in Figure 1.1. On the vertical axis of Figure 1.1 is measured the ratio of the balance of payments to GDP, and on the horizontal axis, the growth of GDP. The solid-line curve gives the negative trade-off curve showing how the balance of payments deteriorates as growth accelerates. The curve is drawn to represent a serious situation where the balance of payments is in deficit (point a) even at zero growth. The objective of trade policy should be to shift the curve upwards to, say, point b on the horizontal axis so that some positive GDP growth is possible without running into balance of payments difficulties.

BP/GDP �

0 �

b � GDP growth (y)

a

�

Figure 1.1 The trade-off between growth and the balance of payments (source: author’s elaboration).

Trade liberalisation in poor countries

19

We have estimated this trade-off curve (using the trade balance/GDP ratio as the dependent variable) for 17 Latin American countries over the period 1977 to 2002 using pooled data (giving 425 observations) to see whether trade liberalisation has resulted in a positive shift (Pacheco-López and Thirlwall, 2007). Fitting a linear (for simplicity) regression line, without controlling for liberalisation, gives the result (t-statistics in parentheses): TB/GDP =

–

(1.5)

The curve cuts the vertical axis in the negative quadrant, which is serious. The average GDP growth for the sample as a whole is 2.76 per cent per annum, with an average trade deficit of –4.69 per cent of GDP. When a liberalisation dummy is included in the equation, the result shows a negative, not positive, effect, i.e. TB/GDP =

–

–

(1.6)

The pre-liberalisation deficit at zero growth is –1.387, and the post-liberalisation deficit is (–1.387) + (–3.610) = –4.997. Liberalisation has worsened the trade-off by 3.6 percentage points. Controlling for changes in the real exchange rate and world income growth reduces the unfavourable impact to –2.0 percentage points, but this is still significant.

1.6 Trade liberalisation and growth performance While it is true that trade liberalisation has improved export performance, liberalisation and export growth are not the same, and should not be confused. As Stiglitz (2006) notes: Advocates of liberalisation cite statistical studies claiming that liberalisation enhances growth. But a careful look at the evidence shows something quite different. . . . It is exports – not the removal of trade barriers – that is the driving force of growth. Studies that focus directly on the removal of trade barriers show little relationship between liberalisation and growth. The advocates of quick liberalisation tried an intellectual sleight of hand, hoping that the broad-brush discussion of the benefits of globalisation would suffice to make their case. Our study of Latin America discussed above is the only one we know that examines the impact of liberalisation on the trade-off between growth and the balance of payments, but there are several time series and cross-section studies of the relation between liberalisation and GDP growth on the one hand and trade openness and GDP growth on the other (although trade openness is not the same as liberalisation). The studies give mixed results, but it can definitely be said that

20

P. Pacheco-López and A.P. Thirlwall

the extravagant claims of the pro-trade liberalisers look hollow when compared with the evidence. Early work by Edwards (1992, 1998) and Dollar (1992) showing a positive relation between openness (or the outward orientation of countries) and growth performance was heavily criticised by Rodriguez and Rodrik (2000) on methodological grounds and for lack of robustness. Similar work by Dollar and Kraay (2004) on ‘globalisation’ and economic growth was likewise criticised by Dowrick and Golley (2004) who show that the faster growth of Dollar and Kraay’s sample of ‘globalising’ countries is entirely due to the fast growth of China and India. Even more telling, the so-called ‘globalising’ countries identified were not the most open or liberalised. Another major study of trade orientation and growth is that by Sachs and Warner (1995) who found that more open economies over the period 1979–89 grew 2.44 percentage points faster than economies identified as closed. Wacziard and Welch (2008) extend the Sachs–Warner study into the 1990s with more countries, and find that their result is not robust; there appears to be no significant effect of openness on growth performance. Greenaway, Morgan and Wright (1998, 2002) examine the relationship between trade liberalisation and GDP growth using an impact dummy variable for the year of liberalisation in a sample of up to 73 countries over the period 1975–93, and find a J-curve effect with growth first deteriorating and then improving. There is no indication, however, of how long the laggedgrowth effect lasts. Rodriguez and Rodrik (2000) conclude their evaluation of studies of trade orientation and economic growth by saying that indicators of openness used are either poor measures of trade barriers or are highly correlated with other determinants of domestic performance. All studies should therefore be treated with great caution. They are particularly concerned that the priority given to trade policy reform has generated expectations that are unlikely to be met, and may preclude other, institutional reforms which would have a greater impact on economic performance. Trade liberalisation, in other words, cannot be regarded as a panacea, or as a substitute for a comprehensive trade and development strategy. To quote Rodrik (2001): ‘Deep trade liberalisation cannot be relied upon to deliver high rates of economic growth and therefore does not deserve the high priority it typically receives in the development strategies pushed by leading organisations.’

1.7 Trade strategy for development So what trade strategy should poor countries pursue? The overriding objective must be to acquire dynamic comparative advantage. For this, the private sector of an economy needs the support of the government in the form of incentives and various types of ‘protection’ to mitigate investment risks. It is one thing to argue against anti-export bias; it is another to argue that poor countries should abandon all forms of protection of domestic industry. Improved market access across developed countries for poor country exports merely perpetuates static comparative advantage. As Rodrik (2001) argued in the lead-up to the recent (failed) Doha round of trade negotiations ‘the exchange of reduced policy autonomy in the

Trade liberalisation in poor countries

21

South for improved market access in the North is a bad bargain where development is concerned’. Poor countries need time and policy space to nurture new (infant) industrial activities as developed countries did historically, and as many newly industrialising economies still do today. As Hausmann and Rodrik (2003) say in their important work on the concept of ‘self discovery’: ‘The fact that the world’s most successful economies during the last few decades prospered doing things that are most commonly associated with failure (e.g. protection) is something that cannot easily be dismissed.’ Hausmann and Rodrik’s argument is that there is much randomness in the process of a country discovering what it is best at producing, and a lack of protection reduces the incentive to invest in discovering which goods and services they are. Poor, labour abundant economies have thousands of things they could produce and trade, but in practice their exports are highly concentrated. Sometimes, over 50 per cent of exports are accounted for by fewer than ten products. Bangladesh and Pakistan are countries at similar levels of development, but Bangladesh specialises in hats and Pakistan in bed sheets. This specialisation is not the result of resource endowment; it is the result of chance choice by enterprising entrepreneurs who ‘discovered’ (ex post) where relative costs were low. Other ‘chance’ investments include cut flowers in Colombia for export to North America, camel cheese in Mauritania for export to the European Union, highyield maize in Malawi and squash in Tonga. The policy implications of the Haussmann and Rodrik observation and model are that governments need to encourage entrepreneurship and invest in new activities ex ante, but push out unproductive firms and sectors ex post. Intervention needs to discriminate as far as possible between innovators and imitators. Normal forms of trade protection turn out not to be the ideal policy instruments because they do not discriminate, and earn profits only for those selling in the domestic market. Export subsidies avoid anti-export bias, but still do not discriminate between the innovators and the copycats, and in any case are illegal under the rules of the World Trade Organization (WTO). The first-best policy is public sector credit or guarantees which can discriminate in favour of the innovator, and be used as a ‘stick’ if firms do not perform well. There is much that the international community can also do to promote trade for development, as opposed to pursuing trade liberalisation for its own sake. The whole world trade system works against the majority of poor developing countries, first because of their dependence on primary commodities (the ‘curse’ of natural resources) and low value-added manufactures, second because the ‘rules of the game’ governing trade between rich and poor countries are rigged and biased in favour of the rich, and third because the agenda for trade reform is largely set by the rich developed countries. The only permanent solution to primary-commodity dependence is structural change which requires the establishment of new, non-traditional industries; but this is what the rich developed nations are hostile to. They want free access to poor countries’ markets, while continuing to protect their own. The most recent example of this is the ongoing debate between the European Union (EU) and the African, Caribbean and Pacific (ACP)

22

P. Pacheco-López and A.P. Thirlwall

countries over Economic Partnership Agreements (EPAs) to replace the trade preferences that the ACP countries used to enjoy under the Lomé Convention. The EU is insisting that poor developing countries reduce restrictions on imports of manufactured goods and service activities in return for continued access to the EU market for their agricultural products. The EU is refusing to look at alternatives to free trade EPAs, but by its own admission it concedes that EPAs could lead to the collapse of the manufacturing sector in many poor countries. As Stiglitz (2006) remarks in his powerful book Making Globalisation Work, ‘the US and Europe have perfected the art of arguing for free trade while simultaneously working for trade agreements that protect themselves against imports from developing countries’. If developed countries really wanted to help poor developing countries they could reduce and eliminate tariffs and barriers against all their goods. Oxfam (2002) estimates that trade barriers against developing country’s goods cost about $100 billion a year; or twice the level of official development assistance. In addition, developing countries might be allowed ‘infant country protection’, which would be equivalent to a currency devaluation, but have the advantage of raising revenue for spending on public goods. One of the severe drawbacks of tariff reductions in poor countries is a loss of tax revenue. If trade is to promote development, the WTO, that now governs world trade, needs radical reform and rethinking. The Agreement establishing the WTO (1995) lists as one of its purposes: Raising standards of living, ensuring full employment and a large and steady growing volume of real income and effective demand, and expanding the production of, and trade in, goods and services, while allowing for the optimal use of the world’s resources in accordance with the objective of sustainable development, seeking both to protect and preserve the environment and to enhance the means of doing so in a manner consistent with their respective needs and concerns at different levels of development The aim is laudable, but unfortunately there is a divorce between rhetoric and reality because the WTO treats trade liberalisation and economic development as synonymous, and yet as we have seen the historical and contemporary evidence is that domestic economic policy, institution building and the promotion of investment opportunities are far more important than trade liberalisation and trade openness in determining economic success in the early stages of economic development. Rodrik (2001) reminds us (like Chang, 2002, 2005 and Reinert, 2007) that: No country has [ever] developed simply by opening itself up to foreign trade and investment. The trick had been to combine the opportunities offered by world markets with a domestic investment and institution-building strategy to stimulate the animal spirits of domestic entrepreneurs. But now, under WTO rules, all the things that, for example, South Korea, Taiwan and other East Asian countries did to promote economic development in

Trade liberalisation in poor countries

23

the 1960s, 1970s and 1980s are severely restricted. Some countries that break the rules are succeeding spectacularly. China is one obvious example, but another would be Vietnam which, while promoting FDI and exports, also protects its domestic market, maintains import monopolies and engages in state trading. The WTO should shift away from trying to maximise the flow of trade, to understanding and evaluating what trade regime will maximise the possibility of development for individual poor countries. A new world trade order is required which acts on behalf of poor countries; and poor developing countries need a louder voice in any reformed structure.

References Arbache, J., Dickerson, A. and Green, F. (2004), Trade Liberalisation and Wages in Developing Countries, Economic Journal, February. Barro, R. (2000), Inequality and Growth in a Panel of Countries, Journal of Economic Growth, March. Bertola, G. and Faini, R. (1991), Import Demand and Non-Tariff Barriers: The Impact of Trade Liberalisation, Journal of Development Economics, November. Bhagwati, J. (2001), Free Trade Today (Princeton, NJ: Princeton University Press). Bourguignon, F. and Morrisson, C. (2002), Inequalities Among World Citizens: 1820–1992, American Economic Review, September. Chang, Ha-Joon (2002), Kicking Away the Ladder: Development Strategy in Historical Perspective (London: Anthem Press). Chang, Ha-Joon (2005), Why Developing Countries Need Tariffs (Geneva: South Centre). Chen, S. and Ravallion, M. (2004), How Have the World’s Poorest Fared Since the Early 1980s? World Bank Research Observer, Fall. Dollar, D. (1992), Outward Oriented Developing Countries Really do Grow More Rapidly: Evidence from 95 LDCs 1976–85, Economic Development and Cultural Change, April. Dollar, D. and Kraay, A. (2002), Growth Is Good for the Poor, Journal of Economic Growth, September. Dollar, D. and Kraay, A. (2004), Trade, Growth and Poverty, Economic Journal, February. Dowrick, S. (1997), ‘Trade and Growth: A Survey’ in J. Fagerberg, P. Hansson, L. Lundberg and A. Melchior (eds), Trade, Technology, and Changes in Employment of Skilled Labour in Swedish Manufacturing (Cheltenham: Edward Elgar). Dowrick, S. and Golley, J. (2004), Trade Openness and Growth: Who Benefits? Oxford Review of Economic Policy, Spring. Edwards, S. (1992), Trade Orientation, Distortions and Growth in Developing Countries, Journal of Development Economics, July. Edwards, S. (1998), Openness, Productivity and Growth: What Do We Really Know? Economic Journal, March. Goldberg, P. and Pavcnik, N. (2007), Distributional Effects of Globalisation in Developing Countries, Journal of Economic Literature, March. Greenaway, D., Morgan, W. and Wright, P. (1998), Trade Reform, Adjustment and Growth: What Does the Evidence Tell Us? Economic Journal, September. Greenaway, D., Morgan, W. and Wright, P. (2002), Trade Liberalisation and Growth in Developing Countries, Journal of Development Economics, February.

24

P. Pacheco-López and A.P. Thirlwall

Grossman, G. and Helpman, E. (1991a), Innovation and Growth in the Global Economy (Cambridge, MA: MIT Press). Grossman, G. and Helpman, E. (1991b), Trade, Knowledge Spillovers and Growth, European Economic Review, April. Hausmann, R. and Rodrik, D. (2003), Economic Development as Self Discovery, Journal of Development Economics, December. Heckscher, E. (1919), The Effect of Foreign Trade on the Distribution of Income, Ekonomisk Tidskrift, Vol. 21. Hirschman, A. (1958), Strategy of Economic Development (New Haven, CT: Yale University Press). Johnson, H. (1964), Tariffs and Economic Development: Some Theoretical Issues, Journal of Development Studies, October. Kaldor, N. (1970), The Case for Regional Policies, Scottish Journal of Political Economy, November. Keynes, J.M. (1930), ‘Evidence to the Macmillan Committee on Finance and Industry’ in D. Moggridge (1981), The Collected Writings of J.M. Keynes Vol. 20: Activities 1929–1931: Rethinking Employment and Unemployment Policies (London: Macmillan). Krugman, P. (1984), ‘Import Protection as Export Promotion: International Competition in the Presence of Oligopoly and Economies of Scale’ in H. Kierzkowski (ed.), Monopolistic Competition in International Trade (Oxford: Clarendon Press). Krugman, P. (1986), Strategic Trade Policy and the New International Economics (Cambridge, MA: MIT Press). List, F. (1885), The National System of Political Economy, translated from the original German edition published in 1841 by Sampson Lloyd (London: Longmann, Green and Co.). Mah, J.S. (1999), Import Demand, Liberalisation and Economic Development, Journal of Policy Modelling, July. Melo, O. and Vogt, M.G. (1984), Determinants of the Demand for Imports of Venezuela, Journal of Development Economics, April. Milanovic, B. (2005a), Can we Discern the Effect of Globalisation on Income Distribution? World Bank Economic Review, January. Milanovic, B. (2005b), Worlds Apart: Measuring International and Global Inequality (Princeton, NJ: Princeton University Press). Mill, J.S. (1848), Principles of Political Economy (London: Longmann, Green and Co.). Myrdal, G. (1957), Economic Theory and Underdeveloped Regions (London: Duckworth). Ohlin, B. (1933), Interregional and International Trade (Cambridge, MA: Harvard University Press). Oxfam (2002), Rigged Rules and Double Standards: Trade, Globalisation and the Fight Against Poverty (Oxford: Oxfam). Pacheco-López, P. and Thirlwall, A.P. (2006), Trade Liberalisation, the Income Elasticity of Demand for Imports and Growth in Latin America, Journal of Post Keynesian Economics, Fall. Pacheco-López, P. and Thirlwall, A.P. (2007), Trade Liberalisation and the Trade-Off Between Growth and the Balance of Payments in Latin America, International Review of Applied Economics, September. Parikh, A. (2002), Impact of Liberalization, Economic Growth and Trade Policies on Current Accounts of Developing Countries: An Econometric Study, WDP 2002/63 (Helsinki: WIDER).

Trade liberalisation in poor countries

25

Ravallion, M. (2006), Looking Beyond Averages in the Trade and Poverty Debate, World Development, August. Reinert, E. (2007), How Rich Countries Got Rich and Why Poor Countries Stay Poor (London: Constable and Robinson). Ricardo, D. (1817), On the Principles of Political Economy and Taxation (P. Sraffa, ed.) (Cambridge: Cambridge University Press, 1951). Rodriguez, F. and Rodrik, D. (2000), ‘Trade Policy and Economic Growth: A Skeptic’s Guide to the Cross-National Evidence’ in B. Bernanke and K. Rogoff (eds), Macroeconomics Annual 2000 (Cambridge, MA: MIT Press). Rodrik, D. (1988), ‘Imperfect Competition, Scale Economies and Trade Policy in Developing Countries’ in R. Baldwin (ed.), Trade Policy Issues and Empirical Analysis (Chicago, IL: Chicago University Press for the NBER). Rodrik, D. (2001), The Global Governance of Trade: As if Development really Mattered (New York: UNDP). Sachs, J. and Warner, A. (1995), Economic Reform and the Process of Global Integration, Brookings Papers on Economic Activity No. 1. Sala-i-Martin, X. (2002), The Disturbing ‘Rise’ of Global Income Inequality, NBER Working Paper, 8904. Samuelson, P. (1962), Economists and the History of Ideas, American Economic Review, March. Santos-Paulino, A. (2007), Aid and Trade Sustainability under Liberalisation in Least Developed Countries, World Economy, June. Santos-Paulino, A. and Thirlwall, A.P. (2004), The Impact of Trade Liberalisation on Exports, Imports, and the Balance of Payments of Developing Countries, Economic Journal, February. Smith, A. (1776), An Inquiry into the Nature and Causes of the Wealth of Nations (London: Straham and Caddell). Solow, R. (1956), A Contribution to the Theory of Economic Growth, Quarterly Journal of Economics, February. Spilimbergo, A., Londoño, J.L. and Székely, M. (1999), Income Distribution, Factor Endowments and Trade Openness, Journal of Development Economics, June. Stiglitz, J. (2006), Making Globalisation Work (New York: W.W. Norton and Co.). Stolper, W. and Samuelson, P. (1941), Protection and Real Wages, Review of Economic Studies, November. Thirlwall, A.P. and Pacheco-López, P. (2008), Trade Liberalisation and the Poverty of Nations (Cheltenham: Edward Elgar). UNCTAD (1999), Trade and Development Report (Geneva: UNCTAD). UNCTAD (2004), Least Developed Countries Report 2004: Linking International Trade with Poverty Reduction (Geneva: UNCTAD). Wacziard, R. and Welch, K. (2008), Trade Liberalisation and Growth: New Evidence, World Bank Economic Review, Vol. 22. Winters, A., McCullock, N. and McKay, A. (2004), Trade Liberalisation and Poverty: The Evidence so Far, Journal of Economic Literature, March. World Trade Organization (WTO) (1995), ‘Agreement Establishing the World Trade Organisation’ (Geneva: WTO Information and Media Relations Divisions).

2

Beyond the Washington Consensus The quest for an alternative development paradigm for Latin America Ignacio Perrotini-Hernández, Juan Alberto Vázquez-Muñoz and Blanca L. Avendaño-Vargas To find a growth performance in Latin America that is even close to the failure of the last 25 years, one has to go back more than a century, and choose a 25-year period that includes both World War I and the Great Depression. (Mark Weisbrot, 2006, p. 3)

2.1 Introduction The trade–development connection has been emphasised in the literature since the inception of economics as a discipline. Adam Smith (1776 [1976: Bk.III, ch.1, p. 405]), for example, regards the evolution from agriculture to manufacturing and ‘last of all to foreign commerce’ as ‘the natural course’ of development. This pattern of development goes on through an increasingly intricate set of transactions between advanced (‘town’) and non-advanced (‘country’) regions and between nations. In the latter case, foreign trade plays the role of a vent for surplus home production, thereby promoting further division of labour and greater accumulation of capital. Smith’s trade–development nexus highlights the principle of absolute advantage in production. Technological progress (enhanced division of labour) changes the pattern of absolute advantage, thereby enhancing the wealth of nations.1 The ‘extent of the market’ limits technological progress, hence the relevance of foreign trade for bringing increasing returns to fruition. The economy, Smith argued, possessed unlimited upward potential. David Ricardo (1821), another classical political economist, held a more sceptical view. He maintained that, since land rent grows as population increases, in the long run the economy follows a path towards a standstill. Ricardo’s model of international trade is composed of two countries and two commodities; it laid out the theory of comparative advantage, which argued that all countries could benefit from free trade, even if a country lacked absolute advantage at producing all kinds of goods. Thus, even if, say, country A is more proficient in producing both goods relative to country B, international trade can profitably continue on

Beyond the Washington Consensus

27

the basis of comparative advantage. According to Ricardo’s doctrine, countries reap gains from specialising in what they are best at producing and trading with each other; hence foreign trade is beneficial at any rate, albeit in the long run increasing differential land rent will ultimately bring about economic stagnation. Building on Ricardo’s theory of comparative advantage, Eli Heckscher and Bertil Ohlin elaborated a theorem maintaining that relative endowments of inputs determines comparative advantage and trade specialisation; goods and factor flows tend to be substitutes (cf. Ohlin, 1933). The theorem essentially says that countries will export goods that utilise their abundant and cheap factor of production and import products that utilise the countries’ scarce factor. The policy lesson that follows from the above is fairly straightforward: in order to attain higher stages of economic development, under the assumption of international immobility of labour and capital, a nation must specialise in the production of those goods that use intensively the input that is relatively more abundant in the domestic market. The Heckscher–Ohlin theorem became a basic constituent of the modern neoclassical theory of international trade, which states that all countries benefit equally from (free) foreign trade, regardless of the nature and quality of the goods being produced and exchanged internationally. After a long process of import-substitution industrialisation – running roughly between 1940 and 1980 – Latin America (and to a lesser extent Sub-Saharan Africa) adopted a model of economic liberalisation in the late 1980s. The neoclassical theory of international trade, in spite of its logical inconsistency (for example, capital as primary factor has no method of measuring it before the determination of profit rate), became the building block of that ‘new’ trade policy reform, encapsulated by the so-called Washington Consensus (henceforth WC). The WC was said to represent a framework of ‘good policies’ of trade and financial liberalisation (Williamson, 1990). While the Heckscher–Ohlin model gave the justification for trade liberalisation in the region in the aftermath of the 1982 foreign debt crisis, McKinnon’s theory (1973 and 1991) provided the strategy for ‘the order’ of financial liberalisation. The new orthodox approach to economic affairs was adopted under the presumption that it would restore growth and development on a sustainable basis, thus bringing about macroeconomic stability, convergence to optimum growth and narrowing the outstanding development gap of the Latin American economies vis-à-vis the leading economy in the world, the US economy. The essential questions to be tackled in this chapter are to enquire: whether the WC strategy – original or ‘augmented’ – contributed to bridging the development gap; in what sense can it be argued that the decision – undertaken in the 1980s – of giving markets free rein improved the performance of the economies under scrutiny if at all; does the development process call for ancillary institutions? If the answer to the latter question is in the positive, as experience appears to suggest (Cimoli et al., 2009a; Chang, 2003 and 2008; Ffrench-Davis, 2005; Rodrik, 2004), one may ask whether industrial policy can play a sensible role. We also expect to contribute to the current debate on an alternative developmental paradigm for the Latin American emerging economies, an urgent need

28

I. Perrotini-Hernández et al.

dramatically prompted by the ongoing economic crisis. Our analysis is focused on Argentina, Brazil, Chile and Mexico, the core economies of Latin America, which altogether account for more than two-thirds of the region’s GDP. China, today’s fastest growing economy, is also included for the sake of comparison; the US economy appears as a benchmark with a view of measuring the evolution of the development gap of the involved countries before and after the inception of the WC model. The chapter is organised in five sections in addition to this introduction. The first section revisits the WC framework and its policy implications for economic development. The second section highlights the performance of some key macroeconomic variables before and after the reform. The next section discusses the evolution of the development gap and the main driving forces behind its trend, the fourth makes the case for industrial policy from a developmental point of view and the last section concludes.

2.2 The Washington Consensus framework revisited The widespread foreign debt crisis of 1982 was a major catalyst for change in Latin America; the depth of the protracted crisis (and the stagflation that ensued) was interpreted as the need for a new development strategy. The Washington Consensus agenda was said to represent such an alternative approach and, therefore, was adopted in the late 1980s under the presumption that it would reignite fast and sustainable growth. The basic thrust of the original WC recipe provided a fairly simple catalogue of policy changes that ‘were needed more or less everywhere in Latin America’ (Williamson, 2004a:1), namely fiscal discipline and reorientation of government expenditure, tax reform, privatisation of state owned assets, deregulation, protected property rights, trade liberalisation, financial liberalisation (indeed, liberalising interest rates), openness to inward foreign direct investment and unified and competitive exchange rates (Williamson, 1990). Getting a minimalist state and prices right was the mantra of the new market-friendly approach to renewed economic development. Even if it went unnoticed at the outset, the advent of the WC implied the crisis of the developmental state that had been responsible, to a great extent, for Latin America’s industrialisation in 1940–1980 (Câmara and Vernengo, 2004). Indulgent fiscal policies leading to excessively large fiscal deficits were considered as the root of high inflation, balance of payments disequilibria and exchange rate instability. Moreover, the failure of import-substituting industrialisation across the region was interpreted as associated to the deleterious effect of government intervention through growing fiscal deficits. Many Latin American governments, seeking macroeconomic stability and sustainable growth in the late 1980s, coalesced at their own volition – combined with external pressure – on conducting market friendly reforms, with the core countries Argentina, Brazil, Chile and Mexico successively excelling as a poster child for the WC strategy. Ocampo (2004: 67) gives an account of the economic reform, worth quoting in full:

Beyond the Washington Consensus

29

Structural economic reforms varied in intensity across sectors and countries. All countries in Latin America significantly liberalized international trade, external capital flows and the domestic financial sector. Policy decisions in these areas included reducing tariffs and their dispersion; dismantling nontariff barriers; eliminating most restrictions on foreign direct investment; phasing out many or most foreign exchange regulations; granting greater or total autonomy to central banks; dismantling regulations regarding interest rates and credit allocation; reducing reserve requirements on domestic deposits; and privatizing several state banks. In the fiscal area, reforms strengthened the value added tax, reduced income tax rates and strengthened tax administration, though with only a limited effect on tax evasion. Social security systems were overhauled in several countries to allow for the participation of private agents and a more clear balance between benefits and (employers’ and workers’) contributions. The original agenda put forth by the WC strategy failed to live up to its goals; it did not produce macroeconomic stability with sustainable growth as had been predicated by its proponents, although Williamson (2004a:8) warned, after 15 years of experience of free-market policies, ‘I have to admit that I too am uncomfortable if it [the WC framework] is interpreted as a comprehensive agenda for economic reform.’ Indeed, while per capita GDP had increased by 2.7 per cent per year during 1950–1980, the ‘lost decade’ (the 1980s) saw a decline of 0.9 per cent annually and the WC era produced a dull recovery of less than 1 per cent per year from 1990 to 2003 (ECLAC, 2002 and 2003). What went wrong? Williamson (2003:2) maintains that ‘the liberalizing reforms of the past decade and a half, or globalisation, can [not] be held responsible for the region’s renewed travails in recent years’. The reasons accounting for the failure are threefold, according to Williamson and Kuczynski (2003). First, while pursuing macroeconomic stabilisation and microeconomic reforms practitioners of the WC incurred misguided macroeconomic policies, such as procyclical fiscal policies and exchange rate overvaluation, which left them exposed to capital reversals. Second, the economic reform was incomplete; and third, the WC’s concern for accelerating growth, alas, did not include equity. Williamson and Kuczynski (2003) addressed these topics and proceeded to propose a set of four second-generation reforms with the aim of restarting and ensuring ‘future growth’. First, with regards the business cycle, countries must build up a countercyclical policy intended for isolating their economies from adverse exogenous shocks and reducing exposure to international financial markets volatility. In this view, it is perfectly appropriate for the government to endorse the operation of the automatic stabilisers and reject deflationary fiscal policies (like the one the current Mexican government has followed, we may add) as a policy reaction to the present financial crisis. The main elements of the proposed crisis-proofing or stabilisation policy are: sufficient ex ante accumulation of budget surpluses, of foreign exchange reserves and stabilisation funds in times of cyclical booms; adoption of a flexible exchange rate regime and a monetary

30

I. Perrotini-Hernández et al.

policy framework of (low) inflation targeting as a cushion against perverse effects resultant from excessive capital inflows; forestalling liability dollarisation (the socalled original sin) and currency mismatch; reinforcement of prudential regulation of the banking sector; and, last but not least, improvement of domestic savings rates. Interestingly, the volume edited by Williamson and Kuczynski (2003:8, 77, 81, passim) endorses Maastricht-type debt and fiscal deficit rules in order to enforce fiscal discipline in Latin America. Second, liberalising reforms should be extended in order to make the labour market more flexible. Excessive rigidity of the labour market ‘constitutes a major obstacle to an expansion of employment in the formal economy’ (ibid.: 10). Inflexibility of the formal labour market, it is argued, impairs acceleration of growth and prevents a reduction of the informal sector, which employs ‘[a] round 50 percent of the labor force in many Latin American countries’ (Williamson, 2004a:11). The net effect of labour market flexibilisation is higher employment rates because it reduces inequality of opportunity in the labour market (ibid.; Saavedra, 2003:237–41, 252–53). Third, the need of building institutions. The original WC was oblivious of the crucial role of institutions in the making of economic development, which is thoroughly acknowledged in the augmented version. The type and quality of institutions suitable for the enhancement of output growth and political stability, for the internalisation of externalities, for efficiently supplying public goods and correcting polarising income distribution, varies from one country to another. In this respect, the government is allowed to play both the ‘old-fashioned’ role of building a good productive infrastructure and the modern role of ‘building a national innovation system’ and promoting R&D (Williamson and Kuczynski, 2003:12). The government can also advocate institutional reforms aimed at enhancing property rights and prudential regulation of the financial sector. While these reforms are not an easy task, if properly designed they tend to reduce transactions costs and the risk premium. Finally, governments should also have income distribution targets through market-incentive mechanisms, i.e. through creation of new assets by providing poor people with more human capital (education), secure property rights through land reforms (in rural areas), microcredit lines (in rural and urban areas alike) and similar instruments that encourage civil society. The crucial tenet here is that these assets, when combined within a sensible programme, may represent a propoor growth strategy that enables outcasts to work their way out of poverty (Williamson and Kuczynski, 2003:14–18; Williamson, 2004a:12).

2.3 Macroeconomic performance before and after the structural reform The WC policy reforms prompted a radical change in the development strategy based on perennial fiscal austerity2 and the laissez-faire paradigm (Davidson, 2003). As pointed out, the WC famously promised that the replacement of the import-substituting closed economy by a neoliberal open economy model in

Beyond the Washington Consensus

31

Latin America would lead to renewed and accelerating economic growth on a more sustainable basis. 2.3.1 Economic growth The observed pattern of output growth varies from one country to another between 1960 and 2009. Yet, some common trends can be highlighted. Argentina, Brazil and Mexico experienced episodes of growth accelerations from 1960 to 1980, albeit growth became highly volatile in Argentina towards the late 1970s, perhaps owing to domestic political instability and adverse oil shocks. Chile’s economic growth rate shows a downward trend from 1961 to 1975 – with some good times in between – and did not return to growth rate levels observed in the early 1960s until 1985. The Chilean economy appears to be an outlier in many ways: import-substituting industrialisation was not as relevant vis-à-vis other Latin American core economies; it adopted the laissez-faire approach to economic development in the mid-1970s, i.e. during the first years of Pinochet’s dictatorship. Chile’s per capita GDP increased over 3 per cent annually in 1980–2002, while Argentina, Brazil and Mexico experienced negligible, even negative, growth rates of per capita GDP. A noteworthy feature about Chile is that her approach to economic reform has been way more pragmatic than other countries’ approach, taming liberalisation with capital and exchange controls and non-tariff restrictions for a certain period of time. Furthermore, beginning in the late 1980s the growth rate of the four core Latin American economies exhibited a short-term upward trend that could not be sustained beyond 1994. Thus, long-term economic activity became slower and more volatile after the economic reforms. Contrary to Latin American countries, the Chinese economy, like many other East Asian economies whose fate has not been subject to the WC agenda, has been more vigorous and stable. If one looks at the more dynamic pattern of behaviour shown by the Chinese economy in the last decades, one may ask, then, what is the advantage of laissez-faire policies for latecomers? Summing up, both GDP and per capita income slowed down in Latin America from 1960–1980 to 1981–2006, whereas China, which kept away from the WC agenda, experienced long-term sustainable growth accelerations (see Figure 2.1a to 2.1e).3 While Latin America’s long-term output performance has been stagnant, China has accumulated three decades of fast growth. 2.3.2  Inflation Price stability through a flexible exchange rate regime and targeting a low rate of inflation is another key condition for restarting growth,4 according to Williamson (2003). Inflation has been conquered in the region but, since monetary anchors had failed to cope with rampant inflation, price stabilisation necessitated a significant role of managed pegs and exchange rate overvaluation. Actually, price stability predates the adoption of an inflation targeting regime by central banks; in nearly all inflation targeters inflation was already on a downward trend prior

Figure 2.1a Argentina’s GDP (annual growth rate), 1961–2009 (source: authors’ calculations using data from the World Bank and the INDEC). Note Hodrick–Prescott Filter: smoothing parameter = 100.

Figure 2.1b Brazil’s GDP (annual growth rate), 1961–2009 (source: authors’ calculations using data from the World Bank and the IBGE). Note Hodrick–Prescott Filter: smoothing parameter = 100.

Figure 2.1c Chile’s GDP (annual growth rate), 1961–2009 (source: authors’ calculations using data from the World Bank and the Banco Central de Chile). Note Hodrick–Prescott Filter: smoothing parameter = 100.

Figure 2.1d Mexico’s GDP (annual growth rate), 1961–2009 (source: authors’ calculations using data from the World Bank and the Banco de México). Note Hodrick–Prescott Filter: smoothing parameter = 100.

34

I. Perrotini-Hernández et al.

Figure 2.1e China’s GDP (annual growth rate), 1961–2009 (source: authors’ calculations using data from the World Bank and the National Bureau of Statistics of China). Note Hodrick–Prescott Filter: smoothing parameter = 100.

to the introduction of the new monetary policy consensus (Rochon and Rossi, 2006). The median inflation rate in Latin America went down from 32 per cent to 14 per cent between 1990 and 1994 (see Figure 2.2a to 2.2e). However, exchange rate-based stabilisation becomes unsustainable in the long-term because of the Impossible Trinity proposition: governments cannot simultaneously keep a fixed exchange rate regime, set the interest rate and have free capital mobility. At the end of the day, when exchange rate crises erupt a nominal exchange rate anchor must be abandoned. The Impossible Trinity helps to explain why Brazil, Chile and Mexico moved from a variety of pegs to a flexible exchange rate regime cum inflation targeting in the 1990s, while Argentina adopted a managed floating exchange rate regime after the financial crisis of 2001 when its currency board was abandoned. Insofar as the monetary-based anchor regime had gone astray in the 1980s and exchange rate-based anchors had collapsed in the mid-to-late 1990s, the only monetary/exchange rate regime left was a model that Knut Wicksell (1898) had originally put forth in which the interest rate regulates the price level. This is the well-known inflation targeting monetary policy framework advocated by the new monetary consensus (Bernanke et al., 1999; Lavoie and Seccareccia, 2005). The transition from the exchange rate-based anchor to the new monetary consensus model was also facilitated by the recent developments in the international capital markets (international securitisation of investment instruments, financial innovation, derivatives, financialisation of the economic activity) which, in the

Figure 2.2a Argentina’s annual rate of inflation, 1961–2009 (source: authors’ calculations using data from the World Bank and the INDEC). Note Hodrick–Prescott Filter: smoothing parameter = 100.

Figure 2.2b Brazil’s annual rate of inflation, 1970–2009 (source: authors’ calculations using data from the World Bank and the Banco Central do Brasil). Note Hodrick–Prescott Filter: smoothing parameter = 100.

Figure 2.2c Chile’s annual rate of inflation, 1961–2009 (source: authors’ calculations using data from the World Bank and the Banco Central de Chile). Note Hodrick–Prescott Filter: smoothing parameter = 100.

Figure 2.2d Mexico’s annual rate of inflation, 1961–2009 (source: authors’ calculations using data from the World Bank and the Banco de México). Note Hodrick–Prescott Filter: smoothing parameter = 100.

Beyond the Washington Consensus

37

Figure 2.2e China’s annual rate of inflation, 1987–2009 (source: authors’ calculations using data from the World Bank and the National Bureau of Statistics of China). Note Hodrick–Prescott Filter: smoothing parameter = 100.

last analysis, led to the ongoing international financial crisis. Nonetheless, given the influence of the volatility of perfect capital mobility on the dynamics of both the domestic interest rate and the exchange rate, the very same transmission mechanism of the international financial markets may powerfully influence the effectiveness of developing countries’ monetary policy. As Rojas-Suárez (2003) put it, financial liberalisation has not guaranteed constant access to international capital markets. Therefore, it appears that a framework composed of inflation targeting cum a flexible exchange rate regime will not necessarily circumvent the constraints caused by the Impossible Trinity nor will it automatically establish an independent monetary policy, though a floating exchange rate will be a better trade shock-absorber than a peg. Moreover, if inflation is not elastic to interest rate changes as a result of exogenously determined distributive shares (real wages and profit margins), in this case setting a low inflation rate as the unique goal of monetary policy may impart negative effects on capital accumulation, investment expenditure, output growth, the rate of employment and income distribution (Palley, 1996; Oreiro et al., 2008). In line with the discussion above, the high pass-through effect of exchange rate fluctuations to the price level is a major reason that makes an arrangement of managed exchange rate and flexible inflation targeting a superior regime than the one proposed by the augmented WC (cf. Galindo and Ros, 2006; Ito and Sato, 2007). Again, as shown in Figure 2.2, China achieved price stability in her own way, as opposed to Latin America, without surrendering herself to the

38

I. Perrotini-Hernández et al.

constraints imposed on domestic monetary policy by international capital markets and the Impossible Trinity. 2.3.3  Trade liberalisation and trade balance Hoping to spur the level of economic activity, the core economies of Latin America moved from import substitution to trade liberalisation in a short period of time. The trade-to-GDP ratio almost doubled between 1986 and 2006 as a consequence of rapid expansion of exports and imports. Free trade in most cases meant severe import competition for many industries that still were at the infancy stage and unable to cope with trade liberalisation. Rodrik (2004) studied 83 growth accelerations events5 in the world economy between 1957 and 1992; he found that the bulk of those accelerations (82 per cent) were not brought about by economic liberalisation. Remarkably, Latin America’s growth accelerations of the post-war period were certainly unrelated to trade liberalisation. Some commentators have argued that, by and large, trade liberalisation in Latin America favoured rapid expansion of exports but at the same time the industrial sector moved to static comparative advantages, while in China and East Asia manufacturing upgrading allowed for dynamic advantages (Shaffaeddin, 2005). Most importantly, trade liberalisation in Latin America led to an export boom in activities characterised by heavy imported inputs content. The export-led growth involved an import-led growth model; the balance of payments constraint to long-term output expansion became even more binding after trade liberalisation as a result of the deterioration of the trade balance (MorenoBrid, 1999; Pacheco and Thirlwall, 2006), although conjuncture improvements in the terms of trade, owing to booming commodity prices such as copper, soja and oil, produced temporary surplus positions which vanished over time. The Chinese economy, instead, improved its current account position while opening to foreign trade (cf. Figure 2.3a to 2.3e).

2.4 Evolution of the development gap Structuralist economic theory emphasises that economic development is basically about structural change, therefore about growth and income distribution (Ros, 2000; Taylor, 1991). There are many factors that can hamper structural change and capital accumulation in a developing economy. These binding constraints often lead to a gap between actual and optimum or desired growth rates of capital accumulation. The structuralist literature has pointed out that ‘a gap signals macroeconomic disequilibrium’ (ibid.:159). The foreign debt crisis of 1982 and the ensuing lost decade of the 1980s signalled a gap in Latin America. Economic liberalisation and market-oriented reforms were expected to remove such a gap (Rodrik, 2007). We now turn to enquire whether economic liberalisation has contributed to narrow the development gap of a core set of Latin American economies and the most developed economy in the world, namely the United States of America

Figure 2.3a Argentina: trade balance–GDP ratio (%), 1960–2009 (source: authors’ calculations using data from the World Bank and the INDEC). Note Hodrick–Prescott Filter: smoothing parameter = 100.

Figure 2.3b Brazil: trade balance–GDP ratio (%), 1960–2009 (source: authors’ calculations using data from the World Bank and the IBGE). Note Hodrick–Prescott Filter: smoothing parameter = 100.

Figure 2.3c Chile: trade balance–GDP ratio (%), 1960–2009 (source: authors’ calculations using data from the World Bank and the Banco Central de Chile). Note Hodrick–Prescott Filter: smoothing parameter = 100.

Figure 2.3d Mexico: trade balance–GDP ratio (%), 1960–2009 (source: authors’ calculations using data from the World Bank and the Banco de México). Note Hodrick–Prescott Filter: smoothing parameter = 100.

Beyond the Washington Consensus

41

Figure 2.3e China: trade balance–GDP ratio (%), 1978–2008 (source: authors’ calculations using data from the World Bank). Note Hodrick–Prescott Filter: smoothing parameter = 100.

(USA). The development gap for Latin America’s core economies is calculated in terms of the per capita income of the country in question as a percentage of the USA’s per capita income. We assess the performance of the development gap over the whole period 1960–2008, and highlight the evolution of the gap before and after economic liberalisation. Also, the evolution of the gap of Latin America’s core economies is contrasted with that of China, undoubtedly today’s fastest growing economy in the world. A perusal analysis of Figure 2.4a to 2.4e shows that Chile is the only Latin American economy in our sample that ends the period on a clear better off position compared with its starting position. Chile had recovered all that it had lost in terms of the development gap by 1993 and somewhat narrowed the gap later on. The worse case is Argentina, followed by Mexico. Brazil lost most of the ground it had achieved in the 1970s thanks to an aggressive industrial policy. By and large, it can be said that import-substituting industrialisation (henceforth ISI) made Brazil and Mexico better off. Yet, Argentina and Chile do not appear to have improved their development gap in the ISI period. The economic record of most Latin American countries after economic liberalisation is weak. Apart from Chile, none of them made great achievements during 1986–2008. However, the case of Chile should be looked at with caution because her big-bang economic liberalisation took place in the 1970s, in particular during 1973–1982, clearly, an era when the development gap was on a

Figure 2.4a Argentina’s development gap, 1960–2008 (source: authors’ calculations using data from the World Bank). Note Hodrick–Prescott Filter: smoothing parameter = 100.

Figure 2.4b Brazil’s development gap, 1960–2008 (source: authors’ calculations using data from the World Bank). Note Hodrick–Prescott Filter: smoothing parameter = 100.

Figure 2.4c Chile’s development gap, 1960–2008 (source: authors’ calculations using data from the World Bank). Note Hodrick–Prescott Filter: smoothing parameter = 100.

Figure 2.4d Mexico’s development gap, 1960–2008 (source: authors’ calculations using data from the World Bank). Note Hodrick–Prescott Filter: smoothing parameter = 100.

44

I. Perrotini-Hernández et al.

Figure 2.4e China’s development gap, 1960–2008 (source: authors’ calculations using data from the World Bank). Note Hodrick–Prescott Filter: smoothing parameter = 100.

widening trend. As a reaction to the foreign debt crisis of 1982, the Chilean government reformed the liberal reforms (Ffrench-Davis, 2005:149), tilting its trade policy towards a more pragmatic and heterodox view (with exchange rate devaluations, antidumping measures, price bands for some agricultural goods, the drawback system, the uniform tariff rate went up from 10 per cent to 35 per cent in 1984, capital controls and multiple exchange rates). So the latter sub-period was not a laissez-faire environment altogether in this country. According to Ffrench-Davis (2005), there was one radical trade reform in 1974–1979 and another moderate, more pragmatic reform in 1983–1991. The first one produced a widening development gap and the second one narrowed it. As for Argentina, Brazil and Mexico, their development gap has widened, in some cases very drastically with a fairly slight improvement towards the end of the period under analysis. In contradistinction, China has continuously been catching up vis-à-vis the US economy; China’s development gap has narrowed rapidly, paradoxically, since it opened up to international trade. What are the driving forces behind Latin America’s poor performance? Needless to say, a number of factors call for the phenomenon. We argue that structural change in Latin America has failed chiefly because the process of economic liberalisation has involved a low gross investment to GDP ratio and a drastic increase of the income elasticity of imports. Actually, in most Latin American countries the gross investment ratio declined vis-à-vis the investment coefficient of the golden age of the ISI model. In Argentina, seemingly the worst case, the gross investment coefficient peaked in 1977 (at 31 per cent of GDP) and reached

Beyond the Washington Consensus

45

a minimum level in 2002 (12 per cent). Brazil’s gross investment ratio peaked in 1989 and has been on a downward trend ever since, whereas the Chilean ratio exhibits an opposite (upward) drift from 1983 (9.8 per cent) to 2009 (25 per cent). Mexico’s gross investment ratio has been low, on average around 20 per cent, during the whole economic reform period. In contrast, China’s fast longterm economic growth has relied on a steadfastly upward drift of its gross investment ratio from the early 1960s (10.8 per cent in 1962) to the present time (close to 40 per cent). Liberalisation of the capital market, an important aspect of the marketoriented economic reform, tends to produce higher interest rates. From a freemarket perspective, higher rates of interest for both lenders and borrowers are viewed as the automatic balancing mechanism that equalises desired saving and desired net investment. The rate of interest sets free the dynamism that turns investment away from inferior projects so as to promote technological progress and economic development (McKinnon, 1973 and 1991). Nonetheless, as Shackle (1983:154) points out, ‘the rate of interest arises from uncertainty about the future prices of the bonds given by borrowers in exchange for loans’. In such a case, the automatic balancing mechanism can be self-destructive. It appears that under deregulated financial markets uncertainty increases the asymmetry between the distribution of costs and benefits of capital account liberalisation, on one hand, and the level of economic and institutional development, on the other, becomes a binding constraint: while the benefits of financial liberalisation have a tendency to increase beyond certain point, the costs associated to free capital flows tend to augment below that dividing line (cf. Prasad and Rajan, 2008). The

Figure 2.5a Argentina’s gross investment and gross savings (% of GDP), 1960–2009 (source: authors’ calculations using data from the World Bank and the INDEC).

Figure 2.5b Brazil’s gross investment and gross savings (% of GDP), 1960–2009 (source: authors’ calculations using data from the World Bank and the IBGE).

Figure 2.5c Chile’s gross investment and gross savings (% of GDP), 1960–2009 (source: authors’ calculations using data from the World Bank and the Banco Central de Chile).

Figure 2.5d Mexico’s gross investment and gross savings (% of GDP), 1960–2009 (source: authors’ calculations using data from the World Bank and the INEGI).

Figure 2.5e China’s gross investment and gross savings (% of GDP), 1960–2008 (source: authors’ calculations using data from the World Bank).

48

I. Perrotini-Hernández et al.

policy implication of this is that the government ought to adopt a pragmatic view on capital account liberalisation rather than an overly laissez-faire approach. The asymmetric experience with capital account liberalisation across emerging market economies reveals that, while financial liberalisation in Latin America has given rise to cycles of ephemeral recovery and prosperity, exchange rate instability, protracted stagnation, recession and sometimes complete growth meltdown, China and other Asian economies tend to exhibit growth acceleration episodes. A second driving force behind the enlargement of the development gap during the period of the market-oriented reforms, we contend, is the high-income elasticity of the demand for imports (henceforth π). The slow growth pattern followed by most Latin American economies6 in the last decades is associated with the distortion of import demand patterns generated by economic liberalisation. A higher π means a tighter balance of payments constraint to economic expansion, therefore a lower growth rate of output consistent with balance of payments equilibrium (Thirlwall, 1979). Following Pacheco-López and Thirlwall (2006), we applied the technique of rolling regressions to assess the performance of p in

Figure 2.6a Argentina’s income elasticity of imports, 1961–2007.

Figure 2.6b Brazil’s income elasticity of imports, 1961–2007.

Figure 2.6c Chile’s income elasticity of imports, 1961–2007.

Figure 2.6d Mexico’s income elasticity of imports, 1961–2007.

Figure 2.6e China’s income elasticity of imports, 1979–2007.

50

I. Perrotini-Hernández et al.

our sample of four Latin American core economies from 1961 to 2007.7 The behaviour of China’s π from 1979 to 2007 is also calculated for the sake of comparison. Rising trends of π for the Latin American countries can be detected during the whole period of trade liberalisation. Opposite to that, China has experienced a drastic reduction in its π. The question arises as to why China has exhibited a declining income elasticity of imports and an ever-diminishing development gap, while Latin America has experienced exactly the opposed trends in the period of the market-friendly reforms. Before the WC agenda, the Latin American economies (Argentina aside) kept a downward drift of π during 1960–1980; several growth acceleration episodes took place along this period, despite the ISI suffering from an anti-export bias. The WC economic reform promoted an export-led growth model with outright import liberalisation. Unfortunately, the propensity to import has increased drastically since the 1980s, thus discouraging the positive effect of booming exports. Hence slow growth. In contradistinction, China (and other East Asian economies) adopted an export-led growth strategy that has triggered growth acceleration events largely because her government prevented π from rising. The expectations of the policy of free markets and sound money, augmented with ‘good governance’ amendments, have not been fulfilled. The growth experience of Latin America under the mainstream consensus, particularly if contrasted with that of China or the successful East Asian economies, rejects the idea that the ‘Victorian’ free markets and sound money approach (Krugman, 1995) is the only and best recipe available in town to economic development. Clearly, an alternative developmental strategy for Latin America should be put in place. This ‘new’ development programme needs not imply an a priori rejection of the export-led model. It must, nonetheless, target a sensible reduction of p. The equation of the alternative developmental strategy can be a strong pro-growth state plus strong market plus diminution of the balance of payments constraint. There is plentiful historical evidence supporting the hypothesis that nowadays’ advanced economies followed a similar path in which industrial policy impelled growth and development (cf. Chang, 2003, 2008; Di Maio, 2009; List, 1885).

2.5 Beyond the Washington Consensus: the case for industrial policy Williamson (2004a:14) admits that ‘of course we need to go beyond it [the Washington Consensus]’ because it ‘did not contain all the answers to the questions of 1989, let alone that it addresses all the new issues that have arisen since then’. Williamson’s idea of going beyond the WC advocates the need of institutional reforms, but he is too willing to ignore the role industrial policy has played in the history of economic development. He goes on to emphatically reject the adoption of industrial policy whereby governments ‘pick winners’. Instead, a ‘cousin of industrial policy’ is welcome, namely a ‘Schumpeterian innovation . . . a national system of innovation’ (Williamson, 2004a:12).8

Beyond the Washington Consensus

51

One may agree that a national system of technological innovation would contribute to substantially improve Latin America’s economic performance, but it is hard to see how Schumpeter’s approach to economic development, with its emphasis on the role of the state and institutions as driving forces of the constitution of markets and industrialisation, can be made compatible with the microeconomic-behavior-based, Washington Consensus laissez-faire approach to growth. The latter assumes a minimal government. The ‘new agenda’ put forth in Kuczynski and Williamson (2003) and Williamson (2004a:5) acknowledges the importance of institutions, albeit it emphasises laissez-faire institutional reforms such as liberalisation of the labour market, privatisation, asset titling and the stipulation of property rights. Interestingly, the new agenda endorses countercyclical policies (allegedly) ‘à la Keynes’ and improvements in the region’s inequality in income distribution. The former requires targeting a budget balance over the business cycle; the latter calls for policies aimed at increasing the poor’s accumulation of human capital. Unlike John Williamson, the recent Schumpeterian literature considers that industrial policy does not necessarily thwart the market mechanism; an industrial policy,9 undertaken as a strategic collaboration or cooperative game between the government and the market, is still viable and reasonable, in particular for the enhancement of latecomers (cf. Aghion and Howitt, 2005; Cimoli et al., 2009a; Dosi et al., 1990; Rodrik, 2007). Industrial policy has been a built-in essential element of every successful development experience (Cimoli et al., 2009a). Williamson’s dismissal of industrial policy runs counter to historical evidence; developmental industrial policies have played a fundamental role in the process of structuring the market and have been part and parcel of institutional building in every historically observed catching-up process of economic development. Importantly, macroeconomic policies must be consistent with developmental industrial policy. As shown above, the free-market reform has induced a deleterious effect on investment, the income elasticity of imports and the level of economic activity. The macroeconomic environment as represented by the WC agenda involves supply-side incentives for entrepreneurship, but, paradoxically enough, it has destroyed industrial production capacity and the domestic economies’ ability to absorb, adapt and apply technological capabilities (Castaldi et al. 2009). The destruction of the domestic production chain is encapsulated in the recent sharp increase in the income elasticity of imports, an unconstructive deindustrialisation effect not seen in the Chinese case.10 The indisputable fact that sometimes industrial policy has been prone to waste, inefficiency and political corruption should not lead to the conclusion that industrial policy never works. The market mechanism may also fail. Privatisation in Latin America and elsewhere has often given rise to picking wrong winners and crony capitalism. As Rodrik (2007:151) correctly points out: It is not true that there is a shortage of evidence on the benefits of industrial policy . . . it is difficult to come up with real winners in the developing world that are not a product of industrial policies of some sort.

52

I. Perrotini-Hernández et al.

Industrial policy from this standpoint, challenges the principle of comparative advantage as conceived by mainstream economics. Inasmuch as industrial policy interacts with adequate institutions, it supplies public goods for improving technological capabilities and knowledge accumulation. Historical experience confirms that in most cases successful industrialisation did not follow a path of productive sectoral specialisation according to the logic of comparative advantage (Imbs and Wacziarg, 2003). Instead, it revolved around industrial policies targeting the acquisition of technological knowledge, highly skilled factors and capabilities that diversified the composition of aggregate output away from given input endowments and from ‘ricardian’ comparative advantages (Cimoli et al., 2009a; Wood, 1995). A Schumpeterian ‘cousin’ of industrial policy, that is, a national system of innovation presupposes that production occurs through a variety of institutions, namely the firm, the market and the government. The existence of market imperfections, externalities, public goods and increasing returns to scale furnish an economic rationale for government interventions through industrial policies and regulation. An adequate industrial policy can help reduce and bridge the technological and development gaps, while regulation frameworks minimise the harmful effects of externalities that the market mechanism either generates or is unable to internalise. Furthermore, it is hard to think of such a national system of innovation without the role of a Developmentist State acting to set up the clustering of diffusion processes of technological innovations resultant from microeconomic behaviour. The view that often the economic impact of a technological innovation is particular and specific in nature is a central feature of Schumpeter’s theory. Technical change remains within the realm of singular firms and/or industries for a certain amount of time before it is spread over a wider socioeconomic environment. If, owing to oligopolistic structures, asymmetric information, differential access to new technologies or effective demand constraints, the market mechanism itself fails to disseminate technical innovations over a broader dominion and at a faster speed, then public institution interventions can enhance technological diffusion, boost the potential economic impact of technical change and induce the multiplier effects of technical innovations, thus leading to higher growth rates. Clearly, only a combination of both technical innovation and institutional innovation can engender constellations of successful industrialisation and economic development. This is the manner in which Schumpeter’s dynamic approach (1934, 1943) envisaged the relevance of technological innovation in a theory of long-term growth, which is far away, we may contend, from the augmented WC interpretation. On the other hand, technological innovation increases productivity. Hence changes the demand for labour per unit of output and the requirement of capital and intermediate inputs in production. Therefore, aggregate supply and aggregate demand change accordingly. The newly augmented productive capacity and the effective demand must increase at the same rate, uniformly, for the economy to grow at full employment and full utilisation of actual productive capacity. However, by and large, demand and productivity grow at a different pace across

Beyond the Washington Consensus

53

industrial sectors and in aggregate terms, in which case economic growth will not necessarily involve full employment. Technological innovation, therefore, in this scenario, produces structural unemployment. The WC’s new agenda recommends labour market flexibility to cope with unemployment and slow growth, under the presumption that flexible wages will provide a full compensation mechanism of technologically induced unemployment. This hypothesis relies on the strong assumption of perfect and complete substitutability of labour and capital. Nonetheless, a number of stylised facts associated with the introduction of technological innovation rule out the prediction of endogenous compensation induced by labour market flexibility as assumed by the new agenda: differential labour skills between displaced labour force and the new input requirements; differential factor intensities between old and new products and industries; heterogeneous capital; changes in the composition of GDP after the introduction of technical innovation; price stickiness in goods markets; and the presence of asymmetric information in goods, labour and credit markets. All these stylised facts are pervasive economy-wide. Therefore, labour market flexibility is no sensible solution to overcome slow growth, structural unemployment, skewed distribution of income and social inequality in Latin America.

2.6 Final remarks The Washington Consensus recommended microeconomic flexibility, macroeconomic discipline and trade and financial liberalisation as a rational method of surmounting stagflation, exchange rate instability and balance of payments constraints. The past 20 years have witnessed that countries that adopted the WC agenda have performed poorly, and the economic prospect of Latin America in the context of a world economy engulfed by a severe financial crisis is certainly not buoyant. Despite the evidence of the last two decades, Williamson (2009:1) went on to state: The microeconomic advice that we have given and the report of the Spence Commission are as valid as ever. It remains sensible to use rather than fight the market, but to be prepared to alter a laissez-faire outcome when one of the classic conditions for market failure arises. These are: when monopoly is unavoidable, when externalities are important, and when the resulting income distribution offends social norms. The problem with the above conclusion is that such ‘classic conditions for market failure’ are ubiquitous in the modern economy; hence the need of an alternative strategy. The question to be asked is where to go from here. We have made the case for a developmental industrial policy. Another question to be raised is how to actually implement such a policy. The main thrust of the WC is fiscal discipline on the ground that budget deficits produce balance of payments crises. The orthodox fiscal policy mix (primary fiscal surplus and nominal fiscal

54

I. Perrotini-Hernández et al.

deficit) coupled with high interest rates resulting from targeting low inflation, is responsible for the stagnant economic path and the growing economic inequality observed over the last 20 years (Câmara and Vernengo, 2004:340, passim). A developmental industrial policy calls for a radical change in the practice of fiscal policy from the WC concept of a primary surplus-dominated macroeconomic discipline to a stabilising-investment-based fiscal policy regime. There is a fundamental contradiction in mainstream fiscal policy: in a context of financial liberalisation and inflation targeting, the primary surplus approach benefits bond holders, reduces public investment, polarises income distribution and hurts economic activity. Hence Williamson’s refusal of industrial policy should not come as a surprise; it is impossible to have both the primary surplus and developmental industrial policy simultaneously. A fiscal policy regime aimed at stabilising investment, output growth and the labour market, congenial with developmental industrial policy, puts forth public investment as the countercyclical tool par excellence. It needs not entail an explosive accumulation of debt, provided the central bank targets low interest rates – Keynes’s euthanasia of the rentier (Keynes, 1936:376) – rather than low inflation, and provided the monetary authority restricts capital movements.11 This alternative policy framework was known to Keynes in the 1930s, and appears to be consistent with Schumpeterian technological and institutional innovation.

Notes 1 To quote Adam Smith (1776 [1976, p. 7]: ‘The greatest improvement in the productive powers of labour, and the greater part of the skill, dexterity, and judgement with which it is any where directed, or applied, seem to have been the effects of the division of labour.’ 2 The WC view is that fiscal deficits crowd out private investment, cause recession, inflation and balance of payments crisis; hence the rejection of fiscal policy as a macroeconomic policy instrument and hence the prescription of perennial primary surplus. 3 Our data in Figure 2.1 are consistent with the growth patterns shown by Solimano and Soto (2005:9–12, passim). They also assert ‘a substantial slowdown in economic growth after 1980’ and identify a high number of ‘growth crises’ (defined as negative growth rates). Solimano and Soto coincide with Titelman et al. (2008), who argue that while the average frequency of shocks to Latin America’s terms of trade diminished from six in 1980–1990 to two in 2002–2006, financial shocks became more frequent between the 1980s and the 1990s. Most importantly, the latter’s magnitude increased from 0.7 per cent of GDP to 3.5 per cent. The authors conclude that during 1980–2006 adverse financial shocks claimed domestic absorption adjustments equivalent to almost –7 per cent of GDP, whereas the impact of terms of trade shocks on absorption declined from 2.25 per cent of GDP in 1980–1990 to 0.40 per cent in 1991–2001 and 0.00 per cent in 2002–2006. Clearly, capital account liberalisation has caused volatility of real economic activity to increase. 4

[L]ow, stable inflation is important for market-driven growth, and . . . monetary policy is the most direct determinant of inflation. Further, of all the government’s tools for influencing the economy, monetary policy has proven to be the most flexible instrument for achieving medium-term stabilisation objectives. (Bernanke et al., 1999:3)

Beyond the Washington Consensus

55

5 Pacheco-López and Thirlwall (2006) summarise Rodrik’s concept of growth accelerations as a positive discrepancy of two percentage points or more between eight years prior to the growth acceleration episode and eight years after the event, with a postacceleration growth rate of 3.5 per cent or higher. 6 As Rodrik (2004) put it: ‘The cold fact is that per capita economic growth performance [in Latin America] has been abysmal during the 1990s by any standards.’ 7 Our results are consistent with those in Pacheco-López and Thirlwall (2006). 8

A national innovation system is about government creating institutions to provide technical education, to promote the diffusion of technological information, to fund precompetitive research, to provide tax incentives for R&D, to encourage venture capital, to stimulate the growth of industrial clusters, and so on. (Williamson, 2004a:11)

9 Industrial policy is defined as a tool that targets specific industrial branches, sectors and firms with the aim of enhancing productivity, rapidly catching up with technological leaders, increasing employment in the formal labour market and accelerating output growth in line with some clearly defined national economic development goals (cf. Bresser-Pereira 2007; Cimoli et al., 2009a; Rodrik, 2007). 10 As Khan and Blankenburg (2009:367) put it, the main effect of liberalisation, across virtually all of Latin America, has been to reinforce Latin America’s commodity bias in the absence of any attempts at ‘Schumpeterian’ dynamic upgrading into higher-technology, higher-value added processes and/or products. Put differently, technological improvements have been limited to certain basic commodities, such as copper concentrates in Chile or iron in Brazil, but no attempts have been undertaken to upgrade to different processes (copper smelting) or product (steel). (emphasis in the original) 11 See also Câmara and Vernengo (2004) for a detailed discussion of the implications of Keynes’s pragmatic separation of the current budget and the capital budget in the practice of countercyclical fiscal policy.

References Aghion, P. and Howitt, P. (2005) ‘Appropriate growth policy: A unifying framework’, The 2005 Joseph Schumpeter Lecture, paper presented to the European Economic Association Congress, Amsterdam, 25 August. Bernanke, B., Laubach, T., Mishkin, F.S. and Posen, A. (1999) Inflation Targeting: Lessons from the International Experience, Princeton, NJ: Princeton University Press. Bresser-Pereira, L.C. (2007) Macroeconomia da Estagnação: Critica da ortodoxia convencional no Brasil pós-1994, São Paulo: Editora 34. Câmara, N.F.A. and Vernengo, M. (2004) ‘Fiscal Policy and the Washington Consensus: A Post Keynesian Perspective’, Journal of Post Keynesian Economics, vol. 27, No. 2, pp. 333–43. Castaldi, C., Cimoli, M., Correa, N. and Dosi, G. (2009) ‘Technological learning, policy regimes and growth: The long-term patterns and some specificities of a “globalized” economy’, in M. Cimoli, G. Dosi and J.E. Stiglitz (eds) Industrial Policy and Development: The Political Economy of Capabilities Accumulation, Oxford: Oxford University Press, pp. 39–77. Chang, H.J. (2003) Rethinking Development Economics, London: Anthem Press. Chang, H.J. (2008) Bad Samaritans, New York: Bloomsbury Press.

56

I. Perrotini-Hernández et al.

Cimoli, M., Dosi, G. and Stiglitz, J.E. (2009a) ‘The political economy of capabilities accumulation: The past and future of policies for industrial development’, in M. Cimoli, G. Dosi and J.E. Stiglitz (eds) Industrial Policy and Development: The Political Economy of Capabilities Accumulation, Oxford: Oxford University Press, pp. 1–38. Cimoli, M., Dosi, G. and Stiglitz, J.E. (eds) (2009b) Industrial Policy and Development: The Political Economy of Capabilities Accumulation, Oxford: Oxford University Press. Davidson, P. (2003) ‘What is wrong with the Washington Consensus and what should we do about it?’ paper presented at conference on ‘Reforming the Reforms: What Next for Latin America’, Rio de Janeiro, 25 July. Di Maio, M. (2009) ‘Industrial policies in developing countries: History and perspectives’, in M. Cimoli, G. Dosi and J.E. Stiglitz (eds) Industrial Policy and Development: The Political Economy of Capabilities Accumulation, Oxford: Oxford University Press, pp. 107–43. Dosi, G., Pavitt, K. and Soete, L. (1990) The Economics of Technical Change and International Trade, London and New York: Harvester Wheatsheaf, New York University Press. ECLAC (Economic Commission for Latin America and the Caribbean) (2002) Latin America and the Caribbean in the World Economy, 2000–2001, Santiago: ECLAC. ECLAC (Economic Commission for Latin America and the Caribbean) (2003) Latin America and the Caribbean in the World Economy, 2001–2002, Santiago: ECLAC. Ffrench-Davis, R. (2005) Reformas para América Latina: después del fundamentalismo neoliberal, Buenos Aires: CEPAL-Siglo XXI. Galindo, L.M. and Ros, J. (2006) ‘Alternatives to inflation targeting: Central bank policy for employment creation, poverty reduction and sustainable growth’, Working Paper no. 7, PERI, University of Amherst, MA, Amherst, September, pp. 1–25. Imbs, J. and Wacziarg, R. (2003) ‘Stages of diversification’, American Economic Review, vol. 93, No. 1, pp. 63–86. Ito, T. and Sato, K. (2007) ‘Exchange rate pass-through and domestic inflation: A comparison between East Asia and Latin American Economies’, RIETI Discussion Paper Series 07-E-040 (May 2007): 1–45. Keynes, J.M. (1936) The General Theory of Employment, Interest and Money, New York: Harcourt and Brace. Khan, M.H. and Blankenburg, S. (2009) ‘The political economy of industrial policy in Asia and Latin America’, in M. Cimoli, G. Dosi and J.E. Stiglitz (eds) Industrial Policy and Development: The Political Economy of Capabilities Accumulation, Oxford: Oxford University Press, pp. 337–77. Krugman, P. (1995) ‘Dutch Tulips and Emerging Markets’, Foreign Affairs, vol. 74, No. 4, pp. 36–7. Kuczynski, P.P. and Williamson, J. (eds) (2003) After the Washington Consensus: Restarting Growth and Reform in Latin America, Washington, DC: Institute for International Economics. Lavoie, M. and Seccareccia, M. (eds) (2005) Central Banking in the Modern World: Alternative Perspectives, Cheltenham, UK, Northampton, MA: Edward Elgar. List, F. (1885) The National System of Political Economy, translated from the original German edition published in 1841 by Sampson Lloyd, London: Longmann, Green and Co. McKinnon, R. (1973) Money and Capital in Economic Development, Washington, DC: Brooking Institution.

Beyond the Washington Consensus

57

McKinnon, R. (1991) The Order of Economic Liberalisation: Financial Control in the Transition to a Market Economy, Washington, DC: Brooking Institution. Moreno-Brid, J.C. (1999) ‘Mexico’s Economic Growth and the Balance of Payments Constraint: A Cointegration Analysis’, International Review of Applied Economics, vol. 13, No. 2, pp. 149–60. Ocampo, J.A. (2004) ‘Latin America’s Growth and Equity Frustrations during Structural Reforms’, Journal of Economic Perspectives, vol. 18, No. 2, pp. 67–88. Ohlin, B. (1933) Interregional and International Trade, Cambridge, MA: Harvard University Press. Oreiro, J.L., Coelho Squeff, G. and de Paula, L.F. (2008) ‘A post Keynesian proposal for a flexible institutional arrangement of inflation targeting regime in emerging economies’, March, paper prepared for the workshop ‘Inflation Targeting: Is There a Credible Alternative?’ at Balliol College, Oxford, 4 April 2008, pp. 1–24. The workshop is organised by the Post Keynesian Study Group, United Kingdom. Pacheco, P. and Thirlwall, A.P. (2006) ‘Trade Liberalisation, the Income Elasticity of Demand for Imports and Growth in Latin America’, Journal of Post Keynesian Economics, vol. 29, No. 1, Fall, pp. 41–61. Palley, T.I. (1996) Post Keynesian Economics: Debt, Distribution and the Macroeconomy, London: Macmillan. Prasad, E.S. and Rajan, R. (2008) ‘A pragmatic approach to capital aAccount liberalization’, Working Paper 14051, NBER, Cambridge, MA. Ricardo, D. (1821 [1951]) On the Principles of Political Economy and Taxation, edited by Piero Sraffa with the collaboration of M.H. Dobb, vol. I of the Works and Correspondence of David Ricardo, Cambridge: Cambridge University Press. Rochon, L.P and Rossi, S. (2006) ‘Inflation Targeting, Economic Performance, and Income Distribution: A Monetary Macroeconomic Analysis’, Journal of Post Keynesian Economics, vol. 28, No. 4, pp. 615–38. Rodrik, D. (2004) ‘Rethinking growth strategies’, WIDER Annual Lecture 8, Helsinki: United Nations World Institute for Development Economic Research. Rodrik, D. (2007) One Economics Many Recipes: Globalization, Institutions, and Economic Growth, Princeton, NJ: Princeton University Press. Rojas-Suárez, L. (2003) ‘Monetary policy and exchange rates: Guiding principles for a sustainable regime’, in P.P. Kuczynski and J. Williamson (eds) After the Washington Consensus: Restarting Growth and Reform in Latin America, Washington, DC: Institute for International Economics, pp. 123–55. Ros, J. (2000) Development Theory and the Economics of Growth, Ann Arbor, MI: University of Michigan Press. Saavedra, J. (2003) ‘Labor markets during the 1990s’, in P.P. Kuczynski and J. Williamson (eds) After the Washington Consensus: Restarting Growth and Reform in Latin America, Washington, DC: Institute for International Economics, pp. 213–63. Schumpeter, J.A. (1934) The Theory of Economic Development, Cambridge, MA: Harvard University Press. Schumpeter, J.A. (1943) Capitalism, Socialism, and Democracy, New York: Harper & Row. Shackle, G. (1983) The Years of High Theory, Cambridge: Cambridge University Press. Shaffaedin, S.M. (2005) ‘Trade liberalization and economic reform in developing countries: Structural change or de-industrialisatioon?’ UNCTAD Discussion Papers, No. 179, pp. 1–25. Smith, A. (1776 [1976]) An Inquiry into the Nature and Causes of The Wealth of Nations, Chicago, IL: University of Chicago Press.

58

I. Perrotini-Hernández et al.

Solimano, A. and Soto, R. (2005), ‘Economic growth in Latin America in the late 20th century: evidence and interpretation’, in Serie Macroeconomía del Desarrollo Num. 33, Santiago, Chile: CEPAL, February, pp. 1–44. Taylor, L. (1991) Income Distribution, Inflation, and Growth: Lectures on Structuralist Macroeconomic Theory, Cambridge, MA: MIT Press. Thirlwall, A.P. (1979) ‘The Balance of Payments Constraint as an Explanation of International Growth Rate Differentials’, Banca Nazionale del Lavoro Quarterly Review (March), pp. 45–55. Titelman, D., Pérez-Caldentey, E. and Minzer, R. (2008) ‘Comparación de la dinámica de los efectos de los choques financieros y los choques de términos del intercambio en América Latina en el período 1980–2006’, Working Paper, ECLAC, Santiago, Chile. Weisbrot, M. (2006) ‘Latin America: The End of an Era’, Washington, DC: Center for Economic and Policy Research, pp. 1–23. Wicksell, K. (1898 [1965]) Interest and Prices: A Study of the Causes Regulating the Value of Money, New York: Augustus Kelley. Williamson, J. (1990) ‘What Washington means by policy reform’, in J. Williamson (ed.) Latin American Adjustment: How Much Has Happened? Washington, DC: Institute of International Economics, pp. 5–38. Williamson, J. (2003) ‘Overview: An agenda for restarting growth and reform’, in P.P. Kuczynski and J. Williamson (eds) After the Washington Consensus: Restarting Growth and Reform in Latin America, Washington, DC: Institute for International Economics, pp. 1–19. Williamson, J. (2004a) ‘A short history of the Washington Consensus’, paper commissioned by Fundación CIDOB for a conference ‘From the Washington Consensus towards a New Global Governance’, Barcelona, 24–25 September. Williamson, J. (2004b) ‘The Strange History of the Washington Consensus’, Journal of Post Keynesian Economics, vol. 27, No. 2, pp. 195–206. Williamson, J. (2009) ‘The Washington Consensus and the global crisis’, paper presented at a conference sponsored by the Johns Hopkins School of Advanced International Studies and the Center for Global Development, ‘New Ideas in Development Finance after the Financial Crisis’, Speeches, Testimony and Papers, Peterson Institute for International Economics, Washington, DC, 22 April. Wood, A. (1995) North–South Trade, Employment and Inequality: Changing Fortunes in a Skill-Driven World, Oxford: Clarendon Press.

3

Foreign trade and per capita income New evidence for Latin America and the Caribbean Humberto Ríos-Bolívar and Omar Neme-Castillo

3.1 Introduction Given the growing importance of goods and services inside global exchanges during the last decades, analysis of relationship between international trade and economic growth takes a central role on present empirical debate. In theoretical scope, neoclassic economics shows out that countries’ share of foreign trade could become a positive strength for stimulating economic dynamism. However, at empirical level, there is not a consensus about the impact of that share and, in most cases, it has only found static effects with limited significance on that relationship. In Mexico and Latin America there are some studies in this sense, nevertheless proxies used for both growth and control variables are limited or exclude important aspects for growth process, such as human capital or imports of goods which enlarge production capacities. The aim of this chapter is twofold. First, it derives a regression equation directly from a neoclassical growth model incorporating human capital and international trade elements following Bidlingmaier’s (2007) model. Second, it applies this equation to the case of Latin America and the Caribbean countries in the period 1970–2007 using a panel data analysis. For this purpose, in the next section aspects of endogenous growth theory concerning international trade are considered with particular attention to underdeveloped countries. These fundamentals are incorporated into an extended neoclassical growth model in the third section of the chapter. The model includes measures of trade openness different from those commonly used, thus establishing whether or not there is a link between international trade and per capita income level, and to directly study the channels through which this effect occurs. The fourth section estimates the mentioned equation in order to test the hypothesis that establishes a positive relationship between international trade and per capita income level differentiated among the considered countries in the region. Finally, conclusions are presented with some implications for international trade theory and for commercial policy.

60

H. Ríos-Bolívar and O. Neme-Castillo

3.2 Theoretical aspects Traditional theory of international trade displays possible benefits of goods exchange among countries, which derives from production specialisation (Ricardian model). If countries specialise according to their comparative advantage, efficient resource allowance would be encouraged; and therefore, welfare of trading countries improves. However, this is only a level effect on the consumption possibilities, that is, it is a ‘one-time increase’. After achieving complete specialisation, productivity will not rise again.1 Therefore, this theory is limited in its explanations of the output growth rate. More recent theories suggest different effects of international trade. In a context, where leading companies enjoy scale economies, the classical explanation of trade and its welfare prediction does not hold any more (Gomory and Baumol, 2001). Consequently, small economies that opened to international trade in a late or slow way and did not acquire required scale may not be able to compete successfully against leading economies, as is the case of most of Latin America and the Caribbean countries. Another argument that limits the validity of classical international trade theory is the possibility of disadvantages arising from increasing specialisation, particularly for developing countries like those of Latin America. If these countries specialise in sectors with lower productivity growths or with low-income elasticity of demand (primary and traditional manufacturing), growth rates will always be below of those of industrialised countries and economic inequality will be widened, falling into what is known as ‘trap of specialisation’ (Redding, 1999). Besides, Lucas (1988) developed a model of human capital accumulation with external effects arising from learning processes, noting that human capital is not only formed by investing in education but also through ‘learning-by-doing’ or training. The model can achieve different growth rates between countries with no differences in productivity rates. If the goods are substitutes, the international exchange of goods increases the comparative advantage through ‘spillover’ effects on labour, restricting countries’ growth with lower learning rates. In the model, ‘learning-by-doing’ process shows diminishing returns. This means that high rates of learning and of human capital formation, and the subsequent economic growth, can only be maintained by reallocating resources to new activities or through developing new products. With diminishing learning rates and with no sector showing permanently higher learning rates than any other sector, a ‘catching up’ of the poor to the rich countries can occur. Nevertheless, if trade liberalisation leads developing countries to import goods of higher quality without producing locally, then learning and growth rates in those countries will be lower; but in contrast, if imports are mainly capital goods used to produce goods of higher quality or goods that, in turn, allow to improve productivity level, then the learning and growth rates will be higher. Likewise, Lucas (1993) pointed out another type of human capital formation based on the idea of global public good and, particularly, on knowledge. He

Foreign trade and per capita income 61 assumes that knowledge diffuses throughout the world, through scientific publications, reverse engineering, blueprints and others. Knowledge can be acquired not only through own efforts of each agent but also through global knowledge spillover. The nature of knowledge is no rivalry and no partial exclusion, which means it can be applied by different people at the same time without incurring additional costs. Economic operators can use new ideas even without contributing to global research and development (R&D) efforts. These knowledge properties allow countries with low human capital stocks to accumulate some of that knowledge quickly through its international adoption and, ultimately, catching up with richer countries. Moreover, Romer (1990) proposed a model with differentiated capital goods, where output level is determined by the number of varieties produced by the domestic economy. The model assumes that capital goods are not perfect substitutes and are produced in the middle sector which uses patents, obtained from R&D processes in the technological sector, as input which, in turn, uses human capital and knowledge for the development of these patents. Because knowledge can be used without cost by any agent through spillover effects, the R&D sector does not exhibit diminishing returns. Thus, the higher the stock of human capital and its use in the R&D sector, the greater the growth rate of an economy. The model predicts positive outcome of international trade. Trade liberalisation both generates level effects and growth effects. The growth effects are reached if knowledge and technology transfer are free of additional costs. One aspect of relevance to Latin American and the Caribbean countries is that with a small stock of human capital and limited ability to employ such trained personnel in the R&D sector restrict its growth, leading to production of only a small range of capital goods (Rivera-Batiz and Romer, 1991) and, thus, establishing the basis for their exclusion from the international economic scenario. This model has two implications for international trade and mainly for the countries in the Latin American region. First, the number of varieties of differentiated capital goods and total output would increase. In countries with different factor endowments, trade in goods increases the growth rate of the economy that grew to a lesser extent and reduces the rate of the most advanced country. Second, if it is true that ideas flow freely worldwide then stock of knowledge in each country that could be employed in the research sector would increase, stimulating the growth of technology and the economy as a whole. Moreover, increasing human capital productivity in the R&D sector would employ more of this factor, boosting even more the growth rate. Therefore, to the extent Latin American countries are able to establish centres and programmes of applied scientific research that use those ideas, per capita income of those countries would be encouraged. Finally, it must be noted that the set of knowledge, procedures, techniques, etc., specific to each firm, generated by persistent R&D activities within the organisations, find greater barriers to spread both to the rest of the domestic economy of the firm carrying out R&D activities, as well as to foreign economies. This idea is supported by the eclectic approach of foreign direct investment (FDI). That is,

62

H. Ríos-Bolívar and O. Neme-Castillo

firms enjoy a property advantage, because they have created that specific knowledge. Moreover, it is inferred that profitability of operating in foreign markets is greater to the sale or licensing of that technology exclusive for the firm, then efficiently internalising that technological advantage. In other words, part of knowledge, the most advanced or strategic, generated by firms that invest productively abroad is excludable in higher grade than the knowledge generated by other organisms (for instance, private research institutions). However, marginal product of innovative activity in the aggregate domestic economy grows and, that while this new knowledge is not totally spread to the rest of economy, it must be recognised that it has impacts on production.

3.3 Model derivation This section provides a model that incorporates the essence of theoretical elements noted above. The model begins from the neoclassical basic production function, which defines output (Y) as a function of capital (K) and labour (L); Y = Ka(A×L)b, with α + β = 1. Technological progress (dA/dt) is labour augmenting and Y exhibits constant returns in all production factors. Population growth is given by the exogenous rate n. Capital accumulation is financed by domestic savings (exogenous saving rate, s = S/Y), that is, dK/dt = I = S = s·Y. Production function is extended by incorporating human capital (H) and international trade, understood as capital goods imports (Z) and the participation of multinational companies (MNCs) operating in countries of the region (W), through their stock of knowledge: (3.1) The idea is that capital goods imports imply a direct transfer of technology. Developing countries with limited capacities for production of capital goods may import these goods from technologically advanced countries. Imported capital stock is modelled as a separate production factor because it is financed through income and net exports and it likely shows higher levels of productivity than domestic capital (Bidlingmaier, 2007). In this sense, exports impact on income only through savings that finance capital goods imports. The change in imports of that kind or goods in time allows imported capital accumulation (M), which is formulated as: dZ/dt = M = NX = x·Y, where NX is net exports, x the participation of net exports in total output (x = NX/Y); XN and x refer only to capital goods. Capital stock directly imported is increasing in M financed by NX. Also, the inclusion of a trade integration variable (W) is based on arguments similar to that of the incorporation of Z. Technological capital stock of developing countries is limited. The latter are integrated economically to industrialised countries with higher production and technological capacities. Particularly, through productive investment flows and a strong commercial relationship, their domestic production capacities and, ultimately, their output will increase.

Foreign trade and per capita income 63 Thus, integration represents two options. First, national firms within the value chain of MNCs that invest abroad, update their technological capacities, generating a ‘catching up’ effect on the domestic economy. Second, stronger trade links between these economies mean that MNCs from an advanced country that employ knowledge, processes, organisational methods, distribution networks, etc., in domestic production, impact to some extent on product growth rate, without necessarily disseminating that knowledge and, therefore, without contributing to human capital formation of the domestic economy. W is included in the model because it depends on foreign investment in capital formation and in systematic R&D activities, which reflects higher productivity levels than the domestic ones. W is understood as expenditures made by other countries in order to build their production capacities, which are nevertheless exported, at least partially, to less developed economies through FDI and represents the available stock that helps them to produce more efficiently, without incurring additional costs. W is interpreted as the international knowledge spillover restricted to related MNCs. That is, knowledge of these firms, or part of them, is limited; it does not flow either with speed or with magnitude one could think if it were a pure public good. Knowledge spillover is partially restricted to links between matrix and subsidiaries of MNCs, which affect the production capacity of host economies of FDI and, ultimately, their growth rate. Thus, trade integration affects income by two different ways. First, to the extent agents in the foreign economy are capable to accumulate physical and technological capital financed by savings. Second, by FDI which operates through MNCs in the domestic economy, namely, to the extent that activities of such companies grow in relation to the national market, firms will import part of this capital stock from their country of origin, with a growing effect in host country output. Formally, dW/dt = (dK/dt)·Λ, where s and Y are the rates of savings and income of the foreign economy (i), respectively, which determine technological capital accumulation in that country; Λ = (FDIij/VDj), is the diffusion rate of this knowledge abroad, where FDIij is foreign direct investment from country i to j and VDj is domestic sales in j. Thus, the second term on the right side represents the share of MNCs that use their ‘know how’ to produce domestically. The magnitude of the ‘spread’ of physical and technological capital by MNCs in foreign markets depends on accumulated amount available in their country of origin. The stock of foreign capital ‘imported’ indirectly is increasing in both terms. On the other hand, human capital can be accumulated by three different ways: education investment, ‘learning-by-doing’ and international knowledge spillover. In the first case, human capital formation through investment in education implies that the share of income a country invests in education infrastructure creates human capital (H). The same production function of physical capital applies to human capital (Mankiw et al., 1992). H is accumulated through investment in education (Ih) which is financed by saving (Sh), that is, dH/dt = Ih = Sh = sh·Y.

64

H. Ríos-Bolívar and O. Neme-Castillo

In the second case, H is accumulated through learning-by-doing during the process of production of each good that faces diminishing returns. The only way to keep learning at high levels, and thus H formation, is through the continued introduction of new differentiated goods. This idea is incorporated into the model by multiplying the stock of human capital formed by investment in education, sh, by an index of differentiated goods produced domestically (p): dH/dt = s·Y·(1 + p), with 0 < p < 1. The more differentiated goods an economy produces, the greater the value of p and, hence greater H. The third option for H formation is international spillover. With the advance of information and communication technologies knowledge is disseminated beyond national borders. Countries that do not make persistent efforts in R&D can benefit from new ideas, processes or inventions through knowledge dissemination. Thus, the more technologically old-fashioned a country is, the greater the possibility of adapting existing ideas through imitation. This offers the possibility of ‘catching up’ with countries in the technological frontier by a faster human capital accumulation. This represents a growth strategy for Latin American countries. This concept is integrated into the model assuming that absorptive capacity of knowledge of a country allows accelerated human capital accumulation. Technically, this is done similarly to ‘learning-by-doing’ by adding the knowledge diffusion factor (m): dH/dt = s·Y·(1 + p)·(1 + m), with, 0 < p < 1 and 0 < m < 1. However, there are barriers to absorption of knowledge between countries. A weak communications infrastructure can hinder the flow of knowledge. Internet access is essential for disseminating global knowledge.2 When an economy has strong barriers to dissemination and application of global knowledge, W takes greater importance since the hypothesis of restriction on the dissemination of knowledge and technology within MNCs and related companies seems to be accepted. The inclusion of variables related to product development (p) and knowledge diffusion (m) allows considering new channels for H accumulation in a neoclassical approach. Multiplying these variables by traditional variables of human capital investment (Ih) provides a basis for a more rapid accumulation of H and, consequently, for a higher growth rate. Thus, in this context the steady-state level of H depends on both the willingness to save for investing in education, and on the ability to develop new products and the knowledge absorption capacity. This model explains in detail the level of production compared to traditional neoclassical model; the basic production function is extended in order to include three types of capital in addition to physical capital (K); human capital (H), imported capital (Z) and, unlike Bidlingmaier’s (2007) model, technological capital ‘available’ from abroad (W).3 Thus, from equation (3.1) the income per unit of efficient labour is obtained:4 (3.2)

Foreign trade and per capita income 65 Assuming the same depreciation rate (δ) for all types of capital, domestic or foreign, a growth rate of population (n) and constant growth of technology (g), the evolution of capital stocks is given by: (3.3) (3.4) (3.5) (3.6) The additional increase in capital stock is offset by the depreciation, population growth and technological progress (n + g + δ). The steady-state values are:5 (3.7)

(3.8)

(3.9)

(3.10)

Substituting these equations into (3.2) gives the equation of income per unit of effective labour of equilibrium and, after applying the logarithmic function for the econometric estimation, obtains:

(3.11) where x = 1 – α – β – γ – η. However, since there are no data of income per unit of effective labour, the previous expression is rewritten as a per capita income equation. Thus, as y = Y/AL = ŷ/A ( ŷ: per capita income), and considering that ln( y) = ln( ŷ) – ln(A), it yields:

66

H. Ríos-Bolívar and O. Neme-Castillo

(3.12) The technology level, A, is expressed as the product of initial stock, A0, and the exogenous growth rate, g, and a country specific shock, namely, A = A0·egt+ε. A0 is country-specific factors such as resource endowments, institutions, business environment, legal framework, climate, etc. With this, the above equation can be rewritten as:

(3.13) Equation (3.13) specifies the factors that determine the level of per capita income of steady-state: initial stock of technology (A0); exogenous technological growth rate (g); domestic savings for investment in education (sh) and in capital (sk); product innovation (1 + p); international knowledge spillover (1 + m) that increase human capital accumulation; net exports (ex) that finance capital accumulation; foreign savings for investment in technological capital (sw) that stimulate the restricted transfer of productive capacities through trade integration; population growth (n) with negative effect on the level of per capita income; and capital depreciation (δ). This specification focuses on the contribution of international trade to accumulation. Trade liberalisation, expressed in (1 + m) and (sw), impacts on accumulation of physical and human capital through knowledge that is disseminated on the whole economy and that is transferred from abroad through multinational companies towards the domestic economy. Thus, the level of per capita income depends on the determinants established in equation (3.13) that are econometrically estimated in the next section for the sample of Latin American and the Caribbean countries considered in this study.

3.4 Empirical application Equation (3.13) is estimated following the panel data technique. This equation is a panel regression model with a general structure that has the form: yit = αit + βkitXkit + uit, where i refers to countries (i = 1, 2, . . ., 21), t represents time (t = 1, 2, . . ., 31), k is independent regressors and uit is the error term given by uit = μt + υit. Where uit is the unobserved individual specific effect, and υit is the rest of the traditional error, identically and independently distributed, that is, iid ~ (0, σ2υ). The dependent variable, yi, is a vector of per capita income in equation (3.13).

Foreign trade and per capita income 67 Similarly, the intercept αit, represents the stock of autonomous technology, ln(A0). Xkit is a matrix containing all independent variables, k, for each Latin American and the Caribbean country. In this case, these variables are the logarithms of the growth rates of population and technology plus the rate of depreciation (n + g + δ), of capital accumulation (sk), accumulation of ‘freely’ imported capital (ex), the accumulation of human capital through education (sh), learningby-doing (1 + p), knowledge transfer (1 + w) and capital accumulation in the foreign countries that spreads ‘exclusively’ through related companies in different countries (sw). Several estimation methods can be used, depending on the variation of the intercept and the coefficients in time and cross-section units. This chapter considers three cases. The first assumes that the intercepts and coefficients are invariant over time and cross-sections (yi = α + βkXkit + uit), which represents the case of pooled data. The second case allows cross-section units have different intercepts (yi = αi + βkXkit + uit). The third possibility considers countries have different intercepts and, further, that they respond in different ways to different independent variables (yi = αi + βkiXkit + uit). Estimating these three options is of interest because it first considers the restrictive assumption of the same technology and, after considering the possibility that variables do not affect per capita income in each country in the same way, it reflects the possibility of heterogeneity in the explanatory variables. In this regard, the assumption of a common intercept for all countries within the panel data regression is restricted because it prevents an estimation of the impact of different initial level of technology (A0) for the countries in our sample. This hypothesis considers that A0 is the same for all countries; which is amply questionable because factors like resource endowment, climate or institutional features could be included in this term. The estimates are done in two ways. On the one hand, by the fixed effects model (FEM) that introduces a dummy variable for each country that incorporates individual effects. On the other, through the random effects model (REM), that divides the error term in a single term, which captures the variance of the dependent variable and a common term, caused by individual random-effect in each country.6 3.4.1 Data The study was performed to a set of 21 Latin American and Caribbean countries (see Table 3.1). The period considered is 1977–2007 for a total of 31 years. The sample is a balanced panel. The data on GDP per capita (pcgdp) were taken from the Penn World Table version 6.2. All terms are expressed by purchasing power parity. Data of population growth were obtained from World Development Indicators (WDI) 2009. Following Mankiw et al. (1992), it is assumed that rates of depreciation and of technology growth together are approximately 5 per cent (g + δ = 5 per cent), which are added to n in order to form the variable popdr. The expected sign of this variable is negative. In addition, the gross capital formation

68

H. Ríos-Bolívar and O. Neme-Castillo

Table 3.1 Latin America and the Caribbean countries considered in the sample Observation 1 2 3 4 5 6 7 8 9 10 11

Country

Code

Observation

Country

Code

Argentina Belize Bolivia Brazil Chile Colombia Costa Rica Dominican Republic Ecuador El Salvador Guatemala

arg bel bol bra chi col cr repd

12 13 14 15 16 17 18 19

hon jam mex nic pan par per tyt

ecu elsal gua

20 21

Honduras Jamaica Mexico Nicaragua Panama Paraguay Peru Trinidad and Tobago Uruguay Venezuela

uru ven

Source: own elaboration.

as a percentage of GDP (gkf_gdp) is used to approximate the investment in physical capital stock (sk). The data were taken from WDI 2008. It is expected that the sign of the estimated coefficient should be positive. The indicator of the rate of accumulation of directly imported physical capital (ex) is mm_gdp, and was constructed by dividing manufacturing imports by aggregate GDP. Manufactured imports include sections 5 (chemicals), 6 (basic manufactures), 7 (machinery and transport equipment) and 8 (miscellaneous manufactured goods) of the Standard International Trade Classification revision 1 (SITC). The series of imports were obtained from the database BADECEL of the Economic Commission for Latin American and the Caribbean (ECLAC) and were combined with those of the UN Comtrade Data Base of the United Nations (UN) for countries and years that were missing.7 Besides, the literacy rate (litr) is considered as an approximation of H formation. Clearly, the expected sign should be positive. The data are in the WDI 2008. The second form of human capital accumulation is through ‘learning-bydoing’, which is approximated by an index of product discovery. The discovery does not necessarily mean inventing a new product, but when a country begins to produce a new good inside its territory, regardless if it was already produced elsewhere, a learning process starts that creates domestic human capital. Given the limited data for the whole set of countries that allows to calculate a rate of introduction of new goods or data related to straight evidence of new goods produced, this study follows the Bidlingmaier (2007) proposal, which employs exports as proxies of ‘learning-by-doing’. The idea is that a good which is systematically exported means a ‘permanent’ discovery that allows the accumulation of knowledge. Thus, the higher the export level, the greater the formation of H.8 Thus, to identify new discoveries we used the series of exports of SITC revision 1 disaggregated to four digits (about 1,100 groups of goods), obtained from

Foreign trade and per capita income 69 BADECEL and the UN Comtrade Data Base. The methodology to be assumed as a new discovery with lasting effects is to identify each product that has been exported for the first time with a value greater than or equal to $100,000 and that they have been exported at least during two consecutive years. The variable is labelled dis and is normalised to obtain values between zero and one. The closer to the unit dis is, the greater the expected impact on pcgdp. The third concept of H is knowledge transfer, associated with the notion of absorptive capacity of a country. It uses two types of proxies of absorptive capacity: (1) the number of telephone lines and cellular, Internet users and number of personal computers (indicator of infrastructure) and (2) coefficient of openness (total trade share in GDP). Data were collected from Social Statistics and Indicators (BADEINSO) of ECLAC. The variables were normalised between zero and one, and combined into an index of knowledge absorption capacity (ikac). Finally, to partially represent excludable knowledge that is appropriated by the companies with R&D activities and that only spread to related companies abroad, a technological capital stock is built for a set of economies with which the listed countries maintain relevant economic relations. The argument is that part of technological capital accumulation in advanced economies spreads out to economies in the region through MNCs.9 That indicator is constructed by weighting the foreign technological capital stock by the share of each country’s trade in total trade of a foreign country in order to obtain ‘available’ foreign technological capital.10 The latter is multiplied by the share of FDI, which flows from these economies, in the value of domestic sales. Thus the stock of ‘available’ foreign technological capital (saftk) entering the domestic economy is obtained through FDI. Formally: (3.14) where stkj is the stock of technological capital in foreign country j, Cij is the total trade between the i home countries and the j foreign country, Cj is the total international trade of j, FDIji is foreign direct investment flowing from j to i and DSi is the domestic sales of i. The data on R&D investment of foreign economies are derived from the OECD Stan Data Base and are complemented by data from WID 2008 and national offices. The series of FDI statistics are taken from UNCTAD and domestic sales of BADEINSO of ECLAC and national statistical offices.11

3.5 Results Three different options were estimated using the pooled, fixed and random effects models. The Hausman test indicates that the best model is the fixed effects (eliminating the random effects model). We report the results for the following estimates: (1) the pooled model (Model I), (2) different constants but with common coefficients (Model II) and (3) different constant and intercepts

70

H. Ríos-Bolívar and O. Neme-Castillo

(Model III). Robust results were obtained for the latter two cases, after corrections to address autocorrelation and heteroskedasticity problems. Although panel data models tend to show lesser problems of multicollinearity among the explanatory variables, the possible presence of multicollinearity was assessed using the correlation coefficient matrix. In 40 per cent of relationships, the highest correlation coefficient is less than 0.50. Further, only in 45 per cent of relationships, half of the coefficients is above 0.50, and less than 10 per cent is higher than 0.85. Then, it is accepted that there are no serious problems of multicollinearity: the correlation coefficient is less than 50 per cent for 40 per cent of the relationships, above 50 per cent in less than 25 per cent of the relationships and higher than 85 per cent for less than 10 per cent of them. Therefore, we conclude that there are no serious problems of multicollinearity. Given that panel data estimations often violate the condition of independent errors and identically distributed with constant variance, an autocorrelation and heteroskedasticity test was conducted. We applied the Wooldridge test for diagnosis of autocorrelation, whose null hypothesis is the non-existence of autocorrelation. In fact, this test rejects the null, indicating that there is a problem of autocorrelation.12 Also, to determine the presence of heteroskedasticity two different tests were considered. First, the Modified Wald test for heteroskedasticity test indicates the rejection of the null hypothesis of constant variance and, therefore, the existence of a heteroskedasticity problem.13 This result was confirmed by the LM test suggested by Greene (2000), which rejected the null of no heteroskedasticity. In order to solve both problems encountered, transformation of Prais–Winston by country was used following the Panel Corrected Standard Errors Method (PCSE). In this respect, Beck and Katz (1995) have demonstrated that standard errors of PCSE are more accurate than errors estimated with Feasible Generalised Least Squares (FGLS). After this correction the problems were eliminated. In general, the explanatory power of the three models is high, implying that other factors not included in the model have limited effects on the income levels of Latin American and the Caribbean countries. On average, the regressors considered explain more than 91 per cent of income changes. Results of the estimations (1) and (2) are shown in Table 3.2. All variables in both models have statistical significance but not all of them show the expected sign. In Model I, fixed effects with common constants and slopes across countries, all variables are significant at 99 per cent. In Model II, fixed effects with equal slopes but differentiated constants across economies, all variables are significant at 99 per cent except the literacy rate which is significant at 95 per cent and the rate of population growth, which is not significant. In Model I, for all countries, population growth, technological progress and depreciation jointly affect negatively the per capita income, as expected. By contrast, the sign found in Model II is positive, though with not statistically significant contribution. The accumulation of physical capital is an element that positively determines this variable in both models. In the first model, it is by far the variable that contributes the most to income; in the second specification it

8.0474* (0.0523) –0.3701* (0.0188) 0.3060* (0.0242) –0.1032* (0.0115) –0.0711* (0.0037) –0.0675* (0.0049) 0.2349* (0.0128) –0.0179* (0.0021) 0.844 503.27 651

I

Models

8.4405* (0.0176) 0.0190** (0.0090) 0.0467* (0.0110) 0.4057* (0.0099) –0.0019*** (0.0013) –0.0080* (0.0014) 0.1629* (0.0053) 0.0084* (0.0012) 0.993 3264.49

II arg bel bol bra chi col cr rdom ecu elsa gua

Country (code) 9.6643 8.2154 7.4571 9.3644 8.9427 8.3309 8.5613 8.4048 8.0375 7.9374 7.7354

0.0585 0.0791 0.0360 0.0468 0.0510 0.0268 0.0325 0.0475 0.0358 0.0375 0.0346

Differentiated SE constants1 hon jam mex nic pan par per tyt uru ven

Country (code) 7.4955 8.1606 9.1042 7.2945 8.7266 7.8788 8.4592 9.3529 9.1477 8.9786

Differentiated constants1 0.0404 0.0331 0.0403 0.0395 0.0506 0.0646 0.0433 0.0566 0.0440 0.0398

SE

Notes 1 Differentiated coefficients by countries in accordance with Model II and all constants are significant at 1%. Model I: same constants and coefficients for all countries, and Model II: differentiated constants. Model I: pooled data model, and Model II: fixed effects model with different intercepts. *, ** and *** are the levels of significance at 1%, 5% and 10%, respectively. SE: standard errors that appear between brackets and below coefficients.

Source: own elaboration.

R2 adjusted F Observations

saftk

ikac

dis

litr

mm_gdp

gkf_gdp

popdr

C

Variable

Table 3.2 Panel data regressions of per capita income

72

H. Ríos-Bolívar and O. Neme-Castillo

appears as the third variable with the largest contribution. In general, this reflects the central role of capital accumulation in the level of per capita income. The coefficient of foreign physical capital, employed in the domestic economy and purchased directly, has an ambiguous effect on income. In Model I, the sign is contrary to the expected one, which means import of capital goods has a negative effect on income levels but relatively low level. This may be due to two reasons: (1) countries with low levels of development (low per capita income) have limited capacity to absorb the new technology (low human capital) and this situation is greater than the positive effect on the more developed countries and (2) the variable is not a good proxy for physical capital, so it can be replaced by more specific variables such as more disaggregated sections of SITC included in the manufacturing imports. However, the results using OLS estimates do not produce the best linear unbiased estimator, due to the specific individual effects, which unequally affect individual countries and are invariant over time. For this reason the model is estimated by fixed effects.14 In contrast, for Model II, capital goods imports are, as expected, significantly positive for income. Slightly less than half of the efforts of technology purchasing from abroad are reflected in the income of each country. Therefore, no matter the size of each economy or of individual capacities of countries, the use of more advanced technology seems to bring positive effects for the economic development of countries. In any case, the relationship between these two variables should be examined more extensively to eliminate the degree of ambiguity found in these two models. Furthermore, increase of literacy rates, contrary to expectations, negatively affects the income level of Latin American and the Caribbean countries, suggesting that favourable impact of human capital and in particular the level of education, is contained in other factors related to H as ‘learning-by-doing’ (dis) or knowledge transfer (ikac). The negative sign means that increasing education of general population does not recover the investment made therein. More people who can read and write means, on average, lower income levels, suggesting the existence of a ‘literacy trap’. Improvements in the way reading and writing is learned, on one side, and acquisition of other skills related to education and work, on the other, could eliminate this perverse circle. The above argument is subject to further exploration using other proxies, and carrying out other studies at sector and individual country levels, which could reduce the level of that trap. Moreover, the variable dis was introduced to measure the impact of ‘learningby-doing’ in per capita income. The sign found in both models is negative, indicating that for all countries of the region, the discovery of new activities adversely affects these economies: the greater the number of products exported, the lower the per capita income. One possible reason for this result lies in the way this indicator was built. It is difficult to determine exactly when there is a new discovery. The new exported goods could be produced domestically for some time before they were exported for the first time, implying that learning rates were reached, laying the bases and skills to start the export process, so that

Foreign trade and per capita income 73 the effect of discovery could be dissipated over time. Moreover, the range of $100,000 and two years was set arbitrarily, and does not include adjustments for inflation, which could consider a larger number of discoveries when in fact they are not reflected in the sign found. This result is subject to further study considering, for example, individual differences in dis for each country. In contrast to the two previous proxies, the composite index of technology transfer, ikac, formed by elements of technology infrastructure and international trade, has the expected positive sign and is statistically significant in both models. The variables within ikac are crucial for the dissemination, acquisition and international knowledge absorption and therefore increase the stock of human capital and income of countries. Certainly, not all aspects of technology transfer are in ikac, such as institutional factors, however a significant portion is captured in such a way that is meaningful and is, in fact, the second largest contribution to income of Latin American and the Caribbean countries in either specification. Finally, contrary to theoretical approach, the coefficient of the stock of ‘available’ foreign technological capital (saftk) appears in the first specification with a negative sign, which is due to a number of reasons. First, the knowledge generated abroad gathered in this variable measures the part which is used in domestic economies just for foreign companies and that, at least for a while, kept without diffusion to other firms, creating a temporary monopoly or some distortion in the market structure which impacts negatively on aggregated production. Second, in the case of Latin America, FDI may not be a disseminator of knowledge and, on the contrary, widens the technological gap vis-à-vis advanced countries, because FDI is oriented to production stages which are not capital or knowledge intensive, generating, ultimately, lower levels of income. However, in the model with differentiated intercepts, the coefficient saftk is significant and positive, albeit low. Anyway, in general terms, the exclusive advantage MNCs own in their home market (skills, knowledge, resources, specific technology, etc.) and that used in some degree in their branches abroad, tend to improve income in the Latin American and the Caribbean countries. While the estimated coefficient is low (0.0084), it should be noted that this technological stock is not built with a view to benefit foreign economies and, nevertheless, it ultimately favours income. Finally, Model II shows that the higher constants are linked to the major economies, while lower intercepts are associated with smaller economies.15 It should be recognised that one limitation of the current model is that all constants are significant and of similar magnitude. Since the panel regression results are not totally satisfactory we tried to mitigate the problem of the assumption of a common intercept among countries by introducing macroeconomic dummies. The idea is that initial technology, represented by term A0, could be similar within the region but different across countries, considering the market size and the necessities derived from this. The countries sample was divided into three groups for each year, depending on the level of GDP (in current dollars) and the average GDP of each group was

74

H. Ríos-Bolívar and O. Neme-Castillo

obtained. Those countries above the average of the first group were assigned a value of 3 (D4); those above average of the second group but below the average of the first group were assigned a value of 2 (D3); countries below the second average but above the third average were given a value of 1 (D2); and countries below the average of group three were given a value of 0 (D1). Only the dummy for D1 is statistically significant at 90 per cent of confidence with positive sign (results not reported). It could be that the variance of initial technology, related to market size, in the other groups was so big that these variables did not produce significant results. Nevertheless, with the additional macroeconomic dummy the fit of the model did not improve at all. Consequently, a third model was considered: a fixed effects model with differentiated constant and intercepts. In general, results remain roughly equal to that of Model II (see Table 3.3). The explanatory power is high (R2 is 0.992) and a high number of coefficients are statistically significant (less than 20 per cent of the regressors are not significant). In this sense, the overall significance is good. For any variable, at least 67 per cent of the estimated coefficients are significant. The coefficient of the population growth rate (popdr) has the expected sign only in four countries (Bolivia, Costa Rica, Panama and Peru). In contrast, the positive sign of this coefficient for most countries in the region is based on the fact that incorporation of labour to productive activity generates a more than proportional increase in the product. Thus, accumulation of capabilities by workers (L) favours income. Note that popdr has no significant effect in Brazil and Chile, two of the largest economies in the area. Also, for most countries the capital stock has negative repercussions on the pcgdp. Argentina is a case in point, with an estimated ratio of –0.828, probably influenced by the strong imbalances experienced at the beginning of this decade. In addition, for Brazil and Trinidad and Tobago, the economies with higher levels of GDP and GDP per capita, respectively, this variable is not significant, likely as a consequence of the relatively inadequate levels of capital goods investment. In contrast, trade liberalisation, represented by imports of capital goods, is remarkably significant for the economies of the region. With the exception of Bolivia, Nicaragua and Paraguay, the directed imported capital stock contributes positively to per capita income. In this respect Chile and Panama excel, which means that their economies have assimilated to a greater extent the impact of this foreign technology through ‘spillover’ effects. The result regarding the literacy rate by this specification is similar to that found in Model II. Most of the coefficients show negative signs, although of low level. In this sense, the existence of a ‘trap of literacy’ is confirmed. Countries that have managed to escape from this are Chile, Ecuador, Mexico, Nicaragua and Paraguay.16 Another element that contributes to formation of H is ‘learning-by-doing’ (dis), this is statistically significant with a negative sign for 10 out of 21 countries, among which are Chile, Mexico and Venezuela. It can be argued, however, when the relatively large participation of foreign capital in these countries is

Foreign trade and per capita income 75 considered over the period, that foreign companies do not generate positive externalities for domestic productivity. However, this indicator does not appear to be a good approximation of learning effects, since the results are significant with positive sign, as indicated by the theory, just for four economies (Belize, Costa Rica, Panama and Paraguay). Anyway, it seems that there is not a strong relationship (positive) between the discovery of new activities and per capita GDP in the countries of the region. The third element inside the stock of human capital is technological transfer measured by ikac; it includes technological infrastructure and commercial exchange with other countries. Results are varied: six coefficients are not significant (Colombia, Ecuador, Guatemala, Honduras, Jamaica and Venezuela) while five are negative (Argentina, Brazil, Mexico, Nicaragua and Paraguay). This result is the effect of two factors: first, inadequate capacities to meet the technological needs of the productive and social sectors in these countries (comparatively, the first three countries maintain the lowest levels of penetration in mobile phones, number of computers and of Internet users); and, second, as they keep trade openness rates relatively high it means that these economies have not absorbed purchases of foreign capital goods into local production processes or, on the other hand, sales to abroad are of low technological content or are focused on markets of low dynamism that, ultimately, means few processes of technological transfer and of human capital formation. In contrast, for ten countries this index is significant and positive, indicating the favourable effect on pcgdp. Finally, the effect of the stock of foreign technological capital ‘available’ for the Latin American and Caribbean countries through the MNCs and that of the stock the latter keep on their property and exploit directly, seem to be ambiguous but significant. Only one coefficient has no statistical significance (Chile). Half of coefficients have a low positive effect on income (the highest coefficient is 0.20). Among the beneficiaries are Brazil, Colombia and Mexico. This verifies the idea that the more closely linked domestic economies are to foreign economies the greater the profits in the domestic economy. That is, higher inflows of FDI, international trade and presumably intra-industry trade in medium and high-tech sectors (as in the specialisation of Brazil and Mexico) have real effects on income. In contrast, this technological stock affects negatively the income of Argentina and Costa Rica, suggesting that foreign firms in these markets internalise the advantage of technological knowledge they have in their country of origin. In other words, the technological stock of MNCs enables them to achieve extraordinary benefits in these countries, possibly through mergers and acquisitions, affecting the market structure of these countries, by crowding out productive investments of domestic firms and, consequently, the level of per capita income.

3.6 Conclusions This chapter derived an extended neoclassical growth model that incorporates factors such as population, physical capital, domestic and imported, human

9.7078* (0.2929) 9.1324* (0.2141) 7.2814* (0.1351) 8.8924* (0.2976) 10.0206* (0.1803) 9.3485* (0.1159) 9.6667* (0.1951) 9.0590* (0.3106) 7.1244* (0.1680) 7.5904* (0.1417) 6.5755* (0.1530) 7.5288* (0.2145) 8.3161* (0.1754) 9.2575*

arg

mex

jam

hon

gua

elsal

ecu

repd

cr

col

chi

bra

bol

bel

c

Country

0.2433* (0.0834) –0.0393 (0.0344) –0.2144* (0.0147) 0.0313 (0.0340) –0.0298 (0.0250) 0.0399** (0.0201) –0.0906* (0.0275) 0.1045* (0.0356) 0.1205* (0.0226) 0.2263* (0.0225) –0.0743 (0.0180) –0.0195 (0.0166) 0.0113 (0.0187) 0.2576*

popdr –0.8277** 0.1423 –0.0923* (0.1085) –0.3725* (0.0632) –0.9477 (0.1835) –0.0730* (0.1034) 0.0437* (0.0555) 0.3300* (0.0884) 0.3147 (0.1319) –0.3452* (0.0577) –0.5013*** (0.0655) –0.3831* (0.0721) –0.3700* (0.0633) –0.6976 (0.0785) 0.7463**

gkf_gdp

Differentiated constants and coefficients: Model III

Table 3.3 Panel data regressions of per capita income

0.4090* (0.03666) 0.3136* (0.1087) 0.4959P (0.0291) 0.4249* (0.0307) 0.9837* (0.0256) 0.4583* (0.0234) 0.5304* (0.0300) 0.5996* (0.0570) 0.3794* (0.0446) 0.6179* (0.0268) 0.3574* (0.0327) 0.2315* (0.0448) 0.7180* (0.0339) 0.3189*

mm_gdp 0.0256 (0.0212) –0.0238* (0.0092) –0.0088* (0.0026) –0.0113* (0.0042) 0.0253* (0.0054) –0.0060*** (0.0038) –0.0156*** (0.0104) 0.0079 (0.0103) 0.0217* (0.0051) –0.0231* (0.0042) 0.0041 (0.0053) –0.0022 (0.0095) –0.0150* (0.0051) 0.0184**

litr –0.0113 (0.0119) 0.1206* (0.0283) –0.0030 (0.0055) –0.0071 (0.0100) –0.0554* (0.0066) 0.0004 (0.0031) 0.0411* (0.0080) –0.0156 (0.0205) –0.0114** (0.0051) –0.01230** (0.0064) –0.02556* (0.0060) –0.0047 (0.0076) –0.0058 (0.0105) –0.0186***

dis –0.1220* (0.0449) 0.5117* (0.0403) 0.1066* (0.0214) –0.0418*** (0.0309) 0.3247* (0.0289) 0.0090 (0.0248) 0.0705* (0.02143) 0.0857** (0.0460) –0.0146 (0.0229) 0.1628* (0.0190) 0.0012 (0.0212) 0.0349 (0.0327) 0.1568 (0.0177) –0.1305*

ikac

–0.0050* (0.0140) –0.2038* (0.0106) –0.0124* (0.0044) 0.0926* (0.0133) –0.3146 (0.0089) 0.0018*** (0.0047) –0.0244*** (0.0161) –0.0150* (0.01005) 0.0814* (0.0050) –0.0548* (0.0057) 0.0569* (0.0186) –0.1313* (0.0179) –0.0841* (0.0140) 0.1541*

saftk

(0.3605) 6.8427* (0.1032) 9.0973* (0.1741) 7.9701* (0.3238) 8.8726* (0.1598) 9.2489* (0.1300) 9.9290* (0.0877) 8.6731* (0.2284) 0.992 497.56 651

(0.0342) 0.1779* (0.0264) –0.0734* (0.0277) 0.5099* (0.1502) –0.1100* (0.0271) 0.0452* (0.0178) –0.0091 (0.0369) 0.1304* (0.0389)

(0.1457) 0.1879* (0.0680) –0.4125* (0.1318) 0.1991* (0.2714) –0.2470* (0.1224) –0.1570 (0.0838) 0.2891* (0.0429) –0.2806** (0.1261)

(0.0199) –0.0552 (0.0803) 0.8416* (0.0711) –0.006 (0.0322) 0.6510* (0.0376) 0.7440* (0.0495) 0.7674* (0.0315) 0.5766* (0.0558)

(0.0109) 0.0358* (0.0076) –0.0919* (0.0123) 0.0739* (0.0141) –0.0399* (0.0149) –0.0545* (0.0138) –0.0840* (0.0120) 0.0069 (0.0147)

(0.0122) –0.0262* (0.0079) 0.0245** (0.0117) 0.0227*** (0.0176) –0.0853* (0.0114) –0.0122*** (0.0087) –0.0961* (0.0133) –0.0408* (0.0111)

(0.0188) –0.0864* (0.0200) 0.4103* (0.0460) –0.3591* (0.0306) 0.1736* (0.03035) 0.2666* (0.0399) 0.0996* (0.0325) 0.0452 (0.0498)

Notes *, ** and *** are the levels of significance at 1%, 5% and 10%, respectively. Standard errors appear between brackets and below coefficients.

Source: own elaboration.

R2 adjusted F Observations

ven

uru

tyt

per

par

pan

nic

(0.0232) 0.0325* (0.0023) –0.0745* (0.0107) 0.1246* (0.0100) –0.0373* (0.0084) 0.0461* (0.0095) 0.0262* (0.0092) 0.0400* (0.0075)

78

H. Ríos-Bolívar and O. Neme-Castillo

capital (direct learning, ‘learning-by-doing’ and product discoveries) and technological capital that foreign firms employ in the Latin American and the Caribbean markets as a consequence of international trade. This is an alternative way to determine the importance of international trade based on neoclassical functions of income. This model was applied to the study of 21 countries in Latin America and the Caribbean during 1977–2007 following a panel data methodology. Regarding human capital, a strong link with income, when structural differences across countries are considered, is observed. With regard to foreign technological stock, it appears that, on average, exclusive technology of foreign firms located in the Latin American and Caribbean countries boosts per capita GDP. However, differentiated estimations are ambiguous, although it seems that the positive effect tends to be present in the larger economies or with greater levels of openness. About the hypothesis that international trade impacts on income levels, it is fulfilled in general terms in the countries of the region, although this assertion must be weighted by country and variable. A notable result is that for most countries the stock of physical capital negatively affects pcdgp, mainly in Argentina. By contrast, imports of capital goods, positively impact on the economies of the region (except for Bolivia, Nicaragua and Paraguay), particularly in Chile and Panama, which have assimilated foreign technology to a greater extent. Thus, the argument that countries can import capital goods they are unable to produce domestically and then benefit in terms of productivity is confirmed. In addition, a negative sign in the relationship between the literacy rate and income was found, so there is a ‘trap of literacy’, except for Chile and Mexico. In contrast, the variable ‘learning-by-doing’, approximated by the new products exported, is statistically significant with a negative sign for half of the countries, including Chile, Mexico and Venezuela. Then it can be argued that discovery of new products or processes (by mere imitation), reduces per capita income in these countries. Regarding intangible transfer of knowledge, which affects productive and organisational processes, it plays an important role in the accumulation of knowledge, human capital and the subsequent income level. It should be remarked that capabilities such as infrastructure and international trade that allow absorption of intangible transfer of knowledge also play an important role. In conclusion, international trade has effects on income, although not all the effects identified by theory. Finally, the inclusion of a variable derived from the region’s opening to foreign productive capital allowed expanding the model to measure the effect of foreign technological capital stock ‘available’ to the region. The results indicate a strong relationship with income but the sign of that relationship is not clear. They prove that the greater the openness to foreign economies, the greater the effect on domestic income.

Foreign trade and per capita income 79

Notes 1 Growth literature calls this situation level-effect, in contrast with growth-effect that is achieved when A, on production function, is growing steadily. 2 Other ways of knowledge dissemination are FDI and imports of advanced technology goods. Thus, the openness of a country, in terms of removing barriers to FDI or tariffs, is essential for global knowledge dissemination within an economy. However, the perspective in this document implies that some knowledge of MNCs is not disseminated to domestic economies, so the impact they may have on human capital formation is lower. 3 The function exhibits neoclassical conventional characteristics like diminishing returns in all production factors and constant returns to scale. 4 kt = K/AL is the stock of physical capital per effective unit of labour, ht = H/AL denotes the stock of human capital per unit of efficient labour, zt = Z/AL is the stock of directly imported capital per unit of efficient labour, wt = W/AL is the foreign technological capital ‘available’ to produce from the domestic economy for each unit of efficient labour and, yt = Y/AL denotes the income per unit of efficient labour, all valid at time t. 5 The steady-state values of k, h, z and w are derived from equations (3.3), (3.4), (3.5) and (3.6) setting them to zero and solving simultaneously. Thus, a system of four equations with four unknown variables is formed. Solving this system of equations yields the steady-state values (k*, h*, z*, w*) expressed as a function of the independent variables sk, sh, sw, and of parameters (1 + p), (1 + m), n, g and δ. 6 The choice between FEM and REM is made using the Hausman test. There is a third specification that can be estimated by restricting pooled panel data, where errors, uit, are independent between time, and individual units have zero mean and constant variance, which would represent the traditional case of regression model that can be estimated by ordinary least squares. The test for determining whether the best specification is a pooled model or FEM is the F test; while the LM test is used for determining the best model between the pooled and REM ones. 7 For Venezuela and Paraguay there is no available data on BADECEL for the years 2006, 2007 and 2008, while for Trinidad and Tobago the series was discontinued from 2004. 8 It should be noted that there is a time lag between discovery and the start of export activity, but it can be argued that when a good is exported for the first time, the country went through a learning process to produce more efficiently, with lower costs, best qualities or different varieties that ultimately, enables the country to produce domestically. 9 The natural way to represent this knowledge partially excludable by multinational companies are data related to the presence of MNCs in Latin American economies. However, these series are neither available for all countries nor for the entire period of interest. 10 The stock of technological capital is calculated through the perpetual inventory method using the expenditures on R&D, that is: stkt = (1 – δ)stkt–1 + It–1, δ is the depreciation rate, It–1, the investment in R&D in the previous period and skt–1 the stock of technological capital in t – 1 that is obtained as skt–1 = [(1 – δ)sk0 + i]/[(1 + υ)], where i is the ratio of investment in R&D to product (GDP), υ the growth rate of investment and sk0 the initial stock of technological capital. 11 Countries included to create the foreign technological stock are: USA, Germany, France, UK, Spain, Italy, Netherlands, Belgium, Portugal, Greece, Switzerland, Sweden, Norway, Finland, Denmark, Australia, New Zealand, Korea and Japan; Latin American countries register a higher degree of trade integration with these economies. 12 The value of the test was 117.64 and the p-value was 0.0038. 13 For the first test, the estimated value of the test was 1.6E12 and the linked p-value was nearly 0.00006 and for the latter, the estimated value was 2062.5.

80

H. Ríos-Bolívar and O. Neme-Castillo

14 In fact, a per capita income equation was estimated, in addition to the pooled model, following the fixed and random effects models. The Breusch-Pagan test was performed to discriminate between REM and OLS, rejecting the null hypothesis which points out there are not significant differences between these models. Thus, the relevance of the random effects is accepted. Also the F test was applied in order to determine the significance of fixed effects; the results rejected the null of absence of fixed effects. Both tests indicate that the model is candidate to be estimated taking into account fixed or random effects. Accordingly, the Hausman test was performed, finding that the best estimate is the fixed effects one. The test results are not reported but are available upon request via email. 15 As the constants reflect the effect of variables not included that precisely make the difference across countries, it is argued that probably variables such as domestic market size, scale economies, poverty and inequality levels, and internalisation, among others, play a leading role in the level of per capita income; thus, an extension of this document should go in that direction. 16 The estimated coefficient is not statistically significant for Argentina, Dominican Republic, Guatemala, Honduras and Venezuela, so that the ‘trap of literacy’ does not apply in these cases either.

References Beck, N. and Katz, J.N. (1995) ‘What To Do (and Not To Do) with Times-Series CrossSection Data’, American Political Science Review, vol. 89, no. 3, pp. 634–647. Bidlingmaier, T. (2007) ‘International Trade and Economic Growth in Developing Countries’, Dynamics, Economic Growth, and International Trade, DEGIT – XII. Melbourne, Australia, June 2007. Gomory, R. and Baumol, W. (2001) Global trade and conflicting national interests. Cambridge, MA: MIT Press. Greene, W. (2000) Econometric analysis (5th edn). New York: Macmillan Publishing. Lucas, R. (1988) ‘On the mechanics of economic development’, Journal of Monetary Economics, vol. 22, no. 7, pp. 3–42. Lucas, R. (1993) ‘Making a miracle’, Econometrica, vol. 61, no. 2, pp. 251–272. Mankiw, N., Romer, D. and Weil, D. (1992) ‘A contribution to the empirics of economic growth’, Quarterly Journal of Economics, pp. 407–437. Redding, S. (1999) ‘Dynamic comparative advantage and the welfare effects of trade’, Oxford Economic Papers, vol. 51, no. 1, pp. 15–39. Rivera-Batiz, L. and Romer, P. (1991) ‘International trade with endogenous technological change’, European Economic Review, vol. 35, no. 4, pp. 971–1,004. Romer, P. (1990) ‘Endogenous technological change’, Journal of Political Economy, vol. 98, no. 2, pp. 71–102.

4

Regional integration and its effects on inward FDI in developing countries A comparison between North–South (Mexico) and South–South (Brazil) integration Thomas Goda

4.1 Introduction During the last 15 years, there has been a proliferation of regional integration agreements (RIAs),1 mostly because of massive increases in North–South2 agreements but also because of more South–South3 agreements (Fiorentino et al., 2007). There are many different motives why more and more RIAs have been signed. Probably the most important motivation for developing countries is the hope to develop economically by achieving higher growth rates, i.e. to catch up with already developed countries. According to the new growth theory, higher growth rates might result from openness to trade and foreign direct investment (FDI) inflows, among other factors. Thus, for developing countries one important reason to engage in a RIA, besides market access, is the expectation that foreign direct investment inflows will increase. In contrast, important motives for developed countries to promote RIAs with developing countries seem to be the aim to lock-in liberalisation reforms in these countries (i.e. to get market access and resource access) as well as to circumvent multilateral negotiations which show slow progress at the moment (Schiff and Winters, 2003). These new motives have led to a new type of RIA, which involves not only deep trade but also deep investment provisions – to attract FDI inflows – and is part of a process called ‘new regionalism’. Another characteristic of ‘new regionalism’ is that, although the provisions favour member countries, external barriers to trade, services, and capital often are also lowered, with the aim of integrating the region in international production networks. Therefore, the focus is not any longer exclusively regional (as it has been in the ‘first regionalism’ agreements) but also global (Schlageter, 2005). Although regulations in North–South and South–South integration agreements are similar, there is a dispute in the literature about whether North–South integration is more beneficial than South–South integration. According to UNCTAD (2007a) there are good reasons to disfavour the engagement of developing countries in a North–South integration agreement. Perhaps the most

82

T. Goda

important problem is the negative ‘impact of these agreements on . . . [the ability of developing countries] to use alternative policy options and instruments in the pursuit of a longer term development strategy’ (ibid., p. 65). Thus, developing countries should cautiously calculate the gains and risks from North–South integration to avoid rushing into integration agreements with possibly overrated expectations. One argument why North–South integration might be preferable is that FDI inflows in a North–South agreement can be expected to be higher than in a South–South agreement (Schiff and Winters, 2003). However, even though comprehensive literature exists on (1) FDI and its possible impacts on developing countries and (2) the effects of RIAs in general, unfortunately, not much research has questioned whether the signing of a North–South agreement has different consequences on inward FDI for developing countries than the signing of a South–South integration agreement. Most of the existing literature about the effects of RIAs is devoted to the impacts on trade, especially trade diversion and creation. The objective of this chapter is to analyse and discuss which impact North– South and South–South RIAs probably have on FDI inflows in developing countries, i.e. to find out if North–South agreements really attract more FDI as conventional wisdom declares. As an example of a developing country in a North–South agreement Mexico – member of the North American Free Trade Agreement (NAFTA) – is taken, and as an example for South–South integration Brazil – member of the Mercado Común del Sur (Mercosur). Both countries are suited very well for a comparison because they have a huge growing internal market, cultural similarities, similar endowments, and they experienced relatively high FDI inflows before they signed and implemented an integration agreement in the 1990s.4 In addition, Mexico and Brazil witnessed structural adjustment programmes in the 1980s and 1990s, respectively. This chapter is structured as follows: in the first section likely impacts of North–South and South–South RIAs on inward FDI in developing countries and more specifically on Brazil and Mexico are discussed. In the second section, an empirical evaluation on the basis of recent FDI inflow and FDI stock data for Brazil and Mexico is made, before the chapter ends with a conclusion of the findings.

4.2 Theoretical considerations: regional integration agreements and possible effects on inward FDI 4.2.1 The possible static effects As already discussed in the introduction, ‘new regionalism’ is expected to lockin reforms and to push further liberalisation. On the one hand, RIAs contain provisions which increase the security for investors who are located inside the region and as a result promote FDI flows within the region. These investmentfriendly regulations normally consist of most favoured nation rules, national

North–South and South–South comparison

83

treatment for regional investors, protection of intellectual property rights, unrestricted capital flows, rules regarding dispute settlement, and the non-existence of performance requirements (Medvedev, 2006). On the other hand, ‘new regionalism’ often also lowers discriminatory barriers against the rest of the world (ROW), i.e. external barriers to trade and capital. As a result, MNCs are more likely to integrate the region in their global production chain and build new affiliates to exploit the locational advantage of the region (te Velde and Bezemer, 2004). Accordingly, Medvedev (2006, p. 2) states: ‘[T]he most immediate impact of . . . [RIAs] on foreign investment is the direct response of FDI to the implementation of various investment provisions in an agreement’. While Dunning (1997) argues that the integration of regions, i.e. the harmonisation of standards and increased investor security, is necessary for countries to attract FDI in the future. This is due to the fact that companies – in order to become more efficient and competitive – increasingly undertake joint ventures. Thus, countries/regions that do not support this new alliance character of capitalism run the risk of being left out. However, there is an emotional discussion if investor protection and the lockin of liberalisation reforms in the course of RIAs are really that important to attract FDI because a more predictable investment climate seems to be only decisive for new investment inflows if ex ante the conditions for investors have been unpredictable (te Velde and Bezemer, 2004) and in most cases restrictions for investors have been lowered considerably and reforms have been undertaken before RIAs have been signed. Accordingly, liberalisation policies and provisions regarding investor protection in the course of RIAs perhaps are the least important conditions for increasing inward FDI flows (UNCTAD, 1998). Nevertheless, after the implementation of RIAs the liberalisation of some sectors might accelerate because countries that have deregulated less than other member countries might be pressured to open up some sectors more radically Therefore, RIAs might be important for FDI inflows because they can open up the service sector – which in most cases consist of non-tradables and accordingly only can be served by foreigners through FDI (Medvedev, 2006). Another reason why inward FDI might increase is that production processes for goods that are aimed to serve the domestic market are different from production processes for goods which serve a regional market.5 Therefore, due to regionalism, such FDI inflows are encouraged which are aimed at the restructuring, i.e. modernisation and enlargement, of existing plants (IADB, 1998). In addition, because of harmonisation and lower trade barriers regionalism contains better possibilities for investors to build clusters for the production of similar goods or goods that consist of similar inputs, respectively. In other words, agglomeration economies can take place regionally (i.e. between bordering countries) and are not limited to one country anymore. Hence, regions become a more attractive location for FDI (UNCTAD, 1998). However, the static effects of RIAs on FDI depend on the pre-existing situation (like pre-existing barriers) and factor endowments. Therefore, the effect might be negative for single countries within the region, while for the region as a

84

T. Goda

whole it is likely to be positive (UNCTAD, 1998). For example, if a country has no locational advantage in comparison to the other member countries and ex ante was a host to a lot of ‘tariff-jumping’ FDI – which tried to circumvent tariff barriers – the removal of such barriers would most probably lead to a decreasing stock of FDI in that country. In contrast, if a country has attractive factor endowments and the economy was not closely linked with the new partners before, the positive effect of a RIA on new FDI inflows presumably is significant (Blomström and Kokko, 1997). Furthermore, some member countries might witness the relocation of existing FDI stocks to other members, if ex ante MNCs had market-seeking FDI in more than one member country and ex post relocate their investment to one location which they use as platform to serve the regional market to exploit economies of scale (Levy Yeyati et al., 2003). Next, with respect to factor endowments and the pre-existing situation, it should be noted that investment provisions, the opening-up of sectors, non-tariff barriers within the region, and external barriers can be seen as important determinants for FDI inflows and changes in FDI stocks. In addition, the regional transport and communication infrastructure is a crucial factor because without sufficient infrastructure transaction costs are so high that platform investment will not take place and (regional) production networks will not emerge (UNCTAD, 1998). In addition, the size of the individual market and plant-level economics of scale still play an important role in the decision making process if, and to what extent, this relocation will take place (Levy Yeyati et al., 2003). However, if one wants to examine the effects of RIA on inward FDI more closely, it makes sense to divide the effects into intra-regional effects6 and interregional effects7 because ‘insiders’ and ‘outsiders’ will be treated differently after the RIA comes into existence. 4.2.2 The intra-regional effects There are two views about the relationship of trade and intra-regional investment. One view postulates that FDI and trade are substitutes. MNCs might serve the regional market through FDI because of barriers to exports, therefore, the abolition of tariffs is supposed to lead to disinvestment of ‘tariff-jumping’ market-seeking FDI from members of the RIAs because they might prefer to serve the whole regional market through exports from their home location to reap economies of scale, i.e. it becomes more efficient to export (Blomström and Kokko, 1997, Ledermann et al., 2003, and Cuevas et al., 2005). However, the other view postulates that FDI and trade can be seen rather as complements than as substitutes because without regional tariff barriers companies are likely to undertake investment in locations with a locational advantage (e.g. best access to all markets of the member countries) and use this location as a platform to serve the whole regional market through exports (Balasubramanyam et al., 2002). Thus, in some countries disinvestment might take place while in countries which have the potential to serve as a platform for the regional market strong intra-regional horizontal FDI inflows can be expected.

North–South and South–South comparison

85

Another argument in regard to the view that trade and FDI are rather complements than substitutes is that RIAs can lead to the establishment of regional production networks. Due to lower transaction costs it is possible to produce only a part of the product in one location and afterwards send it to another location in the region for further processing or assembling. Therefore, regional production networks emerge if factor endowments between member countries and factor requirements in certain stages of production differ (Levy Yeyati et al., 2003). The exploitation of these comparative advantages might require the internalisation of production, i.e. to engage in vertical FDI (te Velde and Bezemer, 2004). In addition, rules of origin (RoO)8 might lead to increased vertical FDI within the region. These rules – which state that a certain part of the production (inputs) needs to take place within (come from) the integration area to have preferential access to the whole regional market – prevent that inputs are sourced from affiliates outside the region. Thus, ‘insider’ MNCs might reallocate production processes that have been undertaken ex ante outside the region ex post to member countries, ensuring preferential access to the markets of the partner countries (ibid.). 4.2.3 The inter-regional effects In contrast to the intra-regional effect, market-seeking ‘tariff-jumping’ FDI that flows into the region might increase if the level of outside protection goes up or, alternatively, foreign investors fear that the protection will increase in the future. In addition, an increase of market-seeking FDI can be expected because, in contrast to the local markets, the regional market might be big enough to make an investment profitable, i.e. it is possible to achieve economies of scale and bear the fixed costs (Blomström and Kokko, 1997). Countries in a RIA tend to share common values and beliefs as well as regional infrastructure and transportation networks. Hence, investment within a RIA has the advantage that it is possible to serve a huge regional market from within the region with low transaction costs. As a result, the region is becoming more attractive as a location for FDI (Sethi et al., 2003). Supplementary to market-seeking FDI, vertical FDI might increasingly flow into the region. As a result of lower discrimination due to ‘new regionalism’, MNCs possibly integrate the region in their global production chain and build new affiliates or extend and modernise existing affiliations to exploit the locational advantage of the region (te Velde and Bezemer, 2004). Rising discrimination against the ROW ex post through the emergence of non-tariff barriers like RoO, quantitative restrictions on imports, and custom procedures, can lead to less vertical FDI or more vertical FDI inflows. The latter two make exporting from outside impossible or more costly. The former means that it is not sufficient to engage only in ‘tariff-jumping’ FDI but also a certain amount of inputs need to come from inside the region (Tuman and Emmert, 2004). If adequate inputs cannot be purchased from local producers RoO might lead to investment with deep production processes (only assembling is not possible)

86

T. Goda

or, alternatively, MNCs might ‘force’ their suppliers to produce as well in the region. Both arguments imply that the amount of FDI inflows will be higher than without RoO (Cuevas et al., 2005). However, RoO could also lead to less FDI inflows into the region. MNCs might not be able or willing to source their inputs from within the region and therefore serve the regional market through exports from outside the region or not at all (Sanguinetti and Bianchi, 2006). 4.2.4 Possible dynamic effects Besides the static effects RIA might also lead to dynamic effects which could ‘prevent’ that FDI inflows decrease after a while (e.g. after the FDI stock has adjusted),9 although these effects are less clear-cut. As discussed above, one motivation to engage in regionalism is the belief that RIAs will raise the economic growth rate of member countries in the medium or long term. If this is the case, one can argue that in the medium and long term also more FDI will flow into the member countries. On the one hand firms within the region will grow stronger and will increase business and on the other hand new companies will emerge (Medvedev, 2006). In addition, the removal of trade barriers supposedly leads to more competition, and as a consequence companies most probably will engage on a larger scale in R&D by which they become more efficient. Along with the new competition situation this might motivate companies to engage in M&As and strategic alliances, i.e. more FDI will occur in the region. The growth of the economies will also lead to an increased market size which makes the region more attractive for FDI. Hence, spillover effects might occur, and specialisation, economics of scale, and agglomeration effects are likely to take place which results in further improvements of growth rates (Blomström and Kokko, 1997). These dynamic effects could lead to a virtuous cycle: the RIA leads to growth, which brings forth new (foreign) investment, which leads to more growth, which attracts additional (foreign) investment . . . However, empirical evidence is not very robust that RIA necessarily increases the growth rate of the member countries (Berthelon, 2004). Moreover, FDI might not lead to positive spillover effects and may even dampen economic growth (e.g. through crowding-out of local investment). Furthermore, there is the danger that some countries be left out and even receive less FDI in the medium and long term due to agglomeration effects within the region (UNCTAD, 1998). Profit repatriation, i.e. a lower net contribution of FDI on gross capital formation, may disturb the dynamic effect as well (Ramirez, 2003). Thus, countries should not count on dynamic effects as a result of the establishment of a RIA. 4.2.5 Possible effects for Mexico The common opinion in the literature is that North–South integration will lead to larger FDI inflows in the developing partner countries than would be the case

North–South and South–South comparison

87

within a South–South agreement. It is argued that if the members of a RIA differ in their factor endowments, the agreement most probably stimulates vertical intra-regional FDI. More specifically in the case of Mexico – which is a member of NAFTA, i.e. a North–South RIA between the US, Canada, and Mexico – the two developed partner countries are likely to shift labour-intensive production to Mexico because preferential access to the markets of the US and Canada is ensured, capital flows are liberalised, and investor protection regulations are part of the agreement (Cuevas et al., 2005). Hence, NAFTA is expected to bring forth more specialisation due to different competitive advantages between the member countries. As a result, a clustering of activities can be expected, i.e. the forming of regional production networks (IADB, 1998). Another scenario could be that the North acts like a kind of flying goose and shifts older technologies to the South (UNCTAD, 2007a). In both cases, theory expects higher flows of vertical FDI from the US and Canada to Mexico, which has much lower labour costs. Furthermore, NAFTA is seen to have positive impacts on Mexico in terms of increased stability (i.e. the lock-in of economic reforms) and accelerated liberalisation (Blomström and Kokko, 1997). Accordingly, higher market-seeking FDI flows to Mexico are expected to take place due to an opening up of the service sector which was embedded in the NAFTA negotiations (UNCTAD, 1998). Moreover, the location advantages of Mexico make it likely that extra-bloc investors will invest in Mexico to produce cheaply and gain access to the markets of the US and Canada, i.e. ‘tariff-jumping’ FDI (Schiff and Winters, 2003). Intra-regional investors as well are likely to use Mexico as a platform for exports in the region and hence to restructure the production processes of former market-seeking affiliates (IADB, 1998), rather than to close these affiliates as expected by Cuevas et al. (2005). However, Kim (2007) developed a model which expects less beneficial consequences. While he agrees that the less developed country can be expected to receive larger efficiency and resource-seeking intra-regional flows, he believes that – if the technology gap is large – inter-regional market-seeking FDI tends to flow into the countries with more sophisticated technology.10 In addition, he argues that technologically intensive production will relocate from the South to the North. This argument is in line with the finding of UNCTAD (1998) that agglomeration will take place. 4.2.6 Possible effects for Brazil In contrast to Mexico, Brazil is engaged in a RIA with developing countries, i.e. in the Mercosur together with Argentina, Uruguay, and Paraguay. However, this South–South agreement is also seen to attract higher inward FDI flows to Brazil. Schiff and Winters (2003) argue that the impact of South–South RIAs on FDI flows to middle-income countries seems to be positive.11 This might be because agglomeration effects (centripetal forces) tend to shift industrial investment to the most developed country within the South–South agreement. However,

88

T. Goda

according to them, Brazil is only likely to attract market-seeking ‘tariff-jumping’ FDI that serves the regional market, while vertical FDI most probably will not increase as a response to the Mercosur agreement. Levy Yeyati et al. (2003) confirm this proposition. They state that due to the fact that Mercosur has increased the size of the regional market, more ‘tariffjumping’ FDI is supposed to take place because now it is more attractive to circumvent the common external tariff than to export from outside the bloc. Furthermore, as for Mexico, the opening up of formerly restricted sectors (mainly the service sector) and the further enhancement of structural reforms are also likely to have led to increased market-seeking FDI inflows to Brazil (Castilho and Zignago, 2005). However, while the IADB (1998) confirms that the privatisation of the service sector was a very important source for FDI inflows to Brazil, it disagrees with the opinion that Brazil will only attract market-seeking FDI as a consequence of the forming of Mercosur. As discussed earlier, ‘new regionalism’ is not only inwardly but also outwardly oriented and lowers the external barriers, which in combination with better investor protection and the liberalisation of capital flows can lead to the growth of the manufacturing industry and the integration of foreign affiliates in global production networks. Nevertheless, intra-regional vertical FDI is seen to develop only slowly as a result of Mercosur. Accordingly, common wisdom is that Mexico should have received considerably more vertical inward FDI as a consequence of the forming of NAFTA than Brazil due to the forming of Mercosur and hence Mexico should have attracted more inward FDI in total as well. 4.2.7 Other possible explanations for increased inward FDI So far the argument was that developing countries should engage in regionalism to attract higher amounts of inward FDI. RIAs might lead to increasing liberalisation (including the opening up of protected sectors), investor protection, access to an increased regional market, the building of (at least) regional production networks, and dynamic effects which attract further FDI. However, it is questionable if RIAs are necessary to attract FDI. East Asian countries are an example in which huge amounts of inward FDI have been attracted due to competitive strength and the lowering of trade barriers within the region, despite the missing of extensive multilateral agreements (UNCTAD, 1993). Therefore, it can be expected that regional dynamics can lead to production networks and higher inward FDI even when the countries have not signed RIAs (UNCTAD, 2007a). Accordingly, other factors might be more important than the signing of RIAs. These other determinants could be resource endowments, infrastructure, a relatively huge domestic market, and strong growth of domestic industries (ibid.). Tuman and Emmert (2004) add to this list per capita growth, the real exchange rate, education, inflation, political stability, and openness to trade. According to Ramirez (2003) unit labour costs are also an important factor to explain higher FDI inflows.

North–South and South–South comparison

89

Another argument why FDI flows might have increased regardless of RIAs is that as a result of structural adjustment programmes, privatisation, the opening up of sectors, and debt conversion programmes would have happened anyway – all can be seen to have contributed significantly to FDI inflows in Mexico and Brazil (Nunnenkamp, 1996). Most of these changes concern the services sector for which a huge regional market does not matter so much because ‘[g]iven the non-tradable nature of most services, FDI in this sector is . . . almost entirely directed towards local markets’ (IADB, 1998, p. 22). Considering the huge market share of Spanish and Portuguese FDI in the Latin American service sector, FDI therefore might be influenced rather by cultural affinity than by RIAs (Guedes and Gómez Olivarez, 2005). Another argument why inward FDI perhaps would have been higher in Mexico and Brazil even without the signing of NAFTA and Mercosur, respectively, could be that since the beginning of the 1990s the structure of worldwide capital flows to developing countries has changed in favour of FDI (UNCTAD, 2005). The reason for this change could be that due to competition-linked considerations, companies more often engage in cooperative agreements (joint ventures) to become more efficient (Dunning, 1997). In addition, companies are seeking to organise their business in a more effective way and accordingly separate their functions like R&D, purchasing, assembling, book keeping, etc., to carry them out at the most efficient location (Chudnovsky and López, 2004). As UNCTAD (2002, p. 121) has phrased it: [Due to better telecommunication and transportation possibilities and] . . . falling barriers to international transactions . . . international production systems have emerged within which TNCs locate different parts of the production processes, including various services functions, across the globe, to take advantage of fine differences in costs, resources, logistics and markets. In addition, low interest rates, increasing share prices on the stock markets, and relatively high growth rates in the USA – which constitutes the lion’s share of Latin American FDI flows – might have played an important role in the emergence of increased FDI flows to Latin America, i.e. more funds have been available to be invested in profitable locations (Ramirez, 2003). To sum up, it can be said that RIAs might influence some variables positively like per capita growth, the regional infrastructure, openness to trade, stability, liberalisation policies, and the emergence of (regional) production networks. However, it is not clear to which extent these factors would have developed without the creation of a RIA because ‘[i]n a liberalizing world of falling barriers, the location advantage of a large regional market may not be what it used to be’ (UNCTAD, 1998, p. 122). Accordingly, RIAs might be less influential in regard to inward FDI than one thinks and they are neither a necessary nor a sufficient condition to attract increasing inward FDI to developing countries (Lederman et al., 2003).

90

T. Goda

4.3 Empirical evaluation of inward FDI in Mexico and Brazil 4.3.1 Changes in inward FDI in Mexico and Brazil Changes in FDI inflows and the FDI stock First of all, it is interesting to compare the amounts of FDI that have flown12 into Mexico and Brazil. In 1985 Brazil ($1.4 billion) starts at a lower level than Mexico ($2 billion). This level stays lower until 1995 (Brazil $4.9 billion and Mexico $9.5 billion) – with the exception of 1988. Then from 1996 to 2000 FDI flows to Brazil skyrocket to about $33 billion (Mexico about $18 billion). From 2001 to 2005 Mexico again has higher inflows than Brazil (around $5 billion p.a. higher). In 2006 both countries end up with FDI inflows of around $19 billion (Figure 4.1). Besides the amount it is interesting to take a look at trends of FDI inflows. In Brazil a massive increase in the mid to late 1990s followed a slump in the 2001–2003 period with a subsequent recovery. However, in 1995 and 2001 changes in the collection method of FDI stock and flow data took place in Brazil13 which had at least some significant influence on the measurement of inward FDI stocks. Thus, it is not clear how far the increase in 1995 can be attributed to the change in the methodology. In addition, it can be expected that the slump after 2000 would have been even higher without the second methodology change. Also notable is that in 2000 an exceptionally large M&A between ‘Telefonica’ and ‘Telecomunicaçoes de Sao Paulo’ took place (around $10 billion). Without this transaction the trend of FDI inflows would have been negative already after 1999. However, FDI flows to Brazil are consistently higher after 1995 than before. Therefore, it seems to be clear that FDI flows to Brazil in

Figure 4.1 Foreign direct investment, net inflows (in billion US$), comparison with ROW (sources: WDI (2007); own calculation).

North–South and South–South comparison

91

general have been higher after Mercosur’s (imperfect) common market came into existence, which is in line with the theoretical expectation. That being said, the trend has been similar in the ROW and therefore the increased FDI inflows might rather reflect a global pattern, not so much the consequence of Mercosur. Mexico clearly has a less volatile upward trend in FDI inflows than Brazil. Between 1987 and 1999 this trend is very similar to other middle-income countries14 (Figure 4.2). The most obvious difference in this period is a peak in 1994, when NAFTA came into existence. However, this peak perhaps can be explained partly by the fact that: ‘The methodology for compiling FDI statistics changed significantly in 1994, so figures for the years before are not directly comparable’ (UNCTAD, 2004, p. 377). The massive increase in 2001 can be explained by one big acquisition of $12.5 billion (in this year the City Group acquired Banacci). Without this M&A Mexico would have had lower FDI inflows in 2001 than in 2000 and accordingly the trend after 1999 would be more similar to other middle-income countries which had a slump in FDI inflows at the beginning of the twenty-first century – possibly due to the stock market crisis at that time. In any case, the overall trend of FDI inflows after the establishment of NAFTA in 1994 clearly is positive until 2004. As a result, FDI inflows have been much higher than before (although the difference perhaps would have been lower without the methodology change in 1994). This is congruent with the expectations of Section 4.2. However, other middle-income countries also had an upward trend during this period and, therefore, as for Brazil the increased FDI inflows to Mexico might rather reflect a global pattern and not the effect of NAFTA. Another way to compare FDI inflows, which is probably more useful, is the FDI inflows to GDP ratio (UNCTAD, 2002). The advantage of this measurement is that the size of the host economy is taken into account and hence countries with a different size can be better compared. Furthermore, this ratio is a good

Figure 4.2 Foreign direct investment, net inflows (in billion US$), comparison with other middle-income countries (sources: WDI (2007); own calculation).

92

T. Goda

instrument to see if inflows are over proportionally higher compared to the increase in GDP, or if the increase in inflows is owed to inflation.15 In any event, the trends of the data in Figure 4.3 are similar to the trends from above. Brazil’s FDI inflows as percentage of GDP have increased tremendously in the period 1995 to 2000 (in 2000 they have reached 5 per cent of GDP). Afterwards there is a sharp downtrend which stops 2003. The movement of Mexico’s FDI/GDP ratio is also relatively similar to the trend of the FDI net inflows in US$ – beside the increase in 1995, which can be explained by a slump in GDP due to the ‘tequila crisis’ (the GDP decreases by over 6 per cent in that year (WDI, 2007)).16 If one makes a cross-comparison of the FDI/GDP ratio it can be stated that the trend of the ROW is similar to Brazil’s from 1992 to 2003, although the sharp increase starts one year later and the decline in 2001 is even sharper – from 1986 to 1995 the ROW has a higher ratio than Brazil and then, from 1996 to 2004 (with one exception in 2000) the ratio is higher in Brazil. These data give some support to the view that Mercosur has led to higher inflows in Brazil. But, when one compares Brazil with other middle-income countries, which is the more interesting reference group because of similar characteristics, Brazil’s FDI/GDP ratio is much below this group’s from 1989 to 1997. Then from 1998 onwards (i.e. three years subsequent the full implementation of Mercosur) the ratio is significantly higher until 2003 when the ratio again is lower. Hence, in the 16 years after the implementation of Mercosur the FDI/GDP ratio in comparison to other middle-income countries only is higher in five years, i.e. less than one-third of the time. In contrast, Mexico outperforms other middle-income countries in most of the post-NAFTA period. Only in the periods 1998–1999, 2005–2006, and in the year 2003 have other middle-income countries had a higher FDI/GDP ratio than Mexico. This means that after 1994 Mexico’s FDI/GDP ratio was higher than that of other middle-income countries 60 per cent of the time. When one

Figure 4.3 Foreign direct investment as percentage of GDP (sources: WDI (2007); own calculation).

North–South and South–South comparison

93

compares Mexico with the ROW or Brazil its superior performance becomes even more obvious (in 1999–2000 and 2006 it has a lower FDI/GDP ratio than the ROW, and in 1998–2000 lower than Brazil). However, when one leaves aside the peaks in the 1994–199517 period and in 2001, Mexico’s FDI/GDP ratio is very similar to other middle-income countries. To sum up, it can be said that after the signing of NAFTA and Mercosur the amount of FDI inflows and the level of the FDI/GDP ratio in Mexico and Brazil have been higher than before (with the exceptions of 1992 for Brazil, and 2003 and 2006 in Mexico) – interestingly an anticipation effect is not obvious.18 This gives support to the theoretical expectation that FDI inflows in middle-income countries increase after RIAs have been signed. On average the FDI/GDP ratio has been higher in Mexico (average 1994–2006: 3 per cent) than in Brazil (average 1991–2006: 2.2 per cent) after the corresponding RIAs came into effect.19 These data can be seen as a confirmation of the expectation that North– South RIAs lead to higher FDI inflows in middle-income countries than South– South RIAs. But, Brazil started from a much lower level (in 1990: 0.3 per cent) than Mexico (in 1993: 1.1 per cent). Thus, the increase of the average FDI/GDP ratio has been similar after Mercosur and NAFTA came into existence. Beside FDI flows, it is useful to consider the changes in the inward FDI stock. Changes in inflow figures ‘do not necessarily reflect changes in the attractiveness of a location . . . [but instead might show only] exceptional one-off investments’ (IADB, 1998, p. 230). Furthermore, inward stocks can be used to see if higher inflows to Mexico and Brazil are ‘just’ signs of stock adjustment. However, as for the flow figures, a comparison with other countries is difficult because different methodologies are used. Furthermore, the changes in methodology most likely have changed the FDI stock to GDP ratio significantly in Mexico and Brazil. Nevertheless, the data from Table 4.1 do not confirm that stock adjustment in Mexico has taken place; rather the changes seem to reflect a global pattern. The same is true for Brazil; however, the case is less clear. After the signing of Mercosur in 1991 the FDI stock in Brazil increases to the same level as developing economies as a group. But, in 1995 the FDI stock as a percentage of GDP falls drastically and accordingly is then below the ratio of developing economies – this decrease most likely is explained by the change in statistics. Afterwards, the FDI stock/GDP ratio steadily increases until 2004 (with the exception 2002). This trend is very similar to the group of developing economies and the South and Central American region. But, the ratio is much lower until 2003. Hence, one could argue that stock adjustment was responsible for the high inflow figures of FDI to Brazil after 1995. However, the data from 2002, 2005, and 2006 do not fit in this picture. Furthermore, in the literature FDI stock adjustment refers to higher FDI inflows in a rather short period. Mexico’s FDI stock in relation to GDP decreases in 1994, the first year of NAFTA – most probably due to the methodology change in this year, but in the subsequent year the FDI stock/GDP ratio increases dramatically. However, this increase is likely explained by the fall of GDP after the ‘tequila crises’. If one leaves aside the exceptional years 1994 and 1995 and the period 1999–2001 the

Market Liberalism, Growth, and Economic Development in Latin America

The principal themes pursued in this book emerge from the great transformation that the Latin American and the Caribbean economies experienced in the aftermath of both the foreign debt crisis of 1982 and the macroeconomic stabilization policies that vividly and painfully produced the so-called ‘lost decade’ of the 1980s. Latin America implemented an economic liberalization process during the late 1980s and the 1990s. The main policy reforms involved in that course can be summarized as privatization of state owned firms, trade openness, deregulation of the foreign direct investment regime, and fiscal discipline. Latin Amer­ ican countries have also embarked in regional trade agreements, the most important ones being Mercosur and the North American Free Trade Agreement. This book compares results from the experience of North–South and South– South moulds of integration. Thus, the impacts of these policies on growth, development, technological progress, poverty, and inequality are analysed. Orthodox and heterodox economic policies and theories are discussed along with relevant empirical evidence with a view to assess, on the one hand, the relative merits of the various policy reforms applied by different countries in the region, and, on the other, the experience of integration into the global economy. There are 13 chapters in this collection linked in varying ways to the series of economic reforms introduced in the region in the last decades. The book will be of interest to academics, researchers, students, and policymakers interested in the study of economic development in emerging economies and in particular in Latin America. Gerardo Angeles-Castro is Head of Research and Postgraduate Studies at the School of Economics at Instituto Politécnico Nacional, México. Ignacio Perrotini-Hernández is Professor and Chair of the Graduate Faculty of Economics at Universidad Nacional Autónoma de México. Humberto Ríos-Bolívar is Research Economist at the School of Economics at Instituto Politécnico Nacional, México.

8 10

10

10 7 10 6 10 19 7

10

9 6 9 7 10 20 7

1992

8 10

1991

Source: WIR Annex Tables (2007).

World 8 Developing 10 economies South and Central 9 America Brazil 8 Argentina 6 Paraguay 9 Uruguay 8 Mexico 9 Canada 20 United States 7

1990

11 8 11 5 10 19 7

10

9 11

1993

10 9 12 6 8 20 7

10

9 12

1994

6 11 8 6 14 21 7

10

9 12

1995

6 12 9 6 14 22 8

11

10 13

1996

8 14 11 6 14 22 8

14

12 17

1997

11 16 15 7 15 24 9

16

14 20

1998

16 22 17 9 16 27 10

21

16 25

1999

Table 4.1 Inward FDI stock as a percentage of GDP, by host region and economy, 1990–2006

17 24 19 10 17 30 13

21

18 26

2000

24 30 18 13 23 30 13

26

20 27

2001

22 42 18 11 24 31 13

28

21 26

2002

26 37 20 16 27 34 13

31

22 27

2003

27 33 17 16 28 33 13

30

23 27

2004

24 30 17 18 27 31 13

29

23 26

2005

21 27 19 23 27 30 14

27

25 27

2006

North–South and South–South comparison

95

FDI stock/GDP ratio is similar to developing countries and/or South and Central America, before and as well as after NAFTA came into existence. Changes in inward FDI by sector For Mexico, in addition to the total FDI inflows and the FDI stock to GDP ratio it is interesting to see if the sectoral distribution of FDI has changed. In the case of Mexico the expectation is that after the implementation of NAFTA both FDI flows into the secondary and tertiary sector should increase. In the secondary sector, this is mainly because of higher vertical FDI inflows from Canada and the United States – which are likely to shift labour intensive production to Mexico as discussed in Section 4.2. In addition, extra-bloc countries might invest in the secondary sector with the aim to produce cheaply in Mexico and use the country as an export platform for the regional market. In the tertiary sector the expectation as well is that investment should increase because Mexico was forced to open up its service sector after the accession to NAFTA (UNCTAD, 1998). The best ‘indication of the geographic and sectoral distribution of FDI’ in general is provided by data about FDI stocks (IADB, 1998, p. 230). Unfortunately, the OECD, the UNCTAD or the Economic Secretariat of Mexico are not publishing stock figures after 2000. Therefore, for the period 2000–2006 inflow figures are taken. These flow figures do not display changes in the stock one to one – because of depreciation and revaluation of capital (IADB, 1998) – but at least they give some hindsight as to how the FDI stocks might have changed after 2000. Another problem with the data is that, as aforementioned, a methodology change took place in 1994. However, it looks as if this change had no great impact on the sectoral distribution of the inward FDI stock because the figures of 1994 are similar to 1993 (Figure 4.4). Anyhow, the available data show that the share of the secondary sector as a percentage of the total FDI stock has increased after NAFTA came into

Figure 4.4 Mexico inward FDI stock, 1990–2000 (% of total stock) (sources: OECD (2004, 2005); own calculation).

96

T. Goda

Figure 4.5 Mexico inward FDI flows, 2001–2006 (% of total inflows) (sources: Secretaría de Economía Mexico (2008); own calculation).

existence. From 1994 to 1998 the share increases, after a decrease between 1990 and 1994. Consequently, despite the decrease in 1999 and 2000, the share of the secondary sector is much higher in 2000 (57 per cent) than in 1993 (44 per cent). The inflows from 2001 to 2006 suggest that, despite higher investments in the tertiary sector at the beginning of the twenty-first century (especially because of the Biancci acquisition in 2001), the level of the secondary sector in 2006 is likely to be still higher than in 1993 (Figure 4.5). Thus, the effect of NAFTA on vertical FDI seems to be most important. These data give support to the view that the lock-in of liberalisation reforms as a result of North–South integration is not that important because many restrictions have been lowered already before the RIAs have been signed. In the case of Brazil the expectation is that, as a response to Mercosur, market-seeking tariff-jumping FDI in the secondary sector should increase. In a later phase and to a much lower extent the same probably is true for vertical FDI. Moreover, the opening up of some sectors as a result of Mercosur should lead to higher inflows, mainly in the tertiary sector. However, it is not obvious in which sector the effect should be stronger according to theory. Unfortunately, like for the data of Mexico, inward FDI stock data is only available until 2000. Therefore, flow figures are demonstrated for the time after 2000. The existing data show that the share of the secondary sector in the FDI stock in Brazil has halved between 1991 (68 per cent), when Mercosur was signed, and 2000 (34 per cent). The only exception in this downward trend is the year 1995 (Figure 4.6), but it is very likely that the methodological change in that year has led to this deviation. Although the flow figures between 2001 and 2006 indicate that the share of the secondary sector as a percentage of the total stock has gone up again, it has most probably not reached near the high level of 1990 (Figure 4.7). Thus, it can be said that in general the tertiary sector has attracted more FDI than the primary and secondary sector in Brazil since Mercosur was signed.

North–South and South–South comparison

97

Figure 4.6 Brazil inward FDI stock, 1990–2000 (% of total stock) (sources: UNCTAD (2008a); own calculation).

Figure 4.7 Brazil inward FDI flows, 2001–2006 (% of total) (sources: UNCTAD (2008a); own calculation).

Changes in inward FDI by home country/region According to theory and the findings from above Mexico should have received more (vertical) FDI from North America since NAFTA was signed and therefore, the share of North America’s inward FDI stock should have increased. This is true for 1994 when the share increased by seven percentage points. However, because of the statistic change in this year and because the sum of regional FDI stocks is bigger than 100 per cent, this increase does not seem to be very meaningful (Figure 4.8). After 1994 the share of North America decreases every year except 2000 and hence the share of 2000 (59 per cent) is much lower than that of 1993 (66 per cent). As for the industry data, there are no stock figures available after 2000. Hence, flow figures are used to consider changes in recent years. If we leave aside the year 2001 – when the Banacci acquisition by the US based City Group in a way distorts the picture – the unweighted average of the inflow figures suggest that the figures have not been changed much since then (Figure

98

T. Goda

Figure 4.8 Mexico inward FDI stock by regions, 1990–2000 (% of total stock) (sources: OECD (2004, 2005); own calculation).

Figure 4.9 Mexico inward FDI flows by region, 2001–2006 (% of total flows) (sources: Secretaría de Economía Mexico (2008); own calculation).

4.9). Thus, surprisingly Europe instead of North America has increased its share significantly after NAFTA came into existence. If one takes a look at country data (OECD, 2004 and 2005; Secretaría de Economía Mexico, 2008; own calculations) it becomes apparent that Canada increased its share between 1994 (1.4 per cent) and 2000 (2.5 per cent), while the United States – even though it remains the country which has by far the highest FDI stock in Mexico – lost shares after 1993 (minus eight percentage points by 2000). This is highly surprising as theory expects that investment from the United States should be over-proportional in comparison to extra-bloc countries.20 The flow figures after 2000 do not change this picture because even with an exceptional year in 2001 the US stock as a percentage of the total stock most likely has not increased back to old levels before NAFTA came into existence. In contrast, Canada probably could hold its percentage of the total stock at the

North–South and South–South comparison

99

level of around 2.5 per cent and accordingly its share has likely increased significantly since NAFTA. Regarding extra-bloc countries it is interesting to notice that in 2006 the share of Japan most probably is much below its pre-NAFTA amount. This is surprising as nearly one-third of Japan’s exports have flown into the NAFTA area since it came into existence (UNCTAD, 2007b; own calculation) and accordingly Japan should be expected to invest in Mexico to produce cheaply and then use the country as an export platform to circumvent tariff barriers. Perhaps, this is not the case because Japan prefers to have regional production networks in Asia, or because of RoO (e.g. for automobiles), or because the external barriers are not high ‘enough’ (e.g. for automobiles). Within Europe most notably Spain21 and the Netherlands improved their shares. Both countries can be expected to have shares around or above 10 per cent. In contrast, Germany’s FDI stock shares in Mexico are most probably lower in 2006 than they were in 1993 (4 per cent). This is just as surprising as the decrease in Japan’s share, even though the share of Germany’s exports to NAFTA has been much lower than that of Japan with an average of 10.5 per cent after 1994 (ibid., own calculation). In contrast to Mexico, Europe has the biggest FDI stock in Brazil. The European stock has decreased by eleven percentage points until 1997, but then recovers to reach 50 per cent in 2000, which is as high as it was in 1990 (Figure 4.10). The flow figures in the subsequent years suggest that the share should be close to this level in 2006 (Figure 4.11). The inward stock of North America develops similarly to Mexico. In the first two years after 1990 the share of the North American stock increases, but then it shrinks constantly and as a result is much lower in 2000 (26 per cent) than in 1990 (34 per cent). The flow figures indicate that the share will have even decreased slightly more by 2006. Hence, because the market shares of other developing countries have shrunk too, the only region able to gain significant shares after Mercosur came into existence is Latin America and the Caribbean. However, the increase of the share from this region

Figure 4.10 Brazil inward FDI stock by regions, 1990–2000 (% of total stock) (sources: UNCTAD (2008a); own calculation).

100

T. Goda

Figure 4.11 Brazil inward FDI inflows by regions, 2001–2006 (% of total inflows) (sources: UNCTAD (2008a); own calculation).

mainly comes from higher investment from the Caribbean22 which will be excluded from the analysis. If one takes a look at the countries with the most important shares of the FDI stock in Brazil it becomes visible that from 1990 to 2000 Spain (from nearly 0 to 12 per cent), the Netherlands (from 3 to 11 per cent), and Portugal (from nearly 0 to 4 per cent) increased their shares significantly, while Germany’s share shrunk dramatically from 15 per cent in 1990 to 5 per cent in 2000. Spain, the Netherlands, and Portugal have all invested mostly in the tertiary sector at the end of the 1990s. However, according to FDI inflow data from 2001 to 2006, Spain can be expected to have lost some of its stock share it gained during the 1990s. In contrast, the Netherlands seems to have increased its share, while Germany and Portugal should have kept their shares more or less stable. In 2000, the Canadian, US, and Japanese FDI stock shares are well below their 1990 shares. But, after 2001 Japan’s and Canada’s shares seem to have recovered a little bit, while the share of the United States is likely shrinking further. In contrast, the shares of Argentina and Uruguay increase until 2000 (from nearly 0 to 1 and 2 per cent, respectively). However, the average of the inflows after 2000 suggests that Argentina and Uruguay have not maintained these shares (data from UNCTAD, 2008a; own calculations). Thus, the data confirms the expectation that FDI within the region has hardly increased as a result of Mercosur. 4.3.2 Possible other explanations for increased inward FDI As aforementioned the increase in global FDI flows and stocks could be the main driver behind the increase of inward FDI in Mexico and Brazil. In addition, as discussed in the theoretical section, changes in inward FDI might have many other different influences besides the signing of a RIA. Therefore, in this section

North–South and South–South comparison

101

of the chapter, some other factors that might have led to higher FDI inflows will be discussed very briefly. They are as follows: (1) a change in the potential to attract FDI inflows, (2) privatisation and the opening up of the service sector, (3) changes in western stock markets, and (4) changes in the exchange rate. This chapter refers to the UNCTAD inward potential index to see how far the potential of Mexico and Brazil to attract FDI has changed. The potential index accounts for some (quantifiable) factors23 that are supposed to influence FDI inflows – the signing of RIAs is not included as a factor. According to this index, the potential of Brazil to attract FDI increases significantly after 1991 and reaches its highest point in 1997–1999 (when FDI inflows also increased dramatically). In Mexico, after a dip in 1994–1996, the potential index is also higher than before 1994 (data from UNCTAD, 2008b). Therefore, one could argue that FDI flows to Brazil after 1991 and to Mexico after 1996 have increased because the overall endowments of both countries have been more attractive for investors than before. However, the potential index is just a benchmark because many factors are not quantifiable (e.g. specific skills of the workforce, quality of local suppliers, etc.) and in the last 15 years the potential index has increased for many other countries as well. Another explanation why inflows might have increased regardless of the RIA status relates to privatisation and the opening up of sectors in which foreigners could not invest before. UNCTAD (2004) states that privatisation in the service sector has led to temporary higher inflows of FDI in Latin America: the telecommunication sector is given as example. It is estimated that in Brazil $64 billion of private capital has been invested in the telecommunications sector. In Mexico the amount has been considerably lower ($38 billion) and most of this investment has come from national sources. Hence, privatisation seems to be more important for Brazil than for Mexico in explaining higher FDI inflows. The data from Table 4.2 confirms these findings. While Brazil saw approximately $31.6 billion of FDI inflows in the late 1990s as a result of the privatisation, Mexico only received around $2.5 billion because of privatisation efforts. Unfortunately, there are no data available as to how much the opening up of formerly closed sectors in general has contributed to FDI inflows in Mexico and Brazil. That Brazil has received much more FDI in the tertiary sector than Mexico and that the share of M&As in total FDI inflows in Brazil between 1990 and 2006 have been much higher than in Mexico (Table 4.3) are hints, however, that the opening up of service sectors has likely been a driving force for FDI

Table 4.2 Transaction values of cross-border M&As of privatised firms (in US$ billion)

Brazil Mexico

1995

1996

1997

1998

1999

Total

– –

2.9 0.1

6.0 2.1

19.9 –

2.8 0.3

31.6 2.5

Source: authors’ elaboration with UNCTAD information.

22 88

14 0

1991

8 22

1992

48 42

1993 17 17

1994

Source: authors’ elaboration with UNCTAD information.

Brazil Mexico

1990 40 8

1995 61 16

1996 64 62

1997 102 24

1998 33 6

1999

Table 4.3 Share of cross-border M&A sales in total FDI inflows in Mexico and Brazil (%)

70 22

2000 31 62

2001

36 37

2002

52 8

2003

37 29

2004

38 21

2005

53 11

2006

North–South and South–South comparison

103

inflows to Brazil as compared to Mexico. This might explain as well why after 2000 – by which time the major privatisation programmes were completed – the flows of FDI to Brazil have decreased more drastically than in Mexico (UNCTAD, 2004). Another reason for a decrease of FDI flows to Latin America might be the devaluation of Brazil in that time and the stock market crises after 2000, which led to liquidity constraints (UNCTAD, 2004). As already discussed in the theoretical section, the exchange rate and the stock market might be important variables in explaining FDI flows. If we compare the average of the composite Dow Jones Index24 with FDI inflow indices from Brazil and Mexico it becomes visible that FDI flows to Brazil have reacted similarly to the stock market in the US, while a correlation is much less visible for Mexico (Figure 4.12). Regarding the exchange rate a clear correlation between the FDI inflows to GDP ratio and changes in the exchange rate is visible neither in Brazil nor in Mexico (Figure 4.13 and 4.14). Therefore, the major reason for decreasing flows to Mexico after 2000 – if one excludes the Banacci M&A – and the stagnation after 2005 might be that countries with lower labour costs than Mexico have become favourable for investors to serve as an export platform to the US (especially China).25 However, an analysis as to how far and for what reasons home countries have shifted their FDI to other locations is beyond the scope of this chapter.

4.4 Conclusions The relationship between regional integration and FDI is complex and the influence of RIAs on FDI inflows depends on many determinants like factor endowments, the pre-existing situation, investment provisions, the opening up of sectors, non-tariff barriers within the region (e.g. RoO), external barriers, and the

Figure 4.12 Dow Jones Index composite average (US$) and FDI inflows in Brazil and Mexico (indices, 1990 = 100), 1990–2006 (sources: Dow Jones Indexes (2008); WDI (2007); own calculation).

104

T. Goda

Figure 4.13 Brazil FDI inflows (as % of GDP) and exchange rate (LCU per US$) (source: WDI (2007)).

Figure 4.14 Mexico FDI inflows (as % of GDP) and exchange rate (LCU per US$) (source: WDI (2007)).

regional transport and communication infrastructure. Moreover, it is not clear how far other determinants besides the signing of a RIA might influence FDI inflows positively. Nevertheless, theory expects both Mexico and Brazil to receive higher FDI inflows as a result of their accession to NAFTA and Mercosur, respectively. While Brazil should receive mainly more market-seeking FDI in theory, Mexico is predicted to receive significantly more market-seeking and vertical FDI (especially from the US and Canada). Thus, in theory NAFTA should have been more beneficial for Mexico in attracting FDI than Mercosur for Brazil.

North–South and South–South comparison

105

The data confirm that in Brazil both FDI inflows and the FDI/GDP ratio are much higher than they were before Mercosur came into existence. The biggest share of this investment has come from European countries and has taken place in the tertiary sector. Therefore, privatisation and the opening up of the service sector seem to have played a decisive role in higher FDI flows to Brazil. However, it is unfortunately not clear how far Brazil would have privatised and opened up sectors without the existence of Mercosur. But, it is very likely that the degree of liberalisation would have been similar without Mercosur because during the 1990s the government of Brazil was in favour of liberalisation policies (Gardini, 2008). In addition to liberalisation, the increase in FDI inflows might have taken place because of changing global patterns, i.e. a worldwide increase in FDI flows. Furthermore, some factors that are believed to attract FDI have changed for the better in Brazil which might explain as well why flows to Brazil have been higher. Above all, the stock market bubble and bust between 1996 and 2002 seems to have been an important determinant for the amount of FDI inflows that Brazil received during this time. In Mexico, FDI inflows also increased significantly after the signing of NAFTA, mainly due to investment activities in the secondary sector – for Mexico privatisation seems to be much less important in attracting FDI than it was for Brazil. Therefore, one can argue that NAFTA has had a bigger impact on Mexico than Mercosur on Brazil. However, the changes in the FDI/GDP ratio do not confirm this view. Interestingly, the share of the inward FDI stock (as a percentage of the total stock) from the United States has decreased, while the European share of the total stock has increased. This finding is not expected by theory and is contrary to the results of previous studies. As for Brazil, there are other factors besides the RIA that might explain the increasing flows of FDI to Mexico, e.g. the FDI inward potential index of Mexico is higher after 1997. The most important factor next to the RIA, however, seems to be the increase of FDI flows to other middle-income countries during the NAFTA period. Mexico’s FDI inflows/GDP ratio is very much in line with these countries if one leaves out years with exceptionally high inflows (which can be explained by other determinants better than by NAFTA). Therefore, one can argue that the increase in FDI flows to Mexico might have taken place due to a global trend of higher investment in middle-income countries and not so much as a result of NAFTA. Ultimately, it should be kept in mind that for developing countries RIA and attracting FDI are only instruments to achieve economic development. If we have a look at some factors in this regard, the performance of Mexico in the last 14 years is rather poor. As UNCTAD (2007a, p. 66) puts it: ‘[NAFTA] does not appear to have helped accelerate output growth, nor does it seem to have contributed significantly to employment growth or to much higher standards of living of the Mexican people’. To achieve economic development, it is important to attract the ‘right’ FDI that brings along beneficial effects for the economy. This may require some regulations. Therefore developing countries should

106

T. Goda

cautiously calculate the gains and risks (i.e. the loss of alternative policy options) that are involved in North–South integration and keep in mind that regional integration is neither a necessary nor a sufficient condition to attract FDI.

Notes 1 RIA is a synonym for: preferential trade agreements, free trade agreements, customs unions, common markets, and economic unions (for a distinction between these agreements please have a look at Bergstrand et al., 2008). 2 North–South integration refers to an integration scheme where at least one developed and one developing country is involved. 3 South–South integration is a synonym for a RIA between developing countries (including countries in transition). 4 NAFTA is a free trade agreement and came into effect on 1 January 1994. Mercosur was founded on 26 March 1991 as a free trade agreement but has been extended to a common market until the 31 December 1994 (although an imperfect one). 5 The label of the product might need to contain different languages, customers in different markets may have different preferences and for this reason the product needs to have different appearances, or rules and regulation within the region are different and hence the production process needs to deviate slightly. 6 The reaction of the investment behaviour from MNCs that are located within the region. 7 The reaction of the investment behaviour from MNCs that are located outside the region. 8 Rules of origin are implemented to prevent trade deflection and will only exist if the RIA is not as deep as a customs union (for more information please consult Sanguinetti and Bianchi, 2006). 9 Stock adjustment means that the FDI stock of countries is below those of countries with comparable attributes and hence FDI inflows will be above the average level for some years to reach equilibrium (Buch et al., 2001). As example for FDI stock adjustment often Portugal and Spain are mentioned: ‘[S]tock adjustment took place after the countries joined the EC, with world investors rebalancing their portfolios in favour of Spain and Portugal in a process that lasted several years’ (Lederman et al. 2003). 10 In the case of NAFTA into the US or Canada. 11 While South–South regionalism is unlikely to have positive effects on FDI inflows into the poorest countries of the agreement. 12 All inflow data are given on a net basis. 13 Before 1995 data were compiled from the Brazilian Central Bank according to registry. In 1995 the system changed and data were compiled from foreign exchange transactions. In the first years after the change, only exchange transactions with the minimum level of $10 million were included in the statistics but this policy was abolished after 2000 (UNCTAD, 2004, p. 161). 14 Other middle-income countries are all countries that are classified from WDI (2007) as middle-income countries minus Brazil and Mexico. 15 In all sources FDI flows are presented only in current US$. 16 For detailed information about the ‘tequila crisis’ please have a look at Taylor (1998). 17 As said before, the increase in 1994 can possibly be explained by the change in methodology and the increase in 1995 by a slump in GDP due to the ‘tequila crisis’. The peak in 2001 can be explained by the acquisition of Banacci. 18 Quite the contrary is true. Both countries have had a lower FDI/GDP ratio in the year before the RIA came into existence. 19 However, unfortunately a comparison is difficult as the changes in methodology in both countries might have had a significant influence on the data. Furthermore, both countries have different methodologies to measure the inflow of FDI.

North–South and South–South comparison

107

20 Some empirical studies found that Mexico has received higher FDI inflows as a response to NAFTA solely because of higher flows from the United States and Canada (e.g. Tekin-Koru and Waldkirch, 2008). 21 Spain has increased its share mainly because of huge investments in the banking and telecommunications sector (Guedes and Gómez Olivarez, 2005). 22 The FDI that comes from the Caribbean offshore centres (i.e. British Virgin Islands and Cayman Islands), can be expected to be investment from the United States or Europe that is only channelled through the Caribbean with the aim to lower tax payments (IADB, 1998). 23

The variables are the rate of growth of GDP; per capita GDP; share of exports in GDP; telephone lines per 1,000 inhabitants; commercial energy use per capita; share of R&D expenditures in gross national income; share of tertiary students in the population; and country risk. (UNCTAD, 2002, p. 24)

The better the unweighted average of these factors in a three year period (to counteract annual fluctuations) the higher is the FDI potential index. 24 The Dow Jones Index is taken, because it is the world’s leading stock market. 25 For example, in the ‘maquiladoras’ factories in Mexico over 200,000 people have been dismissed in the period 2000–2001 (UNCTAD, 2003).

References Balasubramanyam, V.N., Sapsford, D., and Griffiths, D. (2002): ‘Regional Integration Agreements and Foreign Direct Investment: Theory and Preliminary Evidence’. Manchester School, Vol. 70(3), pp. 460–482. Bergstrand, J.H., Estevadeordal, A., and Evenett, S.J. (2008): ‘Introduction: The Sequencing of Regional Economic Integration’. World Economy, Vol. 31(1), pp. 1–4. Berthelon, M. (2004): ‘Growth Effects of Regional Integration Agreements’. Central Bank of Chile Working Paper No. 278, Santiago de Chile. Blomström, M. and Kokko, A. (1997): ‘Regional Integration and Foreign Direct Investment’. NBER Working Paper No. 6019, Cambridge. Buch, C.M., Kokta, R.M., and Piazolo, D. (2001): ‘Does the East Get What Would Otherwise Flow to the South? FDI Diversion in Europe’. Kiel Working Paper No. 1061, Kiel. Castilho, M. and Zignago, S. (2005): ‘Foreign Direct Investment, Trade and Regional Integration in Mercosur’. In: Graham, E.M. (ed.): Multinationals and Foreign Investment in Economic Development. Basingstoke: Palgrave Macmillan, pp. 145–162. Chudnovsky, D. and López, A. (2004): ‘Transnational Corporations’ Strategies and Foreign Trade Patterns in MERCOSUR Countries in the 1990s’. Cambridge Journal of Economics, Vol. 28(5), pp. 635–652. Cueveas, A., Messmacher, M., and Werner, A. (2005): ‘Foreign Direct Investment in Mexico since the Approval of NAFTA’. World Bank Economic Review, Vol. 19(3), pp. 473–488. Dow Jones Indexes (2008): ‘Dow Jones Averages: Composite’. Retrieved 26 June 2008 from: www.djindexes.com/mdsidx/index.cfm?event=showavgIndexData. Dunning, J.H. (1997): ‘Reconfiguring the Boundaries of International Business Activity’. In Boyd, G. and Rugman, A.M. (eds): Euro-Pacific Investment and Trade. Aldershot: Edward Elgar, pp. 1–18. Fiorentino, R.V., Verdeja, L., and Toqueboeuf, C. (2007): ‘The Changing Landscape of Regional Trade Agreements: 2006 Update’. WTO Discussion Paper No. 12, Geneva.

108

T. Goda

Gardini, G.L. (2008): ‘Who Invented Mercosur?’ Diplomacy and Statecraft, Vol. 18(4), pp. 805–830. Guedes, C.M. and Gómez Olivares, M. (2005): ‘Portuguese Investment in Brazil: The Challenges of an Iberian Logic’. In: Graham, E.M. (ed.): Multinationals and Foreign Investment. Basingstoke: Palgrave Macmillan, pp. 212–233. IADB (1998): Foreign Direct Investment in Latin America: Perspectives of the Major Investors. Madrid: IADB. Kim, Y.-H. (2007): ‘Impacts of Regional Economic Integration on Industrial Relocation through FDI in East Asia’. Journal of Policy Modeling, Vol. 29(1), pp. 165–180. Lederman, D., Maloney, W.F., and Servén, L. (2003): Lessons from NAFTA for Latin American and Carribean (LAC) Countries: A Summary of Research Findings. Washington, DC: The World Bank. Levy Yeyati, E., Stein, E., and Daude, C. (2003): ‘Regional Integration and the Location of FDI’. Inter American Development Bank Research Department Working Paper No. 492, Washington, DC. Medvedev, D. (2006): ‘Beyond Trade: The Impact of Preferential Trade Agreements on Foreign Direct Investment Inflows’. World Bank Policy Research Paper No. 4065. Nunnenkamp, P. (1996): ‘The Changing Pattern of Foreign Direct Investment in Latin America’. Kiel Institute for the World Economy Working Paper No. 736, Kiel. OECD (2004): ‘International Direct Investment Statistics Yearbook: 1990–2001’. Retrieved 3 June 2008 from: www.sourceoecd.com. OECD (2005): ‘International Direct Investment Statistics Yearbook: 1992–2003’. Retrieved 3 June 2008 from: www.sourceoecd.com. Ramirez, M.D. (2003): ‘Mexico under NAFTA: A Critical Assessment’. Quarterly Review of Economics and Finance, Vol. 43(5), pp. 863–892. Sanguinetti, P. and Bianchi, E. (2006): ‘Implementing PTAs in the Southern Cone Region of Latin America: Rules of Origin’. In: Cadot, O., Estevadeordal, A., Suwa-Eisenmann, A., and Verdier, Thierry (eds): The Origin of Goods: Rules of Origin in Regional Trade Agreements. Oxford: Oxford University Press, pp. 213–237. Schiff, M. and Winters, L.A. (2003): Regional Integration and Development. Washington, DC: The World Bank. Schlageter, K.M. (2005): Strategien der Automobilindustrie in Südamerika: Eine kritische Diskussion vor dem Hintergrund des neuen Regionalismus. Lohmar: Eul. Secretaría de Economía Mexico (2008): ‘Dirección General de Inversión Extranjera’. Retrieved 8 June 2008 from: www.si-rnie.economia.gob.mx/cgi-bin/repie.sh/reportes/ selperiodo. Sethi, D., Guisinger, S.E., Phelan, E., and Berg, D.M. (2003): ‘Trends in Foreign Direct Investment Flows: A Theoretical Empirical Analysis’. Journal of International Business Studies, Vol. 34(4), pp. 315–326. Taylor, L. (1998): ‘Lax Public Sector, Destabilizing Private Sector: Origins of Capital Market Crises’. CEPA Working Paper Series III, Working Paper No. 6, New York. te Velde, D.W. and Bezemer, D. (2004): ‘Regional Integration and Foreign Direct Investment in Developing Countries’. Retrieved 30 April 2008 from: www.unctad.org/en/ docs/iteiit20062a3_en.pdf. Tekin-Koru, A. and Waldkirch, A. (2008): ‘North–South Integration and the Location of Foreign Direct Investment’. MPRA Paper No. 6912, Munich. Tuman, J.P. and Emmert, C.F. (2004): ‘The Political Economy of U.S. Foreign Direct Investment in Latin America: A Reappraisal’. Latin American Research Review, Vol. 39(3), pp. 9–28.

North–South and South–South comparison

109

UNCTAD (1993): World Investment Report 1993: Transnational Corporations and Integrated International Production. New York: United Nations. UNCTAD (1998): World Investment Report 1998: Trends and Determinants. New York and Geneva: United Nations. UNCTAD (2002): World Investment Report 2002: Transnational Corporations and Export Competitiveness. New York and Geneva: United Nations. UNCTAD (2003): World Investment Report 2003: FDI Policies for Development: National and International Perspectives. New York and Geneva: United Nations. UNCTAD (2004): World Investment Directory: Volume IX Latin America and the Caribbean 2004 Parts 1 and 2. New York and Geneva: United Nations. UNCTAD (2005): World Investment Report 2005: Transnational Corporations and the Internationalisation of R&D. New York and Geneva: United Nations. UNCTAD (2007a): Trade and Development Report, 2007: Regional Cooperation for Development. New York and Geneva: United Nations. UNCTAD (2007b): UNCTAD Handbook of Statistics 2007 On-line. Retrieved 20 June 2008 from: www.unctad.org/Templates/Page.asp?intItemID=1890&lang=1. UNCTAD (2008a): ‘FDI/TNC Database’. Data received 12 February 2008 via email from Kotte, D. (Head, Macroeconomic and Development Policies Branch, UNCTAD). UNCTAD (2008b): ‘Inward FDI Potential Index’. Retrieved 7 June 2008 from: www. unctad.org/Templates/WebFlyer.asp?intItemID=2472&lang=1. WDI (2007): ‘The World Bank Group: World Development Indicators Online’. Retrieved May and June 2008 from: www.worldbank.org. WIR Annex Tables (2007): ‘Key Data from WIR Annex Tables’. Retrieved 3 June 2008 from: www.unctad.org/Templates/Page.asp?intItemID=3277&lang=1.

5

Trade blocs as determinants of trade flows in South American countries An augmented gravity approach Clemente Hernández-Rodríguez

5.1 Introduction In the last years trade flows have been evaluated in the literature, analysing the effects of such flows on the trade agreement itself and the influence on the multilateral trade system.1 Those works focus on the evaluation of the good and bad results of the trade agreements and identify the determinants of trade flows. As a facet of this, is the need to ascertain just how effective trading blocs are in promoting trade. To answer such questions, gravity models2 have been extensively used in the past (see, for example, Tinbergen (1962), Linnemann (1966), Aitken (1973), Frankel et al. (1995), Martinez-Zarzoso and Nowak-Lehmann (2003)). Gravity equation is the most frequent utilised approach in order to evaluate the determinants of the trade flows. Gravity equation assumes that the volume traded between two trade partners is an increasing function of their income and a decreasing function of the transportation cost. Distance between the economic centres of the countries involved is considered to approximate the transportation costs. McCallum (1995), Frankel (1997), Breuss and Egger (1999), Egger (2000), Soloaga and Winters (2001), Feenstra et al., (2001), Carrillo and Li (2002), and Martinez-Zarzoso and Nowak-Lehmann (2003), figure among the many empirical works that have used the gravity equation to evaluate the international trade flows. Latin American countries have embarked in regional trade agreements: the two most important being the Common Market of the South (Mercosur) and the Andean Community (CAN) in South America. In this work the impact of regional integration in South American countries is studied. The Mercosur and the CAN countries are chosen to verify if the pattern of the trade flows is similar or different in both regions. On 26 March 1991 the Heads of State of Argentina, Brazil, Paraguay, and Uruguay met in the Paraguayan capital to sign the Treaty of Asunción, with a view to forming the Mercosur. The Treaty of Asunción (1991) established Mercosur’s objectives: the liberalisation of intraregional trade; a common external tariff; harmonisation of laws and regulations concerning rules of origin; and the mutual consultation on macroeconomic policies. The rationale for Mercosur was

Trade blocs as determinants of trade flows

111

the notion of ‘Open Regionalism’. Unlike the protectionist, inward-oriented model of regional integration of the 1960s and 1970s, Open Regionalism is often presented as part of a broader agenda of the neoliberal economic reforms. On the other hand, the Andean Community, which began with the Cartagena Agreement in 1969, as the Andean Pact, is one of the oldest free trade movements still existing today. The original membership – Bolivia, Chile, Colombia, Ecuador, and Peru – has changed slightly with the inclusion of Venezuela (1973) and the break of Chile (1976), but the Pact remains generally intact. For nearly 20 years, its goals of economic growth and the creation of a regional common market floundered. Over the past several years, however, changes to democratic governance and developments in trade and investment liberalisation have prompted renewed interest and actions among Pact members. These actions have brought the Andean Pact to the forefront of the integration movement in Latin America. The Andean Pact has progressed beyond a free trade agreement, as it begins to implement a customs union similar to the European Union. This chapter seeks to evaluate the extent to which the establishment of the Mercosur and the CAN has led to an improvement of intra-regional trade, as promised by Open Regionalism, avoiding at the same time the reduction of the inter-regional trade. This analysis is interested only in the static effects in the trade flows. It does not contemplate other aspects as the negotiation process, policy coordination, credibility of the members, and the dynamic gains obtained from the trade flows that may be of interest. A gravity approach is utilised because this approach: (a) can explain the variations in the trade flows among a variety of countries and in different periods of time, (b) includes the geographic dimension, (c) improves the theoretical foundations using the imperfect substitutes theory, and (d) helps to examine the convenience of trade agreements, and its effect on the trade pattern, level, and direction of trade flows. The sample of data, 1987–2000, includes some years before the beginning of Mercosur (1991), and then allows the chance to see if the level and direction of the trade flows changed. Regarding CAN countries, it gives us the opportunity to see if the pattern of trade flows was altered with the renovation of the Andean Pact in 1988. One limitation of this analysis is that it is constrained to the South American region, excluding other regions such as the European Union. However, at the same time, that is an advantage because both the Mercosur and the CAN region share a similar political, social, economic, and even cultural centennial tradition. One of the contributions of this chapter is that it shows that the establishment of Mercosur and the CAN led to an increase of intra-regional trade, as the Open Regionalism view proposes, but did not lead to a contraction of extra-regional trade. This chapter consists of four additional sections. The next section shows the impact of the regionalisation in the Mercosur and CAN countries, observing the direction of the trade flows in the past decade. The third section presents the theoretical setting of the gravity model and provides a literature review about the gravity approach. In the fourth section, using an augmented gravity model and

112

C. Hernández-Rodríguez

Panel Analysis, we identify the determinants of the trade flows. The last section concludes.

5.2 Impact of the regionalisation in the level and direction of the trade flows in the Mercosur and CAN countries This work also compares results from the experience in the Mercosur and the CAN. So the extent to which the establishment of the Mercosur and the CAN has led to an improvement of intra-regional trade, as promised by Open Regionalism, avoiding at the same time the reduction of the inter-regional trade, is evaluated. Therefore, I present an analysis of the static effects in the trade flows. However, I first introduce some historical facts about the trading blocs under study. 5.2.1 The Mercosur countries Argentina, Brazil, Paraguay, and Uruguay signed the Mercosur agreement in 1991 (see Table 5.1). Mercosur went into effect in 1995, becoming a customs union. Following the entry into force of the common external tariff (CET) on 1 January 1995, the Mercosur countries must maintain a common commercial policy. Bolivia and Chile are associated countries of Mercosur without full membership status. Bolivia and Chile signed the association agreements with Mercosur in 1995 and 1996, respectively. Mercosur has also been trying to promote Chile’s full membership and inclusion into the Mercosur customs union in 2000. Hugo Chavez’s Venezuela has also been negotiating an inclusion to an expanded Mercosur. Mercosur is considered an emerging market offering good investment opportunities, with a population over 200 million people (it represents half of the population, 58 per cent of the GDP, and 40 per cent of the total foreign trade of Table 5.1 Mercosur and the CAN Agreement

Membership evolution

Type

Common Market of the South (Mercosur)

1991: Argentina

Customs union

Andean Community (CAN)

1988: Bolivia

1991: Brazil 1991: Paraguay 1991: Uruguay

1988: Colombia 1988: Ecuador 1988: Peru 1988: Venezuela

Preferential trading agreement

Source: the list of countries, and regional groupings, is given in an Appendix in Frankel et al. (1995).

Trade blocs as determinants of trade flows

113

Latin America and Caribbean). Mercosur is the third largest trading bloc in the world (after the European Union (EU) and the North American Free Trade Agreement (NAFTA)). The four Mercosur countries embrace an area larger than the continental United States. In 1998 the EU accounted for some 33 per cent of Mercosur’s imports and 39 per cent of its exports. In 2002, the EU imported five times more from Mercosur than the US, making it the group’s main trading partner. Trade in goods between EU and Mercosur has risen considerably in the last years, with the total value of trade flows between the two blocs rising from €18.8 billion in 1990 to €42.5 billion in 1998, an increase of almost 125 per cent. When compared with NAFTA, Mercosur is a distinct model of regional integration. Mercosur seeks to establish a European-style common market with a CET and coordinated macroeconomic policies. NAFTA, by contrast, is essentially a free trade area, and does not contemplate adopting a CET. By 31 December 1994, Mercosur was a free trade area covering 95 per cent of intra-regional trade. The increase from US$5.2 billion in 1991 to US$20.3 billion in 1997 in intra-Mercosur trade (fourfold increase) showed that in spite of divergent exchange rate policies it can be an established vibrant and successful free trade agreement. Due to the crisis in Brazil which resulted in a devaluation of the real in 1999, Argentina, Paraguay, and Uruguay sought exceptions from the CET. This development not only weakened Mercosur as a customs union but also had a negative effect on future negotiations. Especially the Argentinian crisis, which led to even more exceptions from the CET, has left doubts regarding the stability of Mercosur as a customs union and the solvency of Argentinian importers. Figure 5.1 consists of a set of different graphs. In those graphs we can observe that in the period 1987–2000 Mercosur increased the flows at the intraregion level. Argentina changed its exports from the EU countries to the Mercosur countries. Even Brazil increased its exports to the Mercosur countries, though its exports to other regions (NAFTA and the EU) are more important in the exports share. Paraguay increased its exports share to Mercosur countries, substituting export to the EU. Uruguay exhibits a similar evolution. 5.2.2 The CAN countries An economic agreement was forged in 1969 between several South American countries (Bolivia, Chile, Columbia, Ecuador, and Venezuela) to assist in reducing trade barriers and fostering the economic development of the members. The document that formalised the establishment of the Andean Pact is the Cartagena Agreement. Venezuela became part of the Andean Pact in 1973.3 In the meantime, Chile withdrew from the Andean Pact in 1976 to pursue more liberal trade policies.4 The initial stage of the Andean Pact was characterised by the Closed Regionalism Model. Intra-regional trade only increased from 1.7 per cent of total exports in 1970 to 4.5 per cent in 1979. This early stage of the Andean Pact

Figure 5.1 (a) Graphic evolution of exports by country member of Mercosur (source: Excel output using data from the IMF (International Monetary Fund), CEPAL (Comisión Económica para América Latina y el Caribe), and national sources). Notes CAN is Andean Community; MS is Mercosur; NAFTA is the North America Free Trade Agreement; EU is the European Union.

Figure 5.1 (b)

Trade blocs as determinants of trade flows

115

Figure 5.1 (c)

Figure 5.1 (d)

became very inefficient and failed for several reasons: many products were exempted from the tariff liberalisation process; a clear consensus about the common external tariff was lacking due to significant differences in the level of protection of each Andean country; the production requirements established by the Andean Pact did not match the trade needs of each country, especially after the foreign debt crisis; the market was too small; and trade activity was directed mainly to the members of the Andean Pact. Therefore, the Andean countries were limited in their capacity to generate new foreign exchange, which became very important for paying the increasing foreign debt. In 1985, the Andean Pact was practically moribund. Intra-regional trade did not follow the initial industrial

116

C. Hernández-Rodríguez

planning band, only about one-third of the investment programmes (metalmechanic, petrochemical, and automobile sectors) were approved. The external debt crisis that Latin America experienced during the 1980s led Andean countries to apply adjustment policies. These new policies reduced the trade preferences that had been established among the Andean countries and, thus, reduced trade during the mid-1980s. However, the Andean Pact reactivated itself with the Quito Protocol, which was signed in 1987 and later modified over the course of several Presidential Meetings. The members eventually established the CAN in which trade restrictions between members have been reduced. The CAN is a preferential agreement signed in 1988 by Bolivia, Colombia, Ecuador, Peru, and Venezuela, only Bolivia being landlocked (see Table 5.1). The most important modification, the Trujillo Protocol of 1996, resulted in the name being changed from the ‘Andean Pact’ to the ‘Andean Community of Nations’ (CAN), a new structural organisation, and a shift in emphasis from Closed Regionalism (inward integration) to Open Regionalism (outward integration) with the rest of the world.5 The Declaration of Santa Cruz of the CAN, signed in January 2002, introduces a new structure of the CET. The CAN expected to complete the definition of the CET in 2003 and of the Andean Common Market in 2005. The CAN is a sub-regional organisation of Latin America with a total population 120 million, an area of 4.7 million square kilometres, and a GDP of US$260 billion (in 2002). The region is rich in petroleum, minerals, agricultural and forestry resources. The CAN has a CET (average 11.7 per cent, in 2001) for imports from third countries. The total imports of the Community in 2000 were US$40 billion and exports US$57.5 billion. The Community coordinates the position of member countries and speaks with one voice in the World Trade Organization and other international and regional forum. In keeping with the Open Regionalism approach, the CAN has supported the agenda of becoming part of wider economic agreements such as Mercosur and the attempt of the Free Trade Area of the Americas (FTAA). Intra-regional trade has already increased tremendously with the existing free trade zone agreement. Trade between Colombia and Ecuador reportedly increased by 76 per cent from 1992 to 1993; meanwhile, Venezuela–Colombia trade grew from $1 billion in 1992 to nearly $2.5 billion in 1993, an astounding increase of 250 per cent. Similar gains are observed for Peru, with full reinsertion into the Community. From 1992 to 1993, Peru experienced an increase of 29 per cent in intra-regional CAN trade. Figure 5.2 is composed of a set of different graphs corresponding to members of the CAN. In those graphs we can observe that from 1987 to 2000 the CAN steadily increased the flows at the intra-region level. However, the most important trade flows still are placed in regions outside the CAN, at the inter-regional level. For the entire sample, NAFTA is the first target of their exports, followed by the EU. The CAN occupies the third place in the exports share, and Mercosur occupies the last place. Figure 5.1 and Figure 5.2 show that the establishment of the CAN led, on the one hand, to an increase of intra-regional trade, as the Open Regionalism view

Figure 5.2 (a) Graphic evolution of exports by country member of the CAN (source: Excel output using data from the IMF (International Monetary Fund), CEPAL (Comisión Económica para América Latina y el Caribe), and national sources). Notes CAN is Andean Community; MS is Mercosur; NAFTA is the North America Free Trade Agreement; EU is the European Union.

Figure 5.2 (b)

118

C. Hernández-Rodríguez

Figure 5.2 (c)

Figure 5.2 (d)

proposes, and, on the other hand, that there is no contraction of extra-regional trade in both cases.

5.3 Gravity model In the gravity model, the antecedents of which are found in Tinbergen (1962), Pöyhönen (1963), and Linnemann (1966), trade between two countries is analogous to the gravitational force between two objects: directly related to the countries’ size (or income), and inversely related to the distance between them. The general hypothesis of the gravity model is that the trade flows between two countries is an increasing function of their income (GDP) and population, and it is a decreasing function of the distance between them.

Trade blocs as determinants of trade flows

119

Figure 5.2 (e)

The hypotheses of the augmented gravity model are: a b c d e f

more income and the population in a pair of countries, increase trade flows; more distance between two countries diminish trade flows; higher indexes of price in a pair of countries, increase trade flows; a common border encourage trade flows in a pair of countries; a trade agreement common to a pair of countries encourage trade flows; the most important determinants of trade flows in a pair of countries are income and distance between the two of them.

Since Tinbergen (1962) and Pöyhönen (1963) applied the gravity equation to analyse international trade flows, the gravity model has become a popular instrument in empirical foreign trade analysis. The model has been successfully applied to flows of varying types such as migration, foreign direct investment, and more specifically to international trade flows. According to this model, exports from country i to country j are explained by their economic sizes (GDP or GNP), their populations, direct geographical distances, and a set of dummies incorporating some kind of institutional characteristics common to specific flows. Theoretical support of the research in this field was originally very poor, but since the second half of the 1970s several theoretical developments have appeared in support of the gravity model. Anderson (1979) made the first formal attempt to derive the gravity equation from a model that assumed product differentiation. Bergstrand (1985, 1989) also explored the theoretical determination of bilateral trade in a series of papers in which gravity equations were associated with simple monopolistic competition models. Helpman and Krugman (1985) used a differentiated product framework with increasing returns to scale to

120

C. Hernández-Rodríguez

justify the gravity model. More recently Deardorff (1998) has proven that the gravity equation characterises many models and can be justified from standard trade theories. Finally, Anderson and Wincoop (2003) derived an operational gravity model based on the manipulation of the CES expenditure system that can be easily estimated and helps to solve the so-called border puzzle. The differences in these theories help to explain the various specifications and some diversity in the results of the empirical applications. There are a huge number of empirical applications in the literature of international trade, which have contributed to the improvement of performance of the gravity equation. Some of them are closer regarding this work. First, in recent papers, Mátyás (1997, 1998), Breuss and Egger (1999), Egger (2000), and Cheng and Wall (2002), improved the econometric specification of the gravity equation. Second, Bergstrand (1985), Helpman (1987), Wei, (1996), Limao and Venables (1999), Bougheas et al. (1999), and Soloaga and Winters (2001), among others, contributed to the refinement of the explanatory variables considered in the analysis and to the addition of new variables. According to the generalised gravity model of trade, the volume of exports between pairs of countries, Xij, is a function of their incomes (GDPs), their populations, their geographical distance, and a set of dummies. For estimation purposes, the gravity model, in log-linear form for a single year, is expressed as:

(5.1) where, Xij are the exports from country i to country j, i is the exporter, j is the importer; Yi is the GDP in country i; Yj is the GDP in country j; Li is the population of country i; Lj is the population of country j; Dij measures the distance between the two countries’ (country i and country j) capitals (or economic centres); dij is a vector of dummy variables, they take the value one when a certain condition is satisfied (e.g. belonging to a trade bloc), zero otherwise; eij is the error term of the bilateral trade flow; aij is the constant in the bilateral trade flow, representing any other factors aiding or preventing trade between pairs of countries; and bk, for k = 1, . . ., 6, is the corresponding coefficient of the explanatory variable or vector of variables.6 The gravity specification, like many models of trade, contains a role for income, asserting that countries with higher income will trade more. A high level of income in the exporting country indicates a high level of production, which increases the availability of goods for exports. Therefore we expect b1 to be positive. The coefficient of Yj, b2, is also expected to be positive since a high level of income in the importing country suggests higher imports. The coefficient estimate for population of the exporters, b3, may be negative or positive signed, depending on whether the country exports less when it is big (absorption effect) or whether a big country exports more than a small country (economies of scale). The coefficient of the importer population, b4, has also an ambiguous

Trade blocs as determinants of trade flows

121

sign, for similar reasons. The distance coefficient, b5, is expected to be negative since it is a proxy of all possible trade cost sources.7 While the basic gravity specification relates trade to income and distance, the ‘full’ gravity model (using Frankel’s (1997) terminology) also allows a role for income per capita, sharing a common border, and membership in a trading bloc. I also incorporated dummy variables for trading partners sharing a common border as well as dummy variables for trading blocs evaluating the effects of preferential trading agreements. The coefficients of all these trade variables are expected to be positive. In this research an extended gravity specification, for a single period t (as in Frankel et al. 1995), is specified as follows:

(5.2) where, (Xit + Xjt) is the trade volume measured in 1995 US dollars from country i to country j. This is the sum of Xit, the exports from country i, to country j at time t, plus Xjt, the exports from country j, to country i at time t; Yit/Yjt is the relative economic size, in terms of GDP, of country i and country j in the year t, measured in 1995 US dollars; Lit/Ljt is the relative size, in terms of population, of country i and country j in the year t; Dij is the distance between the capital (economic centres) of country i to the capital (economic centre) of country j; Pit/Pjt are the relative prices, in terms of consumer price indexes, of country i and country j in the year t; Aij is a dummy variable for adjacency, it is equal to one if country i and country j share a common border, zero otherwise; Tijt is a dummy variable equal to one if country i and country j are members of the same trade bloc in the year t, zero otherwise; eijt is the error term of the bilateral trade flow in time t; aijt is the specific effect associated to the trade flow. These effects allow us to control for omitted variables in the bilateral trade flow; bk, for k = 1, . . ., 6, is the corresponding coefficient of the explanatory variable. Income enters positively in the equation, implying that trade between two medium-sized countries should exceed trade between a small and a large country (Cyrus 2002). Such an outcome would result from a Helpman and Krugmantype (1985) model of monopolistic competition, but Deardorff (1998) has shown that a standard Heckscher–Ohlin framework can produce the same outcome, so the empirical success of the gravity specification cannot be used as evidence to support a particular theory of trade.8 Distance is a proxy for transportation costs and should have a negative coefficient. The dummy variables for the common border and for trading bloc membership help to determine to what extent trade is due to geographic or political, as opposed to economic, factors.

5.4 Empirical research Gravity models are almost always estimated using ordinary least squares (OLS). This specification contains a potential problem, however: the causality between

122

C. Hernández-Rodríguez

income and trade is not clear-cut. The gravity equation suggests that high income causes high trade, but perhaps it is trade that instead causes income to be high. Alternatively, it is also possible that another factor, such as free-market government policies, pushes up both income and trade. In that case, the gravity equation is mis-specified, for income will be correlated with the error term in the regression; thus, OLS will not provide consistent estimates. Instead, the OLS estimates will overstate the importance of income. In addition, the bias to the income coefficient may also bias the other coefficients in the equation; it cannot be assumed that any of the OLS coefficients can be reliably estimated. There are two important factors that we want to capture. First, heterogeneity across countries in trade flows. Second, how the business cycle, or time, will affect bilateral trade flows. To identify these effects, and hence correctly specify the econometric model, one requires a pooled time-series of cross-sections (panel data) of the countries of interest, in this case countries of the Mercosur and CAN. In the context of the gravity model, we can identify the business cycle or local or exporting country effects, by treating and estimating them as constants in a fixed effects model. Although gravity models have been criticised for their lack of theoretical underpinning, empirically (especially in forecasting) they seem to perform particularly well, and are therefore well suited for policy analysis. In constructing the empirical model I consider a sample of nine countries, four countries from Mercosur (Argentina, Brazil, Paraguay, and Uruguay) and five from the CAN (Bolivia, Colombia, Ecuador, Peru, and Venezuela). The time period under study goes from 1987 to 2000. Countries are ordered by pairs in order to capture the bilateral flows. The data set includes the period 1987–2000. I use the database built by Frankel et al. (1995), to extract the sample by restricting on observations with no missing values over the period 1987 to 2000. I obtain a database covering nine countries, and 36 country-pairs. We thus have a balanced panel structure. I estimated the gravity model of trade described in equation (5.2), in a panel data framework. The use of panel data methodology has several advantages over cross-section analysis. First, panels make it possible to capture the relevant relationships among variables over time. Second, a major advantage of using panel data is the ability to monitor the possible unobservable trading-partner pair individual effects. When individual effects are omitted, OLS estimates will be biased if individual effects are correlated with the regressors. I constructed three different balanced panels. The first panel includes the 36 combined pairs, which includes both the Mercosur and CAN countries. The second panel includes only pairs of Mercosur countries. The third panel includes only pairs of the CAN countries. The last two panels are constructed to test the gravity equation in the intra-regional level. In these panels, I include six variables belonging to the gravity equation. Those variables are: (1) Exports, (2) Population, (3) Distance, (4) Price Indexes, (5) Adjacency, and (6) Trade Agreement. On the one hand, there is Exports (endogenous variable), and, on the other hand, there are GDP, Population, Distance, and Price Indexes (exogenous variables). Additionally, there are two other exogenous variables: those are the dummy variables Adjacency and Trade

Trade blocs as determinants of trade flows

123

Agreement. The dummy variable Adjacency is not included in the second panel given that adjacency is only present in a pair of countries (Paraguay–Uruguay). Trade Agreement is included only as an explanatory variable in the first panel. It does not make sense to include it in the other two panels given the definition of those panels. 5.4.1 Results Equation (5.2) is estimated taking into account both fixed and random effects.9 The results are presented in Table 5.2. Fixed effects turn to be the best approach given that the variance for the estimated coefficients within groups is less than the one for the estimated coefficients between groups. In the fixed effects approach we can see the increase in the commercial flows over time.

Table 5.2 Gravity equation for the panels of Mercosur and CAN: random effects Variables GDP (Y) Population (L) Distance (D) Price indexes (P) Adjacencies (A) Trading bloc (T) Constant-α 87-α 88-α 89-α 90-α 91-α 92-α 93-α 94-α 95-α 96-α 97-α 98-α 99-α 00-α R2 Adjusted-R2 Durbin Watson Statistic Observations

Coefficient (fixed effects) 0.37** (0.07) 0.12 (0.08) 0.97** (0.14) 0.27* (0.15) 1.74** (0.16) 2.06** (0.19) – 1.61 1.46 1.56 1.83 2.13 2.33 2.45 2.56 2.98 2.99 3.09 3.26 3.01 3.10 0.76 0.75 2.00 504

Coefficient (random effects) 0.43** (0.08) 0.06 (0.09) 1.00** (0.16) 0.15 (0.17) 1.78** (0.19) 2.09** (0.22) 2.20 (1.29) – – – – – – – – – – – – – – 0.53 0.52 1.53 504

Sources: Eviews output using data from the IMF (International Monetary Fund), CEPAL (Comisión Económica para América Latina y el Caribe), and national sources. Notes Standard error is in parentheses; ** significance of 5%, * significance of 10%.

124

C. Hernández-Rodríguez

Table 5.2 shows that GDP, distance, CPI, and the two dummy variables are significant. The signs of the coefficients are the expected signs, with the clear exception of distance (distance was expected to be negative in the gravity equation). The positive sign in distance means that distance does not affect inversely the increase in the trade flows. Then the sign of the variable distance does not undermine the theories of international trade, but challenges the traditional gravity equation. However, population is not significant. In order to clarify the non-significance of population, we look at the correlation matrix and see that population is correlated to GDP ( ρ = 0.84). Collinearity is another problem in the gravity equation, and population seems to be the cause. Table 5.2 in the fixed effects column shows that the establishment of the South American trade blocs (the Mercosur and the CAN) led to an increase of total regional trade (from the positive sign in the Trading Bloc dummy variable), as the Open Regionalism view proposes. The second and third panels are estimated to test the gravity equation for each Trade Agreement. Table 5.3 shows the results of the fixed effects approach. Table 5.3 Gravity equation for the panels of Mercosur and CAN: fixed effects Variables

Mercosur

GDP (Y) Population (L) Distance (D) Price indexes (P) Adjacencies (A) 87-α 88-α 89-α 90-α 91-α 92-α 93-α 94-α 95-α 96-α 97-α 98-α 99-α 00-α R2 Adjusted-R2 Durbin Watson Statistic

0.69** (0.39) –0.79* (0.38) 0.54* (0.19) 0.61 (0.75) – 8.48 8.47 8.87 8.79 8.92 9.11 9.49 9.74 9.99 10.11 10.21 10.25 10.01 10.17 0.46 0.33 3.26

Observations

84

CAN –0.07 (0.10) 0.46** (0.07) 0.11 (0.49) 0.14 (0.48) 2.00** (0.41) 10.26 9.29 9.21 9.56 10.11 10.42 10.46 10.82 11.16 11.06 11.33 11.62 11.49 11.67 0.50 0.49 0.43 140

Sources: Eviews output using data from the IMF (International Monetary Fund), CEPAL (Comisión Económica para América Latina y el Caribe), and national sources. Notes Standard error is in parentheses; ** significance of 5%, * significance of 10%.

Trade blocs as determinants of trade flows

125

Table 5.3 shows that, for Mercosur, GDP, population (L), and distance (D) were significant. But price index (P) was not significant due to collinearity with the GDP. The only expected sign belongs to the GDP. The coefficient of population (L) was expected to be positive, and is negative. One possible explanation is the high correlation with the GDP ( ρ = 0.92). Distance was expected to have a negative coefficient and had a positive sign. Table 5.3 shows that, for the CAN countries only population (L) and adjacencies (A) were significant. The coefficient of population has the expected positive sign. The coefficient of GDP is negative but not significant. The coefficient of the dummy adjacency is highly correlated with distance ( ρ = 0.81). But price index (P) was not significant due to collinearity with the GDP. The only expected sign belongs to the GDP. The coefficient of population was expected to be positive, and is negative. One possible explanation is the high correlation with the GDP (0.92). Distance was expected to have a negative coefficient and also had a positive sign. Given the results showed in Table 5.3, the gravity equation fails to support that the main determinants of the trade flows are income and distance. In both panels the sign of the coefficient of distance is positive and in the CAN panel income is negative and not even significant. Figure 5.3 shows a graphic evolution of the fixed effects for both South American trading blocs. This figure confirms the Open Regionalist view that the establishment of a trading bloc (in this case the Mercosur and the CAN) leads to an increase of intra-regional trade flows.

Figure 5.3 Graphic evolution of gravity equation fixed effects for Mercosur and CAN (source: Fixed effects coefficients (Eviews output) using data from the IMF (International Monetary Fund), CEPAL (Comisión Económica para América Latina y el Caribe), and national sources).

126

C. Hernández-Rodríguez

5.4.2  Why is the coefficient of distance positive? Robustness and  specification tests One hypothesis for the positive coefficient on distance is that distance is working as an instrumental variable. To test if distance is such an instrumental variable we use the Hausman Exogeneity Test. Thus, the next step is to determine whether distance is an instrumental variable and if it is valid. The first requirement of good instruments is that they be highly correlated with the variable for which they are instrumenting. The Hausman Exogeneity Test consists in the contrast of a regression in which distance is the only regressor yt = α0 + α01Dt + εt, and yt = α + βDt + δut + εt, where u are the residuals of the regression of the variable D on the instrumental variable DR, ut = D – (θ0 + θ1DRt). This contrast will be done assuming that the regressor is non-stochastic (null hypothesis). The alternative hypothesis is that the regressor is stochastic (only the instrumental variables have this property). The first-stage regression in Table 5.4 shows that distance is indeed highly significant in explaining trade flows. Hausman Tests were performed for the coefficients and regressions. Examining Hausman Tests for particular coefficients allows us to see whether we can trust the OLS results in all cases. The rejection of the null hypothesis implies that we use distance as an instrument. The purpose of the Hausman Test is to determine whether there is indeed correlation between GDP (or, more generally, the regressors) and the error term. The null hypothesis is that there is no correlation, so OLS provides consistent and efficient estimates; if this is true, then the Instrumental Variables (IV) estimates should be similar to the OLS estimates (the IV estimates should equal the OLS estimates plus noise). So, the Hausman Test is a test of equality between the OLS estimates and the IV estimates. In this case, the null hypothesis is rejected at a 90 per cent level; in other words, we do not fail to reject the OLS specification. It implies that distance must be an endogenous variable. The second requirement of good instruments is that they be uncorrelated with the error term. To determine this, tests of over-identifying restrictions were run. This test involves regressing the residual of the OLS regression on the instruTable 5.4 Exogeneity Hausman Test for distance in the gravity equation Variable

Coefficient

Std. Error

t-Statistic

Prob.

DR RESID01 C R-squared Adj. R-squared F-statistic Prob. (F-statistic)

1.632954 –1.624571 –3.697832 0.462191 0.429596 14.18001 0.000036

0.306643 0.370665 2.351061

5.32526 –4.38285 –1.572836 Akaike AIC Schwarz SC

0.0000 0.0001 0.1253 2.923188 3.055148

Durbin–Watson stat

2.505928

Sources: Eviews output using data from the IMF (International Monetary Fund), CEPAL (Comisión Económica para América Latina y el Caribe), and national sources.

Trade blocs as determinants of trade flows

127

ments. I analysed the individual explanatory capacity of the variable distance. It is very difficult for the goodness of fit of this regression to be highly significant.10 Table 5.5 shows that distance is a relevant variable to explain the trade flows. Interestingly in this case the regression shows the expected negative sign. It may perhaps be better to examine the adjusted-R2 of this regression, which is never above 0.100; this indicates that, while the individual instrument is significant, it has virtually no power to explain the error term. Recently Bougheas et al. (1999) showed that transport costs are a function not only of distance but also of public infrastructure. They augmented the gravity model by introducing additional infrastructure variables (stock of public capital and length of motorway network). Their model predicts a positive relationship between the level of infrastructure and the volume of trade, which is supported using data from European countries. Martinez-Zarzoso and Nowak-Lehmann (2003) took a further step in this direction by introducing a new infrastructure index (taking information on roads, paved roads, railroads, and telephones) and differentiating between exporter and importer infrastructure as explanatory variables of bilateral trade flows. Due to the unavailability of data in this case I was unable to reproduce a similar index.

5.5 Conclusions In order to study the impact of regional integration in South America – which includes the Mercosur and the CAN – an augmented gravity model approach was utilised. Data consist of a balanced panel data of 36 trading pairs during 14 years, in the period 1987 to 2000. A gravity equation for trade flows was estimated. Trade flows from country i to country j was used as the dependent variable. Explanatory variables are GDP for both source and destination country (market size), population of both countries, distance between the two countries, and dummy variables for adjacency and trade agreement. The specification is log-linear and the procedure estimation identified collinearity, caused by price index (P) and GDP. This could cause some coefficients

Table 5.5 Distance as the only regressor in the gravity equation Variable

Coefficient

C 19.2276495 D –1.10523307 R-squared 0.10029443 Adj. R-squared 0.09850218 F-statistic 55.960309 Prob. (F-statistic) 0

Std. Error

t-Statistic

Prob.

1.05685158 0.14774535

18.1933299 –7.48066234

0.0000 0.0000

Durbin–Watson stat

0.894435

Sources: Eviews output using data from the IMF (International Monetary Fund), CEPAL (Comisión Económica para América Latina y el Caribe), and national sources.

128

C. Hernández-Rodríguez

like population to have a sign not expected. One of such counterintuitive coefficients is a negative sign for distance. It seems that in this aspect – a positive coefficient in the distance variable – the gravity model does not fit the empirical evidence, at least for this panel. The econometric results, nevertheless, suggest that the factors that influence the most the flows of trade are geographic factors (adjacency), followed by size (income). This work is aimed at exploring how these attempts of regional integration have increased trade volume and trade flows. The following results are found: the establishment of Mercosur and the CAN led to an increase of intra-regional trade, and it has not resulted in a contraction of extra-regional trade. In terms of volume, with the exception of Brazil, for countries that are members of Mercosur, their trading bloc is the most important partner. NAFTA is the most important trading bloc partner for countries of the Andean Pact, and then the EU and Mercosur. The estimations reveal that both Mercosur and the CAN have increased the total regional trade as the Open Regionalism view proposes. For the Mercosur case, there is an increase in the trade flows and trade volume both at the intraregional level and the extra-regional level. These findings are in line with the importance of Mercosur, the customs union, at the intra-regional level. However, for the CAN extra-regional trade flows are more important, pointing to the fact that there is little evidence of a significant effect of this Preferential Agreement in the intra-regional trade as may be expected. In this context, it seems that the South American trade blocs’ characteristics must be studied to understand the reasons of the success of Mercosur when compared to the CAN.

Notes 1 Trade flows are the trade transactions among countries, and are represented by imports and exports. 2 The reason for the name of ‘gravity model’ is the analogy to Newton’s law of gravity: Just as the gravitational attraction between any two objects is proportional to the product of their masses and diminishes with distance, the trade between any two countries is, other things equal, proportional to the product of their GDPs and diminishes with distance. (Krugman and Obstfeld, 2009: 14) 3 The Andean Pact was formed in 1969 to reverse the stagnation of the Latin American Association of Free Trade and to address the development needs of the Andean countries (Venezuela, Colombia, Chile, Ecuador, Peru, and Bolivia). 4 The Pact sought to harmonise policies, define a common external tariff, liberalise intra-regional trade, regulate foreign direct investment in the region, and to organise production across member Andean countries by encouraging the development of promising industries. This strategy was consistent with the Import Substitution, or Closed Regionalism, Model that predominated in Latin America during the 1970s. According to this model, the government must coordinate economic policies and regional development plans in order to direct the market towards proposed goals. The consequence of this model is that protected rent activities develop, mainly in the industrial sector, which are financed in part by the resources generated by primaryresource-intensive exports.

Trade blocs as determinants of trade flows

129

5 The establishment of the Andean Free Trade Zone (AFTZ) in 1993 and the Andean tariff union or the Andean common external tariff in 1995 gave rise to private initiatives and innovative rent-seeking activities – instead of protected rent activities – that aimed at achieving an efficient allocation of resources and exploiting the competitive advantages of the region. This increasing efficiency and innovation is the main reason behind the shift towards the Open Regionalism model. The next step for the CAN is the establishment of the Andean Common Market in the year 2005, as ratified in the presidential meeting of the Andean countries in June of 2000. This market would enable the free movement of goods, services, capital, and people. 6 An alternative formulation of equation (5.1) uses per capita income instead of population, where YHi (YHj) are the exporter (importer) GDP per capita. This alternative formulation is equivalent to the formulation in (5.1). The specification given by equation (5.1) is often used to estimate aggregated exports (Endoh, 2000). 7 Traditionally, the gravity model uses distance to model transport costs. 8 Income per capita is included in a gravity equation in order to gauge the importance of development rather than mere size. Even if bigger countries trade more, it also seems that richer countries, whatever their size, engage in more trade, so we expect the coefficient on GDP per capita also to be positive. 9 Under fixed effects we control years. That is, we try to see the individual effect for each pair of countries in each year. 10 Since the high number of observations, even a tiny amount of correlation between the residual and the instruments will cause the instruments to appear invalid.

References Aitken, N.D. 1973. ‘The Effect of the EEC and EFTA on European Trade: An Empirical Cross Section Analysis’, American Economic Review 63 (5): 881–892. Anderson, J.E. 1979. ‘A Theoretical Foundation for the Gravity Equation’, American Economic Review 69 (1): 106–116. Anderson, J.E. and Wincoop, E.V. 2003. ‘Gravity with Gravitas: A Solution to the Border Puzzle’, American Economic Review 93 (1): 170–192. Bergstrand, Jeffrey H. 1985. ‘The Gravity Equation in International Trade: Some Microeconomic Foundations and Empirical Evidence’, Review of Economics and Statistics 67 (3): 471–481. Bergstrand, Jeffrey H. 1989. ‘The Generalized Gravity Equation, Monopolistic Competition, and the Factor-Proportions Theory in International Trade’, Review of Economics and Statistics 71 (1): 143–153. Bougheas, S., Demetriades, P.O., and Morgenroth, E.L.W. 1999. ‘Infrastructure, Transport Costs and Trade’, Journal of International Economics 47 (1): 169–189. Breuss, F. and Egger, P. 1999. ‘How Reliable are Estimations of East–West Trade Potentials based on Cross-Section Gravity Analyses?’ Empirica 26 (2): 81–95. Carrillo, C. and Li, C.A. 2002. ‘Trade Blocks and the Gravity Model: Evidence from Latin American Countries’, Working Paper (downloaded), Department of Economics, University of Essex, UK. Cheng, I-Hui and Wall, Howard J. 2002. ‘Controlling for Heterogeneity in Gravity Models of Trade and Integration’, revision of Federal Reserve Bank of St Louis Working Paper 1999-OlOC, p. 33. Cyrus, Teresa L. 2002. ‘Income in the Gravity Model of Bilateral: Does Endogeneity Matter? International Trade Journal 16 (2), 161–180. Deadorff, Alan V. 1998. ‘Determinants of Bilateral Trade: Does Gravity Work in a

130

C. Hernández-Rodríguez

Neoclassical World?’ In: Jeffrey A. Frankel (ed.), The Regionalization of the World Economy, University of Chicago Press: Chicago, IL, pp. 7–31. Egger, P. 2000. ‘A Note on the Proper Econometric Specification of the Gravity Equation’, Economics Letters 66 (1): 25–31. Endoh, M. 2000. ‘The Transition of Post-War Asia-Pacific Trade Relations’, Journal of Asian Economics 10 (4): 571–589. Feenstra, R. 2003. Advanced International Trade: Theory and Evidence, Princeton University Press: Princeton, NJ, Chapter 5. Feenstra, R., Markusen, James A., and Rose, Andrew K. 2001. ‘Using the Gravity Equation to Differentiate among Alternative Theories of Trade’, Canadian Journal of Economics 34 (2): 430–447. Frankel, J. 1997. Regional Trading Blocs in the World Economic System, Institute for International Economics: Washington, DC. Frankel, J., Stein, Ernesto, and Wei, Shang-Jin. 1995. ‘Trading Blocs and the Americas: The Natural, the Unnatural, and the Super-Natural’, Journal of Development Economics 47 (1): 61–95. Hamilton, C. and Winters, L.A. 1992. ‘Opening up International Trade with Eastern Europe’, Economic Policy 14 (April): 77–116. Helpman, E. 1987. ‘Imperfect Competition and International Trade: Evidence from Fourteen Industrial Countries’, Journal of Japanese and International Economics 1 (1): 62–81. Helpman, E. and Krugman, P. 1985. Market Structure and Foreign Trade, MIT Press: Cambridge, MA. Krugman, P. and Obstfeld, M. 2009. International Economics: Theory and Policy, Pearson-Addison Wesley: Boston, MA. Limao, N. and Venables, A.J. 1999. ‘Infrastructure, Geographical Disadvantage and Transport Costs’, Policy Research Working Paper 2257, World Bank. Linnemann, H. 1966. An Econometric Study of International Trade Flows, NorthHolland: Amsterdam. Martinez-Zarzoso, Inmaculada and Nowak-Lehmann, Felicitas. 2003. ‘Augmented Gravity Model: An Empirical Application to Mercosur-European Union Trade Flows’, Journal of Applied Economics 6 (2): 291–316. McCallum, John. 1995. ‘National Borders Matter: Canada–US Regional Trade Patterns’, American Economic Review 85 (3): 615–623. Mátyás, L. 1997. ‘Proper Econometric Specification of the Gravity Model’, World Economy 20 (3): 363–368. Mátyás, L. 1998. ‘The Gravity Model: Some Econometric Considerations’, World Economy 21 (3): 397–401. Pöyhönen, P. 1963. ‘A Tentative Model for the Volume of Trade Between Countries’, Weltwirtschaftliches Archiv 90 (1): 93–99. Soloaga, I. and Winters, L.A. 2001. ‘Regionalism in the Nineties: What Effect on Trade?’ North American Journal of Economics and Finance 12 (1): 1–29. Tinbergen, J. 1962. Shaping the World Economy: Suggestions for an International Economic Policy, Twentieth Century Fund: New York. Wei, Shang-Jin. 1996. ‘Intra-National versus International Trade: How Stubborn Are Nations in Global Integration?’ NBER Working Paper 5531

Part II

Trade reforms and development experience Case studies in Latin America

6

Downhill or the long agony of Argentinian development Alcino Ferreira Câmara-Neto and Matías Vernengo

6.1 Introduction The performance of the Argentine economy during the twentieth century is usually seen as an atypical case of persistent economic decline.1 Income per capita fell from levels equivalent to Western Europe during the Belle Époque, or from 80 per cent of the income of the United States to around one-third in more recent times (see Figure 6.1). Yet, it is important to note, from a long period historical perspective, that decline was not uniform and was basically associated to two specific events with different causes. The extraordinary performance of the economy in the last decades of the nineteenth and the beginning of the twentieth century was the result of the increased integration with the centre based on the exports of primary commodities (beef and grains) (Cortés, 1998), and in a very simplified way the decline

Figure 6.1 Income per capita (% of US GDP per capita) (source: Maddison (2001), IFS/ IMF and authors’ calculations).

134

A. Ferreira Câmara-Neto and M. Vernengo

that ensued can be seen as the result of the deconstruction of the process of globalisation that tied the Argentine economy to the world through trade and capital flows. However, it is important to note that the collapse of the primary-export model of development in Argentina, as much as in other parts of Latin America and the peripheral world, were the result of the collapse of the system in its centre. It was the collapse of the hegemonic position of the United Kingdom that disorganised the international division of labour, and the Gold Standard system that underpinned and gave sustainability to British domination. In other words, the internal conditions in Argentina, in contrast to other countries in the region (e.g. Mexico, that suffered a revolution), were not central for the change in the development strategy. In contrast to conventional wisdom, the second relative decline of income per capita in Argentina cannot be attributed to the failure of the import substitution industrialisation (ISI) model, in the economic plane, or to the failures of the populism of Peronism or the developmentalist programme of Frondizi, in the political plane, simply because the economic performance in the immediate post-war period up to the 1970s is rather satisfactory.2 Only in the 1980s, the second important decline in the relative income of Argentina actually took place, that is, well after the import substitution model was abandoned by the liberalising policies of the last military government (1976–1983). The policies initiated in 1976 were designed to restore the primary-export model of the Belle Époque, promote growth and reduce what was perceived as the excessive power of the trade unions. In contrast to the first change in development strategy, the move away from ISI was a policy decision of the local elites. In other words, the relative decline in income per capita took place in two distinctive periods, associated to the crisis of the export commodity and the ISI model. For that reason it seems more promising to treat the Argentine story not as one of continual decline, but as one in which for external and internal circumstances the model of development is overturned. The results of the new development strategy might be reduced levels of growth and divergence of income per capita with respect to the centre, but that might not be the overarching reason for the change in the first place. In the second case, reduced power for trade unions and lower levels of social conflict might have been seen as more relevant than high rates of growth, for example. In this chapter we seek to analyse the relative performance of the Argentine economy during the ISI period and the subsequent neoliberal model implemented after 1976 and, with the truncation of the Alfonsín government (1983–1989), maintained during the 1990s up to the collapse of Convertibility in 2001–2002. Finally, the period of recovery between 2003 and 2008 is shown to fall short of a full break with the neoliberal model implemented after 1976. The rest of this chapter is subdivided in four sections. The next section discusses the general tendencies of output, investment and productivity for the whole period starting in 1950. The following one deals with the macroeconomic policies. The third section analyses the coordination of the process of investment and the international insertion of the Argentine economy, while the fourth deals

Argentinian development 135 with income distribution. In the conclusion we discuss the current Argentine development strategy, and the effects of the international economic crisis that started in 2008.

6.2 From the hegemonic tie to the commodity boom In the period that goes from the first government of Juan Domingo Perón (1946–1955) up to the last military government, which is approximately the same as the so-called Golden Age of capitalism, the economic performance of the Argentine economy is not stellar, if compared to some other peripheral countries, but not dismal either. The rate of growth of income per capita was around 2 per cent (see Table 6.1). During the whole period, as we have seen, the income per capita remains constant at around 50 per cent of the American one. There is no convergence with the centre, but the disparities that had increased during the Great Depression at least levelled off. Given the difficulties imposed by the commodity-export model and the inevitability of industrialisation after the Great Depression, which had limited the ability to import manufactured goods, the economy performance was quite reasonable. The economic policies associated to the developmental programme of expansion of the domestic market allowed for the acceleration of labour productivity. The explanation of that is associated to the so-called Kaldor-Verdoorn’s Law that implies that demand expansion drives labour productivity. Also, the acceleration principle indicates that capacity adjusts to the level of activity. Both relations seem to be supported by Table 6.1 in which there is a clear correlation between output growth and productivity and investment behaviour. It must be noted that the post-war period was an extremely volatile period in political terms, in many respects worse than in other countries in the region, that were also quite convoluted. In the Argentine case, the reasons for political instability are often associated to what Juan Carlos Portantiero (1973) referred to as the hegemonic tie, that produced a sort of stop-and-go cycle, as suggested by Kalecki (1971 [1943]) for developed economies, but also more profound questioning about the pertinence of the strategy of development. The political instability is reflected in the economic fluctuations as shown in Figure 6.2. Table 6.1 Growth and productivity Growth and productivity

Rate of investment (%) GDP growth (%) GDP per capita growth (%) Labour productivity growth (%)

1950–1975

1976–2002

2003–2008

3.1 3.4 2.1 2.1

1.3 1.7 –0.1 0.6

9.0 8.5 7.5 1.6

Source: World Bank and Maddison (2001).

136

A. Ferreira Câmara-Neto and M. Vernengo

Figure 6.2 Business cycle (source: Maddison (2001) and World Bank).

The consolidation of Peronism in the second half of the 1940s led to increasing political force for the process of industrialisation, which according to Portantiero lacked economic force. The incredible economic power of the sectors associated with primary exports, given their economic success in the past, gave them sufficient veto power, so to speak, on any development strategy. The tension between the trade unions and the emerging industrial bourgeoisie, on the one hand, and the old agrarian elite, on the other, were reflected in the cycles of political rupture and democratic restoration. The political instability might be one of the factors that explain the relatively less dynamic performance of the economy in this period when compared with Brazil and Mexico.3 The change in economic strategy in 1976 despite its relative success, in particular during the decade that preceded the military coup, indicates that the decision to arrest the process of industrialisation and weaken the social groups connected to that process should be seen as a political and not an economic decision.4 It must be noted that Argentina had already done the transition associated with the easy stage of ISI, and the industrial sector occupied a preponderant place in the economic system in the 1930s. In other words, it could be argued that faced with the difficult stage of ISI, associated with the installation of a capital goods sector, the increasing balance of payments problems5 during the 1960s, and the exacerbated class conflict, the captains of industry, some with ties with the old agrarian elite, decided that the industrial project in Argentina was not viable.

Argentinian development 137 The economic performance in the subsequent period is usually seen as inferior to the ISI period, but the reasons are not uniquely associated to the change in the development strategy and the turn towards neoliberalism in 1976. The 1970s were associated to the disorganisation of the international monetary system, competitive devaluations, the oil shocks and the acceleration of inflation. In Argentina, after the so-called ‘Rodrigazo’ – named after Celestino Rodrigo the Finance Minister at the time – in June 1975, with a maxi-devaluation of 160 per cent of the peso in nominal terms, and an increase in utilities and energy prices of around 180 per cent, in addition to nominal wage adjustments, inflation accelerated out of control.6 After the mid-1970s the problem of stabilisation dominated all macroeconomic debates and long run questions were relegated to a secondary plane until the Convertibility Plan in 1991. In that sense, the liberalising policies of Martínez de Hoz starting in 1976 were defended fundamentally as a part of a stabilisation plan, and the change in the development strategy were seen as instrumental in promoting an environment less prone to inflationary pressures. In addition the debt crisis, which originated in the Mexican default of August 1982, and exacerbated in the Argentine case by the appreciation of the currency that was part of the stabilisation plan, led to a collapse of economic growth. In terms of income per capita the economy actually contracted in this period. During the 1980s, after redemocratisation, the inability to renegotiate the debt on a sustainable level imposed a severe external constraint on the economy, which led, as in the rest of Latin America, to the so-called ‘lost decade’. The external restriction and the need for permanent devaluations in a context of wage indexation explain a great deal of the inflationary acceleration during the period. In the early 1990s after the renegotiation of the external debt, the re-entry in international financial markets and the stabilisation of prices, economic growth picked up, but it was a short lived expansion. The frequent crises in Mexico, Asia, Russia and Brazil starting in 1995 and the straitjacket of Convertibility implied moderate rates of growth and a profound recession starting in 1998. For the period as a whole, which goes from 1976 to the collapse of Convertibility in the 2001–2002 biennium, the rate of growth of income per capita was nil (see Table 6.1). If we break the period in two, in the first part from 1976 to 1989 income per capita was negative 1.4 per cent, while the second sub-period from 1990 to 2002 it was of only 0.9 per cent per year on average. The crisis of the Convertibility Plan opened up a new chapter in Argentine economic policy, and the performance of the economy, up to the 2008 crisis, was exceptional by historical standards. The rate of growth of income per capita of more than 7 per cent was higher than the rates associated with the commodityexport model. Ironically, even though growth was associated to the increased utilisation of excess capacity, the expansion was ultimately correlated to the commodity boom, which represents a partial return to the old commodity-export model.7

138

A. Ferreira Câmara-Neto and M. Vernengo

6.3  External constraint, inflation and financialisation Macroeconomic policy in peripheral countries is more often than not determined by structural factors associated to the balance of payments, than to short run issues related to the smoothing of the business cycles, and that is also the case in Argentina. The overriding short run preoccupation is often related to price stability. Demand management policies are, then, often limited by the current account performance, since the recurrent external deficits would lead to external debt accumulation and foreign exchange crises. The decomposition of the elements of demand permits to understand whether internal or external components of demand are central for the expansion of output.8 The decomposition used here follows the methodology presented in Lance Taylor (2006). Aggregate supply (X) is defined as the sum of consumption (C), investment (I ) and exports (Ex). The national rate of savings (s) is defined as income minus consumption over aggregate supply and the propensity to import (m). The internal and external stances are, then, given by: (6.1) The decomposition is shown in Figure 6.3, and it is shown that Argentina almost always grew below the balance of payments constraint, with external demand being central for expansion. With the exception of two periods in the late 1950s/early 1960s and the 1990s internal demand was of secondary importance, or at least it never expanded significantly to the point that it threatened the balance of payments position. This suggests that the balance of payments imposed a severe restriction on the possibility of expanding domestic demand, and, as a result, of the domestic markets itself. In other words, growth was always limited by the external constraint and demand management policies were fundamentally used to stay within the limits imposed by the balance of payments. It should be noted that the decomposition does not explain the rhythm of growth, but only its composition between internal and external forces. The evolution of output (X ) does, however, show the rate of growth of the economy. As it can be seen in Figure 6.3, output expands relatively fast up to the 1980s, stagnates for about a decade after that, to briefly recover and completely collapse at the end of the last century, and shows a brisk upturn in the current century. In that sense, the 1980s, when growth was pushed by external forces, can be seen as a decade of export-led stagnation. In other words, the fact that the external market was the dynamic component of demand does not imply that the economy was growing fast. In contrast with East Asia, in Argentina, as in the rest of Latin America, exports were necessary to service foreign debt obligations, while imports collapsed during the 1980s, and the external market was incapable of promoting accelerated expansion of the economy. Only after the Brady Plan, with the

Argentinian development 139

Figure 6.3 Output decomposition (source: authors’ calculations).

re-entry into external financial markets, did the economy start to recover. However, the recovery was short lived with the economy plunging into its worst crisis in history, even counting the Great Depression, the debt crisis of 1982 and the 1989 hyperinflation. To a great extent, from a macroeconomic point of view, the inability to break with the external restriction was associated to exchange rate policy. During the ISI period there were extensive foreign exchange controls, and multiple exchange rates, with pressures from groups associated with the process of industrialisation for appreciated exchange rate in order to facilitate the importation of capital and intermediary goods. Devaluations tended to be contractionary, and to favour exporters, often linked to traditional commodity producing sectors (DíazAlejandro, 1965). The conflict between those favouring an appreciated versus a depreciated exchange rate produced a great political instability, creating exchange rate volatility with bouts of appreciation followed by maxidevaluations (see Figure 6.4). In addition, the foreign exchange rate was often used as an anchor in stabilisation programmes. Figure 6.4 shows two such processes of exchange rate appreciation linked to stabilisation plans. In the mid-1970s the military regime introduced a system of pre-announced devaluations (the infamous tablita), and after April 1991, the Convertibility Plan, which collapsed in 2002, tied the peso to the dollar. It should be noted that after the collapse of Bretton Woods it became considerably more difficult to manage the exchange rate, and the pressures from the

140

A. Ferreira Câmara-Neto and M. Vernengo

Figure 6.4 Real exchange rate (source: authors’ calculations).

International Monetary Fund (IMF ) and the United States for a more open financial account became stronger. It should not be a surprise that the use of an exchange rate anchor as an instrument of stabilisation, in the 1970s and 1990s, was accompanied by financial liberalisation and speculative capital inflows that proved to be excessively volatile to maintain the exchange rate arrangement in the long run. In the same vein, the option for a more open financial account also reflects the interests of the economic groups that prefer an economy more integrated with international markets, and with a smaller role for domestic industries in the composition of the economy, with a reduced role for trade unions. It should be noted that the process of liberalisation and deindustrialisation did not imply, at least not initially, a reduced role for the state in fiscal terms. Table 6.2 shows the primary and global fiscal balances and the debt servicing spending throughout the decades, beginning with the 1960s. It is quite clear that in the transition from a developmentalist to the neoliberal model (1970s and 1980s) the primary and global deficits increase, and that a significant adjustment Table 6.2 Fiscal policy (% GDP)

1961–1970 1971–1980 1981–1990 1991–2000 2001–2008

Primary balance

Global balance

Interest

–3.4 –6.0 –5.1 0.1 2.4

–4.0 –7.0 –7.0 –2.1 0.1

0.6 1.0 1.9 2.2 2.3

Source: Damill et al. (2003) and ECLAC.

Argentinian development 141 only takes place in the 1990s. Further, the fiscal adjustment is tightened in the current century, after the collapse of Convertibility. Table 6.2 also shows the increasing financialisation of public spending with more than 6.2 per cent of GDP being transferred as interest payments to owners of public bonds, mostly banks and corporations, and their owners.9 This shows that the nature of the state’s intervention changed after the debt crisis and the opening up of the financial account of the balance of payments, and that the transfer of resources to rentier groups became a central part of public policy. Finally, it should be noted that monetary policy was passive for most of the time that exchange rate policy was a central element of stabilisation policy that was at last successful in the 1990s, in the context of global price stability.10 In that sense, exchange rate policy had always a short run bias related to stabilisation instead of a long run preoccupation with external competitiveness.

6.4 The state, foreign capital and spurious competitiveness The characteristics of the model of development in the post-war period are relatively well known. Fundamentally, the economy was less open, characterised by higher tariffs, quantitative controls and bureaucratic restrictions to trade, favouring capital imports and discouraging imports of superfluous consumer goods. Also, a larger participation of the state in the economy, through direct production of goods and services, more active purchasing policies and through the public financial sector, was the norm. For example the Banco de Crédito Industrial Argentino (BCIA), created in 1944, provided at its peak 80 per cent of all the credit to the manufacturing sector.11 In several areas, the entry of foreign direct investment was stimulated as a way to promote transfer of technology, in particular, in the petrochemical and metal-mechanic complex, like the automobile industry (Sourrouille et al., 1985, p. 39). It must be noted that the relative openness with respect to foreign capital began in 1953, still during Perón’s government, allowing foreign firms greater freedom to repatriate profits to their country of origin. It is in this period of the 1950s that several foreign multinational groups like Fiat, Mercedes-Benz, Siemens, Bayer, established their local affiliates. Further, the national oil company, Yacimientos Petroliferos Fiscales (YPF ), signed, in this period, contracts for exploitation of oilfields with foreign companies, in particular Standard Oil, which reveals that, in contrast to Brazil during Vargas or Mexico during Cardenas, the participation of foreign capital in the energy sector during the government of Perón was significant. In this sense, it is important to avoid the typical simplification that equates the Peronist period with the dominance of national developmentalist groups with populist tendencies. Even though the influence of foreign capital was smaller than in the so-called developmentalist government of Arturo Frondizi (1958–1962), it is still not true that foreign capital was excluded in the Peronist period. The industrialisation process in Argentina, as much as in the rest of Latin America, was more dependent on foreign capital and imported technology than

142

A. Ferreira Câmara-Neto and M. Vernengo

in other peripheral regions. The participation of foreign capital was not central in terms of its volume, as noted by Altamir et al. (1967), but for its strategic character connected to sectors with high value added and complex technologies.12 The boom in foreign direct investment (FDI) in the 1990s is related to the process of privatisation rather than capital formation per se.13 Table 6.3 shows the participation of public capital in the process of investment during the ISI period, and its subsequent decline after liberalisation. It is important to note that the participation of the public sector in gross capital formation in the last period, associated to the commodity boom, did not lead to a recovery of the levels of the ISI period. Another important element of the temporal trajectory of investment is that the relative fall after 1976 can be entirely attributed to the decline in public investment. In fact, in the last period private investment increased, but the change in private investment was insufficient to compensate the fall in public capital formation. Finally, another characteristic of investment in Argentina during the ISI process is related to the dimension of the firms that, when compared with enterprises of other countries, were relatively small if measured by number of employees or by energetic capacity per establishment (Vitelli, 1999). Part of the problem of smaller firms is that for scale and scope reasons they tend to be less technologically dynamic. The larger dependence on foreign capital and the larger number of small domestic firms might be part of the explanation for the reduced innovativeness of Argentine firms. The process of industrialisation until the 1970s did not lead to a heavy burden of foreign debt accumulation. The disproportional growth of foreign debt, and the use of public firms as a vehicle for external borrowing and of the central bank for the nationalisation of private debt were a phenomenon of the 1970s and of the international context associated with the recycling of the petrodollars. The liberalisation policies of this period exacerbated the problems by facilitating imports and the movements of funds, which eventually materialised in capital flight. In Table 6.4 we can see that the external debt as a proportion of gross national income was approximately 19.1 per cent in 1970 and grew to 85 per cent during the 1970s, more than double its growth during the following decade. Additionally, we can see that debt not only grew in the 1970s, but continued to grow in the 1980s, after the debt crisis, and 1990s after further liberalisation policies had

Table 6.3 Investment composition

Public investment Private investment Total investment

1960–1975

1976–2002

2003–2006

7.2 15.3 22.5

4.3 15.5 19.8

2.3 17.1 19.4

Source: World Bank and Indec.

Argentinian development 143 Table 6.4 Debt sustainability indicators

Debt/GNP Debt/exports Interest/exports

1970

1980

1990

2000

2007

19.1 ND ND

35.6 242.4 37.3

46.0 373.7 37.0

53.3 380.4 70.5

50.0 174.0 13.0

Source: World Bank.

been implemented.14 It is only after the 2002 default that the burden of debt eases up, and debt servicing falls from 70 per cent to 13 per cent of exports. In the last period, the fall in the interest to export ratio, not only results from the default and renegotiation of the debt by the Kirchner government, but also from the export performance. Table 6.5 shows the performance of manufacturing exports and of the exports of high technological content within total manufacturing exports. An increase in manufacturing exports took place in the 1980s and it was maintained in the following decade, and an increase in the export of manufactures intensive in the use of technology. It must be noted that the majority of Argentinian exports are still of traditional products, and that the exports of goods intensive in the use of more complex technology are small when compared to other peripheral countries. To some extent, the lack of export dynamism reflects the difficulties connected to what Fajnzylber (1989) called spurious competitiveness. In his view, competitiveness resulted less from technological capabilities than with the advantages associated with the specialisation in natural resources and primary commodities, low wages, devalued exchange rates and the combination of domestic protection with excessive export subsidies. However, it is important to qualify Fajnzylber’s argument for the Argentine case. While it is true that the Argentine economy is still very much specialised in commodity exports, and that income distribution had an impact in the changes in the productive structure, it is far from clear that the exchange rate or commercial policies were active in the expansion of exports. Arguably, Argentinian governments made an explicit effort to return to the old commodity export model, in which the comparative advantages would be explicitly associated to the Ricardian competitiveness of the agro-pastoral Table 6.5 Manufacturing exports (%)

Manufactures High technology Source: authors’ calculations.

1980

1990

2000

35.0 26.8

52.2 23.6

52.1 38.5

144

A. Ferreira Câmara-Neto and M. Vernengo

sector.15 In that sense, the question of distribution, and lower wages, should not be separated from the strategy of reprimarisation of the Argentine economy.

6.5 Inequality and reprimarisation If there were a question that shows the negative consequences of the productive transformation of the last 30 years that would be the evolution of income distribution. Income per capita was at its highest at the beginning of the twentieth century, but only at the end of the ISI period that income distribution reached levels that were comparable with those of more equalitarian developed countries. Figure 6.5 shows the evolution for Argentina and the average of selected Latin American countries (Brazil, Chile, Colombia, Mexico, Peru and Venezuela). In the 1950s the Gini coefficient was around 40, but by the 1970s it had fallen to around 35, not very different from Southern European countries. However, after the coup d’état of 1976, there is a significant increase in the Gini, converging to the Latin American average over time. The Gini coefficient usually shows the difference between wages, rather than the evolution of wages in total income. Figure 6.6 shows the evolution of wages with respect to income per capita. It is in the 1950s that we find the highest remuneration for workers. Also, the graph suggests that the inflationary acceleration of the 1970s was instrumental for wage contraction. In that sense, the second Peronist government in the 1970s was incapable to promote wage redistribution. It is important to note that the recovery of wages since 2003 was insufficient to bring them back to the levels of the 1960s. The recovery in wages should be seen as a modest concession, by the Kirchner administration, on a model of development that still favours an open economy that is integrated mainly through

Figure 6.5 Gini coefficients (source: authors’ calculations).

Argentinian development 145

Figure 6.6 Wage share (source: Llach and Gerchunoff (2004) and Indec, and authors’ calculations).

the exports of commodities and maintains a relatively devalued currency to facilitate those exports. Finally, it should be noted that distribution of land in Argentina remains relatively unequal. The Gini coefficient for land distribution is of approximately 80, a level significantly higher than that of more equalitarian countries like Cuba after the Revolution, with a Gini of slightly less than 60, or Mexico after Cárdenas’ agrarian reform, also below 60 (IFAD, 2001). In that sense, the pro-agrarian bias of the development strategy tends to maintain and exacerbate the deeply entrenched inequalities of the Argentine society.

6.6 Concluding remarks The recent economic history of Argentina can be seen as composed of three distinctive periods. The first period can be described as a successful, in terms of growth but not necessarily in terms of income distribution, integration to the world economy on the basis of commodity exports, which was not sustainable after the disintegration of the international division of labour during the interwar period. In particular, after the Great Depression it was clear that the export commodity model was incapable of incorporating surplus labour, in part because it was highly productive. The second period was also very successful in terms of growth, albeit less than the previous, but considerably more equalitarian, and more capable of incorporating surplus labour. The limits to the import substitution development strategy that characterised the second period were fundamentally political in nature. The inherent difficulties of the process of industrialisation were exacerbated by the strength of labour, the increasing social conflicts of the 1960s, and the pressures of the Cold War and the fears of the Cuban Revolution, leading in the 1970s to a change in the model of development and the implementation of the liberalisation strategy.

146

A. Ferreira Câmara-Neto and M. Vernengo

The new development strategy presupposed that a return to the old export commodity model was feasible and desirable, in spite of the significant changes in the global economy. For example, while for the first globalisation the integration was with a hegemon that imported commodities, the United Kingdom, in this renewed phase the hegemonic power is a competitor that produces commodities, the United States. In addition, it seems that the countries that succeeded, at least to some extent, to close the gap with the developed world have followed the path of manufacturing exports and the development of the domestic market (Amsden, 2001). From that point of view, even if the change in development strategy is comprehensible, the possibilities for success are limited at best. The good performance in the 2003 to 2008 period should not be seen as a new phase, but simply as the result of the functioning of a development strategy that was implemented in the 1970s, and that worked fundamentally as the result of the positive terms of trade shock. In this case one must hope for prices of commodities to continue to go up forever. Hope springs eternal!

Notes 1 According to Della Paolera and Gallo (2003, p. 373) the Argentine decline remains a puzzle, ultimately explained by micro and macro institutional failures. 2 Diaz-Alejandro’s (1970, p. 129) classic book is paradigmatic in its enthronisation of Peronism as the villain of the Argentine process of development. According to him the Peronist government was keener on redistribution and higher rates of consumption for the masses that were pursued at the cost of lower rates of capital accumulation. These views are much favoured by current Argentine historiography (e.g. Llach and Gerchunoff, 2004). 3 According to Basualdo (2006), even though there might have been a hegemonic tie until the early 1960s, by the mid to late part of that decade it had dissipated. That would, in part, explain the accelerated industrial expansion between 1966 and 1974. 4 Jorge Schvarzer (1983, p. 15) argues that not only in Argentina, but also in the whole Southern Cone, the political logic overpowered economic reasoning in the determination of the development strategy. 5 Jorge Katz and Bernardo Kosacoff (1989, p. 16) noted the relevance of the inelasticity of primary exports as a permanent restriction to economic growth. 6 See Rapoport (2005, pp. 571–573) for a discussion of the economic policies in this period. 7 See Pérez Caldentey and Vernengo (2008a) for a discussion of the current model of development in Latin America. 8 The assumption is that the level of output is determined by the autonomous components of demand along Keynesian lines. The idea of the external constraint to accumulation is based on Prebisch (1949), and has been more recently formalised by Thirlwall (1979). 9 It should be clear that the simple analogy between developmentalist and Keynesian policies, the latter in the conventional sense of favouring deficits, lacks support. For a more accurate description of the role of fiscal deficits in the process of economic development, and Keynes’ views on the issue see Câmara-Neto and Vernengo (2004–2005). 10 For different interpretations of the Convertibility Plan see Della Paolera and Taylor (2001) and Pérez Caldentey and Vernengo (2008b).

Argentinian development 147 11 For example, the imports of non-durable consumption goods fell from 23.3 per cent to less than 6 per cent in the first government of Perón. On the other hand, the imports of capital goods increased from 3 per cent to 17 per cent, at the peak of the investment process in the post-war period (Rapoport, 2005, p. 358). 12 During the period of the commodity export model, foreign capital was certainly higher. The participation of foreign capital in investment fell from 38 per cent between 1900 and 1909 to around 3 per cent in 1953. After the fall of the first Peronist government in 1955, there was a significant increase in foreign participation in investment reaching 13 per cent in 1959 (Altamir et al., 1967). 13 Chudnovsky and López (2002) provide a lengthy discussion of the role of multinational firms in the economy during the 1990s. 14 It must be noted that the debt crisis in Argentina, as in the rest of Latin America, was the result of the international situation characterised by large liquidity stocks and higher rates of interest in the United States. Internal conditions were considerably less relevant (Calcagno, 1988, p. 45). 15 It must be noted that since the 1970s there has been an agriculturisation of the Argentine primary sector, with an increasing role for grain production, particularly soybeans, and a reduced role for cattle. The good performance of the agribusiness in the 1990s took place despite the overvaluation of the currency, and of the increase in wages and retentions to exports in the current decade (Barsky and Gelman, 2001).

References Altamir, O., Santamaría, H. and Sourrouille, J. (1967) ‘Los instrumentos de promoción industrial en la posguerra’, Desarrollo Económico, 7(27), pp. 361–376. Amsden, A. (2001) The rise of the rest, New York: Oxford University Press. Barsky, O. and Gelman, J. (2001) Historia del agro argentino, Buenos Aires: Sudamericana. Basualdo, E. (2006) Estudios de Historia Económica Argentina, Buenos Aires: Siglo XXI. Calcagno, A.E. (1988) La perversa deuda, Buenos Aires: Legasa. Câmara-Neto, A.F. and Vernengo, M. (2004–5) ‘Fiscal policy and the Washington Consensus’, Journal of Post Keynesian Economics, 27(2), pp. 333–343. Chudnovsky, D. and López, A. (2002) ‘The strategies of the multinationals in the 1990s Argentina’, Revista de la CEPAL, No. 76, April, pp. 151–166. Cortés Conde, R. (1998) Progreso y declinación de la economía argentina, Buenos Aires: Fondo de Cultura Económica. Damill, M., Frenkel, R. and Juvenal, L. (2003) ‘Las cuentas públicas y la crisis de la convertibilidad en la Argentina’, Desarrollo Económico, 43(170), pp. 203–229. Della Paolera, G. and Gallo, E. (2003) ‘Epilogue: the Argentine puzzle’, in G. Della Paolera and A. Taylor (eds) The new economic history of Argentina, Cambridge: Cambridge University Press. Della Paolera, G. and Taylor, A. (2001) Straining at the anchor, Chicago, IL: University of Chicago Press. Díaz-Alejandro, C.F. (1965) Exchange-rate devaluation in a semi-industrialized country: the experience of Argentina, Cambridge: MIT Press. Díaz-Alejandro, C.F. (1970) Ensayos sobre la historia económica argentina, Buenos Aires: Amorrortu. Fajnzylber, F. (1989) ‘Industrialización en América Latina: de la “caja negra” al “casillero vacío” ’, Cuadernos de la CEPAL, No. 60, Santiago de Chile.

148

A. Ferreira Câmara-Neto and M. Vernengo

IFAD (International Fund for Agricultural Development) (2001) Rural Poverty Report, Rome: IFAD. Kalecki, M. (1971 [1943]) ‘Political aspects of full employment’, in Selected essays on the dynamics of capitalist economies, Cambridge: Cambridge University Press, pp. 138–45. Katz, J. and Kosacoff, B. (1989) El proceso de industrialización en la Argentina: evolución, retroceso y prospectiva, Buenos Aires: CEAL/CEPAL. Llach, L. and Gerchunoff, P. (2004) Entre la equidad y el crecimiento: Ascenso y caída de la economía argentina, 1880–2002, Buenos Aires: Siglo XXI. Maddison, A. (2001) The world economy: a millennial perspective, Paris: OECD. Pérez Caldentey, E. and Vernengo, M. (2008a) ‘A tale of two monetary reforms: Argentinean convertibility in historical perspective’, Studi e Note di Economia, 12(2), pp. 139–170. Pérez Caldentey, E. and Vernengo, M. (2008b) ‘Back to the future: Latin America’s current development strategy’, Ideas Working Paper Series, No. 07/2008. Portantiero, J.C. (1973) ‘Clases dominantes y crisis politica en la Argentina actual’, in O. Braun (ed.) El capitalismo argentino en crisis, Buenos Aires: Siglo XXI, pp. 73–118. Prebisch, Raúl (1949) ‘O desenvolvimento da América Latina e seus principais problemas’, Revista Brasileira de Economia, 3(3), pp. 47–111. Rapoport, M. (2005) Historia económica, política y social de la Argentina, Buenos Aires: Ariel. Schvarzer, J. (1983) Martínez de Hoz: La lógica política de la política económica, Buenos Aires: CISEA. Sourrouille, J. (1985) Trasnacionalisación y Política Económica en Argentina, Buenos Aires: Centro Editor de América Latina. Taylor, L. (2006) ‘External liberalization, in Asia, socialist Europe and Brazil’, in L. Taylor (ed.) External liberalisation, in Asia, socialist Europe and Brazil, New York: Oxford University Press. Thirlwall, A. (1979) ‘The Balance of Payments Constraint as an Explanation of International Growth Rate Differences’, Banca Nazionale del Lavoro Quarterly Review, 32(128), pp. 45–53. Vitelli, G. (1999) Los dos siglos de la Argentina, Buenos Aires: Prendergast.

7

The determinants of FDI in Chile A gravity model approach Matteo Grazzi1

7.1 Introduction Since 1980, foreign direct investment (FDI) flows have grown at remarkable rates. From 1991 to 2000 the outflows averaged a yearly rise of over 28 per cent. Even if the FDI trends slowed considerably in the early 2000s, in the last years FDI has started to increase at substantial rates again. In 2005 the growth was of 29 per cent and in 2006 FDI was 38 per cent, reaching a considerable amount of US$1.3 trillion (UNCTAD, 2007). According to preliminary estimates, global FDI flows in 2007 reached the record level of US$1.8 trillion (ECLAC, 2007), which would correspond to an increase of 29 per cent with respect to the previous year. These figures reflect not only a greater level of cross-border mergers and acquisitions (M&As) among developed countries, but also an impressive surge in flows towards developing countries. With a net flow of over $300 billion in 2006, FDI has become the most important source of foreign financing for emerging economies. Furthermore, the overall benefit of FDI is well documented in literature. Given a correct host-country policy and a minimum level of development, most scientific work shows that FDI generally triggers technology spillovers, helps to create human capital, contributes to international trade integration, leads to a more competitive business environment and enhances enterprise development. All these factors stimulate economic growth, which is a prerequisite for alleviating poverty in developing countries. Furthermore, beyond the strictly economic benefits, if addressed correctly, FDIs may help to improve environmental and social conditions in the host country, for example by transferring less polluting technologies and leading to more socially responsible corporate policies (OECD, 2002). In the global situation described above, it is of primary importance for developing countries to find ways to look more and more attractive in the eyes of investors, in order to receive higher FDI flow amounts, especially of the kind that will bring the greatest benefit to a country in terms of investment levels, job creation, higher-value-added activities and innovation. Consequently, an urgent research topic for the development economist is to analyse the foreign investment flows into developing countries and clearly identify their main determinants, in order to offer useful indications for policy-makers.

150

M. Grazzi

The aim of this chapter is to analyse the determinants of FDI inflows into Chile from 1985 to 2005 through the estimation of a gravity equation. Chile is a relatively small and resource-rich country, and has been highly successful in attracting FDI. In fact, FDI has represented an important contribution to the sustained economic development of Chile in the last decades. Since 1990, when the country returned to democracy, Chile has undertaken an active foreign policy targeted to the full integration in the international arena. A main pillar of this strategy has been constituted by the signature of a high number of bilateral economic agreements, such as Bilateral Investment Treaties (BITs), Double Taxation Treaties (DTT) and Free Trade Agreements (FTAs). The validity of such treaties as instruments to raise the level of FDI into a country has been questioned by several scholars and, to date, there is still controversial empirical evidence. This work contributes to this debate. The chapter is structured as follows. The second section analyses trends and characteristics of FDI inflows into Chile, with a particular attention on the geographical origin and the sector distribution. In the third section the empirical methodology is presented. The fourth section contains data and variables description. Then, the results are presented and commented in the fifth section. Finally, some conclusions and policy implications are provided in the last section.

7.2  FDI to Chile: facts and figures With a population of 16.3 million people as of 2005, Chile is often referred to as one of South America’s most stable and prosperous nations. However, despite being Latin America’s fastest-growing economy during the 1990s, the country experienced moderate economic downturns at the end of the past decade, as a result of the unfavourable economic conditions that the Asian financial crisis brought with it. Mining is the dominant sector of the national economy, accounting for the largest share of GDP: in 2006, copper mining alone2 contributed for 21.6 per cent of GDP while manufacturing industry for 12.8 per cent. The agricultural sector accounts for a mere 5.5 per cent of GDP, while services and the industrial sectors account respectively for 47.7 per cent and 46.8 per cent of GDP (World Bank, 2005). Mine products are the main export product as they account for about half of the total export value. Apart from minerals, Chile also exports wood products, fish and fishmeal, fruits and wine. Main imports are petroleum, wheat, capital goods, spare parts and raw materials. Its chief trading partners are European Union nations, the United States, Japan and Brazil. Specifically between the years of 1998 and 2006, the share of export of goods and services of Chilean GDP has impressively increased (from 26.3 per cent to 45.4 per cent), growing not only in quantity but also in variety (Economist Intelligence Unit, 2007). Although the mining sector remains prevalent, with a dominating copper industry (43 per cent of merchandise export earnings in 2002–2006), the diversification of trade has stimulated the developing of new export-oriented industries,

FDI in Chile 151 such as cellulose, salmon, fruit, meat, wine, ethanol. In fact, the dependence of the economy on mineral prices, together with the production of adequate food for its population, is the major economic problem of the country. Since the 1970s, Chile has based its national development strategy on openness to foreign investment. The Chilean Constitution grants no discrimination against foreigners, and national treatment is generally guaranteed in the legislation. Nevertheless, it is only since the return of full democracy in the country, in 1990, that investment flows have markedly increased. The country’s businessfriendly environment, based on certainty of law and transparency, coupled with political stability and the signing of numerous investment treaties effectively attracted a large amount of foreign capital during the 1990s. The FDI inflow was $287 million in 1980, 661 in 1990, 4,860 in 2000 and 6,667 in 2005. Today, at $475.76 FDI per capita, Chile is the third largest recipient of foreign investment in Latin America (after Brazil and Mexico) in absolute terms and the second in per capita figures after Trinidad and Tobago. According to the 2006 World Investment Report, published by the United Nations Conference on Trade and Development (UNCTAD), the FDI stock in Chile passed from 30.0 per cent of GDP in 1990 to 64.6 per cent in 2005. These numbers are impressive, especially if compared with the average world percentage of 22.7 per cent in 2005 and that for developing countries, 27.0 per cent. Between 1974 and 2005, the total gross materialized FDI in Chile was US$78.1 billion, 89 per cent of which entered the country after 1990. If we consider net inflows, US$52.1 billion entered the country during this period. In the 1990s,

Figure 7.1 FDI flows into Chile 1985–2006 (in nominal US$ million) (source: author’s elaboration on data provided by the Chilean Foreign Investment Committee).

152

M. Grazzi

FDI gross inflows represented an annual average of 6.4 per cent of Chile’s GDP, rising to an annual average of 8 per cent between 1995 and 2000. After this uninterrupted surge in FDI, the amount of foreign investment entering Chile decreased significantly in 2000 as a consequence of difficult international economic conditions, which affected global FDI flows in almost all countries. The amount of FDI inflows dramatically fell from the record high of US$9.9 billion in 1999 to a negative peak of US$5 billion. In the early 2000s, the negative trend continued as a consequence of the global collapse of the M&As market, the main driving force of FDI flows around the world and in Chile in the previous decade. Many reasons explain such a phenomenon: not only the global economic uncertainty, but also a drop in share prices, lower corporate earnings and a cut in the budget of multinational companies for expansion. Furthermore, various deep financial crises provoked in these years heavy losses to a broad range of investors across several Latin American countries. As a result the region was perceived as a high risk investment area exactly in the moment when risk-adverse shareholders were pushing multinational firms to perform safer investments. The Chilean Foreign Investment Committee (2006) indicated that, in the case of Chile, a greater use of the local capital market by foreign investors could have distorted recent FDI figures. The high liquidity and the dynamism of the Chilean financial sector, combined with historically low interest rates, encouraged a growing number of foreign companies to raise finance locally, through either borrowing from local banks, issuing bonds on the domestic market or reinvesting locally their profits. Such a trend could be reflected negatively in the FDI inflows into the country. However, starting from 2004, FDI inflows into Chile showed an increase that reflected a renewed interest in mergers and acquisitions and the development of new projects in mining, telecommunications and infrastructure. In 2006, the provisional figures for gross FDI inflows had reached US$5.9 billion. Table 7.1 presents Chilean FDI gross inflows by sector of destination, from 1974 to 2006.3 Considering the entire period, mining accounted for 33.1 per cent; services for 19.6 per cent; electricity, gas and water industries for 19.2 per cent; manufacturing for 12.7 per cent; transport and communication for 11.6 per cent; construction for 2.3 per cent and agriculture, forestry and fishing for 0.5 per cent each. However, on evaluating the variation of percentages over time, it is interesting to note that the mining sector, traditionally the most important recipient sector by far, represented almost 47 per cent of total inflows until 1990, but progressively lost importance thereafter. In the 2001–2005 period its share fell to 22.04 per cent, and in 2006 FDI gross inflows directed to the mining sector amounted to 23.1 per cent of the total. This decrease has been offset by higher inflows into the industries involved in the privatization process: the transport and communication sector (28 per cent in the 2001–2005 period, even if just 16 per cent in 2006) and the electricity, gas and water sector (28.9 per cent in the 2001–2005 period and 28 per cent in 2006). From 1997 to 2001, because of the privatization process, Chile saw a dramatic surge in M&As activity, mainly in the services, electricity and telecommunications sectors.

1,313,636 470,693 7,675,990 657,763 750,290 276,832 1,826,667 4,069,726 1,410,331 270,938 505,946 626,736

88,004 105,930 152,803 6,766,117 1,115,662 243,005

28,327,069

$

1996–2000

Source: author’s elaboration on data provided by the Chilean Foreign Investment Committee.

100

524,511 315,902 0 122,953 153,659 21,061 283,110 779,061 37,015 22,963 299 28,887 5,111,987

10.26 6.18 0 2.41 3.01 0.41 5.54 15.24 0.72 0.45 0.01 0.57

79,626 10,695 17,853 2,398,769 215,471 100,152

Agriculture and livestock Forestry Fishing and aquaculture Mining and quarrying Food, beverages and tobacco Wood and paper products, printing and publishing Chemical, rubber and plastics Other manufacturing industries Electricity, gas and water supply Construction Wholesale and retail trade Transport and storage Communications Financial services Insurance Engineering and business services Sewage, sanitation and similar services Other services

Total

1.56 0.21 0.35 46.92 4.22 1.96

$

Sector

%

1974–1989

Period

100

4.64 1.66 27.1 2.32 2.65 0.98 6.45 14.37 4.98 0.96 1.79 2.21

0.31 0.37 0.54 23.89 3.94 0.86

%

918,210 289,105 6,067,620 598,763 313,173 198,329 5,688,993 292,413 391,499 169,052 5,845 245,864

13,164 4,590 25,239 4,622,950 520,144 613,909

20,978,862

$

2001–2005

Table 7.1 FDI inflows into Chile by sector, 1974–2006 (in nominal US$ million and percentages)

100

4.38 1.38 28.92 2.85 1.49 0.95 27.12 1.39 1.87 0.81 0.03 1.17

0.06 0.02 0.12 22.04 2.48 2.93

%

49,305,931

2,231,846 759,798 13,743,610 1,256,526 1,063,463 475,161 7,515,660 4,362,139 1,801,830 439,990 511,791 872,600

101,168 110,520 178,042 11,389,067 1,635,806 856,914

$

2006

100

4.53 1.54 27.87 2.55 2.16 0.96 15.24 8.85 3.65 0.89 1.04 1.77

0.21 0.22 0.36 23.1 3.32 1.74

%

154

M. Grazzi

After 2001, the trend in FDI changed again, shifting towards projects that require smaller amounts of capital but have a high impact in terms of job creation and technology transfer. These smaller, high-impact projects range from software development initiatives, call centres and shared services centres, to new investment in the manufacturing and agribusiness sectors. Unfortunately, a large part of these investments do not pass through the mechanism of the DL600, therefore are not considered in the Chilean Foreign Investment Committee data (see Note 3). From 2004, in line with the regained dynamism of the M&As market on the global stage, the Chilean market experienced a new wave of acquisitions of local companies by foreign firms. As for geographical origin of the investments, in the period between 1974 and 2006, materialized FDI through D.L.600 mechanism has been mainly originated in the European Union (41 per cent), in the United States (25 per cent), in Canada (16 per cent), Australia (5 per cent) and Japan (3 per cent). However, considering data on disaggregated time periods, the share of investment originated in the EU is growing, while that from the United States is decreasing. In 1995–1998 the percentages of the European Union and of the US were similar: 35.33 per cent and 35.58 per cent (see Table 7.2). Yet in 1999–2002 the gap became consistent, with 51.99 per cent for the EU against 21.94 per cent for the US. In 2003–2006, while the EU maintained its investment level (51.79 per cent), the United States accounted only for 7.78 per cent of the total investment that arrived in the country. Within the European Union, the main investor is Spain, followed by the United Kingdom. Finally, it is worth

Figure 7.2 FDI inflows into Chile by geographical origin 1974–2006 (percentages) (source: author’s elaboration on data provided by the Chilean Foreign Investment Committee).

FDI in Chile 155 Table 7.2 FDI inflows into Chile by geographical origin, 1974–2006 (in nominal US$ million and percentages) Period

1995–1998

Origin

$

European Union United States Canada Australia Japan South America Others Total

1999–2002 %

4,631,610 4,665,013 2,279,577 330,942 339,270 319,219 544,834

$

35.33 10,747,520 35.58 4,536,305 17.39 2,316,572 2.52 847,965 2.59 509,425 2.43 247,652 4.16 1,467,834

13,110,465 100.00 20,673,273

2003–2006 % 51.99 21.94 11.21 4.10 2.46 1.20 7.10

$ 5,620,957 844,658 2,476,485 677,905 148,117 0 1,085,943

% 51.79 7.78 22.82 6.25 1.36 0.00 10.00

100.00 10,854,065 100.00

Source: author’s elaboration on data provided by the Chilean Foreign Investment Committee.

noting that the investment originated from other South American countries is very low, and it has decreased over time.

7.3  Empirical methodology The gravity equation is a common well-established formulation for statistical analyses of bilateral flows between different geographical entities (Head, 2003). It is based on the relationship described in the ‘Law of Universal Gravitation’ postulated by Isaac Newton in 1687, which states that the attractive force between two point masses is proportional to the product of the two masses and inversely proportional to the square of the distance between them. Tinbergen (1962) was the first to propose that the same functional form could be utilized to describe international trade flows and, since, the gravity equation has gained increasing popularity thanks to its remarkable explanatory capacity. The most commonly used version of the gravity equation (Bergstrand, 1985) is presented in equation (7.1): All displayed equations in this chapter should have thin spaces around  mathematical symbols. (7.1) where Xij,t is the amount of exports from country i to country j, at time t. The variable Yi,t is the GDP of country i at the time t, while Yj,t is the GDP of country j at the time t. Dij is the distance between the two countries i and j. The variable Aij represents various factors that may either stimulate or reduce trade between country i and country j. Finally, ζij,t is a log-normally distributed error term, with E (ln (ζij,t)) = 0.

156

M. Grazzi

In this elementary version, the equation indicates that the volume of export between two countries depends positively on their economic size and negatively by the transport costs, captured by the absolute distance between their economic centres. This specification is usually presented in a logarithm format, where logarithms are all natural logarithms. Therefore, the coefficients of the independent variables represent the elasticities of the export flows to host and source country’s GDPs and distance between the countries. Even this basic version of the gravity equation is a quite simple but powerful instrument to explain bilateral flows between countries. However, a large part of variation in these flows remains unexplained. In order to refine the estimation of the gravity equation, most scholars ‘augment’ it, adding other variables, with various theoretical justifications. The most commonly used include: income per capita, adjacency, common language, colonial links and border effects. In the last 40 years the gravity equation has been one of the most popular techniques to analyse bilateral trade flows, but only recently it has been applied to the analysis of cross-border capital movements or cross-border multinational activities. A plausible reason appears to be the fact that, while in the trade literature the gravity model has a robust conceptual basis, the use of this model for the case of FDI is still somewhat ad hoc (Stein and Daude, 2007), although there have been recent developments in laying the theoretical underpinning of the gravity equation applied to cross-border investment (e.g. Kleinert and Toubal, 2005). However, given the empirical similarity of FDI trends with those of trade flows, the gravity instrument has often been employed in estimating bilateral FDI flows, usually with good results (e.g. Brainard, 1997; Brenton et al., 1998; Eaton and Tamura, 1994). In this chapter, the gravity model is used to predict the volume of FDI flows into Chile, and it implies a particular version of the basic gravity equation, where the dependent variable is now specified as inward FDI flows into Chile, while the variables regarding the host country on the right hand side do not vary by country, but just over time. In this first basic model, we add just the per capita GDP of the investor country as an additional variable, to proxy for the development level of a country. So the gravity equation becomes: (7.2) where the dependent variable is the flows of FDI into Chile from country j at time t and the independent variables are: Chilean GDP at time t, GDP of country j at time t, GDP per capita of country j at time t and geographical distance between Chile and country j. A main problem of this specification is that FDI flows between two countries are often equal to zero and they may also be negative (e.g. due to repatriation of profits). Yet the gravity equation predicts that flows between countries are always positive, even if they may be small. Furthermore, the usual natural logarithm transformation cannot operate on zero and negative values. Several strategies have been suggested in the literature to handle the presence of zero flows.4

FDI in Chile 157 A common solution is to simply reduce the considered sample to the positive observations, in order to avoid the estimation problem related to zero and negative values (e.g. Rose, 2000). But this approach does not consider the fact that also zero flows may convey important information for the analysis. A second possible approach suggests substituting the non-positive values with a small constant, in order to be able to utilize these observations in the log-linear model. Wang and Winters (1991) and Raballand (2003), among others, followed this strategy. However, this approach seems generally unsatisfactory, because the inserted value is arbitrary and does not necessarily reflect the underlying expected value (Linders and de Groot, 2006). Moreover, in the case of bilateral trade, zero flows mostly occur between very small or very distant countries, specifically for FDI, zero or negative values are much more frequent. For example, in our sample, 281 observations are non-positive and omitting them from the analysis could seriously underestimate the effects of variables with a negative impact on the flows. A third solution implies the application of the so-called ‘Inverse Hyperbolic Sine Function’ to the dependent variable, instead of the natural logarithm function. Such a transformation, first proposed by Johnston (1949), does not truncate or eliminate values of the dependent variable and it allows performing the regression on the entire available sample. This way of imposing the inverse hyperbolic sine (IHS) function to the dependent variable while imposing natural logarithm on the dependent variables has been used in studies on household wealth (Burbidge et al., 1988 and Carroll et al., 1999) and it has been recently proposed by Kristjánsdóttir (2005) as applicable in gravity modelling. In this investigation, the estimation strategy will consist of two different stages. In the first one, the basic equation (7.2) is estimated using the three procedures5 described above in a panel-estimation framework. Even if the large majority of studies estimate the gravity equation using ordinary least squares (OLS) either on cross-sectional or pooled cross-sectional data, two distinct advantages in using a panel data model rather than the traditional linear regression model can be identified. First, a panel data model captures both crosssection and time-series variation of the dependent variable. Second, it allows the measurement not only of the effects that observable variables have on the dependent variable, but also of the effects of relevant unobservable or nonmeasurable variables. While observable variables are normally considered into the model, the unobservable variables are incorporated into the model depending upon whether a fixed-effect (FE) or random-effects (RE) model is used in estimation. In the RE model, the unobservable factors that differentiate cross-section units are assumed to be characterized as randomly distributed variables. Even if Matyas (1997) and Egger (2000) claim that the correct econometric specification should be estimated as a fixed effect model rather than as a random model one, given that the cross-sectional units of this analysis are the countries which are the source of FDI flows into Chile and that these partners vary considerably by culture, religion, political system and many other factors, it seems

158

M. Grazzi

quite reasonable to assume here that the differences between them are randomly distributed. Among others, Bevan and Estrin (2004) used random effects to estimate a gravity equation for FDI flows. Moreover, in all our specifications we perform the Breusch-Pagan test and in all cases the result is a rejection of the null hypothesis of no-random effect. We have performed also the Hausman test to assess the independence of the random effects from the explanatory variable and the result is the acceptance of the null hypothesis. Considering that, the adoption of the random-effects estimation model method is justified. An additional econometric problem arises from the fact that we are estimating a single-country panel model. In fact, the variable GDPC,t is an exclusively timevariant regressor, and then the direct estimation of the gravity equation would prevent us to identify the real impact of this variable on FDI inflows. A possible solution, offered by Földvári (2006) is to include in the regression a polynomial time trend, in order to capture the time-variant effects without losing the possibility to identify the coefficients of the only time-variant regressor. Consequently, a polynomial time trend of fourth order is included in the regression equation. Then, applying the natural logarithm transformation6 and adding the time trend, the basic equations to be estimated becomes for each of the three specifications, respectively:

(7.3)

(7.4)

(7.5) where zj,t is replaced by ej,t, so that E(ln zj,t) = E(ej,t) = 0. The estimation of equations (7.3), (7.4) and (7.5) constitutes the first step of our empirical analysis, while the second step is the estimation of an ‘augmented’ version of the basic equation, designed to take into consideration the impact of the entry into force of different models of international economic agreement, such as the FTAs, DTTs and BITs, and the membership of APEC (Asian-Pacific Economic Cooperation). Moreover, we control for trade flows, classic ‘gravity dummies’ (common language and sharing a common border)7 and a set of ‘trade blocs’ dummy, which indicate the membership of the partner country in a regional economic agreement (EU, NAFTA, CAN and Mercosur). Thus, the augmented equations for the three specifications are:

FDI in Chile 159

(7.6)

(7.7)

(7.8)

7.4  Data and variables description Data on FDI used in this chapter have been provided by the Chilean Foreign Investment Committee. These data cover FDI flows into Chile from 47 different countries over a 16 years period, from 1990 to 2005, covering almost 98 per cent of the total inflows through the Decree 600 Mechanism. The data cover the annual FDI flows into Chile from: Argentina, Australia, Austria, Belgium, Bolivia, Brazil, Canada, China, Colombia, Costa Rica, Denmark, Dominican Republic, Ecuador, Egypt, Finland, France, Germany, Greece, Honduras, Iceland, India, Ireland, Israel, Italy, Japan, Republic of Korea, Liberia, Luxembourg, Malaysia, Mexico, Netherlands, New Zealand, Norway, Panama, Paraguay, Peru, Portugal, Romania, Singapore, South Africa, Spain, Sweden, Switzerland, United Kingdom, United States, Uruguay and Venezuela. Inflows from Bahamas, Barbados, Bermuda, Cayman Islands and Netherlands Antilles have not been considered, because they are presumably not the real source of the investment but just a channel used for fiscal reasons. The number of observations is therefore 47 times 16, equal to 752. As for trade data, the chapter makes use of the IMF data set ‘Direction of Trade’, which provides bilateral merchandise trade between over 180 countries from 1948 to 2004. The FDI and trade values are reported in current US dollars, which should approximate a correction for the different exchange rates across countries. In order to obtain real figures, these values have been deflated by using the US GDP deflator furnished by the Global Development Finance database of the World Bank, whose base year (i.e. where GDP deflator value is equal

FDI flows into Chile from country j in time t through DL600 Mechanism Natural log transformation of FDIj,t Inverse hyperbolic sine transformation of FDIj,t GDP of country j in time t Natural log transformation of GDPj,t GDP per capita of country j in time t Natural log transformation of GDPPCj,t GDP of Chile in time t Natural log transformation of GDPC,t Distance between Santiago de Chile and the most important city in country j Natural log transformation of Dj Dummy variable = 1 if country j shares a border with Chile Dummy variable = 1 if the language of country j is Spanish Dummy variable = 1 if an FTA is in force between Chile and country j in time t Dummy variable = 1 if DTT is in force between Chile and country j in time t Dummy variable = 1 if a BIT is in force between Chile and country j in time t Dummy variable = 1 if country j was member of the EU in time t Dummy variable = 1 if country j was member of the CAN in time t Dummy variable = 1 if country j was member of the Mercosur in time t Dummy variable = 1 if country j was member of the NAFTA in time t Dummy variable = 1 if country j and Chile were both member of APEC in time t Trade flows from country j into Chile in time t Natural log transformation of IMPj,t Trade flows from Chile into country j in time t Natural log transformation of EXPj,t Polynomial time trend

FDIj,t ln (FDIj,t) Sinh–1 (FDIj,t) GDPj,t ln(GDPj,t) GDPPCj,t ln (GDPPC) GDPC,t ln (GDPC,t) Dj ln (Dj) Adjj Lanj FTAj,t DTTj,t BITj,t EUj,t CANj,t Mercosurj,t NAFTAj,t APECj,t IMPj,t ln (IMPj,t) EXPj,t ln (EXPj,t) t, t2, t3, t4

Source: author’s elaboration.

Description

Variable

Table 7.3 Variable description

Trillions of US dollars (2000 base) Natural logarithm US dollars (2000 base) Natural logarithm Trillions of US dollars (2000 base) Natural logarithm Kilometres Natural logarithm Dummy variable Dummy variable Dummy variable Dummy Variable Dummy variable Dummy variable Dummy variable Dummy variable Dummy variable Dummy variable Thousands of US dollars (2000 base) Natural logarithm Thousands of US dollars (2000 base) Natural ;ogarithm

Billions of US dollars (2000 base) Natural logarithm

Unit of measurement

FDI in Chile 161 Table 7.4 Summary statistics Variable

Obs.

Mean

Std. Dev.

FDIj,t Ln (FDIj,t) Ln (FDIj,t +1) Sinh–1 (FDIj,t) GDPj,t ln(GDPj,t) GDPPCj,t ln (GDPPC) GDPC,t ln (GDPC,t) Dj ln (Dj) IMPj,t ln (IMPj,t) EXPj,t ln (EXPj,t)

752 471 752 752 752 752 752 752 752 752 752 752 752 721 752 716

73.49 9.03 5.63 5.83 563 25.47 13,512.36 8.87 67.9 24.91 9,899.73 9.01 328 18.21 305 18.1

301.15 2.72 4.85 5.68 1,440 1.93 11,625.46 1.38 15.3 0.24 4,761.72 0.71 626 2.03 629 2.14

Min. –390.36 0.69 0 –13.57 121 18.61 56.52 4.03 40.8 24.43 1,128.32 7.03 0 8.64 0 9.06

Max. 4,720 15.37 15.35 16.06 11,100 30.04 49,979.78 10.82 93.9 25.27 19,079.88 9.86 5,540 22.44 4,540 22.24

Source: author’s elaboration on own calculations.

to 100) is 2000. Population and real GDP data have been obtained from the World Bank’s World Development Indicators. Distance refers to the geographical absolute distance between the most important cities/agglomerations (in terms of population) of two countries. It has been obtained from the CEPII (Centre d’Etudes Prospectives et d’Informations Internationales) database. Information on common borders and common language has been taken from the same database. As for the list of BITs, DTTs and FTA signed by Chile and currently in force, information has been obtained from the Chilean Foreign Investment Committee. Variables description is summarized in Table 7.3, while descriptive statistics are provided in Table 7.4.

7.5 Results In Table 7.5 the main empirical results obtained from the estimation of the basic gravity model (equations 7.3, 7.4 and 7.5) are presented. The results obtained are generally consistent with the theoretical expectations, and they are generally not dependent on the adopted estimation methodology. Coefficients of source countries’ GDP and GDP per capita and distance are positive and highly significant in all the three columns, while the coefficient of Chilean GDP is not significant, irrespective of methodology. The variable distance is statistically significant at the conventional level and negative in all the estimations. The estimates suggest that a 1 per cent increase in distance leads, ceteris paribus and on average, to a decline in FDI that varies between 0.79 per cent and 1.66 per cent, depending on the estimation methodology. The fact that transport costs impact negatively on the level of investment

162

M. Grazzi

Table 7.5 Random effects estimation of the baseline equation Dep. variable

Ln (FDIj,t)

Ln (FDIj,t +1)

Sinh–1 (FDIj,t)

Variables

(1)

(2)

(3)

ln(GDj,t)

0.827*** (0.12) 0.596*** (0.18) –5.109 (5.66) –0.795** (0.31) 0.321 (0.58) 0.112 (0.12) –0.0153 (0.013) 0.000521 (0.00041) 471 47 0.4197 318.12 0 7.4 0.1925

1.256*** (0.23) 0.909*** (0.32) –5.521 (7.83) –1.664*** (0.57) 1.316* (0.79) –0.0629 (0.16) –0.00383 (0.016) 0.000225 (0.00053) 752 47 0.4288 923.9 0 7.29 0.2946

1.313*** (0.26) 0.929*** (0.36) –2.612 (10.7) –1.453** (0.63) 1.228 (1.08) –0.11 (0.21) 0.000815 (0.022) 0.0000895 (0.00072) 752 47 0.3411 504.61 0 4.27 0.6401

ln (GDPPC) ln (GDPC,t) ln (Dj) tt tt2 tt3 tt4 Observations Number of partner Adjusted R-squared Breush-Pagan Prob> chi2 (1) Hausman Test Prob> chi2 (6)

Source: author’s elaboration on own calculations. Notes Robust standard errors in parentheses. * Significant at 10%; ** significant at 5%; *** significant at 1%.

flows into Chile confirms the theoretical expectations. In fact, Chile is a naturalresource economy and the main recipient of foreign investment is the mining sector. Therefore, one could expect that the major share of the FDI is ‘vertical’ in nature and that transport costs should affect FDI negatively. The coefficients for the market size of the source economies are highly statistically significant and also in line with the theoretical predictions. An increase in the source country market size of 1 per cent would increase FDI inflows by the range 0.82 per cent and 1.31 per cent, depending on the estimation methodology. A positive coefficient that is statistically significant is also obtained for the estimates regarding the GDP per capita of the source countries in all the three columns. The theoretical assumption that more developed economies engage in more FDI is therefore confirmed from the data. Naturally, richer countries generally source international investments. As for the coefficients corresponding to the variable related to the size of the host country market (Chilean GDP), they are not statistically significant. Therefore the hypothesis that foreign investors are

FDI in Chile 163 attracted to a larger domestic market is rejected. It strengthens the argument that FDI in Chile has been basically resource-seeking rather than market-seeking. Further confirmation of these findings is offered by the results of the estimation of the augmented model, which are presented in Table 7.6. In fact, the coefficients for GDP and GDP per capita of the source country maintain a positive sign and high significance independently of the estimation methodology adopted. Moreover, the dimension of their impact on FDI flows is very similar to that estimated for the basic model: the coefficient for source country GDP now varies from 0.92 to 1.3, while that regarding source country GDP per capita varies from 0.82 to 1.07. The negative sign of the coefficient on distance (not significant only in the case of the estimation performed in the restricted sample of positive observation) and the non-significance of Chilean GDP also in the augmented model estimation confirm the hypothesis that FDI into Chile is basically a vertical, resourceseeking investment. The coefficients for the trade blocs dummies are not significantly different from zero in all the three regressions. It implies that it is not possible to identify an impact on the FDI flow into Chile deriving from belonging to a particular regional organization. Not statistically significant are also the coefficients regarding the border dummy and the language dummy. It seems reasonable to assume that in the case of capital flows, adjacency is not as important as in trade flows. As for the coefficients of trade flows, neither the coefficients on Chilean imports nor those on Chilean exports are statistically significant at conventional levels. The lack of significance of the import’s coefficient is consistent with the hypothesis of prevalent ‘vertical’ FDI,8 while one would have expected a significant positive relation between FDI inflows and Chilean exports. In fact, if the firms choose to invest abroad to seek resources or lower factor prices and the investment is designed to supply the source country’s market, the theoretical prediction is a positive and significant coefficient. Finally, we analyse the coefficients related to the ‘Chilean Economic Foreign Policy’. The entry into force of a BIT between Chile and a FDI source country is found to have a considerable significant and positive impact in both the estimation methodologies which use the entire sample. The predicted positive variation is given, on average and ceteris paribus, by [exp(coefficient) – 1]*100. The size of this impact is then computed to be either an increase of 144.73 per cent or 216.76 per cent. A small set of papers has empirically assessed the impact of BITs on FDI. UNCTAD (1998) has not found any statistical evidence that they could attract FDI in addition to traditional determinants, using a cross-section analysis based on about 100 countries. Hallward-Driemeier (2003), using a 20 years panel dataset, confirmed the lack of an independent effect of BITs of FDIs, after having controlled for other determinants of country attractiveness (especially institutional quality). However, Egger and Pfaffermayer (2004) in a recent study provided a more optimistic vision, finding a positive impact. In this chapter, given the single-country approach adopted, the institutional quality is not as important as in a multi-country analysis, and our results are in

Table 7.6 Random effects estimation of the augmented equation Dep. variable

Ln (FDIj,t)

Ln (FDIj,t +1)

Sinh–1 (FDIj,t)

Variables

(4)

(5)

(6)

ln(GDP j,t) ln (GDPPC) ln (GDPC,t) ln (D j) ln (IMP j,t) ln (EXP j,t) Adj j Lan j EU j,t NAFTA j,t Mercosur j,t CAN j,t APEC j,t BIT j,t DTT j,t FTA j,t tt tt2 tt3 tt4 Observations Number of partner Adjusted R-squared Breush-Pagan Prob> chi2 (1) Hausman Test Prob> chi2 (15)

0.916*** (0.26) 0.825*** (0.25) –5.309 (5.75) –0.77 (0.689) –0.114 (0.18) 0.208 (0.14) –0.289 (1.06) 0.744 (0.8) –0.284 (0.58) 0.473 (0.66) –0.268 (0.93) –0.77 (0.92) –0.436 (0.42) 0.0937 (0.28) 0.408 (0.46) –0.0372 (0.4) 0.284 (0.61) 0.14 (0.13) –0.0189 (0.013) 0.000642 (0.00042) 450 45 0.45 215.13 0.00 18.2 0.1502

1.280*** (0.31) 0.969*** (0.35) –3.666 (8.13) –2.384** (1.09) –0.0155 (0.18) 0.087 (0.17) –2.505 (1.76) 0.43 (1.25) –0.394 (0.93) 1.504 (1.07) –1.91 (1.38) –1.529 (1.44) –0.586 (0.59) 0.895** (0.38) 1.194* (0.61) 0.0354 (0.5) 1.021 (0.87) –0.0454 (0.17) –0.00449 (0.017) 0.000228 (0.00055) 701 47 0.51. 327.53 0.00 32.73 0.112

Source: author’s elaboration on own calculations. Notes Robust standard errors in parenthesis * Significant at 10%; ** significant at 5%; *** significant at 1%.

1.309*** (0.38) 1.073*** (0.4) –0.218 (11.2) –2.858** (1.32) –0.0525 (0.23) 0.193 (0.22) –3.571 (1.98) 0.14 (1.43) –0.783 (1.07) 1.351 (1.41) –3.290* (1.83) –1.682 (1.65) –0.464 (0.8) 1.153** (0.51) 0.973 (0.84) 0.315 (0.69) 0.94 (1.2) –0.112 (0.23) 0.00212 (0.024) 0.00006 (0.00076) 701 47 0.42 151.15 0.00 21.28 0.1281

FDI in Chile 165 line with the findings of Egger and Pfaffermayer. A possible area for further research in the field is the study of the impact of signing a BIT on the probability of establishing an FDI flow, passing from zero to a positive investment. With regards to other economic treaties, no statistical significance for the entry into force of a FTA has been found, while there is a small evidence of a positive impact of DTT, with only one of the adopted estimation methodologies. Finally, the simultaneous membership of Chile and the source country in the APEC Forum does not seem to have a positive impact on the amount of FDI flows.

7.6 Conclusions We have examined the determinants of FDI inflows into Chile from 1990 to 2005 through the estimation of a gravity model. First, the obtained results confirm the validity of the gravity equation as a valuable instrument to analyse not only bilateral trade flows but also bilateral capital flows. FDI is found to be negatively affected by distance and positively affected by the source country GDP and GDP per capita. Chilean market size, proxied by GDP, does not influence FDI inflows. These findings confirm the hypothesis that investments in Chile are mostly of the vertical type, i.e. resource-seeking, rather than marketseeking. Second, I have used the gravity instrument to evaluate the impact of the Chilean Economic Foreign Policy, that is, the signature of bilateral economic agreements, such as BITs, DTTs and FTAs, on the amount of FDI flows. The entry into force of a BIT is found to have a positive and significant impact on the FDI inflows, while there is little or no evidence of a significant effect of DTTs and FTAs. The finding that the signature of a BIT with a country increases the investment flows originated in that country is particularly interesting for the connected strong policy implication, not only for developing countries, but also for industrialized countries devoted to global development. In fact, BITs represent one possible way for a developed country to boost private investment flows towards poor countries. But ‘rich countries do not have many direct policy instruments to improve the amount of FDI received by poor’ (Mayer, 2006), because it usually implies policy measures that need to be implemented in the host country rather than in the source country. Consequently, BITs are a valuable instrument to drive foreign investment flows.

Notes 1 The contents are the sole responsibility of the author, and do not necessarily reflect the position of the Inter-American Development Bank. 2 Chile is the world’s largest producer of copper with a mine output in 2009 of nearly 5.4 million tonnes, corresponding to over one-third of world copper mine production (ICSG, 2010). 3 FDI can enter into Chile through different legal mechanisms. The detailed sectoral data used in this section covers only the investment made through the Decree Law 600 (DL600). Under this regime, whose use is optional, foreign investors bringing capital,

166

4

5 6 7 9

M. Grazzi

physical goods or other forms of investment into Chile may ask to sign a foreign investment contract with the state of Chile. Although it is the channel through which most of the investment enters the country, this fact could potentially bias the analysis: investment in sectors where using this legal mechanism is particularly favourable may be overestimated, while other sectors where a large part of the inflows arrives through other legal channels may be underestimated. Besides the solutions described in this section, other methodologies used in the literature to solve the problem are: the estimation of a TOBIT model (Eaton and Tamura, 1994; Wei, 1998); a Heckman selection model (Razin et al., 2003; Razin et al., 2005) and of a Poisson model (Silva and Tenreyro, 2006). Following Razin et al. (2003), a one-dollar value (with the log equalling zero) as a common low value for the value of the FDI flows in the zero or negative FDI observations. The natural logarithm transformation is applied both on the sample constituted just by the positive observation and on the entire sample, having substituted the zero values with one. In this case, the 14 negative observations have been considered as zero values. The common colonization dummy has not been considered, because, in the case of Chile, it is equal to the common language (Spanish) dummy. In case of ‘horizontal’ FDI, countries would substitute exports to the partner market by localizing the production in that country.

References Bergstrand, J. (1985), ‘The gravity equation in international trade: some microeconomic foundations and empirical evidence’, Review of Economics and Statistics, 67 (3): 474–481. Bevan, A. and Estrin, S. (2004), ‘The determinants of foreign direct investment into European transition economies’, Journal of Comparative Economics, 32 (4): 775–787. Brainard, S.L. (1997), ‘An empirical assessment of the Proximity-Concentration Tradeoff between Multinational Sales and Trade’, American Economic Review 87 (4): 520–544. Brenton, P., Di Mauro, F. and Lucke, M. (1998), ‘Economic integration and FDI: an empirical analysis of foreign investment in the EU and in Central and Eastern Europe’, Empirica, 26 (2): 95–121. Burbidge, J.B., Magee, L. and Robb, A.L. (1988), ‘Alternative transformations to handle extreme values of the dependent variable’, Journal of the American Statistical Association, 83 (March): 123–127. Carroll, C., Dynan, K. and Krane, S. (1999), ‘Unemployment Risk and Precautionary Wealth: Evidence from Households’, Balance Sheets. Federal Reserve Board, Finance and Economics Discussion Paper No. 1999–15. Chilean Foreign Investment Committee (2002), online, available at: http://ww.foreigninvestment.cl (accessed October 2009). Eaton, J. and Tamura, A. (1994), ‘Bilateralism and regionalism in Japanese and U.S. trade and direct foreign investment patterns’, Journal of the Japanese and International Economies, 8 (4): 478–510. Egger, P. (2000), ‘A note on the proper econometric specification of the gravity equation’, Economics Letters, 66 (1): 25–31. Egger, P. and Pfaffermayer, M. (2004), ‘The effect of bilateral investment treaties on FDI’, Journal of Comparative Economics, 32 (4): 788–804. ECLAC (2007), ‘Foreign Investment in Latin America and the Caribbean’, Santiago de Chile.

FDI in Chile 167 Economist Intelligence Unit (EIU) (2007), ‘Country Report Chile’, London. Földvári, P. (2006) ‘The economic impact of the European integration on the Netherlands: A quantitative analysis of foreign trade and foreign direct investments’, Tekst, Proefschrift Universiteit Utrecht. Hallward-Driemeier, M. (2003), ‘Do bilateral investment treaties attract foreign direct investment? Only a bit . . . and they could bite’, World Bank Policy Research Working Paper No. 3121. Head, K. (2003), ‘Gravity for beginners’, online, available at: www.economics.ca/keith (accessed October 2009). International Copper Study Group (ICSG) (2010), ‘The World Copper Factbook, 2010’. Kleinert, J. and Toubal, F. (2005), ‘Gravity for FDI’, CEGE Discussion Paper no. 46, Göttingen University. Kristjánsdóttir, Helga (2005), ‘Determinants of foreign direct investment in Iceland’, CAM Centre for Applied Microeconometrics Working Paper no. 15, University of Copenhagen. Johnston, N.L. (1949), ‘Systems of frequency curves generated by methods of translation’, Biometrika, 36 (1/2): 149–176. Linders, G.M. and de Groot, H.L.F. (2006), ‘Estimation of the gravity equation in the presence of zero flow’, Tinbergen Institute Discussion Paper 2006-072/3. Mayer, T. (2006), ‘Policy coherence for development: A background paper on foreign direct investment’, OECD Centre for Development Working Paper No. 253. Matyas, L. (1997), ‘Proper econometric specification of the gravity model’, World Economy, 20 (3): 363–368. OECD (2002), ‘Foreign direct investment for development: Maximising benefits, minimising costs’, CIME Report. Raballand, G. (2003), ‘Determinants of the negative impact of being landlocked on trade: An empirical investigation through the Central Asian case’, Comparative Economic Studies, 45 (4): 520–536. Razin, A., Rubinstein, Y. and Sadka, E. (2003), ‘Fixed costs and FDI: The conflicting effects of productivity shocks’, NBER Working Paper 10864, National Bureau of Economic Research, Inc. Razin, A., Sadka, E. and Tong, H. (2005), ‘Bilateral FDI flows: Threshold barriers and productivity shocks’, NBER Working Papers 11639, National Bureau of Economic Research, Inc. Rose, A. (2000), ‘One money, one market: The effect of common currency on trade’, Economic Policy, 15 (30): 7–46. Silva, J.M.C. and Tenreyro, S. (2006), ‘The log of gravity’, Review of Economics and Statistics, 88 (4): 641–658. Stein, E. and Daude, C. (2007), ‘Longitude matters: Time zones and the location of foreign direct investment’, Journal of International Economics, 71 (1): 96–112. Tinbergen, J. (1962), Shaping the World Economy, New York: Twentieth Century Fund. UNCTAD (various issues), ‘World Investment Report’. Wang, Z.K. and Winters, L.A. (1991), ‘The trading potential of Eastern Europe’, CEPR Discussion Paper 610. Wei, S. (1998), ‘How taxing is corruption on international investors?’ NBER Working Paper 6030. World Bank (2005), ‘Country Profile for Chile’, online, available at: http://web.worldbank.org/WBSITE/EXTERNAL/COUNTRIES/0,,pagePK:180619~theSiteP K:136917,00.html (accessed October 2009).

8

Assessment of the distributive impact of trade reforms in Uruguay1 Fernando Borraz, Daniel Ferrés and Máximo Rossi

8.1 Introduction Marked trade liberalisation and deeper regional integration have characterised Uruguay’s trade policy since the early 1970s. As Uruguay entered Mercosur in the early 1990s, the country embarked in a plan for tariff reduction at the regional and extra-zone levels. The coming years may imply further trade opening, particularly as the WTO Doha Development Agenda, the EU–Mercosur Association Agreement and other initiatives under negotiation will enter the implementation phase. Even in the textbook case, traditional trade theory acknowledges that although the gains from trade might be positive for a country as a whole, they might not be distributed evenly across all the groups. There is nowadays an increasing concern throughout the region over the asymmetric distribution of costs and benefits of trade integration. One of the initial objectives of Doha was to ameliorate inequalities between rich and poor countries. In this context, it is fundamental to determine whether trade integration can be regarded as poverty reduction policy or, on the contrary, if it may be associated with intensified poverty effects. Regressive outcomes are more likely in the absence of complementary domestic reforms and policies that would help maximise gains from trade, protect the most vulnerable from transitional costs and ensure an equitable distribution of net gains. In order to design a domestic complementary agenda, it is therefore of the utmost importance to generate empirical evidence to determine the distributional impacts of trade liberalisation. Trade reforms cause direct changes in local relative prices, which indirectly affect household’s income, expenditure and welfare. On the expenditure side, net effects depend on product structure of the consumption basket and on whether individuals are net producers or net consumers. Changes in household’s income are explained by the fact that the trade reforms imply a reallocation of resources between sectors, resulting in changes in factor prices, particularly wages. As we analyse both changes in prices and variations in income, we are able to determine the overall change in household welfare. Recently, promising trade economics literature is attempting to precisely measure the net effect of trade

Trade reforms in Uruguay 169 integration on income distribution and poverty, taking into consideration both income and expenditure effects (Giordano and Florez, 2007). By trade reforms we mean both national and foreign trade reforms. We consider that national trade reforms imply the removal of tariff protection on Uruguayan imports. Foreign trade reforms refer to the possibility of local exports to access those markets in the developed countries (or elsewhere). For small open economies, like Uruguay, theory indicates that changes in world prices translate immediately to local price levels. Therefore, when tariff reductions and importquotas removals take place in third countries, the price of Uruguayan exports to developed countries is positively affected. But trade liberalisation plus enhanced market access does not necessarily equal poverty reduction. As a means to measure the effect of trade liberalisation on poverty, we plan to evaluate the impact of both national and foreign trade reforms on the head count ratio. The objective of this technical research is to assess the linkages between trade, poverty and inequality by analysing the impact of trade liberalisation through two main transmission channels: prices and income. Following the methodology developed by Porto (2006), the study first assesses the implications of a given trade shock, that is, a national or a foreign trade reform, in relative domestic prices of traded goods (imports and exports). Second, the study will analyse the response of labour income and consumption channels at the household level. This leads to the third step, which is the induced change in the head count poverty ratio. This methodology will allow us to identify the new income that individuals would earn as a result of a policy change, in order to determine to which extent trade liberalisation contributes to poverty reduction. Detailed data at the household level will be used to assess how inequality and poverty have evolved over time, across regions (e.g. urban areas compared to the rest of the country) and across different household types (e.g. ranked according to the education level). Obtained results evidence that (1) the decrease of tradable goods’ prices largely benefited the lower-income segment of the Uruguayan population, (2) the dynamics of the non-tradable goods’ prices had a clear pro-rich impact and (3) trade liberalisation had a clear positive impact for both the highly paid and for those with the lower positions in the salary distribution.

8.2 Trade reform in Uruguay The trade liberalisation process that took place in Uruguay since 1980 can be classified in four periods. Average formal tariff decreased from a 30 per cent level in 1980 to around 16 per cent in 1983. From 1983 to 1990 it remained quite constant (it actually grew on a year-to-year basis in 1985 and 1990). The 1990s marked an intense period of tariff reduction: average tariff went from 14 per cent in 1990 to 8 per cent in 1994. Since then, the average tariff level has been stable. We can conclude that in the last 25 years Uruguay has experienced two periods of marked tariff reduction (1980–1983 and 1991–1994) and two periods of tariff stability (1985–1990 and 1995–2005). When we observe the trade statistics, it is direct to conclude that Uruguayan trade flows show a sharp increase in the last

170

F. Borraz et al. Table 8.1 Trade openness coefficient In constant terms (%) 1970–1979 1980–1989 1990–1999 1995–2004

39.6 47.4 76.8 80.5

Source: Central Bank of Uruguay.

35 years. As shown in Table 8.1, trade has particularly intensified in the last 10–15 years. It is interesting to note that while imports were 12 per cent higher in 2004 than in 1995, exports grew by 28 per cent in the same period. In 1991, Uruguay entered Mercosur, a Trade Agreement signed between Argentina, Brazil, Paraguay and Uruguay (Treaty of Asuncion). The creation of Mercosur marked the acceleration in the fall of import tariffs and the long-term commitment that Uruguay would continue the liberalisation process. At the early years of Mercosur, the administration in place made strong emphasis on finally concluding the liberalisation process (started in the mid-1970s), removing the remnants of the protectionism apparatus. For example, since 1991 the number of tariff categories was reduced from five to three. In this scenario, the Uruguayan trade policy imaged those requirements of the country’s regional partners. From January 1995, Mercosur began to operate like an imperfect customs union. Table 8.2 shows the intra and extra Mercosur trade flows. Ideally, Mercosur would enable Uruguay to obtain preferential access to a large and close market. But at present there is a level of disenchantment with the integration process at Mercosur. Many Uruguayans feel that the integration process has been slow-paced, responding to specific interests from industrial lobbying groups from Brazil and Argentina. As an example, the proliferation of non-tariff barriers shows the low level of commitment to trade disciplines. This phenomenon has operated in a way that production specialisation has not really occurred in Uruguay. Additionally, another sign of Mercosur modest results at the extrazone level is related to the low number of trade agreements signed with third parties (countries or regions). In fact, it was only in 2007 that Mercosur (as a group) signed its first trade agreement with an extra-zone party (Israel). Table 8.2 Intra and extra Mercosur trade flows (US$): simple average

Intra-Mercosur trade Extra-Mercosur trade Total Mercosur trade Intra-Mercosur trade (%) Source: ALADI.

1995–2000

2001–2006

35,464,482 148,903,829 184,368,311 19%

34,620,294 202,954,670 237,574,964 15%

Trade reforms in Uruguay 171

8.3 Inequality and poverty in Uruguay: the stylised facts It is important to make clear that income inequality and poverty are different concepts. While income inequality refers to income distribution (a relative term), poverty refers to the relationship between (absolute) individual income and the poverty line. Poverty reduction may be associated to either higher income inequality or a more equal income distribution. It is broadly accepted that economic researchers and policy-makers should be concerned about both indicators of social welfare, when evaluating alternative policies. Various studies have described the stylised facts of income distribution and poverty for Uruguay in the time horizon that we are considering. Bucheli and Rossi (1994) concluded that inequality was quite constant during the period 1984–1992. Moreover, Rossi (2001) examined the evolution of inequality and poverty in Uruguay between 1989 and 1997.2 His results show that wage inequality increased since 1991 and poverty levels increased between 1993 and 1997. This result is confirmed both for Montevideo and the rest of the urban country, especially since 1991. Also, Miles and Rossi (1999) and Gradin and Rossi (2000) obtained results indicating higher levels of inequality in Uruguay during the 1990s. With respect to poverty, the evolution for the 1989–1997 period, based on the headcount ratio, shows that poverty decreased before 1993 and then (slightly) increased (Rossi, 2001). Authors attributed that increase in poverty levels during the 1990s could be related to growth problems, increased openness of the Uruguayan economy or to the process of decentralisation in wage negotiation. Men and women show similar evolutions but women have an increase in their poverty relative to men. Finally, poverty is more intensive in cities outside Montevideo than in the capital city. The poverty (headcount ratio) shows stability between 1997 and 2001, an increasing trend until 2004 and then a decreasing trend up to now (Vigorito 2007). The figures show that the period of relative stability in tariffs (the 1980s) coincides with a period of relatively unchanged income distribution. As trade liberalisation occurred (the 1990s), wage and income inequality grew. Additionally, trade liberalisation coincided with a period of increase in poverty in Uruguay. It is fundamental that we may investigate whether there is a statistical relation between changes in import tariffs and changes in income distribution and poverty. In particular, proper econometric analysis will allow us to identify the direction in which causality runs.

8.4 Methodology 8.4.1 Effects of national trade reform From a theoretical perspective, the impact of trade on wage inequality could go in either direction. In a Heckscher-Ohlin model, workers should see wages increase relative to capital owners’ rents (alternatively, unskilled wages should

172

F. Borraz et al.

go up relative to skilled wages) in a developing country relatively well endowed with labour (or unskilled labour). In that case, workers would benefit relative to capital owners (or more skilled workers) and income distribution would improve. Under a specific factors model, however, workers that are unable to relocate to labour-intensive industries would lose, and the distributional impact of trade liberalisation is ambiguous. Moreover, empirical studies show that the wage gap between skilled and unskilled workers may increase after trade and investment reform. This could occur, for example, if foreign-owned firms that begin operating in a developing country bring with them technology that increases the demand for skilled workers. In that case, the distributional impact is adverse. The project will study the link between trade, poverty and inequality by analysing the impact of trade liberalisation through two main transmission channels: prices and income. The first possibility is that the new tariff levels that result from trade reforms explain price changes. Price changes may affect individuals in different ways, for example, depending on the share of each good in their consumption basket, as suggested earlier, or if individuals are net producers (as in the case of farmers) or net consumers. A second possibility is changes in household income. This effect is explained by the fact that trade liberalisation implies a reallocation of resources between sectors, resulting in changes in factor prices in the process. In this study we restrict the analysis to four trade goods: food and beverages (FB), clothing and footwear (CF ), house equipment and electronics (HQ), other traded goods (OT); and four non-traded goods: health and education (HE), transport and communications (TC), housing (HO) and other non-traded goods (ON). To analyse the distributional impact of Mercosur on Uruguayan households we use a model based on Dixit and Norman (1980). The variation in exogenous income (Y0) needs to compensate household i to keep the same utility after a change in the price of trade good k (k = 1, . . ., 4) because of the trade reform can be approximated by the following equation: (8.1) where Yi0 is the exogenous income of households i, tk is the tariff for traded good k, Sik is the budget share spent on the good k by household i, Pk is the price of trade good k, Pn is the price of non-traded good n, Sin is the budget share spent by household i, ewiPk is the wage price elasticity with respect to traded good k and qwi is the share of labour income in total household income. The first term in equation (8.1) shows that for a given increase in the price of the trade good k, the higher the share the higher will be the income necessary to compensate the consumer. The budget share approximates the consumption effect. The second term of (8.1) shows the compensation generated by the change in the price of non-traded good that is explained by the trade reform. Their importance is related also to the share spent on non-traded goods. The first

Trade reforms in Uruguay 173 and second term in (8.1) approximate the consumption effect of the Mercosur. Finally, the last term is the labour effect. The trade reform, change the price of trade goods that change household wages. In order to assess the distributional effect to Mercosur we have to estimate the three terms of the previous equation. Impact of tariffs on prices of traded goods Initially, the project will estimate the impact of tariffs on prices. Following Deaton (1997) it is possible to approximate the change in consumption explained by the changes in prices using the expenditures shares of each of the goods. Therefore, only the direct impact will be considered and not other indirect effects. In order to quantify the distributional effects of these price changes there are two possibilities. The first one consists in the estimation of price indices for each individual in the survey, based on pre-trade reform expenditures shares with both prices. In a second step, the effects on individuals of the price change that is explained by the reforms will be quantified. The second approach following Deaton (1997) consists in a non-parametric estimation of expenditure shares across the entire distribution of consumption, and computing average market shares for different incomes. When using the second approach, results are highly dependent on a proper choice of the bandwidth. In particular, the induced change in the price of trade good k after the trade reform is: (8.2) where slk is the expenditure share of the sub-category l in traded good k, δlm is the fraction of imports of good l coming from Mercosur and δkrw is the fraction coming from the rest of the world. Equation (8.2) estimates the price change of traded goods from Mercosur. Impact of prices of traded goods on the price of non-traded goods In order to estimate the impact of the prices of traded goods on the prices of nontraded goods we will estimate the following translog equation:

(8.3) We regress the prices on traded goods on monthly prices of the traded goods and their interactions. In order to avoid a spurious regression we check for cointegration between the variables included in equation (8.3).

174

F. Borraz et al.

Impact of prices on income Some of the papers in this literature focus only on distribution effects of price changes after the reforms, without considering some import effects on the factor markets. This proposal seeks to quantify the impact of openness on total income. In addition, the wage–price elasticity will be estimated. In particular we will regress the log of the real wage earned by person i against completed years of schooling (s), variables (z) such as age, marital status, children at home, region, etc., and the log prices of traded goods interacted with schooling and region.

(8.4) 8.4.2 Effects of external trade reform In order to analyse the impact of external trade reforms over the Uruguayan economy, we focus on a major exported good: beef. Although we only analyse how changes in the global market for beef affect specific variables of the local economy, we believe that these results could be generalised to other exportable goods items. Specifically, we will quantify the impact of trade liberalisation in the global beef markets over labour income, employment and poverty levels in Uruguay. First, we estimate how the change in global price impacts the price level in the local market. We follow a methodology similar to that developed by Porto (2006) in order to identify different effects across education levels and industries. Second, using results obtained in the first stage, we estimate the impact of alternative scenarios of trade liberalisation over labour income, employment and poverty. We study the link between trade, poverty and inequality by analysing the impact of trade liberalisation through two main transmission channels: prices and income. The first possibility is that the new tariff levels that result from trade reforms explain price changes. Price changes may affect individuals in different ways, for example, depending on the share of each good in their consumption basket, as suggested earlier, or if individuals are net producers (as in the case of farmers) or net consumers. A second possibility is changes in household income. This effect is explained by the fact that trade liberalisation implies a reallocation of resources between sectors, resulting in changes in factor prices in the process. The impact of changes in international prices on domestic prices In this section, we aim to estimate the impact of variations in international prices on local price levels: what fraction of the change in global prices is transmitted to the local price levels? And, how long does the transmission process take? In this respect, we will test the long-term cointegration between international and domestic prices.

Trade reforms in Uruguay 175 Given the fact that Uruguay is a geographically small and homogeneous country, we do not consider that it is necessary to work with prices per region. So, we work with an average national price. We estimate the following regression: (8.5) Equation (8.5) allows us to identify the long-term relationship between local and international prices. β1 allows us to determine the referred relationship. In order to estimate cointegration, we conduct the Augmented Dickey–Fuller (ADF ) test over equation (8.5) residuals. Also, we are interested in testing the short-term price dynamics so that we can identify the duration of the transition process. We do this by estimating the following error correction model:

(8.6) where local prices vary between t–1 and t due to changes in international prices for that period (response is indicated by δ) and due to the adjustment to the ‘long term equilibrium’ level with a velocity of γ. In case a cointegration relationship exists, equation (8.6) is valid since it deals only with stationary variables. Based on equations (8.5) and (8.6) we obtain the local prices adjustment after a change in global prices (in an n-months time horizon). The interpretation is as follows: as world prices increase by 1 per cent, local prices vary by δ per cent. In the second period, a term for error correction (γ) is considered. The time horizon for the adjustment of local prices after a shock in the world prices can be estimated as follows: (8.7) The impact of changes in domestic prices on labour income Some of the papers in this literature focus only on distribution effects of price changes after the reforms, without considering some import effects on the factor markets. In our work, we seek to quantify the impact of openness on total income. In addition the wage–price elasticity will be estimated. In particular we will regress the log of the real wage earned by person i against completed years of schooling, individual variables, specific variables indicating geographic location of the household (per Department), and the log prices of traded goods interacted with a sub-group of independent variables. We estimate the following model at the individual level: (8.8)

176

F. Borraz et al.

where wi is the logarithm of real wage per hour, p indicate domestic beef prices, D indicates geographic variables (per Department) and X is idiosyncratic individual variables. We indicate whether the individual is the household head, education level, employment status (and industry), marital status, number of children in the household with age six or below, number of people in the household with age between six and 14. Since the dependent variable, wi, is a zero-censored variable the estimation of (8.4) should not be conducted using OLS. In that case, we would have obtained biased and inconsistent estimators of the impact of beef prices and of individual and geographic variables over labour income. Instead, we estimate the bias selection correction factor based on a Probit model in order to estimate labour market participation. Then incorporate the referred term into equation (8.8) but only for those wage levels that are strictly greater than zero.

8.5 Estimation 8.5.1 Estimation of the effects of national trade reform Impact of tariffs on traded goods Table 8.3 shows the evolution of tariff in Uruguay since 1985. By the mid-1980s, the tariff levels in Uruguay ranged between 43 per cent and 55 per cent. The early years of the 1990s implied a drastic reduction in tariffs: rates went down to an average of 22 per cent in 1992.3 In 1996, Mercosur imposed a sharp reduction in the intra-zone tariff and a slight decrease in the non-Mercosur tariff. The most significant decrease in the intra-zone tariff rate was in FB category (from 21 per cent in 1992 to 5 per cent in 1996). The reduction in the ‘other traded goods’ category was the smallest, from 23 per cent in 1992 to 11 per cent in 1996. In 1999, only four years after the initial Mercosur intra-zone tariff reduction, for all the categories of goods the intra-zone tariff was approximately zero. There were only a few exemptions like the sugar sector. Mercosur was an effective regional trade agreement to rapidly eliminate almost all intra-zone tariffs. The situation is different with respect to the common external tariff (extrazone tariff ), where the reduction was minor. Only for FB we observed an important reduction in the extra-zone tariff following Mercosur, from 21 per cent in 1992 to 15 per cent in 1996. We expect this decrease to cause an important improvement in income distribution and poverty alleviation because poor households have a higher consumption share for FB than the rich households. For the other three categories of goods we observe a minor reduction (around three points) in the extra-zone tariff. However in 1999, we observe a reversal in the trend of reduction of the extra-zone tariff and the tariff for CF, HE and OG return to their pre-Mercosur levels. There is a reversal in the trend toward integration to the world economy. In the FB sector the extra-zone tariff increased only two points from 15 per cent in 1996 to 17 per cent in 1999. In this case we expect an effect of the extra-zone tariff reduction on income inequality and

Trade reforms in Uruguay 177 poverty because of the small tariff reduction. In 2006 Uruguayan extra-zone tariff was 14 per cent for FB and approximately 18 per cent for the other categories of goods. In Table 8.4 we estimate the induced change in tradable prices after Mercosur for the four categories of traded goods considered. We estimate the price change for the 1992–1996 period. Mercosur causes a decrease in the price of the four traded goods considered. It is remarkable that the price reduction was very similar across goods. The highest decrease was for the other traded goods (6.1) and the lowest was for house equipment (4.7 per cent). Figure 8.1 shows the consumption effect for each of the traded good categories. Estimations are made as a Kernel regression. The effect is positive for all of the individuals. However, for FB, HE and OG the consumption effect is propoor. For the poor individuals the consumption gain is higher than for richer individuals. Figure 8.2 shows the pro-poor consumption effect of traded goods. Impact of tariffs on non-traded goods To avoid the spurious regression problem we apply the Engle–Granger cointegration test (based on residuals) to determine the long-term equilibrium cointegrating

Figure 8.1 Consumption effect: compensating variation as per cent of income by income distribution ($U) (source: authors’ estimations).

43 21 14 15 12

Extrazone 1985 1992 1996 1999 2006

44 21 15 17 14

Extrazone 1985 1992 1996 1999 2006

Source: ALADI and Secretaría del Mercosur.

44 21 5 0 0

Intrazone 1985 1992 1996 1999 2006

Food and beverages

Weighted average by expenditure shares

43 21 4 0 0

Food and beverages

Intrazone 1985 1992 1996 1999 2006

Simpled average

Table 8.3 Tariff structure: Uruguay

53 24 21 23 20

53 24 9 0 0

Clothing and footwear

55 23 21 22 19

55 23 7 0 0

Clothing and footwear

48 21 18 21 18

48 21 6 0 0

House equipment and electronics

53 21 19 21 18

53 21 5 0 0

House equipment and electronics

50 23 18 21 17

50 23 11 0 0

Other traded goods

49 22 19 22 17

49 22 11 0 0

Other traded goods

62 15 13 10

1994–95

1992

21 24 21 23

Consumption share

Tariff

Note The price change in the last column is computed using equation (8.2).

Source: authors’ estimations.

Food and beverages Clothing and footwear House equipment Other traded goods

Category

Table 8.4 Price changes from Mercosur

5 9 6 11

1996

Intrazone tariff

15 21 18 18

1996

Extrazone tariff

–5.1 –4.8 –4.7 –6.1

Price changes from Mercosur

180

F. Borraz et al.

&RPSYDULDWLRQ



 

3HUFDSLWDLQFRPH



Figure 8.2 Compensating variation as per cent of income by income distribution ($U): traded good (source: authors’ estimations).

relationship between each of the prices of non-tradable goods and the prices of the traded goods. In the first step, we use the ADF unit root test to analyse the stationarity of the prices. Table 8.5 indicates that all the price variables are non-stationary with a unit root. Next, we proceed to estimate the equation (8.3) by OLS and check for stationarity of the residuals. The result of the Engle-Granger based on residual cointegration tests is shown in Table 8.6: prices of non-traded and prices of traded goods are cointegrated. In other words, there is a stable long-run relationship between both prices. Figure 8.3 shows that the consumption effect of non-traded goods is pro-rich. This fact can be explained by the effect of the change of the price of traded goods in the housing price. Wage–price elasticities Since it is likely that there is a large number of individuals who do not work (especially women) and therefore report zero wages, it would not be appropriate to estimate equation (8.4), the wage equation, using OLS. Since the dependent variable is censored at zero, we only observe the wages of the employed individuals and estimation of the wage equation by OLS will simply yield inconsistent estimates. We allow the impact of the price of traded goods on wages to vary according to individual characteristics including schooling, age and geographical location of the household. This implies that the elasticities of wage and labour market participation with respect to prices vary from one individual to another, according to her age, schooling and geographic location. This is mandatory to

–2.11 –2.10 –11.00 –2.57 –3.43*** –4.33***

–1.73 –2.30 0.03

–3.05 –1.86 –2.81***

–2.08 –4.39*** –5.38***

–1.67 –1.69 0.38 –3.65** –3.74*** –4.70***

–2.43 –2.77* 0.91

–3.21* –3.13** –3.89***

–1.42 –3.29** 1.52

TC

–1.75 –2.76* –4.48***

–1.40 –1.08 –0.58

HO

–3.23* –4.99*** –6.43***

–1.66 –1.80 0.19

ON

Notes * Statistically different from 0 at the 10% level or better; ** statistically different from 0 at the 5% level or better; *** statistically different from 0 at the 1% level or better.

–3.90** –4.13*** –4.59***

–1.50 –1.74 0.41

OT

HE

HQ

FB

CF

Non-tradable goods

Tradable goods

Source: authors’ estimation.

Constant and trend Constant None Log difference Constant and trend Constant None

Level

Table 8.5 Unit-root test: tradable and non-tradable prices (ADF performed with 12 lags)

182

F. Borraz et al.

Table 8.6 Prices co-integration Engle–Granger Cointegration Test ADF performed with 12 lags Constant and trend Health and education Transport and communications Housing Other non-tradable

–6.07*** –4.25*** –4.16** –4.85***

Source: authors’ estimation. Notes *** Statistically different from 0 at the 1% level.

&RPSYDULDWLRQ



 

3HUFDSLWDLQFRPH



Figure 8.3 Compensating variation as per cent of income by income distribution ($U): non-tradable goods effect (source: authors’ estimations).

estimate the impact of changes in prices on household wages at different points of the whole income distribution. We estimate a Heckman selection model using maximum likelihood. All regressions include year and geographic location dummies. Estimates from this model allow us to calculate the impact of the price of traded goods on labour income and the impact of changes in prices of traded goods on the labour market participation of each individual in the sample. We also take into consideration the fact that men and women’s labour market rewards may differ and we therefore separately estimate wage equations by gender. Our wage equations are limited to individuals aged 18 through 55. Figure 8.4 shows that the labour effect is pro-poor. This fact can be explained by the effect that the change of the price of traded goods has the highest impact on the wage of the low-income individuals.

Trade reforms in Uruguay 183

&RPSYDULDWLRQ





 

3HUFDSLWDLQFRPH



Figure 8.4 Compensating variation as per cent of income by income distribution ($U): labour income effect (source: authors’ estimations).

Estimation of total effect Figure 8.5 presents the estimation of the consumption and labour income effects. Trade liberalisation had a clear positive impact for both the highly paid and those with the lower positions in the salary distribution.

&RPSYDULDWLRQ



 

3HUFDSLWDLQFRPH



Figure 8.5 Compensating variation as per cent of income by income distribution ($U): total effect (source: authors’ estimations).

184

F. Borraz et al.

Poverty and inequality effects We use the wage–price elasticities estimated above to quantify the change in the head count ratio and income inequality indicators after Mercosur. In Tables 8.7a and 8.7b we observe a reduction in poverty for low educated persons located in the border and central regions of Uruguay. We do not observe differences by gender. There are no significant changes in income inequality after reform. It is interesting to note that we observe a decrease in poverty but income inequality remains constant. 8.5.2 Estimation of the effects of external trade reform First, we present results related to the price transmission. Second, we show results related to the labour market participation and labour income. Price transmission We aim to determine whether there is a permanent and long-term relationship between domestic prices (paid to producers) and global prices in the beef sector. We conducted a unit-root analysis, using the ADF test. Table 8.9 presents the ADF results for variables expressed in levels and in differences. We analysed both a model incorporating constant and trend and an alternative model without constant. Results indicate that we cannot reject the null hypothesis of existence of unitroot for the following series: the log of the price paid to the beef producer; the log of the export price in Brazil; the log of the export price in New Zealand. So, we conducted ADF test for the growth rates of the prices levels. At this time, we Table 8.7a Poverty: before and after trade reform Headcount ratio (P0), poverty gap index (P1) and squared poverty gap index (P2) Total (men + women)

Change P0

Change P1

Change P2

Total Education ≤6 years Education 7–12 years Education >12 years Montevideo Border South Central

–0.018 (**) –0.028 (***) –0.017 (**) –0.002 –0.006 (**) –0.041 (**) –0.017 (**) –0.036 (**)

–0.004 (**) –0.008 (*) –0.003 (**) –0.000 –0.001 (**) –0.001 (**) –0.003 (**) –0.007 (**)

-0.002 (**) –0.002 (**) –0.000 –0.000 –0.000 –0.003 (**) –0.001 (**) –0.002 (**)

Source: authors’ estimations. Notes * Statistically different from 0 at the 10% level or better; ** statistically different from 0 at the 5% level or better; *** statistically different from 0 at the 1% level or better. Poverty line = half of mean laboural income

Trade reforms in Uruguay 185 Table 8.7b Poverty: before and after trade reform Headcount ratio (P0), poverty gap index (P1) and squared poverty gap index (P2) Change P0

Change P1

Change P2

1 Men Total Education ≤6 years Education 7–12 years Education >12 years Montevideo Border South Central

–0.020 (**) –0.036 (**) –0.018 (**) –0.004 (**) –0.008 (**) –0.049 (**) –0.018 (**) –0.043 (**)

–0.000 (**) –0.009 (**) –0.004 (**) –0.000 –0.002 (**) –0.011 (**) –0.005 (**) –0.010 (**)

–0.002 (**) –0.004 (**) –0.001 (**) –0.001 –0.001 (**) –0.004 (**) –0.002 (**) –0.004 (**)

2 Women Total Education ≤6 years Education 7–12 years Education >12 years Montevideo Border South Central

–0.015 (**) –0.027 (**) –0.013 (**) –0.001 (**) –0.005 (**) –0.039 (**) –0.014 (**) –0.029 (**)

–0.103 (**) –0.006 (**) –0.002 (**) –0.000 –0.001 (**) –0.009 (**) –0.003 (**) –0.005 (**)

–0.001 (**) –0.001 (**) –0.001 (**) –0.000 –0.000 (**) –0.002 (**) –0.001 (**) –0.001 (**)

Source: authors’ estimations. Notes ** Statistically different from 0 at the 5% level or better. Poverty line = half of mean laboral income.

Table 8.8 Change in income inequality Before and after trade reform Gini index Total 1 Men Total Education ≤6 Education 7–12 Education >12 Montevideo Border South Central

No effect No effect No effect No effect No effect No effect No effect No effect No effect

Source: authors’ estimation.

Total 2 Women Total Education ≤6 Education 7–12 Education >12 Montevideo Border South Central

No effect No effect No effect No effect No effect No effect No effect No effect No effect

186

F. Borraz et al.

Table 8.9 Unit-root test: ADF

Domestic price paid to producer Export prices Export prices in Brazil Export prices from New Zealand to USA

Levels

Differences

–3.01 –1.63 –1.99 –1.6

–5.47*** –3.86*** –7.43*** –10.48***

Source: authors’ estimation. Notes *** Null hypothesis is rejected at 1% significance level. Critical values are –3.14 at 10%, –3.43 at 5% and 4.00 at 1%. The number of lags was estimated using Akaike

were able to reject the null hypothesis at the 1 per cent significance level. We conclude that series (in level) are integrated of order 1. This is to say that we are dealing with no stationary time series. So we proceeded to analyse the cointegration hypothesis between domestic and international prices. We estimated cointegration for three relationships: domestic prices and export prices in Uruguay; domestic prices and export prices in Brazil; domestic prices and export prices in New Zealand; in Table 8.10 we present results for the cointegration test of (real) domestic prices and the (real) export prices in Uruguay. We find that both prices are cointegrated at 1 per cent significance level. We conclude that both international prices and domestic prices move together. Although the transmission is not perfect, β1 from equation (8.1) is 0.76, the relationship is statistically significant at 1 per cent level. We also analysed the short-term price dynamics. We find that adjustment to the long-term equilibrium price level takes four years. We note that after three months that the external shock has appeared, only one-third of the total impact has occurred; only two-thirds of the total impact takes place after one year from the shock. We conclude that price adjustment occurs, but definitely at a pretty low pace. Selection models estimation We used Heckman models for estimating wages for both men and women (and for the entire sample). Obtained results have the expected signs and are statistically significant. Interestingly, results suggest that impact of global beef prices over wage levels have opposite signed for the cases of men and women. Export prices negatively affect female average wage levels while they positively impact wage levels among men (the impact over male wage is marginally more important). In both cases results are statistically significant. As we pay attention to the selection variable, we find that the impact is positive both for men and women. Again, the marginal effect is stronger among men. Total effect is also affected by interactions of different variables and beef prices.

Trade reforms in Uruguay 187 Table 8.10 Engle-Granger: cointegration test Dependent variable: average domestic price paid to producer Variable Constant Export prices in Uruguay ADF test on residuals Critical value at 1% level is 4 Error Correction Model Constant Short-term adjustment Long-term adjustment

Coefficient

Std. Error

Ratio-t

Prob.

0.35 0.76

0.10 0.04

3.45 21.35 –6.10

0.0007 0 0.0000

0.00 0.22 –0.06

0.00 0.06 0.04

0.89 3.91 –1.69

0.38 0.00 0.09

Speed of adjustment 3 months 0.32 6 months 0.40 1 year 0.51 1 and a half years 0.59 2 years 0.65 3 years 0.71 4 years 0.75 Source: authors’ estimation.

Global price variations: simulations In this chapter, we consider three scenarios suggested by Olarreaga (2006). • • •

Scenario 1: Doha Agreement (favourable), price increase of 3.9 per cent. Scenario 2: Free Trade Agreement with Europe, price increase of 5.6 per cent. Scenario 3: Free Trade Agreement with USA, price increase of 7.6 per cent.

We analyse the impact of an increase in the price of beef under the above mentioned scenarios. In general terms, we find that only in Scenarios 2 and 3 there are statistically significant effects. For example, under Scenario 3 (Free Trade Agreement with USA) the impact of beef price changes over men and women are qualitatively different. In the case of men, there is a positive impact in labour income across education levels and industry. In particular, those who are highly educated and work in the agro-industrial sector are highly benefited in this scenario. In the case of women, there is a positive effect for labour income for those working in the agricultural sector and related industries. Still, that effect is clearly lower than in the case of men. For women working in other industries, Scenario 3 implies a marked negative impact on wage levels. We conducted simulations in order to evaluate the impact of variations in prices over poverty and inequality levels. We found that none of the considered scenarios imply a clear impact on poverty levels (Table 8.11).4 Instead, we observe changes on income inequality indicated by changes in the Gini and Theil

188

F. Borraz et al.

Table 8.11 Poverty and inequality effects of liberalisation in the beef international trade Alternative scenarios Base scenario

Scenario 1 3.9%

Scenario 2 5.6%

Scenario 3 7.6%

Inequality Gini men Theil men Gini women Theil women

0.4052 0.2662 0.3513 0.2004

0.402 0.262 0.355 0.204

0.401* 0.260* 0.3565* 0.206*

0.399* 0.258* 0.359* 0.209*

Poverty Poverty ratio (P0) Poverty gap (P1) Poverty gap – squared (P2)

0.296 0.116 0.060

0.296 0.117 0.061

0.296 0.117 0.061

0.297 0.117 0.061

Source: authors’ estimation. Note * Indicates that result is statistically different from the base scenario at 5% significance level.

coefficients under Scenarios 2 and 3. In those cases, income becomes more evenly distributed among men and more concentrated in the case of women. Table 8.12 presents the change in the probability of being employed under Scenario 3 (Free Trade Agreement with USA). In general, we observe that variations are minor (less than 1 per cent). For the case of men employed in the Table 8.12 Change in the probability of being employed after a Free Trade Agreement with USA By activity sector and educational level Education Low

Medium

High

Men Industry/sector Agroindustrial Beef industry Other industry Other sector

–0.28 0.15 0.30 0.22

–0.24 0.05 0.20 0.15

–0.13 0.00 0.13 0.14

Women Industry/sector Agroindustrial Beef industry Other industry Other sector

0.29 0.47 0.61 –0.04

0.15 0.34 0.50 –0.13

0.03 0.09 0.28 –0.24

Source: authors’ estimation.

Trade reforms in Uruguay 189 agricultural sector we see a negative change in the probability of being employed. This result is also observed for the case of women across a broad range of economic activities.

8.6 Conclusions and policy implications Although it is commonly believed that trade liberalisation results in higher GDP, little is known about its effect on poverty and inequality. As many developing countries embrace trade integration as the remedy for all diseases, it is fundamental that liberalisation could be analysed from a broad range of perspectives (GDP growth, employment, poverty, inequality, etc.). In our work we focused on the poverty and inequality effects of tariff reduction in Uruguay for the 1986–2006 period. We measured the variation in income needed to compensate each household to keep the same utility after a change in the price of tradable goods. A positive change in the referred variable means that the household has improved when compared to the pre-liberalisation scenario. We analysed the impact of trade integration on households’ welfare through various transmission channels: (1) reduced tariffs affect the price of tradable goods, (2) reduced tariffs impact the prices of non-tradable goods and (3) reduced tariff cause a reallocation of productive resources and changes on labour income. As said, when interpreting results, it is important to bear in mind that while intra-zone tariffs were slashed after Mercosur was in place, extra-zone tariffs slightly decreased in the 1992–2006 period. Also, note that while tariffs for the ‘food and beverage’ category were drastically reduced in the initial Mercosur years, tariffs affecting other industrial sectors experienced a more ‘gradual’ reduction. Obtained results evidence that: (1) the decrease of tradable goods’ prices largely benefited the lower-income segment of the Uruguayan population, (2) the dynamics of the non-tradable goods’ prices had a clear pro-rich impact and (3) trade liberalisation had a clear positive impact for both the highly paid and for those with the lower positions in the salary distribution.5 Going further, one could say that the evolution of the prices of housing, health and education negatively affected the lower income population, while the decrease of the ‘food and beverages’ prices positively affected them. We think that these findings could have clear policy implications: as tariffs are reduced, the price of non-tradable goods became burdensome for the poor; if public authorities aim to develop propoor policies, then efforts should target the housing, health and education categories.6 We also analyse results at the aggregate level (when changes of the prices of tradable and non-tradable goods and labour income are considered together). Results show that average income (actually, compensating income – as defined in equation (8.1)) increased along the liberalisation process across the entire income distribution. We think that this result is important, indeed. For the case of Uruguay, talking about the income effect of trade liberalisation should not be associated with the typical ‘winners and losers’ scheme. Evidently, specific

190

F. Borraz et al.

groups obtained higher benefits than others, but we could not find any evidence about absolute losers resulting from Mercosur. In sum, the question about the impact of trade liberalisation over poverty and income can be answered with a common place: (mild) gains from trade. While not evenly distributed among the income distribution, benefits from trade spread in every Uruguayan household. With respect to external trade reform we focused on the most important Uruguayan export: the beef sector. The adjustment of local beef prices after an external shock to the worldwide price levels is imperfect. Estimations indicate that 76 per cent of a certain shock to the export prices is transmitted to the price paid to the local producers. Short-term local price dynamics show that the transmission is pretty low paced. One year after the shock, only two-thirds of the long-term effect has occurred. Price changes after trade liberalisation (under alternative scenarios) imply that men become better off (in terms of earned real wages), in particular those who are highly educated and work in the agricultural sector. For the case of women, increases in labour income after trade liberalisation are mild (or negative, in specific cases). We do not observe poverty impacts after trade liberalisation (in the three proposed scenarios). Additionally, changes in employment levels are almost immaterial. Yet, we find that there are specific inequality effects under some scenarios. In particular, for Scenarios 2 and 3, we conclude that income concentration is lower in the case of men and higher for the case of women.

Appendix: data To undertake this study we used the annual Uruguayan national household survey, Encuesta Continua de Hogares (ECH), conducted by the Instituto Nacional de Estadística (INE). We used ECH data for estimating the price–wage elasticity for the 1990–2001 period. We also used data from Encuesta Nacional de Ingresos y Gastos de los Hogares (ENIGH), the national household expenditure and income survey (we used the 1996 wave). This survey identifies the consumption structure of an average family in Uruguay. ENIGH also contains socio-economic information about Uruguayan households. This fact is crucial for us, because it allows us to identify the consumption structure of households of the same socio-economic group. We used this information in order to assess the impact of change in prices on changes in the value of the consumed basket of each household. Asociación Latinoamericana de Integración (ALADI) and Uruguay’s Ministry of Finance (MF ) provided historical information about the Mercosur common external tariffs for the period between 1986 and 2006. Secretaría del Mercosur (SM) provided data about intra-zone tariff levels (for the same time horizon). Both ALADI and SM provided raw data at a per-item desegregation level. Additionally, ALADI and the Central Bank of Uruguay (BCU) sourced our information about trade flows for the four-product categories with Mercosur and the rest of the world. We used this information in order to determine the impact of change in tariffs on prices of tradable and non-tradable goods. Information about price levels comes from the Consumer Price Index, constructed by INE.

Trade reforms in Uruguay 191

Notes 1 We thank Marcela Arnaiz for excellent research assistance. We also thank the Trade and Poverty Trust Fund of the Inter-American Development Bank for financial support. 2 Rossi used the Gini coefficient, the Theil index and the coefficient of variation to measure inequality. 3 In those years, tariff rates differentials across imported goods became significantly more uniform. 4 We calculated the poverty line by dividing the average income of the referred year by two (for each scenario). 5 This explains the U-shaped curve in Figure 8.5. 6 The negative impact for the poor through the non-tradable goods’ prices is explained by the evolution of the housing prices.

References Bucheli, M. and Rossi, M. (1994) ‘Distribución del ingreso en Uruguay: 1984–1992’, Facultad de Ciencias Sociales, Universidad de la República. Deaton, A. (1997) The analysis of household surveys: a micro econometric approach to development policy, Baltimore, MD: John Hopkins University Press. Dixit, A. and Norman, V. (1980) Theory of international trade: a dual general equilibrium approach, Cambridge: Cambridge Economic Handbooks. Giordano, P. and Florez, V. (2007) Assessing the trade and poverty nexus in Latin America, Washington, DC: Inter-American Development Bank. Gradín, C. and Rossi, M. (2000) ‘Income distribution in Uruguay: the effects of economic and institutional mimeo’, Facultad de Ciencias Sociales, Universidad de la República. Miles, D. and Rossi, M. (1999) ‘Geographic concentration and structure of wages in developing countries: the case of Uruguay’, Working Paper No. 13/99, Economics Department, Social Sciences Faculty. Olarreaga, M. (2006) ‘Estimating changes in the export price of meat received by Uruguayan exporters under different trade shocks’, The World Bank Group. Porto, G. (2006) ‘Using survey data to assess the distributional effects of trade policy’, Journal of International Economics, 70:140–160. Rossi, M. (2001) ‘Poverty in Uruguay: 1989–1997’, Facultad de Ciencias Sociales, Universidad de la República. Vigorito, A. (2007) ‘Estadísticas de Distribución del Ingreso y Pobreza’, Mimeo Instituto de Economía.

Part III

Economic liberalization, development and growth in Mexico

9

Economic liberalisation and income distribution Theory and evidence in Mexico Gerardo Angeles-Castro

9.1 Introduction In Mexico, the debt crisis of 1982 signalled the end of the import-substitution industrialisation model (ISI) and the predominance of protectionist policies. Over the subsequent years a number of structural reforms and market-oriented policies were undertaken. In 1985 the government eliminated some import licences and reduced the number of tariff categories. In 1987 the elimination of import licences was extended, the degree of tariff dispersion was reduced, and a stabilisation programme was put in place. Between 1988 and 1990 the government liberalised the financial system, reformed the FDI regime, eliminated some restrictions to portfolio investment, and opened the stock market and the money market to foreign investors; in addition, the external debt was renegotiated. The privatisation process was initiated in 1982 and was intensified during the late 1980s and early 1990s. Negotiations on the North America Free Trade Agreement (NAFTA) commenced in 1990 and it became effective in 1994. On the basis of the Stolper-Samuelson theorem (SST) we can expect that trade liberalisation in Mexico can increase demand for unskilled labour, as this is considered an abundant factor in this country. The introduction of trade reforms therefore should lead to a rise in the relative return to unskilled labour and to a narrowing of inequality. However, the empirical evidence shows that income distribution worsened in Mexico following economic liberalisation (Feliciano, 2001; Cortez, 2001; Tanski and French, 2001; Ros and Bouillon, 2002). Globalisation is sometimes presented in the relevant literature as a cause for the deterioration of income distribution in recent decades across developed countries (Smeeding, 2002). Furthermore, some empirical studies show a positive relationship between the increase in trade and income dispersion (Baldwin and Cain, 2000; Haskel and Slaughter, 2001). This trend in many developed countries is in keeping with the SST. On the other hand, several studies attribute the rise in inequality to the skill-biased technological change (SBTC) (Berman et al., 1998; Acemoglu, 2002). According to this argument, countries tend to experience a fall in relative demand for unskilled labour and an increase in that for skilled labour, due to an acceleration of technical change over the past few decades, this process is expected to exacerbate inequality. Both explanations

196

G. Angeles-Castro

(trade and technical change) dominate the literature dealing with the study of inequality in industrialised countries and have been dubbed the ‘transatlantic consensus’. Some evidence from the developing world is also consistent with the idea that trade openness can lead to more income inequality, despite the opposite SST prediction (Litwin, 1998; Flemming and Micklewright, 2000; Ros and Bouillon, 2002; Mah, 2002). An approach to explain this relationship is the idea that greater competition leads to a reduction of producer rents in the traded sector; to the extent that these rents are shared with workers, wages will decline postliberalisation. The skill enhancing trade hypothesis (SETH), based on empirical evidence, is another explanation about the expansion of the wage gap in developing countries (Robbins, 1996, O’Connor and Lunati, 1999). It takes arguments that, to some extent, can be similar to those used in the ‘transatlantic consensus’. This hypothesis claims that economic liberalisation and the intrinsic adoption of new technologies are accompanied by a relative increase in demand for skilled labour, which can worsen inequality. The relevant literature offers substantial support to both approaches (Feenstra and Hanson, 1997; Arbache et al., 2004). Thus, trade and technological change are also relevant causes of inequality in developing countries. The rise of service argument has also been advanced to explain the rise in income dispersion. It holds that globalisation fosters demand for specialised services; this process can increase income dispersion, as the service sector can be considered skill-biased in developing countries (Gordon and Gupta, 2003; Sinha, 2005). In this chapter we explore whether trade, technological change, and the rise of services can be a cause of increasing inequality in post-reform Mexico. On the other hand, there is some evidence that since the late 1990s inequality has levelled and even decreased slightly. In this respect, we also explore whether these three arguments can remain in force during the period of income distribution improvement in Mexico. The data source, the Encuesta Nacional de Ingresos y Gastos de los Hogares (ENIGH), is a household income and expenditure survey produced by the Mexican government’s statistical office, Instituto Nacional de Estadística, Geografía e Informática (INEGI). It has been carried out in 1984, 1989, and subsequently every two years since 1992 in randomly selected households. The period of analysis is from 1984 to 2002; this time frame allows us to examine the sub-period of rising inequality and the later sub-period in which inequality reverses. We find that income inequality worsens after liberalisation, mainly because of an increase in skill premium, an expansion of the income gap between the service and the agricultural sectors, and a negative relationship between market openness and wages in the traded sector. On the other hand, there is evidence that inequality decreases after 1998 and, potentially, the factors driving this trend are the decrease in returns to skill and a weaker effect of trade on wages in the traded sector. In this study we also identify two main factors that help to mitigate

Theory and evidence in Mexico 197 inequality along the period, transfer income and the re-composition of households, whereas the deterioration of the agricultural sector is a persistent source of inequality. The Mexican case is particularly interesting for the following reasons. First, this country has long been known for its unequal distribution of income.1 Second, in a few years Mexico moved from protectionism to market liberalism; moreover, it has signed a number of free trade agreements and was the first developing country to implement one (NAFTA) with two developed countries. Finally, the era of market openness in Mexico has now lasted for more than two decades. Therefore the Mexican case offers a good time span for assessing whether market-oriented policies can reduce high levels of inequality or produce a different effect. The chapter is organised as follows. Section 9.2 gives a discussion of theoretical issues supporting distributional effects under conditions of market openness and also discusses contesting arguments. Section 9.3 explores individual income distribution and wage inequality through a descriptive approach. The analysis is extended in Section 9.4 by using parametric methods. Section 9.5 explores additional forms of income distribution, at the level of households and at the level of income source. Finally concluding remarks are provided in Section 9.6.

9.2 Theoretical debate and complementary arguments 9.2.1 Standard theory Since the 1980s a number of developing countries, especially in the Meso-south American subcontinent, have adopted an economic model that places special emphasis on market forces. The set of policies involved in this development paradigm can be summarised as deregulation, privatisation, liberalisation of markets, and macroeconomic discipline. This prescription is intended to create preconditions for the expansion of trade and flow of investment across countries and finds theoretical support in familiar neoclassical theory (Jones and Barry, 1988: 30–33; Corden, 1993), which claims that trade, investment, and the market mechanism in general boost growth and facilitate development. Proponents of the model maintain that improvements in income distribution can be achieved for two main reasons. First, emphasis on outward-looking growth fosters exports, employment, and output, and thus provides additional resources for redistribution. Second, economic liberalisation facilitates the operation of market forces and the price mechanism, which allows resources to be allocated more efficiently. The basis for tracing distributional effects of market liberalism in developing countries is the SST (FitzGerald, 1996: 32; Litwin, 1998: 3). Within this twofactor (capital and labour) neoclassical model, liberalisation of foreign trade increases demand for the abundant factor, as exports and imports adjust according to the orthodox principle of comparative advantages, and redirects demand away from the scarce factor. This mechanism increases the return of the factor

198

G. Angeles-Castro

which is relatively most used in the export sector and which is also more abundant – this factor is conventionally assumed to be low-wage, unskilled labour in developing countries – and leads towards factor price equalisation; by the same token income distribution improves. 9.2.2 Contesting arguments The skill-enhancing trade hypothesis According to this proposition, increasing openness in developing countries can accelerate inflow of foreign technology due to a rise in imports and FDI. Robbins (1996) finds that the skill gap tends to widen in a sample of developing countries and shows that there is a high correlation between increasing demand for skill and imports of technology. He calls this effect ‘skill-enhancing trade hypothesis (SETH)’ and argues that trade liberalisation may sometimes widen wage dispersion instead of compressing relative wages, as more openness permits or encourages the acceleration of imported physical capital stock. In this sense, Arbache et al. (2004: 76–77) argue that the new inflowing technology can be skill biased because it is designed through relatively skillintensive methods in more industrialised countries and because its implementation and operation involves new procedures and techniques. As a result, technological change can increase demand for skilled workers. Moreover, they point out that the reduction in demand for skilled labour predicted by orthodox theory can be surpassed by this process depending on the magnitude of the shift. New technology is not only considered skill biased in developing countries but also in developed economies (Berman et al., 1998). The rise of services An alternative argument undermining basic predictions of neoclassical theory is the idea that the service sector, which has traditionally been considered a sector with higher wages than some of the conventional economic activities in developing countries, is likely to expand faster than other sectors, under conditions of economic liberalisation, and can also be considered skill-biased. This is because the globalisation and internationalisation of the economy brings with it increasing demand for financial, communication, IT, transport, and business services among others. These activities clearly require workers relatively more qualified, on average, than workers in some of traditional economic activities in developing countries, such as primary production and labour-intensive manufacturing. Gordon and Gupta (2003) show that factors playing an important role in accelerating services growth are high-income elasticity of demand, increased input usage of services by other sectors, and deregulation and economic reforms. Sinha (2005) shows that although employment in the service sector in India during the 1990s remained steady, its share of GDP rose substantially. She also stresses that the pattern and composition of growth acceleration of services

Theory and evidence in Mexico 199 creates further inequality between rural and urban areas, and between the skilled and the unskilled. Reduction of rents in the traded sector Arbache et al. (2004) hold that the reduction or elimination of trade barriers and tariffs turns protected markets into more contestable ones, which induces lower prices and therefore a reduction of producer rents; if rents are shared with employees it is expected that wages fall after liberalisation. They show that contrary to the predictions of the SST, wages fell substantially in the traded sector after trade liberalisation in Brazil, consistent with the reduced rents argument, as industries faced greater competition. Temporary adverse effects Some authors have analysed the view that when a country begins to adjust to a more competitive environment serious dislocations are encountered as the economy adapts to the shifting patterns of employment and resources. As a consequence, income dispersion may widen and absolute poverty increases in the short run. However, this effect is considered to be temporary because as the period of adjustment continues, individuals adapt and markets react, boosting investment in the traded sector and generating greater employment. In addition, growth provides additional resources for poverty alleviation and redistribution. Eventually, there may be a decrease in unemployment and income gap, and inequality begins to decrease in the long run (Jacobsen and Giles, 1998: 419–420; FitzGerald, 1996: 32). Even leading globalisers have stressed that in the short run, structural reforms may increase unemployment and worsen inequality; however, as long as these policies are consistent with sustainable economic growth they can reduce poverty and improve equity over the longer term (Camdessus, 1998: xiv–xv). From a more theoretical approach Pissarides (1997) illustrates that the rising gap between skilled and unskilled wages observed in developing countries, that have adopted market-oriented policies, can be explained by increasing trade that acts as a means for the transfer of technology across countries. The difference to the skill-enhancing hypothesis is that either the new technology or the importation and assimilation process can be skill biased and give a temporary and relative advantage to skilled labour that leads to higher relative wages only during the period of transition towards a higher level of technology. He also argues that the response of relative supply of skilled and unskilled labour to trade openness can also explain a temporary increase of wage differentials. In addition, Goldin and Katz (1998) hold that within firms, demand for skill rises when new technologies are introduced, but it declines once the other workers learn to use the new equipment; hence, this process can follow a technological cycle.

200

G. Angeles-Castro

9.3 Individual income distribution The data source is ENIGH by INEGI, as outlined earlier. We use data from six periods 1984, 1989, 1994, 1998, 2002, and 2006. Initially, the selected sample comprises individuals reporting monetary income, aged between 16 and 65 inclusive, and there are no restrictions for the number of hours employed in the corresponding economic activity. In a first stage the analysis involves the main source of income only and is not restricted to labour earnings. In other words, individuals whose main source of income comes from entrepreneurial and financial activities, property rents, and transfers are also included in the sample. The hourly income is computed as monthly income in the respondent’s main economic activity, divided by weekly hours employed in the corresponding economic activity multiplied by 4.33. The hourly income is deflated by the consumer price index based in 2002 pesos in order to obtain real hourly income. Between 1984 and 1998 the Gini coefficient for individuals increased from 0.512 to 0.632 and then dropped to 0.548 between 1998 and 2006. To understand the forces driving this pattern we present a decomposition of income by educational levels, economic sectors, and deciles, and conduct a comparative analysis between the period of increasing inequality and the latter one. In addition, a parametric analysis is conducted using labour income data. 9.3.1 Returns to education This section analyses whether skill premium is likely to increase after economic liberalisation and also explores whether returns to skill can decrease over a longer period. We present the average real hourly income for three different levels of education; primary, secondary, and tertiary.2 From panel 1 in Table 9.1 we observe that the average hourly income for both primary and secondary education tends to decline along the whole period. As for tertiary education, this indicator has a substantial increase between 1984 and 1994 but then shows a sharp fall over the last two periods. The average hourly income for the three educational levels has a slight recovery in 2006. It is worth noting that the percentage change between 1984 and 2006 is negative for the three educational levels but it decreases less in the tertiary level, as illustrated in panel 2. In order to explore how these changes on average income of educational levels have affected income dispersion between skilled and unskilled individuals, Table 9.1 presents the ratio of average hourly income in panel 3. We observe that marginal returns to tertiary education in relation to primary and secondary levels increased between 1984 and 1998 and declined between 1998 and 2006, but remained above their original levels. On the other hand, although returns to secondary education fluctuated, they actually decreased slightly in relation to the first period. Therefore, only income premium paid to high-skilled individuals has expanded, although there is evidence that this trend has reversed over the last few years. Table 9.1 also reports share of the three education categories under two different considerations. Panel 4 presents labour share weighted by hours3 and

Theory and evidence in Mexico 201 Table 9.1 Average real hourly income (2002 pesos) per level of education

(1) Income Primary Secondary Tertiary

1984

1989

1994

1998

2002

23.12 39.16 63.04

22.91 32.65 61.92

20.77 34.97 73.01

16.82 29.57 66.92

16.76 27.08 59.09

(2) Income variation 2006 vs 1984 Primary Secondary Tertiary

2006 17.03 27.43 59.68 –26.35% –29.95% –5.33%

(3) Ratio of income Primary * 1.69 Secondary ** 1.61 Tertiary *** 2.73

1.43 1.90 2.70

1.68 2.09 3.52

1.76 2.26 3.98

1.62 2.18 3.53

1.61 2.18 3.50

(4) Share of hours Primary 82.38 Secondary 10.23 Tertiary 7.39

73.07 15.01 11.92

75.49 14.31 10.19

72.99 16.42 10.6

69.04 17.43 13.53

67.31 18.44 14.25

(5) Share of income bill Primary 68.21 Secondary 14.71 Tertiary 17.08

55.29 17.38 27.33

51.31 17.7 30.99

49.75 18.54 31.71

47.19 19.54 33.27

45.87 20.11 34.02

Source: author’s computation with information from INEGI (various years). Notes * Comparison between secondary and primary education; ** comparison between tertiary and secondary education; *** comparison between tertiary and primary education.

panel 5 displays income bill share of individuals. In both panels we observe that the share of tertiary education increased along the whole period, the share of secondary education also increased, but the variation is more moderate; in contrast, the share of primary education fell gradually. Table 9.2 makes comparisons between the rising inequality period and the later period, by exploring annualised changes in the labour share weighted by hours, income bill share of individuals, and in the average hourly income for both primary and tertiary educational categories. Adopting Autor et al. (1998) assumptions,4 and following Airola and Juhn (2005), we interpret changes in income bill share as relative demand shifts. As for the first period, the simultaneous increase in relative income and relative supply of individuals with tertiary education suggests that demand for highly educated individuals also increased. In fact, labour share increased at an annual pace of 2.57 per cent, whereas income bill share expanded at an annual rate of 4.38 per cent. Although highly skilled labour supply increased over the first period, its expansion was not enough to meet the larger increase of demand;5 this fact explains the rise in income of this educational category.6 On the other hand,

202

G. Angeles-Castro

Table 9.2 Changes in average hourly income and share of educational levels (annualised log change × 100) 1984–1998

1998–2006

Primary Average income Hours Income bill

–2.26 –0.86 –2.25

–0.08 –1.43 –1.34

Tertiary Average income Hours Income bill

0.43 2.57 4.38

–3.06 6.02 1.24

Source: author’s computation with information from INEGI (various years).

changes in labour share and income bill share of the least educated group show that relative unskilled labour demand decreased faster than relative supply. Not surprisingly, average income of this category fell at an annual pace of 2.26 per cent. During the second period, income bill share of the most educated individuals continued to increase but slowed to the rate of 1.24 per cent per year, and labour share continued its expansion and even accelerated to the pace of 6.02 per cent. Consequently, income of the tertiary education category decreased at an annual pace of approximately 3.06 per cent. Finally, changes in income bill share and labour share of the primary education category continued to decrease; however, the former slowed to the rate of 1.34 per cent and the latter accelerated to the pace of 1.43 per cent; as a result, average income of the least educated group continued to fall but at a negligible annualised rate of 0.08 per cent. 9.3.2 Decomposition of income by economic sectors Here we evaluate whether the service sector is likely to raise post-liberalisation in relation to other income sources and its evolution over further periods. Income is decomposed into three main economic sectors – agriculture, manufacturing, and services. Initially, panel 4 in Table 9.3 illustrates changes in the sectoral composition of employment by reporting labour shares weighted by hours. It can be observed that the share of agriculture declined throughout the period and the decrease is more severe between 1998 and 2006. The manufacturing sector remained more or less steady. In contrast, the share of the service sector increased and the largest variation is registered during the latest period too. Average real hourly income per sector is presented in panel 1 and the percentage change throughout the period is displayed in panel 2. We observe that individual income fell 7.95 per cent in the service sector whereas it fell 24.37 and 31.63 per cent in the manufacturing and agricultural sectors respectively. Panel 3 presents ratios of average hourly income. This indicator shows that income dispersion between the service and the other two sectors expanded

Theory and evidence in Mexico 203 Table 9.3 Average real hourly income (2002 pesos) and educational attainment per sector

(1) Income Agriculture Manufacturing Services

1984

1989

1994

1998

2002

21.78 27.98 30.57

19.59 25.91 33.57

19.46 24.69 34.05

16.75 19.46 28.62

14.61 20.94 27.9

(2) Income variation 2006 vs 1984 Agriculture Manufacturing Services (3) Ratio of income Agriculture* 1.40 Manufacturing** 1.09 Services

2006 14.89 21.16 28.14 –31.63% –24.37% –7.95%

1.71 1.30

1.75 1.38

1.71 1.47

1.91 1.33

1.89 1.33

(4) Share of hours Agriculture 22.73 Manufacturing 17.99 Services 59.28

20.49 18.29 61.21

22.11 17.18 60.71

19.77 17.68 62.55

13.82 18.01 68.17

13.17 18.26 68.57

(5) Share of income bill Agriculture 16.76 Manufacturing 18.9 Services 64.34

13.13 16.86 70.01

13.57 16.14 70.28

12.05 16.76 71.19

7.56 16.29 76.15

7.03 16.56 76.43

(6) Years of education Agriculture 2.94 Manufacturing 6.69 Services 7.13

3.54 7.61 8.30

3.52 7.2 8.00

4.41 7.58 8.36

3.85 7.83 8.94

3.93 8.14 9.30

Source: author’s computation with information from INEGI (various years). Notes * Comparison between the service sector and the agricultural sector; ** comparison between the service sector and the manufacturing sector.

between 1984 and 1998; however, during the last period the income gap with respect to the manufacturing sector fell, whereas it continued to increase in relation to the agricultural sector. Panel 5 illustrates that income bill share of the service sector rose along the whole period. It remained more or less steady in the manufacturing sector after a decrease between 1984 and 1989; in contrast, income bill share of the agricultural sectors dropped permanently. Finally, panel 6 shows that individuals in the service sector have higher educational attainment and their skill upgrading is faster than in the other two sectors. In contrast, individuals in the agricultural sector have the lowest educational attainment and their skill upgrading is the slowest. Table 9.4 separates income bill shares for individuals by sector and education category and reveals that in the service sector skill demand increased permanently, whereas demand for unskilled individuals fell in relative terms. Relative

204

G. Angeles-Castro

Table 9.4 Income bill share per sector and level of education 1984

1989

1994

1998

2002

2006

Agriculture Primary Secondary Tertiary

15.63 0.64 0.49

11.98 0.42 0.73

12.30 0.65 0.62

10.66 0.77 0.62

6.52 0.43 0.61

6.02 0.57 0.71

Manufacturing Primary Secondary Tertiary

12.82 2.28 3.79

10.00 3.07 3.79

9.47 2.90 3.78

9.20 3.00 4.57

9.16 3.53 3.61

8.63 3.37 4.17

Services Primary Secondary Tertiary

39.76 11.78 12.80

33.31 13.89 22.81

29.97 14.22 26.10

29.88 14.77 26.53

31.51 15.58 29.06

29.90 16.06 30.57

Source: author’s computation with information from INEGI (various years).

demand for unskilled labour was expected to increase in the manufacturing and agricultural sectors, but it fell gradually in the former and substantially in the latter. From these two sectors, relative skill demand seems to remain steady throughout the period. From the descriptive analysis provided above, changes in inequality can be explained as follows. Between 1984 and 1998 both employment and skill demand increased in relative terms in the service sector. In contrast, relative employment and demand for unskilled individuals did not increase in the manufacturing and agricultural sectors, as predicted by standard theory; this fact can explain the increasing income gap between the service sector and the other two sectors, and can be consistent with the rise of services argument. Following the Arbache et al. (2004) industry classification, we can consider the agricultural and the manufacturing sectors as the traded industry and the service sector as the non-traded industry; in this sense, we observe a sharp fall of income in the traded industry relative to the non-traded industry, and this pattern is in keeping with the argument supporting the reduction of rents in the traded sector. Furthermore, in this period demand for skill increased faster than supply, whereas relative demand for unskilled individuals dropped; this fact can explain the upturn in skill premium and is consistent with the SETH. Hence, the increase of relative income in the service sector, the sharp fall of income in the traded sector, especially in the agricultural sector, and the expansion of the skill premium can contribute to explain the growth of overall inequality between 1984 and 1998. However, it is important to explore whether the non-traded sector keeps a relative increase in average income, once education and other variables are controlled for. Between 1998 and 2006 the increase in overall skill demand slowed down and educational attainment, on average, increased faster, although the increase

Theory and evidence in Mexico 205 does not necessarily occur among individuals with relative low income and low educational achievement.7 As a result, skill premium declined and this fact seems to be an important reason explaining the decrease in inequality in this period. Both employment and skill demand continued to increase in relative terms in the service sector, whereas in the manufacturing sector relative employment had negligible improvements and relative demand for unskilled individuals fell slightly. Bearing this in mind, we should expect further income dispersion between these two sectors; nevertheless the income gap dropped, the most plausible reason for this drop is thus a reduction of skill premium. However, as noted before, it is important to explore whether, allowing for education and other variables, the changes of relative income between sectors persist.

9.4 Econometric analysis with disaggregate data (labour income) This section extends the preliminary analysis by conducting parametric methods. It uses labour income data and applies standard Mincerian earning functions, in which the log of real monthly wages are regressed on personal characteristics and different variables in order to analyse the effect of the skill premium, returns to labour by sector, and openness on wage dispersion. So as to explore the effects of economic liberalisation over different stages in time, the analysis follows a before–after (liberalisation) approach as in Arbache et al. (2004), and also splits the sample in different periods. A (0,1) dummy variable is created; it takes a value of one for the post-liberalisation period, which is defined as after 1984. In addition, the impact of liberalisation is explored separately for the different sectors; we also focus on returns to education and the effect of market openness pre- and post-liberalisation by applying the corresponding interactions. 9.4.1 Returns to labour by sectors Column 1 of Table 9.5a shows an OLS regression which decomposes the log wage between sectors (agriculture and manufacture vs services) and distinguishes trade regime. On average, workers in the agricultural and manufacturing sector were paid 50.80 per cent less and 6.37 per cent more respectively than those in the service sector before liberalisation.8 However, as anticipated in panels 1 and 2 of Table 9.3, in the post-liberalisation period average wages in agriculture and manufacturing fell more than those in services.9 In agriculture average wages dropped 20.18 per cent and in manufacturing they dropped 26.85, whereas they fell 13.80 per cent in services.10 Once age, gender, education attainment, and unionisation are controlled for in column 2, we observe that higher human capital and higher unionisation rates in the service sector contribute to increase average wages compared to those in the other sectors. By comparing columns 1 and 2 we notice that before liberalisation the wage gap between the agricultural and service sectors changes from –50.80 per cent to –42.61 per cent and between the manufacturing and service sectors it

206

G. Angeles-Castro

changes from 6.37 per cent to 11.02 per cent. Moreover, after liberalisation the drop of average wages in the service sector is larger and the fall of average wages in the manufacturing sector is more moderate. Nevertheless, column 2 illustrates that average wages post-liberalisation increase in relative terms in the service sector, as they fall 19.62 per cent, whereas wages in agriculture and manufacture drop 21.11 and 24.13 per cent respectively. In addition, the wage gap of agriculture widens slightly from –42.61 per cent to –43.68 per cent and the wage gap of manufacturing decreases from 11.02 per cent to 4.79 per cent, between the pre- and post-liberalisation periods.11 This result is consistent with the rise of services argument; it is also consistent with the reduction of rents in the traded sector argument, if we consider both the agricultural and the manufacturing sectors as the traded sector. We also observe that there is an inverted U-shaped age-earning profile with a peak at around 45 years, women earn 28.40 per cent less than men with similar age and education, union workers earn 31.30 per cent more than equivalent nonunion workers, and returns to education increase with higher education levels. This wage equation explains 38 per cent of total variation in earnings between workers. Using the before and after methodology the post-liberalisation period is disaggregated in five sub-periods in order to examine any differential effect of liberalisation over time; results are presented from column 3 to column 7 in Table 9.5b. The sharpest fall in wages in the immediate post-liberalisation period occurs in the manufacturing sector and the most moderate occurs in the agricultural sector. Over the subsequent periods wages continue to fall, there is some recovery in the manufacturing and service sectors between 1998 and 2002, but Table 9.5a Performance of sectors (labour income) (1)

(2)

Pre-lib. Services Agriculture Manufacture Age Age2 Female Union Secondary education Tertiary education Constant Observations R2

–0.709 0.062

Post-lib. –0.148 –0.935 –0.251

Pre-lib. –0.555 0.105 0.080 –0.001 –0.334 0.272 0.424 0.913

7.968 78,815 0.11

6.637 78,815 0.38

Source: author’s computation with information from INEGI (various years).

Post-lib. –0.218 –0.793 –0.172

1.045 6.469 17,108 0.41

6.538 16,137 0.33

–0.091 –0.536 –0.694 0.117 –0.085 0.083 –0.001 –0.345 0.263 0.503

0.770

–0.113 –0.593 –0.683 0.103 –0.059 0.083 –0.001 –0.324 0.185 0.420

Post-lib.

6.461 15,311 0.40

0.988

–0.375 –0.523 –0.877 0.109 –0.308 0.084 –0.001 –0.340 0.350 0.473

Pre-lib.

(5) 1984, 1998

Notes Results corrected for heteroskedasticity, all coefficients are significant at the 1% level.

Services Agriculture Manufacture Age Age2 Female Union Secondary education Tertiary education Constant Observations R2

Post-lib.

Pre-lib.

Pre-lib.

Post-lib.

(4) 1984, 1994

(3) 1984, 1989

Table 9.5b Performance of sectors (labour income)

Post-lib.

6.567 22,281 0.38

0.868

–0.275 –0.552 –0.962 0.099 –0.222 0.080 –0.001 –0.328 0.326 0.394

Pre-lib.

(6) 1984, 2002

Post-lib.

6.501 27,535 0.40

0.842

–0.269 –0.547 –0.963 0.106 –0.209 0.080 –0.001 –0.308 0.310 0.382

Pre-lib.

(7) 1984, 2006

208

G. Angeles-Castro

wages do not return to their pre-liberalisation levels in any of these two sectors. In the agricultural sector the fall is permanent. When education and other variables are controlled for, we observe that a source of inequality between the pre- and post-liberalisation periods is the change in the wage gap of agriculture relative to services, as it widens from –42.61 per cent to –43.68 per cent, although the variation is slight. On the other hand, the relative increase in wages in the service sector tends to equalise wages in relation to the manufacture sector, since manufacturing wages are originally higher but then the wage gap between these two sectors falls from 11.02 per cent to 4.79 per cent. Thus, the rise of services and the relative reduction of rents in the traded sector have an unequalising effect in relation to the agricultural sector only. The evolution of returns to labour by sectors does not contribute to explain the reversal of inequality between 1998 and 2000, because the wage gap relative to services continues to widen in agriculture and remains relatively stable in manufacturing. Table 9.6 focuses on the returns to education pre- and post-liberalisation. The first column illustrates the results obtained from the whole sample and the last five columns show the results obtained once the post-liberalisation period is disaggregated over time. Three main findings emerge from this analysis. First, average income tends to be lower in every level in the post-liberalisation periods and this is consistent with decreasing real wages as noted previously. Furthermore, as anticipated in panels 1 and 2 of Table 9.1, average wages for the primary and secondary levels fall relative to the tertiary level. In the postliberalisation period workers with primary and secondary education are paid 21.20 per cent and 25.82 per cent less respectively, whereas workers at the highest educational level are paid 9.22 per cent less.12 Second, the marginal returns to education – comparing each education level with those below – tend to be greater along the post-liberalisation periods only for high-skill workers or those with tertiary education, but not for those with secondary education. The point estimate of the marginal return to tertiary level rises from 119.10 per cent to 152.42 per cent and from 34.74 per cent to 64.89 per cent in relation to the primary and secondary levels respectively, between the pre- and post-liberalisation periods.13 This finding confirms the trend observed in panel 3 of Table 9.114 and is in keeping with the skill-enhancing trade hypothesis.15 Finally, the marginal returns to tertiary education peak by 1994 and then decline but remain higher than in the pre-liberalisation period. When controlling for sectors, unionisation, and personal characteristics, we observe that the evolution of skill premium post-liberalisation is a factor that has a clear effect on changes in inequality because returns to tertiary education increase after 1984 and this fact widens the income gap between skilled and unskilled labour. This trend is due to a faster increase in skill demand than supply, whereas relative unskilled demand decreased faster than supply between 1984 and 1998, as shown in Table 9.2. Hence, the evidence illustrates that during periods of economic liberalisation and its intrinsic technological change, relative

0.663

0.800 –0.585 0.038 6.523 17,108 0.41

0.786

0.494

0.083 –0.001 –0.348 0.261

0.928

0.347

–0.160

Notes Results corrected for heterosedasticity, all coefficients are significant at the 1% level.

Source: author’s computation with information from INEGI (various years).

–0.576 0.068 6.526 16,137 0.33

0.295

–0.098

0.508

0.083 –0.001 –0.325 0.184

–0.509 0.081 6.486 15,311 0.40

0.782

0.483

0.084 –0.001 –0.344 0.348

0.643

0.066

–0.407

–0.655 0.064 6.590 22,281 0.38

0.760

0.455

0.080 –0.001 –0.330 0.321

0.580

0.077

–0.308

–0.659 0.060 6.524 27,535 0.40

0.781

0.475

0.080 –0.001 –0.310 0.305

0.574

0.081

–0.298

Post-lib.

(6) 1984, 2006

Post-lib. Pre-lib.

(5) 1984, 2002

Post-lib. Pre-lib.

(4) 1984, 1998

Post-lib. Pre-lib.

(3) 1984, 1994

Post-lib. Pre-lib.

(2) 1984, 1989

Post-lib. Pre-lib.

Age 0.080 Age2 –0.001 Female –0.335 Union 0.271 Primary –0.238 education Secondary 0.486 0.188 education Tertiary 0.784 0.688 education Agriculture –0.571 Manufacture 0.052 Constant 6.394 Observations 78,815 R2 0.37

Pre-lib.

(1)

Table 9.6 Returns to education (labour income)

210

G. Angeles-Castro

demand for skill tends to increase and this pattern supports the SETH. Moreover, variations in skill premium also contribute to explain the fall in overall inequality after 1998 because skill premium tends to fall, especially after this year. The cause of this trend is also shown in Table 9.2, where we observe that the increase in skill demand slows down whereas the increase in supply accelerates, and unskilled supply falls faster than demand. Hence the rise in skill premium is temporary and cyclical. We also use the data-set comprising all income sources in the analysis of educational levels and sectors, and find that the general conclusions are similar to those using labour income only.

9.5 Additional forms of income distribution (all income sources) 9.5.1 Household inequality vs individual inequality Initially, a simple comparison between households and individuals in terms of income and Gini coefficients is presented in Table 9.7. From the first panel we observe that household Gini is lower than individual Gini and the last column reveals that the former grew slower than the latter throughout the period. Moreover, the rise of household Gini started to reverse slightly after 1994, whereas individual Gini started to drop after 1998. The second panel shows that real hourly individual income declined 12.08 per cent, whereas real monthly household income fell 0.45 per cent over the whole period. It is worth noting that household income increased 18.57 per cent when it is expressed in per capita terms. An important reason for mitigation of inequality and income fall among households is presented in the bottom panel. We observe that the average number of members per household dropped 17.41 per cent, whereas the number of income receivers increased 33.87 per cent. As a result, the proportion of income receivers per household increased, from 31.66 per cent to 51.31 per cent between 1984 and 2002. Although the upper quintiles have kept a higher proportion of income receivers over time, the lower quintiles have increased the proportion faster and therefore the percentage of income receivers tends to converge across income levels. This families’ reaction counteracts the increase in inequality and the general trend of declining real income; this in fact raises per capita household income. 9.5.2 Gini decomposition by income source Table 9.8 presents the decomposition of the household Gini coefficient by three main income sources – labour, transfers, and business and finance (B&F ) – applying the Yao (1999) method. The first panel reveals that the transfer income is the most equally distributed and its Gini has fallen markedly, as recorded in column 7. In contrast, the Gini of B&F income is the largest and has expanded

4.97 1.68 33.86 22.71 29.30 35.12 42.27 47.59

23.03 27.21 31.41 36.95 47.84

7,146 1,437 29.30

0.530 0.551

5.07 1.61 31.66

6,441 1,270 28.03

0.485 0.512

1989

Source: author’s computation with information from INEGI (various years).

(3) Household composition Household members Receivers per household Receivers per household % Receivers per quintile % 1st 2nd 3rd 4th 5th

(2) Income Monthly income per household Monthly income per member Hourly individual income

(1) Gini Household Gini Individual Gini

1984

29.82 33.93 40.13 46.78 54.98

4.72 1.86 39.52

6,928 1,469 29.02

0.553 0.587

1994

37.90 41.59 45.53 52.11 56.42

4.40 2.00 45.51

5,859 1,331 24.40

0.549 0.632

1998

46.50 44.50 47.81 54.31 59.26

4.23 2.10 49.63

6,348 1,499 24.55

0.515 0.560

2002

47.75 46.17 49.14 55.11 59.93

4.19 2.15 51.31

6,412 1,505 24.65

0.504 0.548

2006

Table 9.7 Average real monthly income (2002 pesos), Gini, and composition of income receivers per household and individuals

107.37 69.69 56.45 49.13 25.28

–17.41 33.87 62.08

–0.45 18.57 –12.08

3.90 7.02

2006 vs 1984

Change %

212

G. Angeles-Castro

sharply over time. As a result, the second and third panel illustrate that the contribution of transfer income to total income is larger and has increased faster over time (column 8) than its contribution to the overall Gini, whereas the contribution of B&F income to total income is lower and has decreased more than its contribution to the overall Gini. As for labour income, its income share has increased slightly more than its Gini share. Consequently, transfer income, which is mainly composed of remittances from emigrant workers and social government expenditure, helps to reduce household inequality for the following reasons. First, its Gini is reasonably smaller than the overall Gini and the gap has expanded over time (panel 4). Second, although the smallest income source out of the three categories, it has increased gradually. In this context and to a lesser extent, labour income helps to reduce household Gini too. On the other hand, B&F income drives inequality up as its Gini is higher than the overall household Gini and the gap has tended to increase over time (panel 4). The last panel summarises the impact of every income category on the overall household income inequality by displaying the ratio of Gini share to income share. If the ratio is greater than one, it means that the corresponding income source can increase inequality, otherwise it helps to decrease the household Gini coefficient. We observe that the contribution ratio of the B&F income is greater than one and has increased over time (column 7), which suggests that this income source is a driving force of household inequality. Labour income is relatively neutral. Finally, the contribution ratio of transfer income is the lowest and has decreased sharply along the period, which indicates that transfer income is an important factor to reduce the household Gini coefficient.

9.6 The effects of market openness on income dispersion The before–after methodology, conducted so far, compares variations in the income structure before and after trade liberalisation, but it does not disentangle the effect of market openness. In order to address this problem, we apply a second strategy in this section. The strategy includes an industry-level index of openness in the analysis; the index allows for intensity and timing in the process of liberalisation across the traded sector. However, due to the nature of the analysis we restrict the sample to the traded sector only. The industry-level index of openness is expressed as follows:

where ejt denotes the effective rate of protection in industry j at time t. There is an inverse relationship between the measure of openness and e because higher tariffs reflect more protectionism, while lower tariffs reflect a more competitive environment. Table 9.9 presents the impact of trade reforms on log real wages using the measure of openness in 17 traded industries across time. In all columns we

10.64 –1.00 –30.52

(4) Gini variation % (income source vs overall) Business and finance 6.18 17.79 Labour –2.89 –10.10 Transfer –7.46 –9.21 1.10 0.98 0.74

Source: author’s computation with information from INEGI (various years).

(5) Contribution ratio (gini/income) Business and finance 1.06 Labour 0.97 Transfer 0.93 1.18 0.90 0.91

35.37 58.52 6.10

42.38 50.97 6.64

37.94 55.00 7.07

(3) Contribution to Gini Business and finance Labour Transfer

32.20 59.53 8.28

35.98 56.70 7.32

(2) Contribution to income Business and finance 35.73 Labour 56.63 Transfer 7.64

0.612 0.547 0.384 0.553

(3) 1994

0.624 0.476 0.481 0.530

(2) 1989

0.515 0.471 0.448 0.485

(1) Gini Business and finance Labour Transfer Total

(1) 1984

Table 9.8 Decomposition of household Gini by income source

1.12 0.96 0.82

12.21 –4.22 –17.76

38.03 53.77 8.20

33.89 56.14 9.97

0.616 0.526 0.452 0.549

(4) 1998

1.17 0.96 0.74

16.78 –3.55 –26.46

34.34 57.87 7.79

29.40 60.00 10.60

0.601 0.496 0.378 0.515

(5) 2002

1.17 0.97 0.68

16.70 –6.16 –26.64

35.26 57.12 7.62

30.07 58.76 11.17

0.587 0.472 0.369 0.503

(6) 2006 (8) Diff.

10.44 0.10 –26.28

14.08 0.30 –17.72 3.79

–2.68 2.12 0.55

–5.66 2.13 3.53

2006 vs 1984 2006–1984

(7) %

Change

214

G. Angeles-Castro

Table 9.9 The effect of market openness on income (traded sector)

Open Age Age2 Female Union Agriculture Secondary education Tertiary education Constant Observations R2

(1)

(2) 1984–1998

(3) 1998–2006

–0.227 0.071 –0.001 –0.337 0.269 –0.387 0.422 0.919 6,478 17,928 0.39

–0.275 0.078 –0.001 –0.339 0.254 –0.367 0.419 0.927 6,523 9,143 0.41

–0.164 0.07 –0.001 –0.335 0.278 –0.394 0.426 0.886 6,469 8,785 0.38

Source: author’s computation with information from INEGI (various years).

control for educational levels, age, gender, and unionisation. The first column reveals that greater openness is associated with lower income. This result represents support for the reduction of rents in the traded sector argument, which stresses that more competition can reduce rents and therefore can reduce wages too. A 10 per cent increase in the index of openness, ceteris paribus, reduces income by 2.03 per cent. Columns 2 and 3 split the whole time period in two sub-periods, 1984–1998 and 1998–2006, which are the periods of rising inequality and decreasing inequality, respectively, as commented before. In the first period we observe that the effect of trade liberalisation on income is more adverse; an upturn of 10 per cent in the openness index, ceteris paribus, reduces income by 2.40 per cent; whereas in the period of decreasing inequality the reduction is smaller, 1.51 per cent. Two main conclusions emerge from these results. First, market openness reduces income in the traded sector and this pattern helps to explain the income polarisation between this sector and the service sector. Second, over the longer run the adverse effect of market openness on income tends to decline and this pattern helps to explain the reduction of income inequality after 1998. To some extent this pattern gives support to the temporary adverse effects argument, which claims that in the longer run markets react and individuals adapt to a more competitive environment, and therefore the dislocations that occur in a country during its early economic liberalisation period tend to banish in subsequent periods.

9.7 Concluding remarks Due to market-oriented reforms in Mexico since the mid-1980s, and on the basis of the SST we might expect a rise in the relative return to low-income, unskilled labour, or an increase in individual income in activities such as agriculture and

Theory and evidence in Mexico 215 labour-intensive manufacturing, and therefore a reduction of income inequality. However, in the post-liberalisation period skill premium and income differential between low- and high-income individuals expanded,16 and relative income in agriculture and manufacturing dropped. Furthermore, overall individual inequality increased, although there is some evidence that has tended to decline after 1998. These trends undermine orthodox theory and provide room for contesting arguments. The analysis, finds various factors driving inequality between 1984 and 1998. An important reason for income dispersion is the fact that marginal returns to education increased, which is consistent with the SETH. Note, however, that the hypothesis applies to tertiary education in particular. In the service sector relative income, employment, and demand for skill increased; consequently, the evidence corresponds with the rise of service argument. This pattern contributes to explain income dispersion, in the sense that the wage gap between the service and the agricultural sectors expanded. Relative income in the traded sector fell following liberalisation, and this is in keeping with the view that market-oriented reforms increased the degree of competition and therefore reduced rents. Income also dropped in the non-traded industry, indicating either a degree of spillover, or the effect of other reforms such as privatisation or deregulation.17 However, the relevant finding is that income in the traded sector fell in relative terms, which is another reason of income dispersion. The evidence also corresponds with the decline of labour market institutions argument to the extent that average wages, union density, and union premium fell. However, changes in the wage gap between union and non-union workers cannot contribute to explain an increase in income dispersion, as the gap decreased in average in the post-liberalisation period. Only around 1998 union premium was higher than its position pre-liberalisation. Nevertheless, the fact that a large number of workers moved away from unions and entered a nonunion sector, characterised by diverse and flexible wages and higher Gini coefficient, represents a source of inequality. The rise in income Gini coefficient reversed between 1998 and 2002 and so did the income gap between upper and lower deciles. The factors that can explain this variation are summarised as follows. The upturn in skill premium started to reverse around 1994 and the downturn was faster around 1998. In addition, by 1998 the wage gap between union and non-union workers had peaked and fell afterwards and the fall in the unionisation rate stopped and reversed slightly. In this respect, some authors have stressed the possibility that income distribution can follow cycles under conditions of market openness and technological change. One of these approaches explains that when a country begins to adjust to a more competitive environment serious dislocations are encountered as the economy adapts to the shifting patterns of employment and resources. As a consequence, income dispersion may widen and absolute poverty increases in the short run. However, this effect is considered to be temporary because as the period of adjustment continues markets stabilise and individuals adapt to the prevailing conditions. Eventually, there may be a decrease in unemployment and

216

G. Angeles-Castro

income gap, and inequality may begin to decrease in the longer run (Jacobsen and Giles, 1998: 419–420, FitzGerald, 1996: 32). In keeping with this approach, evidence in the Mexican case shows that over the longer term, individuals react by achieving higher educational attainment or increasing movements towards higher income activities.18 In addition, transition and adjustment in labour unions seem to come to an end, or at least changes are less marked. Finally we observe that individuals tend to increase the number of income receivers and to reduce the number of members in their households, which leads to higher per capita income, especially in low-income sectors. In terms of technological change Pissarides (1997) shows that in developing countries, that have adopted market-oriented policies, the importation and assimilation process of new technology can be skill biased and give a temporary and relative advantage to skilled labour only during the period of transition towards a higher level of technology. He also argues that the response of relative supply of skilled and unskilled labour to trade openness can also explain a temporary increase of wage differentials. In addition, Goldin and Katz (1998) hold that within firms, demand for skill rises when new technologies are introduced, but it declines once the other workers have learned to use the new equipment. Around 1998 the evidence starts to correspond with these ideas since we observe higher levels of educational achievement and an acceleration of skill supply in relation to previous periods, whereas skill demand falls substantially. Although we have found factors that can contribute to lessen inequality in the longer term, there are adverse effects lasting the whole period of study, for instance the deterioration of the agricultural sector. We also found two main factors that can contribute to mitigate adverse effects; they are the recomposition of households and transfer income. The study identifies reactions of individuals that can help to reduce inequality. However, the results suggest that solutions for income inequality can also rely on government action. Some of the main policies implied are to increase expenditure in the form of transfers, to take strategic action to develop the agricultural sector, and to facilitate access to education, especially to the vulnerable and those at low-income levels. Furthermore, the boost of employment in unskilled, labour-intensive activities, combined with the reduction of supply of unskilled individuals by increasing educational levels can encourage factor price equalisation. However, heavy reliance on low-wage employment is not a desirable long-term solution as it does not encourage domestic markets and sustained growth; in this context, gradual and strategic industrialisation can be a complementary strategy. Finally, income redistribution can be encouraged by introducing a progressive taxation policy at the highest income levels.

Notes 1 From a sample of 49 countries including different definitions of Gini coefficients over time, Li et al. (1998) show that the Mexican average Gini, 54.59, is the second highest of the sample. 2 The classification is conducted under the following criteria. Primary level comprises

Theory and evidence in Mexico 217

3 4 5

6

7

8 9

10 11

12 13 14 15

individuals with some elementary or completed elementary education. In Mexico, the first nine years of the educational system are considered elementary education. Secondary level includes individuals with some education after the basic level but with no university education. The tertiary level comprises individuals with university education, completed or incomplete, also includes individuals with one or more years of postgraduate education. We also calculated the educational distribution weighted by individuals, and it was found that it does not differ substantially from that weighted by hours. Autor et al. (1998) show that under the assumption that the elasticity of substitution between skilled and unskilled workers equals one and the production function is CobbDouglas, relative demand shifts can be represented by changes in wage bill share. By conducting an international comparison Cortez (2001) shows that Mexico’s performance in terms of educational expansion is poor, as the reduction in the percentage of low-education workers is slow and the increase in the percentage of workers with higher education is small compared to countries like South Korea and Sweden. Although the average hourly income of the highly educated group rose between 1984 and 1998, the annualised change was relatively low, 0.43 per cent; however, between 1984 and 1994 the rate was higher, 1.47. This can be explained because supply of skilled individuals fell between 1989 and 1994 and then continued to increase in the later periods. Nevertheless, the analysis considers periods before and after 1998, because the rest of the changes in the trend of supply and demand of skilled and unskilled individuals, and changes in individual income Gini, appear to be stronger around this year. We also decompose income by deciles and find that the lower the income level of individuals the fewer the years of schooling received in every period. Furthermore, throughout the period the speed of skill upgrading is the lowest in the first quintile whereas it is the highest in the fifth quintile. We also find that between 1998 and 2006 human capital increased faster in the last fourth quintiles compared to the previous periods, only the first quintile showed a decrease. Information computed by deciles is available upon request. Figures computed as (exp(β) – 1)*100 where β is the coefficient on the corresponding sector dummy variable. Although Table 9.2 comprises all sources of income, in 2002 labour income accounted for 60 per cent of total income; hence, this table can be representative of the pattern followed by this income source over time. Moreover, we construct Table 9.2 using labour income only and also observe that relative wages in agriculture and manufacturing fell more than in services. The change in average wages after liberalisation is computed as (exp(b1 – b2) – 1)*100, where b1 and b2 are the coefficients of the corresponding sector post- and preliberalisation respectively. In the post-liberalisation period the wage gap of agriculture and manufacturing in relation to services is computed as (exp(b1 – b2) – 1)*100, where b1 is the coefficient of the corresponding sector and b2 is the coefficient of the service sector postliberalisation. Fall post-liberalisation is computed as (exp(b1 – b2) – 1)*100, where b1 and b2 are the coefficients of the corresponding educational level post- and pre-liberalisation respectively. Marginal returns to education comparing two levels of education can be obtained as (Exp(bupper – blower) – 1)*100, where bupper and blower are the coefficients on the education level dummy variable for the upper and lower level respectively for a specific period. Although Table 9.1 is constructed from all income sources it shows a good approximation of the trend in labour income as this income source represents 60 per cent of total income, as noted in Note 10. Note that the hypothesis applies to tertiary education in particular. This finding is

218

G. Angeles-Castro

similar to that obtained by Arbache et al. (2004) for the case of Brazil, as they conclude that the SETH applies to college-educated labour only. 16 By decomposing the overall income by deciles, it is found that the bottom nine deciles lost income share and decreased average real hourly income between 1984 and 1998; furthermore, the lower the income level, the higher is the loss. In contrast, the top decile gained income share and increased average income in this period. Consequently, the ratios of the tenth decile to the first decile, in both indicators, increased until 1998, and they actually doubled, as they passed from 32 to 64. Information computed by deciles is available upon request. 17 Arbache et al. (2004) reached similar conclusions for the case of Brazil. 18 By decomposing income by quintiles and economic sectors we observe that between 1998 and 2002 the employment share in services and manufacturing increased, but the former had the highest increase in the first and second quintile and the latter in the first quintile.

References Acemoglu, Daron (2002) Technical Change, Inequality and the Labour Market. Journal of Economic Literature, 40(1), pp. 7–72. Airola, Jim and Juhn, Chinhui (2005) Wage Inequality in Post-Reform Mexico. Discussion paper series, No. 1525, IZA – Institute for the Study of Labour. Arbache, Jorge S., Dickerson, Andy, and Green, Francis (2004) Trade Liberalisation and Wages in Developing Countries. Economic Journal, 114(493), pp. 73–96. Autor, David H., Katz, Lawrence F., and Krueger, Alan B. (1998) Computing Inequality: Have Computers Changed the Labour Market? Quarterly Journal of Economics, 113(4), pp. 1169–1213. Baldwin, R.E. and Cain, G.G. (2000) Shifts in US Relative Wages: The Role of Trade, Technology, and Factor Endowments. Review of Economics and Statistics, 82(4), pp. 580–595. Berman, Eli, Bound, John, and Machin, Stephen (1998) Implications of Skill Biased Technological Change: International Evidence. Quarterly Journal of Economics, 113(4), pp. 1245–1279. Camdessus, Michel (1998) ‘Income Distribution and Sustainable Growth: The Perspective from the IMF at Fifty’, in: Vito Tanzi and Ke-Young Chu (eds) Income Distribution and High-Quality Growth (Cambridge, MA: MIT Press), pp. xi–4. Corden, W.M. (1993) Protection and Liberalisation: a Review of the Analytical issues. Occasional paper, No. 54, International Monetary Fund. Cortez, Willy W. (2001) What is Behind Increasing Wage Inequality in Mexico? World Development, 29(11), pp. 1905–1922. Feliciano, Zadia M. (2001) Workers and Trade Liberalisation: The Impact of Trade Reforms in Mexico on Wages and Employment. Industrial and Labour Relations Review, 55(1), pp. 95–115. Feenstra, R.C. and Hanson, G.H. (1997) Foreign Direct Investment and Relative Wages: Evidence from Mexico’s Maquiladoras. Journal of International Economics, 42, pp. 371–394. FitzGerald, E.V.K. (1996) ‘The New Trade Regime: Macroeconomic Behaviour and Income Distribution in Latin America’, in: Victor Bulmer-Thomas (ed.) The New Model in Latin America and its Impact on Income Distribution and Poverty (London: Macmillan Press), pp. 29–52. Flemming, J.S. and Micklewright, John (2000) ‘Income Distribution, Economic Systems

Theory and evidence in Mexico 219 and Transition’, in: Anthony B. Atkinson and Francois Bourguignon (eds) Handbook of Income Distribution, Vol. 1 (Amsterdam: Elsevier), pp. 843–918. Goldin, Claudia and Katz, Lawrence F. (1998) The Origins of Technology-Skills Complementarity. Quarterly Journal of Economics, 113(3), pp. 693–732. Gordon, Jim and Gupta, Poonam (2003) Understanding Indian Service Revolution, Paper prepared for the IMF-NCAR conference, International Monetary Fund, November 2003. Online, available at: www.imf.org/external/np/apd/seminars/2003/newdelhi/ gordon.pdf (accessed August 2008). Haskel, J. and Slaughter, M.J. (2001) Trade, Technology and UK Wage Inequality. Economic Journal, 111(468), pp. 163–187. INEGI, Instituto Nacional de Estadística Geografía e Informática (1985, 1990, 1995, 1999, 2003) ENIGH, Encuesta Nacional de Ingreso y Gasto de los Hogares, 1984, 1989, 1994, 1998, 2002, CD-Rom, Mexico, DF. Jacobsen, Peter W.F. and Giles, David E.A. (1998) Income Distribution in the United Sates: Kuznets’ Inverted-U Hypothesis and Data non-Stationary. Journal of International Trade and Economic Development, 7(4), pp. 405–423. Jones, R. and Barry, J. (1988) ‘Liberal Political Economy’, in: R. Jones (ed.) The Worlds of Political Economy (London: Pinter Publishers), pp. 27–56. Li, Hongyi, Squire, Lyn, and Zou, Heng-Fu (1998) Explaining International and Intertemporal Variations in Income Inequality. Economic Journal, 108(446), pp. 26–43. Litwin, Carol (1998) Trade and Income Distribution in Developing Countries. Working papers in economics, No. 9, Department of Economics, Göteborg University. Mah, Jai S. (2002) The Impact of Globalization in Income Distribution: The Korean Experience. Applied Economic Letters, 9(15), pp. 1007–1009. O’Connor, D. and Lunati, M.R. (1999) Economic Opening and Demand for Skills in Developing Countries: A Review of Theory and Evidence. Technical Papers, No. 149, OECD Development Centre. Pissarides, Christopher A. (1997) Learning by Trading and Returns to Human Capital in Developing Countries. World Bank Economic Review, 11(1), pp. 17–32. Robbins, Donald. J. (1996) HOS Hits Facts: Facts Win; Evidence on Trade and Wages in the Developing World. Discussion paper series, No. 557, Harvard Institute for Development, Harvard University. Ros, Jaime and Bouillon, César (2002) ‘Mexico: Trade Liberalization, Growth, Inequality and Poverty’, in: Rob Vos, Lance Taylor, and Ricardo Paes de Barros (eds) Economic Liberalization, Distribution and Poverty: Latin America in the 1990s (Cheltenham: Edward Elgar Publishing, Inc.), pp. 347–389. Sinha, Aseema (2005) Globalization, Rising Inequality and New Insecurities in India, Paper presented at the conference on Difference on Inequality in Developing Societies, University of Virginia, April 2005. Online, available at: www.apsanet.org/imgtest/ TaskForceDiffIneqDevSinha.pdf (accessed August 2008). Smeeding, Timothy M. (2002) Globalisation, Inequality, and the Rich Countries of the G-20: Evidence from the Luxembourg Income Study (LIS). Working Paper, No 48, Syracuse University, Center for Policy Research. Tanski, Janet M. and French, Dan W. (2001) Capital Concentration and Market Power in Mexico’s Manufacturing Industry: Has Trade Liberalization Made a Difference? Journal of Economic Issues, 35(3), pp. 675. Yao, Shujie (1999) On the Decomposition of Gini Coefficients by Population Class and Income Source: a Spread Sheet Approach and Application. Applied Economics, 31(10), pp. 1249–1264.

10 How risk factors affect growth in Mexico A free-market liberalism approach Francisco Venegas-Martínez

10.1 Introduction Nothing should matter more to a country and its inhabitants than the behavior of its rate of economic growth in the long run. In this regard, a large number of investigations have been produced in the specialized literature for the last two decades. Hundreds, or perhaps thousands, of theoretical and empirical studies on many countries have highlighted the correlation between economic growth and its determinants. However, it remains to explain the correlation between growth and risk factors. The goal of this chapter is to stress the connection between economic growth and the currency, market, debt, and fiscal risks. Economic growth is quantified as the increase in the amount of the goods and services produced by a country in a given period of time. It is, conventionally, calculated as the percent rate of increase in real gross domestic product (GDP). An increase in GDP of a country is generally associated with an increase in the standard of living of (all) its inhabitants. Therefore, the issue of setting up the determinants of growth and how risk factors impact growth are two questions of great interest to policy makers. A market liberal revolution is reaching most of the countries in the world, and Mexico is not the exception. The notion of free-market liberalism emphasizes the support of free-market economies with personal freedom and human rights. The concept of market liberalism can be used as a synonym to economic liberalism when both the economic aspects of the classical liberalism and the individual aspects of freedom are relevant; the market-liberal order is also ethical, in the sense that it is based on personal freedom under the principles of law. One of the objectives of this chapter is to develop, under a free-market liberalism framework, a stochastic model of endogenous growth in which the inhabitants of a country have expectations of depreciation of the exchange rate driven by a diffusion process combined with Poisson jumps; other investigations on stochastic models of growth can be found in Canton (2001) and Gokan (2002). It is supposed, in our modeling, that the economy has no contingentclaims markets to hedge against future exchange-rate depreciation. It is worthwhile to mention that a study where exchange-rate derivatives are available can be found in Venegas-Martínez (2005a). Other studies for the Mexican case

How risk factors affect growth in Mexico 221 within a stochastic framework can be found in Venegas-Martínez (2004, 2005b, 2006, 2008). It is supposed, in the proposed model, that an uncertain tax rate on both wealth and taxes are governed by a geometric Brownian motion. Assuming risk adverse agents, we examine the growth rates of consumption and output in the presence of taxes on wealth and consumption. The production function has constant return to capital (Rebelo, 1991); we combine this technology with the optimizing behavior of consumers and firms to obtain the stochastic per capita growth rates of consumption, capital, and output. This research provides a meanvariance description of the growth rate of output, showing explicitly the underlying risk factors that affect growth. Moreover, the proposed model is used to carry out a simulation experiment that reproduces the observed mean and variance of the growth rate of output in Mexico from 1930 to 2002.

10.2 The setting of the economy In this section, we establish the main characteristics of both the economy and its inhabitants. Let us consider a small open economy with identical (in preferences and endowments) and infinite-lived households, with personal freedom and human rights, in a world with a single, internationally tradable and perishable consumption good. The individuals are both consumer and producers and share out a technology showing a constant marginal product. In what follows, we assume that the generic good is freely traded, and its domestic price level, pt, is determined by the purchasing power parity condition, namely (10.1) where st is the foreign-currency price of the good in the rest of world, and et is the nominal exchange rate. We will assume, for the sake of simplicity, that st = 1. We suppose that the number of atypical movements in the exchange rate, that is, the jumps in the exchange rate, per unit of time, follows a Poisson process Qt with intensity h, so P(N){dQt = 1} = hdt and P(N){dQt = 0} = 1 – hdt + o(dt). In such a case, (10.2) Let us consider now a Brownian motion, dVt, that is, dVt ~ N(0, dt) E[dVt] = 0 and Var[dVt] = dt. We assume that the consumer perceives that the expected inflation rate, dpt/pt, and consequently the expected rate of depreciation, det/et, follows a geometric Brownian motion with Poisson jumps in accordance with (10.3)

222

F. Venegas-Martínez

where e is the mean expected rate of depreciation conditional on no jumps, sP is the instantaneous volatility of the expected price level, and m is the mean expected size of an exchange-rate jump. Process Vt is supposed to be independent of Qt. In what follows, e, sP, h, and m are all supposed to be positive constants. The agent holds real cash balances, mt = Mt/pt, where Mt is the nominal stock of money. The stochastic rate of return of holding real cash balances, dRm, is given by the percentage change in the price of money in terms of goods, dmt/mt. By applying Itô’s lemma for diffusion-jump processes to the inverse of the price level, with (10.3) as the underlying process, it can be shown that (10.4) The agent also holds capital, kt, that pays a risk-free real interest rate r, which is constant for all terms, satisfying dkt = rktdt, where k0 is given. The representative agent takes r as given. Let us consider now a Brownian motion dWt, that is, (10.5) We assume that the representative consumer perceives that his/her wealth is taxed at an uncertain rate, tt, in accordance with the following stochastic equation: (10.6) where t0 > 0 and (10.7) _

_

with κ ∈(–1, 1). Here, τ is the mean expected growth_ rate of the taxes on wealth, sP is the volatility of the tax rate on wealth, and κ is the correlation between changes in inflation and changes in wealth taxes. Notice that an increase in the rate of depreciation will produce a higher depreciation in real cash balances. This, in turn, will reduce real assets, which could lead to the fiscal authority to modify tax rates. Processes Qt, Vt, and Wt are supposed to be pairwise independent. Consider a cash-in-advance constraint of the form: (10.8) where ct is consumption, and a > 0 is the average time that money must be held to finance consumption. Condition (10.9) is critical in linking exchange-rate dynamics with consumption. Finally, observe that

How risk factors affect growth in Mexico 223 (10.9) In the sequel, we will assume that the error o(a) is negligible.

10.3 The consumer’s decision problem In this section, we characterize the optimal decisions of a representative agent on consumption and portfolio shares through the Hamilton–Jacobi–Bellman condition (necessary condition for a maximum) of the continuous-time stochastic dynamic programming. The stochastic consumer’s wealth accumulation in terms of the portfolio shares,

and consumption, ct, is given by the following stochastic system:

(10.10) where dRk = dkt/kt is the return of capital, and τ̂ is a resident-based ad valorem tax rate on consumption. By substituting (10.4), (10.5), and (10.9) into the first equation of (10.10), we find (10.11) where b = (1 + τ̂ )a–1 + r + e – s P2. The von Neumann–Morgenstern (expected) utility at time t, Vt, of the competitive consumer is assumed to have the timeseparable form: (10.12) Notice that the agent’s subjective discount rate has been set equal to the constant real international rate of interest, r, to avoid unnecessary technical difficulties. Of course, in the real world, when the subjective discount rate is equal to the real rate of interest is due only to coincidence, We consider the logarithmic utility function, u(ct) = log(ct), in order to derive closed-form solutions and make the analysis easy to manage. In this case, the Hamilton– Jacobi–Bellman equation for the stochastic optimal control problem of maximizing the agent’s lifetime expected utility subject to the intertemporal budget constraint is:

224

F. Venegas-Martínez

(10.13) where

is the agent’s indirect utility function (or welfare function) and Lt (xt, tt, t) is the co-state variable. Given the exponential time discounting in (10.14), we specify L(xt, tt, t) in a time-separable form as (10.14) where (10.15) Here d0, d1, and J(tt, d2, d3) are to be determined from equation (10.15). Coefficients d2 and d3 must satisfy the following conditions: (10.16) By substituting (10.14) into (10.13), we have

(10.17) The first-order conditions of the intertemporal optimisation of the risk averse representative agent lead to a time-invariant wt ≡ w, and (10.18)

How risk factors affect growth in Mexico 225 Figure 10.1 shows optimal w* as a function m and h. We choose now J(tt) as a solution of the second-order ordinary non-homogeneous differential equation (10.19) Coefficients d0 and d1 are determined from (10.15) after substituting optimal w*. Thus, d1 = r–1, so the coefficient of log(xt) in (10.17) becomes zero, and

(10.20) Logarithmic utility implies that w depends only upon the parameters determining the stochastic characteristics of the economy, and hence w is constant. In other words, the consumer’s attitude toward currency risk is independent of his/ her wealth, that is, the resulting level of wealth at any instant has no relevance for portfolio decisions. Moreover, due to the logarithmic utility, the correlation

Figure 10.1 Optimal w* as a function m and h (source: own estimations based on INEGI).

226

F. Venegas-Martínez _

coefficient, κ, plays no role in the consumer’s optimal portfolio, it only matters the trend and volatility components of the stochastic processes driving the dynamics of the exchange rate and the tax policy. Finally, it is important to point out that equation (10.18) is cubic, therefore it has at least one real root. Notice also that from the fact that d1 = r–1, it can be shown that the solution of (10.19) is

where

_

and a = (2τ – s2t ). Coefficients d2 and d3 are determined in such a way that J(t0) = 0 and Jʹ(t0) = 0. Equation (10.18) is cubic with one negative and two positive roots. This can be seen by intersecting the straight line defined by the right-hand side of (10.18) with the graph defined by the left-hand side of (10.18). In such a case, there is only one intersection defining a unique, perfectly viable, steady-state share of wealth set apart for consumption such that w * ∈ (0, 1).

10.4 Wealth dynamics Through this section, we derive the stochastic process that generates an individual’s real wealth when the optimal rule is applied: such a stochastic process follows a stochastic differential equation that with three components; one which provides the physical trend, another which models small movements observed every day (diffusion part), and another describing atypical movements (jump part). Thus, after substituting the optimal share w * into (10.11), we get

(10.21)

where (10.22) and (10.23)

How risk factors affect growth in Mexico 227 We also have that conditional expectation is (10.24) and the conditional variance satisfies (10.25) It can be shown that the solution of the stochastic differential equation in (10.21), conditional on x0, is (10.26) where (10.27) (10.28) and (10.29) as usual, P(×) denotes a Poisson distribution. The stationary components of the parameters of the above distributions are:

and

Notice also that the conditional expectation satisfies (10.30) and the conditional variance is (10.31) Moreover, it readily follows that (10.32)

228

F. Venegas-Martínez

and

(10.33) Finally, according to (10.26), the last two equations determine the mean and variance of the growth rate of real assets.

10.5 Consumption dynamics We now study the stochastic consumption demand. In virtue of equations (10.9) and (10.26), the stochastic process for consumption can be written as (10.34) This indicates that, in the absence of contingent-claims markets, the exchangerate depreciation risk has an effect on wealth via the uncertainty in xt, that is, uncertainty changes the opportunity set faced by the consumer. On the other hand, the depreciation risk also affects the composition of portfolio shares via its effects on w *. Thus, a policy change will be accompanied by both wealth and substitution effects. On the other hand, from (10.34), we can compute the probability that, in a given time interval, certain levels of consumption occur. It is also important to note, regarding (10.34) and (10.12), that the assumption that the agent’s timepreference rate is equal to the world’s interest rate does not ensure a steady-state level of consumption. However, we do have a steady-state share of wealth set aside for consumption. We may conclude that uncertainty is the clue to rationalize richer consumption dynamics that could not be obtained from deterministic models. Finally, in virtue of (10.34), equations (10.32) and (10.33) determine the mean and variance of the growth rate of consumption. Figure 10.2 shows consumption as a function of x0 and t0. We suppose that technology is of the form yt = Akt where A > 0. That is, the marginal product of capital is constant and equal to A. We assume that capital does not depreciate. The condition for profit maximisation require r = A. Since 1 – w * = kt/xt, we have yt = A(1 – w *)xt. In virtue of (10.26), we obtain yt = A(1 – w *)x0ext. Notice first that the production–consumption ratio remains constant according to (10.35) and due to the cash-in-advance constraint, the money–consumption ratio, mt/ct = a, remains also constant. On the other hand, from equations (10.34) and (10.35), we have that

How risk factors affect growth in Mexico 229

Figure 10.2 Consumption as a function of x0 and t0, (x0 in 1011 pesos of 1993) (source: own estimations based on INEGI).

and since mt = w *xt, we obtain mt = w *x0ext. Therefore, (10.36) where, during in the instant [t, t + dt], we have qt|tt ~ N[[B[(v ]*) – tt]t, G([v w]*) dt], dft = D(w *)dQt, and dQt ~ P(hdt). Hence, if dt = T – t, in virtue of (10.36), the expected growth rate of output, in [t, T], satisfies

(10.37) Thus, yt,T depends upon the parameters determining risk factors (uncertain fiscal and monetary policies), which shows significant qualitative differences with

230

F. Venegas-Martínez

respect to the deterministic framework. Also, from (10.31), the variance of the growth rate of output, in [t, T], is given by (10.38) Finally, from (10.36), the expected growth rates of consumption and real cash balances, as well as their variances, are also determined by (10.37) and (10.38), respectively. The above equations provide a mean-variance description of the growth rate of output, showing explicitly the underlying risk factors that affect growth.

10.6 Simulation exercise The following experiment is intended to replicate, via Monte Carlo methods, the mean and its variance of the observed annual growth rate of output, E[dyT/yt] = E[dxT], and its variance, Var[dyT/yt] = Var[dxT], by using equations (10.33) and (10.34), in Mexico between 1930 and 2002. In Figure 10.3, we show the observed annual growth rate of output from 1930 to 2002. Table 10.1 presents a vector of diffusion and jump parameter values, (e, s –1 p, h, µ), that replicate the mean and variance of the annual growth rate. In order to choose such a vector, we tried about 800 different feasible combinations of parameter values. For simulation purposes, we have used a standard discretetime version of (10.38) with an appropriate unit of time, see, for instance, Ripley (1987). Results are based on 10,000 iterations.

Figure 10.3 Observed growth rate from 1930 to 2002 (pesos of 1993) (source: own estimations based on INEGI).

How risk factors affect growth in Mexico 231 Table 10.1 Optimal consumption shares, parameters, and estimates ω* = 0.430004 ε = 0.300000 σ = 0.009999 h = 0.100000 μ = 0.300000 F(ω*) = 0.011026 G(ω*) 0.000019 A(ω*) = –0.004539 ω* = 0.430004 ε = 0.300000 σ = 0.009999 h = 0.100000 μ = 0.300000 F(ω*) = 0.011026 G(ω*) 0.000019 A(ω*) = –0.004539 Estimated growth rate mean = 0.0470 Estimated growth rate mean = 0.0473 Estimated growth rate variance = 0.0223 Estimated growth rate variance = 0.0222 Source: own estimations based on INEGI.

10.7 Conclusions Most of the existing models of endogenous growth ignore uncertainty, providing elaborate justification why uncertainty does not need to be considered. We have shown, under a free-market liberalism framework that risk factors may lead to significant qualitative changes in the determinants of growth in contrast with the deterministic setting. The considerations of uncertainty in the expected dynamics of both the exchange rate and the tax policy have led to more complex transitional dynamics, but results were certainly richer. On the other hand, it is important to mention that our investigation has provided a stochastic model of endogenous growth that explains how risk factors, such as currency, market, debt, and fiscal risk factors, affect economic growth. This extends the literature by including stochastic determinants of growth. Our stochastic framework, in which a Brownian motion and a Poisson process drive the expectations of exchange-rate jumps, and a geometric Brownian motion guides a tax rate on wealth, has provided new elements to carry out simulation experiments and empirical research. In particular, our stochastic model was capable of explaining the average and variance of economic growth for the Mexican case in a given period of time.

References Canton, E. (2001) ‘Fiscal Policy in a Stochastic Model of Endogenous Growth,’ Economic Modeling, vol. 18, No. 1, pp. 19–47. Gokan, Y. (2002) ‘Alternative Government Financing and Stochastic Endogenous Growth,’ Journal of Economic Dynamics and Control, vol. 26, No. 3, pp. 681–706.

232

F. Venegas-Martínez

Rebelo, S. (1991) ‘Long-Run Policy Analysis and Long-Run Growth,’ Journal of Political Economy, vol. 99, No. 3, pp. 500–21. Ripley, B.D. (1987) Stochastic Simulation, Wiley: New York. Venegas-Martínez, F. (2001) ‘Temporary Stabilisation: A Stochastic Analysis,’ Journal of Economic Dynamics and Control, vol. 25, No. 9, pp. 1429–49. Venegas-Martínez, F. (2004) ‘Reforma fiscal incierta y sus efectos en las decisiones de consumo y portafolio: impacto en el bienestar económico,’ Problemas del Desarrollo, Revista Latinoamericana de economía, vol. 35, No. 136, pp. 137–50. Venegas-Martínez, F. (2005a) ‘Bayesian Inference, Prior Information on Volatility, and Option Pricing: A Maximum Entropy Approach,’ International Journal of Theoretical and Applied Finance, vol. 8, No. 1, pp. 1–12. Venegas-Martínez, F. (2005b) ‘Política fiscal, estabilización de precios y mercados incompletes,’ Estudios Económicos, vol. 20, No. 1, pp. 3–18. Venegas-Martínez, F. (2006) ‘Stochastic Temporary Stabilisation: Undiversifiable Devaluation and Income Risks,’ Economic Modeling, vol. 23, No. 1, pp. 157–73. Venegas-Martínez, F. (2008) Riesgos financieros y económicos. Productos derivados y decisiones económicas bajo incertidumbre, 2nd edn, Cengage Learning: México.

11 Anti- inflationary policy and financial fragility A microeconomic analysis case study of Mexico, 1990–2004 Ignacio Perrotini-Hernández, Blanca L. Avendaño-Vargas and Juan Alberto Vázquez-Muñoz

11.1 Introduction The Mexican economy has undergone a number of structural changes since the foreign debt crisis of 1982. While the most radical reforms were introduced with the Brady Plan (1988), trade liberalisation (TL henceforth), financial liberalisation (FL), monetary policy and exports became the driving forces of economic growth. The Brady Plan helped ease the debt burden and foreign saving and investment flows increased with the aid of financial deregulation and price stabilisation policies based on a nominal exchange rate anchor through 1988–1994. Yet, the stabilisation strategy also produced interest rate and asset price hikes, exchange rate appreciation, deindustrialisation and slow output growth, deteriorating saving and investment coefficients as portfolio short-term investment crowded out productive investment and, last but not least, current account disequilibrium. Our task in this chapter is to assess whether the monetary policy framework of low inflation and the investment dynamics derived thereby have triggered financial fragility in the Mexican economy.

11.2 The financial fragility hypothesis The backbone of Keynes’s theory, according to Minsky (1975: 94), is his ‘theory of investment and why it is so prone to fluctuate’. Minsky maintains that financial factors determine the pace of investment, which is the key to economic fluctuations. Minsky’s analysis of the role of financial factors in the process of capital investment is based on a twofold approach, namely the existence of two prices and of a complex financial structure. The former extends Keynes’s model with a view to construct a theory of systematic financial fragility where income flows and the net supply of real capital assets determine the relationship between the

234

I. Perrotini-Hernández et al.

real and financial sectors of the economy. Capitalised income flows determine investment positions and monetary policy and interest rates play a relevant part in the process of financial instability. The supply price of investment is: (11.1) Given the term structure of interest rates, rl, . . . rn, the supply price of capital investment will be: (11.2) Where Ki and Yi denote capital and income flows in period i respectively. Assuming Q1 = Q2 = . . . = Qn, the capitalisation rate can be expressed as: (11.3) where expected returns (Qn) are determined by the future price of produced goods, the volume of sales, input costs and firms’ financial costs. Minsky uses Keynes’s (1936) concept of capitalisation of investment returns to determine the demand price of capital assets (PKi), which depends on the parameter of capitalisation of quasi-rents (fi) (11.4) The rate of capitalisation crucially depends on the monetary and financial conditions of the economy, that is, on the capitalisation rate of the loan borrowed to finance investment. Therefore, the price of liabilities, determined in the debt market, bear an influence on that of capital assets. Thus, the rate of capitalisation of Qn also depends on the level of uncertainty prevailing in financial markets. Then, ‘the capitalisation rate of capital assets is some ratio, 0 < µ < 1, of the capitalisation rate of money loans’ (Minsky, 1975: 102). Let µ denote financial uncertainty and fL the capitalisation rate of the bank loan or the financial liability of firms, then: (11.5) and (11.6) The rate of capitalisation of quasi-rents depends on the performance of the financial market, whereas the capitalisation rate of the loan used to finance investment hinges on the central bank’s monetary policy (the money supply). Since firms

Microeconomic analysis of Mexico 235 get loans in order to finance capital asset positions, the rate of interest becomes a key variable in the process of capitalisation of Qn. In sum, there is a negative relationship between the monetary rate of interest and f. Hence the effect of monetary policy on investment through the effect of the interest rate on the parameter of capitalisation: (11.7) Equation (11.5) is an unstable function – this is a ‘fundamental fact’ (Minsky, 1975) – because the monetary and financial conditions of the economy may affect Pki through the effect of the interest rate on expected quasi-rents. The central bank can affect the balance sheet of those firms that have been undertaking debt in the credit markets with the aim of financing fixed capital asset positions. The relationship between r and fi depends on the market assessment of the probability distribution of actually getting a certain and assured stream of income flows, Qi, vis-à-vis the probability of getting it at a fluctuating market rate of return, Qn, plagued by uncertainty. Investment, then, is determined by the discrepancy between the demand price and the supply price, while the adjustment process induces instability. As for the financial structure approach, Minsky (1982) focuses on the methods used by enterprises to finance investment. Potential sources are: (1) cash and financial assets, (2) internal finance (after-tax and dividends profits) and (3) external finance (loans and equity issuance). Investors consider these different methods of finance and must include the financial cost of capital,1 apart from wage and input costs, in the supply price of produced goods. Investors are bound to forecast their future income streams and the specific conditions which will prevail in the financial markets, as fixed capital investment is a long-term action, while the means available to finance investment positions are short term in nature. Minsky’s financial fragility hypothesis states that there exists an inherent tendency for the economy to become financially fragile as a result of the negative influence of financial variables on firms’ capital structure throughout the business cycle. He then goes on to establish a taxonomy of financial structures according to firms’ balance sheets in terms of flows of income and liabilities and debt payments. Thus, income sources, revenue flows from operations plus new debt (D), equal expenditure streams – investment (I ) plus payment commitments on debt (V ): (11.8) If R < 0, the firm will face a loss; if D < 0, the firm will be repaying its debt, and V < 0 makes the firm a net creditor. The net wealth of a firm (W) is measured as: (11.9) where A denotes total assets and B is the value of its outstanding debt. W increases when investment is increased and/or debt gets reduced. A firm becomes

236

I. Perrotini-Hernández et al.

insolvent and goes bankrupt when W ≤ 0 and its creditors are unable to rescue the capital involved in the unit. A firm is said to hold hedge finance when its cash income flows from operation are expected to be larger than its payment commitments on debts. A hedge firm can be troubled whenever R gets reduced during downswing periods and/ or V increases in a credit crisis. A firm can be characterised as speculative if R ≥ V and R < V + I, which implies D ≥ 0, though D < I. Typically, speculative firms run financial deficits during expansion periods when they engage in investment opportunities that exceed their internal financial capacity. In this case W increases, and the rate of return on investment determines whether a speculative unit survives or goes out of business. In addition, Minsky defines as Ponzi finance a firm that meets its cash payment commitments on debt by augmenting the amount of debt outstanding. In this situation, R < V and D > I. While the credit profile of a Ponzi unit depends on its ability to persuade creditors that its income streams will increase in the near future, the increment in the outstanding debt will make it harder for the former to find voluntary lenders. By and large, business enterprises depend on expectations about future interest rates and financial market conditions. The latter can force hedge firms alternatively into speculative and Ponzi financing. Unless effective demand (sales) rises and/or the interest rate falls, the probability of a firm falling into Ponzi financing will arise.2 The financial structure of firms changes along the business cycle, usually from hedge to Ponzi finance; hence macroeconomic instability. Foley (2003), following Minsky, suggests a typology in terms of the growth rate of firms’ assets (A) and liabilities (B) and the rate of returns on assets: g = I/A is the growth rate of assets, r* = R/A is the rate of return and i* = V/B is the interest payments to debt ratio. A particular combination of i*, r* and g sets the stage for financial fragility of firms (see Table 11.1). The modus operandi of financial fragility can be summarised as follows (cf. Minsky, 1982, 1986; Wolfson, 1989): 1

Expected returns (PE) depend on expected income flows of sales, which in turn depend on effective demand. The actual flow of returns (PR) validate the amount of outstanding debt securities. PE induces fluctuations in the rate of capital investment.

Table 11.1 Typology of the growth rate of firms Typology

Condition

Hedge Speculative Ponzi

r > g > i or r > i > g g>r>i i>r

Source: authors’ own elaboration.

Microeconomic analysis of Mexico 237 2

3

4

Expected returns are stable during the upswing. Since the safety margins increase with economic expansion, refinancing of debt contracts in secondary markets works out properly. Therefore, business optimism produces boom investments that tend to exceed firms’ own internal financial means; keen investors engage into debt financing of economic activity. As the economy gains momentum, the most aggressive firms accumulate debt; increase their leverage rate and the potential for balance sheet destabilisation as the business cycle reaches a turning point. Once the business cycle turns to a downswing PE will diminish, PR will tend to decline with the reduction of effective demand and sales, and financial fragility sets in. Thus, financial instability results endogenously from economic success. The theory of financial instability explains how the accumulation of debt impairs investment: while financial leverage expands aggregate demand at the outset of the business cycle, it also alters the capital structure of those corporations that rely heavily on loans through the effect of the interest rate on income flows and the balance sheet of indebted firms. Finally, the influence of debt financing and higher interest payments on the level of economic activity brings about effective demand constraints, underutilisation of productive capacity and unemployment of the labour force.

According to Minsky, monetary policy rules can also contribute to financial instability. Clearly, in an open economy setting where local firms borrow foreign savings to finance economic activity, an anti-inflationary monetary policy may add to financial fragility through interest rate and exchange rate volatility. A fixed nominal exchange rate may anchor inflation and the net wealth of firms. However, it also invites speculative attacks against the domestic currency and increases the liability burden on highly indebted firms because of interest rate hikes used to cope temporarily with speculative runs against the domestic currency. At the end of the day, as recent experience has shown, this antiinflationary strategy collapses and the central bank (CB), in need of an independent monetary policy, must choose an alternative framework. Many developing countries have adopted an inflation targeting monetary policy framework (IT) where the interest rate has become the policy instrument to achieve both price stability and exchange rate stability. This is the so-called Taylor rule (Taylor, 1993).3 From a Minskyan point of view, the fact that developing economies suffer from a high pass-through effect of exchange rate fluctuations on to the price level, imply that the IT model does add to financial instability, because the CB must increase the interest rate in order to prevent (actually, just to postpone) exchange rate depreciations and attain the expected inflation rate. Thus, the sequence of booms and recessions along the business cycle results from specific monetary policy strategies (Minsky, 1986). Hence anti-inflationary monetary policies may trigger financial instability.4 The IT model is no exception, given the inter-play between the interest rate, the exchange rate and the rate of inflation.

238

I. Perrotini-Hernández et al.

11.3 Empirical evidence, a microeconomic analysis Our empirical analysis is based on data for 47 non-financial firms quoted in the Mexican stock exchange market during 1990–2004. We consider two subperiods, the first from 1990 to 1994 when there existed a fixed nominal exchange rate anchor, and the second from 1995 to 2004 when an IT strategy was introduced.5 Mexico shows a high pass-through effect from exchange rate (e) variations to the rate of inflation (ṗ). The IT strategy brought with it a stronger correlation between the exchange rate and the interest rate (i) on CETES (28 days). The analysis of the effect of the Banco de Mexico’s anti-inflationary policy on firms’ financial profile is focused on the real rate of interest (r) and the real exchange rate (q), owing to the former is the relevant adjustment variable and the latter impacts the real value of dollar-denominated liabilities. Figure 11.1 shows the behaviour of g, r* and i* for our sample of firms as a whole. Variables were built as follows: g = I/A, where I denotes the quarterly increment of total assets and A is the value of total assets; r* = R/A, where R is net income from operation plus the amount of capital invested by shareholders; and i* = V/B, where V is the quarterly integral cost of borrowing (interest payments less interest earnings) and B is total liabilities. All variables have been adjusted by the 2002 consumer price index. As shown above, the firms in the sample as a whole can be characterised as speculative during 1990–1994: g and r* performed positively, by and large g > r*, which can be explained by the transition of the Mexican economy to a more stable environment. The maximum value of g (17.7 per cent) was attained

Figure 11.1 Growth (g), returns (r) and interest rate (i): aggregate levels, 1990:02–2004:04 (source: authors’ own elaboration based on data from Banco de México and Bolsa Mexicana de Valores).

Microeconomic analysis of Mexico 239 Table 11.2 Average values for the sample of firms as a whole Variable

1990:2–2004:4 (%)

1990:2–1994:4 (%)

1995:1–2004:4 (%)

r* r*1 r*2 g i* b

0.78 0.42 0.36 1.30 0.54 1.84

2.31 1.55 0.76 3.80 0.79 4.51

0.06 –0.11 0.17 0.11 0.42 0.57

Source: authors’ own calculations on the basis of data from Banco de México and the Mexican stock exchange market.

in the last quarter of 1994. Table 11.2 shows that the returns from shareholders’ contributions to asset accumulation (r*2) is responsible for one-third of the total rate of returns, whereas returns from operations of the firm (r*1) is responsible for 42 per cent, signalling optimistic investors. After the financial crisis of 1994–1995, the average growth rate of firms (g) shrank from 3.80 per cent during 1990–1994 to 0.11 per cent during 1995–2004. The rate of return, in turn, averaged 2.31 per cent throughout the nominal exchange rate anchor period and 0.06 per cent during the IT era. The fall in r* signals the loss of momentum in the pace of capital accumulation. On the other hand, i* remained low and stable through 1990–1993, but increased sharply after the speculative attack against the Mexican peso in 1994 and has declined throughout the years of the flexible exchange rate regime (see Figure 11.1 and Table 11.2). The rate of investment and the growth rate of private debt (b) rose during the fixed exchange rate regime, partly because financial liberalisation enhanced the supply of loanable funds (see Figure 11.2 and Table 11.2). As Minsky (1975, 1982) points out, an increasing rate of return spurred by financial leverage, does generate an optimistic environment that prompts booming investment and higher indebtedness. The financial crisis of 1994–1995 was followed by a long episode of credit rationing which was not offset by financial sources other than the formal banking sector: the quarterly growth rate of Mexican firms’ debt (b) averaged 4.51 per cent from 1990 to 1994 and 0.57 per cent from 1995 to 2004. The aforementioned financial crisis also produced a sharp increase in i* and a devaluation of the exchange rate, which led to a higher b. The overall effect on our sample of firms was a lower net wealth (W = g – b) and a faster transition from speculative to Ponzi finance, since the higher rate of inflation that erupted from the financial crisis eroded the real value both of total assets and total liabilities (see Figure 11.3). In fact, it can be argued that attainment of price stability in Mexico has meant a reduction of W from –0.71 per cent throughout 1990–1994 to –0.46 in 1995–2004, on the one hand, and a negative accumulation of assets in aggregate terms, on the other. Figures 11.4 and 11.5 show the relative composition of both firms and assets according to a classification of hedge (h), speculative (s) and Ponzi (p) finance.

Figure 11.2 Growth (g) and firms’ indebtness (d): aggregate levels, 1990:02–2004:04 (source: authors’ own elaboration based on data from Banco de México and Bolsa Mexicana de Valores).

Figure 11.3 Net wealth (w), growth (g) and debt (d): aggregate levels, 1990:02–2004:04 (source: authors’ own elaboration based on data from Banco de México and Bolsa Mexicana de Valores).

Microeconomic analysis of Mexico 241 And Table 11.3 presents average values of data shown in Figures 11.4 and 11.5 for the whole period and the two sub-periods considered in the analysis. The percentage composition of firms and assets does not vary significantly, except for speculative and Ponzi units during the fixed exchange rate sub-period. Interestingly, the number of hedge units represented the smallest proportion from 1990 to 1994. The share of hedge units increased after the financial crisis of 1995 because several Ponzi firms were unable to recover financially and left the MSM. This explains why the share of Ponzi units declined starting in 2003, even though their absolute number increased during the IT sub-period. We now turn to the analysis of the influence of BM’s anti-inflationary policy on financial fragility. As mentioned, our analysis is focused on r and q, their relationship being given as: (11.10) where D is a quarterly differential operator, q is the quarterly real exchange rate depreciation, µt is white noise and bi are estimate parameters.6 A high correlation between e and q can be confirmed throughout the IT sub-period. Thus, given this high correlation between the nominal depreciation of e and the rate of inflation (which implies a real depreciation), the central bank reacts, with the aim of diminishing aggregate demand through higher r, via a further increase in i whenever e rises. We estimate the effect of changes in r and q on b using the following equation: (11.11)

Figure 11.4 Composition of firms according to financial structure (%), 1990:02–2004:04 (source: authors’ own elaboration based on data from Banco de México and Bolsa Mexicana de Valores). Note The size of the sample varies in each period as some firms either left the Mexican stock market or did not provide information of their operations. h: hedge, s: speculative, p: Ponzi.

242

I. Perrotini-Hernández et al.

Figure 11.5 Composition of firms’ assets according to financial structure (%), 1990:02–2004:04 (source: authors’ own elaboration based on data from Banco de México and Bolsa Mexicana de Valores). Note The size of the sample varies in each period as some firms either left the Mexican stock market or did not provide information of their operations. h: hedge, s: speculative, p: Ponzi.

where j refers to the j firms, Wi are the parameter estimates, vtj denote white noise and Dr˜ represents, on the one hand, quarterly changes in r from 1990:2 to 1994:4 and, on the other, the residual from estimation of equation (11.10) from 1995:1 to 2004:4, in other words, variations in r which are unexplained by depreciations of q. Estimation of equation (11.11) was conducted using the fixed coefficients approach and unbalanced panel data for 47 non-financial firms. Our results are as expected: variations in the interest rate and exchange rate depreciations increase outstanding debt of firms (see Table 11.4).7 It is worth noting that the debt elasticity with respect to firms’ growth (gtj) is greater than one (1.67), signalling an endogenous risk of financial fragility. Moreover, the high and negative (–1.97) debt elasticity with respect to firms’ rate of returns (r*tj) highlights the importance of internal finance j for capital investment. Finally, bt–1 is significant but close to zero. Table 11.3 Average percentage composition of the companies according to their financial structure Period

1990:2–2004:4

1990:2–1994:4

1995:1–2004:4

Financial structure By number of firms By assets

h 27 26.1

h 23.4 22.4

h 28.8 27.8

s 30.1 33.5

P 42.8 40.5

s 39.3 45.3

p 37.3 32.2

s 25.8 27.8

p 45.4 44.4

Source: authors’ own calculations on the basis of data from Banco de México and the Mexican stock exchange market.

0.25 0.16 –1.97 1.67 –0.04 R2 adjusted DW F-Statistic

2.34 5.81 –39.73 61.82 –3.38 0.64 2.17 1,118.12

0.02 0 0 0 0

–0.02 0.19 –2.69 2.93 –0.04 R2 adjusted DW F-Statistic

–0.13 2.41 –37.65 56.04 –2.67 0.82 2.14 874.6

t-Statistics

Coefficient

Pro.b

Coefficient

t-Statistics

1990:2–1994:4

1990:2–2004:4

0.9 0.02 0 0 0.01

Prob. 0.02 0.13 –1.75 1.46 –0.04 R2 adjusted DW F-Statistic

Coefficient

Note For the sake of brevity we omit the fixed values of the coefficients.

0.17 4.68 –31.17 50.14 –3.12 0.65 2.16 803.87

t-Statistics

1995:1–2004:4

Source: authors’ own calculations on the basis of data from Banco de México and the Mexican stock exchange market.

Dr˜t θ·t r*ti g ti i bt–1

Independent variable

Table 11.4 Dependent variable: b ti

0.87 0 0 0 0

Prob.

244

I. Perrotini-Hernández et al.

The relevance of the effect of real exchange rate depreciations on firms’ indebtedness is also confirmed when we estimate equation (11.11) for subperiods 1990–1994 and 1995–2004, although in these cases interest rate variations became insignificant perhaps because FL and reprivatisation of the banking sector flooded the economy with booming credit (in the first sub-period) and the impact of interest rate variations on b was somewhat captured by depreciations of q (during the IT sub-period). Both g tj and r*tj were higher throughout the first sub-period than in the second one and greater than one in both cases (see Table 11.4), confirming a high propensity of Mexican non-financial firms to financial fragility when the BM pursues anti-inflationary monetary policies, either in the form of a nominal exchange rate anchor or an IT. Following the same estimation method as in equation (11.11), we then estimate the effect of monetary policy variables on the increment of firms’ net debt (–w), where Yi are the parameter estimates and u tj is white noise: (11.12) Results from equation (11.12) are as expected (see Table 11.5): r and q bear a positive correlation effect on firms’ accumulation of net debt, while r* bears a negative relationship with net debt accumulation. However, greater effects of variations in r and depreciations of q during 1990–1994, as opposed to those of the IT sub-period, can be observed, possibly because of better expectations and the acceleration of investment and debt accumulation in those years. The elasticity of –w with respect to r*, both in the period as a whole and across sub-periods, remains close to one. It is also interesting to note that, throughout the IT subperiod, variations in the interest rate which are explained by factors other than exchange rate depreciations are not significant. The following ordered model with panel data estimates the probability of a firm being hedge, speculative or Ponzi finance: (11.13) where v tj denotes the rate of debt growth of firm j which is not explained by the independent variables in equation (11.11), and F* is a latent variable used to order the dependent variable F as follows: 0 if 1 if 2 if F tj represents the observed financial structure of firm j in period t: the firm is said to be hedge if F tj = 0, speculative if F tj = 1 and Ponzi if F tj = 2. Furthermore, we assume a ‘normal’ probability function for the determination of a specific

0.36 0.14 –0.95 –0.07 R2 adjusted DW F-Statistic

2.93 4.48 –30.74 –4.24 0.29 2.09 342.72

0 0 0 0

0.62 0.5 –0.95 –0.09 R2 adjusted DW F-Statistic

2.39 3.73 –10.27 –2.45 0.13 2.03 55.13

t-statistics

Coefficient

Prob.

Coefficient

t-statistics

1990:2–1994:4

1990:2–2004:4

0.02 0 0 0.01

Prob. 0.06 0.11 –0.98 –0.08 R2 adjusted DW F-Statistic

Coefficient

Note For the sake of brevity we omit the fixed values of the coefficient

0.47 3.75 –31.78 –3.98 0.38 2.09 358.95

t-statistics

1995:1–2004:4

Source: authors’ own calculations on the basis of data from Banco de México and the Mexican stock exchange market.

Dr˜t θ·t r*ti i –wt–1

Independent variable

Table 11.5 Dependent variable: w ti

0.64 0 0 0

Prob.

2.28 –14.49 6.22 Limit values of γi –0.76 0.18 Statistic LR index LR index

Coefficient

0.12

–25.09 6.52 620.97

8.44 –21.44 19.11

Z-Statistic

0.000 0.000

0.000 0.000 0.000

Prob.

LR index

2.26 –12.1 5.2 Limit values of γi –0.91 0.32 Statistic LR 0.11

–16.04 6.43 186.21

2.79 –12.25 8.53

Z-Statistic

Prob.

0.000 0.000

0.01 0.000 0.000

LR index

2.19 –14.26 6.26 Limit values of γi –0.68 0.11 Statistic LR

Coefficient

0.11

–19.32 3.45 409.79

7.65 –17.14 15.79

Z-Statistic

No. observations: 1708

No. observations: 814 Coefficient

1995:1–2004:4

1990:2–1994:4

Source: authors’ own calculations on the basis of data from Banco de México and the Mexican stock exchange market.

Limit: γ0 Limit: γ1

θ·t r*ti g ti

Independent variable

(Unbalanced 1990:2–2004:4 panel) No. observations: 2452

Table 11.6 Dependent variable: F ti

0.000 0.000

0.000 0.000 0.000

Prob.

Microeconomic analysis of Mexico 247 financial structure of particular firms, since variations in the firm’s debt ratio may augment the probability of that firm falling into Ponzi finance, but the greater the variations in the debt ratio the smaller the increment in such probability. Table 11.6 summarises the results from estimating equation (11.13) with the maximum likelihood method.8 The probability of a firm falling into Ponzi (hedge) financing increases (diminishes) with the expansion of the firm and real exchange rate depreciations, while, conversely, such probability decreases (augments) with increments (reductions) in the firm’s rate of return. By and large, anti-inflationary monetary policies add to Ponzi financing of firms. On the other hand, it is difficult to determine, a priori, the influence of the relevant variables on the probability of a firm falling into speculative finance, because, according to empirical evidence, that effect tends to zero. Figures 11.6 and 11.7 show the probability with which firms may be classified as hedge, speculative or Ponzi finance. As shown in Figures 11.6 and 11.7 and Table 11.7, the average probability of speculative finance was greatest when BM’s monetary policy focused on a fixed nominal exchange rate anchor; the probability of Ponzi finance was second and that of hedge third. The adoption of the IT model changed the probability distribution of the various financial structures inasmuch as Ponzi finance became paramount among Mexican firms during 1995–2004. All in all, it appears that according to empirical evidence for a sample set of Mexican non-financial firms, Minsky’s financial instability hypothesis can help

Figure 11.6 Probability of hedge (h), speculative (s) and Ponzi (p) finance among firms, 1990:02–1994:04 (source: authors’ own elaboration based on data from Banco de México and Bolsa Mexicana de Valores). Note Data are ordered firm-wise.

248

I. Perrotini-Hernández et al.

Figure 11.7 Probability of hedge (h), speculative (s) and Ponzi (p) finance among firms, 1995:01–2004:04 (source: authors’ own elaboration based on data from Banco de México and Bolsa Mexicana de Valores). Note Data are ordered firm-wise.

Table 11.7 Average of the estímate probability of financial structure of firms Period

1990:2–2004:4

1990:2–1994:4

1995:1–2004:4

Financial structure By number of firms

h 26.4

h 22.5

h 28.1

S 30.8

p 42.8

S 40.7

p 36.8

s 26.5

p 45.5

Source: authors’ own calculations on the basis of data from Banco de México and the Mexican stock exchange market.

us understand the evolution of capital structures from hedge to speculative to Ponzi financing when the central bank pursues an anti-inflationary monetary policy. Therefore, inflation targeting matters for financial fragility.

11.4 Conclusion The present chapter aimed at arguing that the structure of capital matters for economic stability, the monetary policy framework is also relevant for financial stability, in other words, money is not neutral and, last but not least, given a high pass-through coefficient, the dynamics between the real exchange rate, the interest rate and the rate of inflation, pursuing an inflation targeting strategy may encourage Ponzi financing. Using data for 47 Mexican non-financial firms quoted in the Mexican stock exchange market from 1990 to 2004, we have assessed Minsky’s financial

Microeconomic analysis of Mexico 249 instability hypothesis. We conclude that the latter provides insights into acrossfirm characteristics of financial fragility, in particular when the central bank narrowly (and uniquely) pursues an inflation target with no regards whatsoever for other economic policy targets, such as growth and employment.

Notes 1 Typically, finance for production purpose is short term and, habitually, take the form of bank loans. 2 Fisher (1933) had described this situation, which triggers money flows due to asset sales, in his theory of debt-deflation: firms cancel debts, thus causing inflation of the value of money as a result of an increasing demand for money and a limited supply. As Fisher argued, the more debt is cancelled, the more debtors owe (Minsky, 1977). 3 Ball (1998) has extended Taylor’s model for the open economy case where the CB follows a monetary conditions index, given by a weighted average of the interest rate and the rate of appreciation of the exchange rate, to attain the targeted rate of inflation. 4 In this scenario, both speculative and Ponzi units increase their demand for loans with a view to refinance debt commitments, and demand is interest rate inelastic. 5 Banco de Mexico (BM) adopted an IT in 2001, though major features of such monetary policy framework had been introduced since 1995. 6 The estimation of equation (11.10) was not significant for the fixed exchange rate subperiod and significant for the IT sub-period. While the results from regressions are not reported in this text, they are available from the authors upon request. 7 The estimation results from equations (11.11) and (11.12) were corroborated by the GLS method. 8 The final estimation excludes parameters for variables r˜t and vt because our initial estimation proved them statistically insignificant.

References Ball, L., 1998, ‘Policy Rules for Open Economies’, National Bureau of Economic Research, Working Paper no. w6760. Bolsa Mexicana de Valores, Indicadores Bursátiles, various years. Fisher, I., 1933, ‘The Debt–Deflation Theory of Great Depressions’, Econometrica, 1, pp. 337–357. Foley, D., 2003, ‘Financial Fragility in Developing Economies’, in Dutt, A. and J. Ros (eds), Development Economies and Structuralist Macroeconomics, Essays in Honor of Lance Taylor, Cheltenham: Edward Elgar. Keynes, J.M., (1936), Teoría General de la Ocupación, el Interés y el Dinero, Fondo de Cultura Económica, México. Minsky, H.P., 1975, John Maynard Keynes, New York: Columbia University Press. Minsky, H.P., 1977, ‘A Theory of Systematic Fragility’, in E.I. Altman and W. Sametz (eds), Financial Crises: Institutions in a Fragile Environment, New York: Wiley. Minsky, H.P., 1982, Can ‘It’ Happen Again? Armonk, NY: M.E. Sharpe. Minsky, H.P., 1986, Stabilizing an Unstable Economy, New Haven and London: Yale University Press. Taylor, J., 1993, ‘Discretion versus Policy Rules in Practice’, Carnegie – Rochester Conference Series on Public Policy, no. 39, pp. 195–214. Wolfson, Martin, 1989, Financial Crises: Understanding the Postwar U.S. Experience, Armonk, NY: M.E. Sharpe.

12 Technological innovation and sectoral productivity in the Mexican economy Regional evidence José Carlos Trejo-García, Humberto Ríos-Bolívar and Ana Lilia Valderrama-Santibáñez

12.1 Introduction There are a large number of studies in economics related to the relationship between technological progress, innovation and economic growth. One of the pioneers was Robert Solow, who published his first work on this subject in 1957. However, to date it should be recognized the difficulty of measuring the proper role of technology and innovation in growth, especially when it comes to empirically demonstrate this fact. This has led economists to focus on the analysis of expenditure on research and development (R&D) for innovation as a close variable to technology and technological innovation. According to the evidence shown by Solow, such spending contributes to technological improvements, so that investment in R&D is considered to have a significant impact on productivity growth. Thus, the empirical analysis of the relationship between R&D for innovation and productivity can be done through the estimate of a production function, where technological capital and innovation are included as explicative variables. In a production model which includes technological capital, Griliches (1979) argues that production function includes, in addition to the usual factors of production, another factor that may be named technological capital, technological innovation or R&D capital. This chapter studies the relationship between the labour productivity growth and investment in this kind of capital for the manufacturing sector, trade and services. For this purpose INEGI’s statistical data from the economic census of 1994 and 2004 is used. The use of cross section data for 2,438 municipalities allows having a large number of data samples in addition to statistical information disaggregated by economic sectors. Until now, papers in Mexico about the relationship between productivity and R&D spending have primarily estimated production functions. These functions determine production elasticity related to production factors: capital and labour, leaving aside technology and innovation factors. It should be mentioned that one of the main problems in this kind of analysis is the need of information about the stock of research and development capital, which is generally not available.

Innovation and productivity in Mexico 251 Therefore it was necessary to estimate the stock of capital in R&D. The perpetual inventory method is frequently applied. That is, the capital stock for each period is calculated from the capital stock (minus depreciation) in period t–1, plus the capital investment in the period t. However, this difficulty can be avoided if we estimate a transformation of the production function that requires knowing only the R&D expenditure in each period. This is what we do in this research.1 This requires a certain degree of novelty compared to previous research conducted in Mexico, because, instead of estimating production functions where the capital in R&D is another factor, productivity growth is directly related with the intensity of R&D spending. To our knowledge, there is no research in this line for Mexican economy. The present chapter is structured as follows. The next section reviews the literature on the subject. The third presents the theoretical model. The fourth describes the data, variables and empirical methodology applied in the econometric analysis. An econometric model is estimated in Section 5. The final section summarizes the paper and highlights the most important conclusions.

12.2 Background While the relationship between productivity and R&D for innovation in Mexico has been discussed by Unger (1996) and Jasso (1998), among others, the common feature of these papers is that they are based on the specification of a production function as well as on the estimation of R&D–capital elasticity in the industry. The present chapter applies a transformation of a production function to avoid the use of technological capital stocks as an independent variable.2 This kind of analysis has been done for a wide number of countries. Some of the most important studies are Griliches and Mairesse (1983), who analyse the influence of R&D expenditure on productivity from individual data for the United States (US) and France between 1973 and 1978; Clark and Griliches (1984) study the relation between productivity growth and R&D expenditure during the period 1970–1980, their statistical sample includes data for 924 US manufacturing companies; Lichtenberg and Siegel (1991) use panel data to study the relationship between R&D and productivity growth in the US industry in the period 1972–1985. Recently, Bessen (2000) used a sample of 471 US companies between 1983 and 1989 to get results on the relationship between productivity and R&D expenditure. While the primary goal of previous researches was to measure the costs for firms for adopting the technology derived from R&D, there are other studies. For example, Odagiri and Iwata (1986) estimated the impact of R&D expenditures on the rate of productivity growth in Japan using data from individual companies in two different periods: from 1966 to 1973 and from 1974 to 1982. Fecher (1990) analyses the influence of R&D spending on productivity, from individual data of companies from Belgium, between 1981 and 1983. Hall and Mairesse (1995) updated their own results of previous research about the link between

252

J.C. Trejo-García et al.

productivity and R&D in the French economy. The study runs from 1980 to 1987 and involves information from 351 companies. Wakelin (2001) examines the relationship between productivity growth and intensity of expenditure on R&D in the UK using information provided by 170 British companies during the years 1988–1996. Finally, for Italy, Parisi et al. (2002) show empirical evidence of the effect of innovation on productivity in the production process, on the one hand, and the impact of innovations in the product, on the other. They also studied the effect of investment in R&D on the probability of making innovations. The information comes from 941 Italian companies and refers to 1992–1997. Estimates of the rate of return provided for these papers are mixed. Overall, the results depend on the way of measuring different variables included in the estimates and data sources used in the study.

12.3 Theoretical aspects The theoretical understanding of the relationship between productivity and R&D expenditure for innovation is based on the model of Griliches (1979). This model, in turn, is based on the accumulation of capital in R&D for technological innovation as an additional factor of production, along with the usual factors of production: physical capital and labour. In this research, the starting point for building the model is a Cobb–Douglas production function with three production factors: technological innovation (H ), stock of physical capital (k) and labour (L). The aggregated production function is written as: (12.1) where subscripts i and t denote the firm and the period, respectively. Q is an output measure (usually sales or value added of economic sectors), L represents labour (usually the number of employees), H and K measure the stocks of technological innovation capital and physical capital, respectively, A is a constant, α, β and g are the corresponding elasticities of output related to R&D, physical capital and labour, respectively, λ is the rate of unincorporated technical change (exogenous changes in production technology over time that cause variations in the growth rate of productivity, common to all economic sectors), μ represents a non-observable specific effect of each economic sector, constant over time, and ε is a random error term. This is the Cobb–Douglas production function approach whose main characteristic is factors divisibility; this allows dividing the effects of the R&D factor and the possibility of estimating a linear model in levels, in differences or under some kind of transformation, for instance logarithmic. The following equations, written in logarithmic terms, result thereby: (12.2)

Innovation and productivity in Mexico 253 The function in first differences: (12.3) where lower-case letters denote logarithms of each variables and Δ represents the first difference of the specified variable. The specific effects of the economic sector, μi, are eliminated when taking first differences. The main disadvantage of such a specification is the need for an appropriate measure of R&D capital stock. To avoid this problem, we can make some changes in the Cobb–Douglas production function. Moreover, under the assumption that the production function presents constant returns to scale related to the standard inputs, a + b + g = 1, subtracting labour logarithm from expression (12.2) in both sides, obtains: (12.4) where g = 1 – a – b Substituting g, we have: (12.5) Rearranging terms: (12.6) Taking first differences in this expression obtains: (12.7) where ui = Deit and D(q – l) is the growth rate of labour productivity, D(h – l) the growth rate of technological capital related to labour and D(k – l) is the growth rate of capital labour. The parameter g is the elasticity of output with respect to R&D.3 Moreover, the growth rate of R&D capital is calculated using the next expression: (12.8) From elasticity g and expression (12.8) it is obtained: (12.9) where Y = (∂Q/∂H)it is the marginal productivity of capital in R&D, Rit is investment or expenditure in R&D of sector i in the period t, and R/Q is the intensity of R&D spending or the level of technological effort.

254

J.C. Trejo-García et al.

Using equation (12.9) and a = 1 – b – g, in order to get aD(h – l)it as the growth rate of technological capital related to labour, we rewrite it as follows: (12.10) Substituting equation (12.10) in (12.7) it is obtained: (12.11) Differentiating: (12.12) Simplifying terms obtains: (12.13) Rearranging terms: (12.13b) Having g – 1 = –a – b, expression (12.13) becomes: (12.14) Expression (12.14) can be used to estimate the R&D spending, instead of the stock of R&D capital. Moreover, under the assumption that market operates under competitive conditions, Y can be interpreted as the rate of return of R&D expenditure. Equation (12.14) allows estimating the value of the rate of return of technological capital Y. This is one of the important points of this research. The estimate of the proposed equation is done mainly from data provided by the Mexican economic census, elaborated by INEGI. These surveys include a set of units listed for each municipality, for three economic sectors: manufacturing, trade and services. Econometric estimations that use cross section data for all regions are conducted in this chapter. Our approach is similar to that adopted by Mankiw et al. (1992) which examines the determinants of growth in terms of R&D. Statistical information was taken from the Municipal Information System Database (SIMBAD), particularly: • •

1994 Economic Census, 2004 Economic Census.

The data fall into three sectors; manufacturing, commerce and services. Variables used are shown in Table 12.1.

Innovation and productivity in Mexico 255 Table 12.1 Variables Name

Abbreviation

Gross domestic product Population Fixed assets R&D expenditure Technological effort Total remunerations Average employed persons Computers Line phones Internet access availability

Q N K H R/Q RT L D1 D2 D3

Source: author’s own elaboration.

12.4 Econometric model The chance of an economy to experience economic growth as well as improvements in technology and innovation is one of the major concerns of the present time. One premise of the theory of technological innovation is that innovation is one of the basic elements for economic growth and for improving a nation’s technology. In this way, it can be argued that spending on R&D could be the indispensable factor for economies towards development. Several empirical models have been used to demonstrate the interaction between growth, technological change and technological innovation and more specifically between output growth, technological change and R&D. One of the studies that focused on such interaction was the seminal paper by Solow (1957) on the US economy in the period 1909–1949. Solow found that over 80 per cent of output growth per hour of labour recorded in that period was due to technical progress. That is, of an annual average growth rate of 2.9 per cent of real GDP, approximately 0.32 per cent of the increase was due to accumulation of capital, 1.09 per cent to increases in the quantity of labour and the remaining 1.49 per cent was attributed to technical progress. In a subsequent study along this line of research, Denison (1974) used data for the period 1929–1969 for the US economy and the results confirmed the findings by Solow. He found that the real GDP average rate was 2.92 per cent; 0.56 per cent caused by the growth of capital, 1.34 per cent due to increases in the quantity of labour and 1.02 per cent to technical progress. Most of the recent empirical studies, based on neoclassical and endogenous growth theories, emphasize the role played by R&D spending in the evolution of technology and economic growth. The first contributions are due to Romer (1986) and Lucas (1988), both authors agreed that spending on R&D is acquired through formal education, informal training and labour experience; they also argued that these factors are conducive to higher growth rates, at least during a transition period.

256

J.C. Trejo-García et al.

Under this research framework, the purpose arose to empirically study the role of education and human capital in the evolution of technological change and economic growth in Mexico. In more concrete terms, the need arises to answer questions such as: What are the components of technological progress? What is the role of technological innovation in technological progress and output growth? What is the role of education in technological progress? among others. In order to respond to this set of questions, the next section presents a model based on Solow’s model.

12.5  Econometric model specification To provide a basis for answering the previous questions, an econometric model represented by a production function is proposed. It establishes the relationship between inputs and output, and indicates the maximum product that is possible to obtain under certain combinations of inputs. As is usual in these models, we assume that these inputs are variable, differentiable and convertible into production. The production function is expressed as: (12.15) where Q is the product level, X is a vector of explanatory variables of the product and Z is a vector of parameters governing the rates at which the explanatory variables are transformed into product. It is assumed that A is a well-behaved, continuous and differentiable function. A functional form that meets these characteristics is the Cobb–Douglas production function. So: (12.16) in logarithmic terms: (12.17) This expression is also known as the log-linear model. In what follows, the functional form of the model, used to empirically estimate the determinants of economic growth, is presented. A proper way to estimate the relationship between output growth is based on the simple Solow model, where the residue accounts for the growth that is not explained by production factors. The function used is a production function with Hicks’ neutral technological progress, expressed in the following equation: (12.18) Re-expressing this equation in terms of the output growth rate, the following equation is obtained:

Innovation and productivity in Mexico 257 (12.19) The model specifies that production is boosted by increase in production factors: R&D capital, physical capital, labour and variables that can be part of the residual, such as human capital or improvements in production factors. The general model to be considered is the determination of the output growth rate, which is expressed in logarithmic terms and through instrumental variables in order to approximate the variable associated with technology. The model is as follows:

(12.20) where the left term determines the logarithm rate of output growth; the first term on the right is the intercept, the second one is the initial value of the product, its coefficient measures the rate of convergence of the economy under study; the following two terms specify the participation levels of capital factors, that is, of R&D and physical capital. The remaining terms reflect the participation of not incorporated variables into the model. Equation (12.20) is used to determine the share of production factors and technological change in the growth rate of output; it is also used to compare the performance of these factors in different sectors of the economy, as discussed below.

12.6 Estimation and model results To analyse the behaviour of the economy, the model (12.20) is estimated with the weighted least squares method. Estimates are made for two general issues; on the one hand, the rate of output growth and, on the other, the growth rate of technological change. Because estimation with the least squares method tends to be affected by heteroscedasticity problems, we used the weighted least squares methodology as a way to correct this problem in the cross section estimation. The weighted term was assigned according to the variable that could be causing problems with the variance of errors. This change does not affect the calculation of the parameters. The estimates about determinants of economic growth are classified into four categories. For each of these four categories, there is a subdivision by economic sectors; manufacturing, commerce and services. The first two subdivisions correspond to: 1

information for municipalities with high population density (urban) and weighted by the product;

J.C. Trejo-García et al.

258 2

information for municipalities with low population density (rural) and weighted by the product.

We used equation (12.20) for estimating these two cases. The specific feature of this approach lies in using the product as a weighting variable in both cases. Both estimates were made simultaneously with the aim of carrying out a comparison between estimated parameters, that is, the performance of a variable under different scenarios can be compared. In this case, the scenarios are urban and rural. The results of these estimates are shown in Tables 12.2 and 12.3. Table 12.2 shows the results of econometric estimates by economic sectors, corresponding to high population density municipalities. The most relevant descriptive statistics of the sample are shown. First, the positive growth rates (Q) in the three sectors is indicated, with the greatest growth in the services sector. There is also a positive role of R&D spending (H), although at very low levels. The level of technological effort (R/Q) is positive and with high levels in all sectors. Regarding the average labour productivity, significant differences between sectors is also perceived. While there is some relationship between productivity and technological effort of the sectors, that relationship is not conclusive: that is, not all sectors with an innovative effort above average have productivity higher than average productivity. The descriptive analysis supports the inclusion of other determinants in the growth of productivity, in addition to investments in R&D, which are collected by the variable A. Table 12.3 shows the results of econometric estimates by economic sectors, corresponding to low population density municipalities. The most relevant Table 12.2 Estimate for municipalities with high density of population (urban) and weighted by the product Descriptive statistics (1994–2004) Coefficient Probability (*) Sector

Q0

H

K

L

R/Q

A

Manufacturing

–0.0021 0.0426 0.027 0.000 –0.0053 0.000

0.0032 0.049 0.0018 0.0012 0.0729 0.000

0.082 0.0073 0.062 0.0452 0.141 0.000

0.042 0.006 0.051 0.000 0.097 0.000

0.73 0.000 0.69 0.053 1.73 0.000

0.24 0.000 1.56 0.000 3.036 0.000

Commerce Services

Source: own estimations based on Economic Census. Note (*) The level of probability to reject the null hypothesis is 0.05.

Innovation and productivity in Mexico 259 Table 12.3 Estimate for municipalities with low density of population (rural) and weighted by the product Descriptive statistics (1994–2004) Coefficient Probability (*) Sector

Q0

H

K

L

R/Q

A

Manufacturing

1.02 0.038 1.439 0.000 1.135 0.000

0.003 0.046 0.000 0.518 0.038 0.000

0.091 0.035 0.0383 0.0652 0.121 0.000

0.051 0.000 0.098 0.000 0.063 0.000

0.003 0.000 0.001 0.064 0.013 0.000

1.13 0.000 0.614 0.000 1.026 0.000

Commerce Services

Source: own estimations based on Economic Census. Note (*) The level of probability to reject the null hypothesis is 0.05.

descriptive statistics of the sample are shown. First, the positive growth rates (Q) in the three sectors are indicated, with the greatest growth in the services sector. There is also a positive role of R&D spending (H), although at very low levels, even lower than in the urban sample. The level of technological effort (R/Q) is positive and with low levels in all sectors. Regarding the average labour productivity, significant differences between sectors are still present and are even deeper. While there is some relationship between productivity and technological effort of the sectors, that relationship is not conclusive: that is, not all sectors with an innovative effort above average have productivity higher than average productivity. As in Table 12.2, the econometric results support the inclusion of other determinants in the growth of productivity, in addition to investments in R&D, which are collected by the variable A. On other hand, when statistical data is weighted by the number of registered units, there are two possibilities for analysis: 3 4

information for municipalities with high population density (urban) and weighted by the number of registered units; information of the municipalities with low population density (rural) and weighted by the number of registered units.

As in the two cases previously discussed, the possibilities (3) and (4) are estimated using the model (12.20), but differing with earlier estimates since the weighting variable is now the number of registered units. The results of these estimates are presented in Tables 12.4 and 12.5. Table 12.4 shows the results of econometric estimations for the three economic sectors considered. As in the results in Table 12.2, there is also evidence

260

J.C. Trejo-García et al.

Table 12.4 Estimate for municipalities with high density of population (urban) and weighted by the number of registered units Descriptive statistics (1994–2004) Coefficient Probability (*) Sector

Q

H

K

L

R/Q

A

Manufacturing

1.017 0.000 1.82 0.000 3.921 0.000

0.01 0.049 0.004 0.591 0.061 0.000

0.091 0.05 0.069 0.038 0.138 0.000

0.071 0.032 0.042 0.000 0.089 0.000

0.69 0.013 0.71 0.053 1.941 0.000

0.21 0.000 0.079 0.000 0.117 0.000

Commerce Services

Source: own estimations based on Economic Census. Note (*) The level of probability to reject the null hypothesis is 0.05.

Table 12.5 Estimate for municipalities with low density of population (urban) and weighted by the number of registered units Descriptive statistics (1994–2004) Coefficient Probability (*) Sector

Q

H

K

L

R/Q

A

Manufacturing

1.021 0.000 1.461 0.000 1.171 0.000

0.051 0.046 0.062 0.518 0.042 0.000

0.0496 0.035 0.062 0.0652 0.0158 0.000

0.062 0.000 0.071 0.000 0.056 0.000

0.003 0.000 0.007 0.064 0.011 0.000

0.734 0.000 0.615 0.000 1.029 0.000

Commerce Services

Source: own estimations based on Economic Census. Note (*) The level of probability to reject the null hypothesis is 0.05.

of a positive contribution of each of the regressors; services shows the biggest participation as long as manufacturing sector has the smallest contribution. Significantly, there is a high share of R&D spending (H) as well as the technological effort level (R/Q) for all the sectors. Finally, the results of the estimates for municipalities with low population density and weighted by the number of units surveyed are shown in Table 12.5. This table shows the results of econometric estimations for the three economic

Innovation and productivity in Mexico 261 sectors. There is also evidence of a positive participation of each of the regressors. In this case, the commerce sector has the biggest share as long as the contribution of the manufacturing sector is the smallest one. Just as in the previous cases, there is a relative high share of spending on R&D (H) as well as the technological effort level (R/Q) for the three sectors.

12.7 Conclusions This chapter has presented a theoretical model that relates the growth of labour productivity with R&D spending. The model allows estimating the rate of return of technological capital from R&D spending flows, without the need to build the stock of research capital. The theoretical model is specified using an econometric model of delays distributed in time. The empirical estimates are based on cross section data. Statistical information was taken from SIMBAD. In particular, economic censuses of 1994 and 2004 were used. Statistical information is for three economic sectors – manufacturing, commerce and services – and two groups of municipalities – urban and rural areas – according to population density. Regarding econometric model estimation, the method of weighted least squares was used. The empirical results achieved are consistent with theoretical expectations, indicating that the investment in R&D by the three economic sectors surveyed has a positive effect and in most cases statistically significant. At first glance, it appears that the contribution of the factors over the growth rate of output is high for urban municipalities; in consequence, it should encourage private and public investment in each of the production factors, especially in R&D. Nevertheless, in practice it is found that this does not happen, since the share of investment in this area is lowered through time. This might be because of the high risk associated with R&D projects and the difficulty in obtaining full benefits from innovation. This can discourage firms about engaging in such activities, despite the high gains expected. In addition, firms with investment intentions could find remarkable problems in financing their investments in R&D. More so if they have already had financing problems, especially for small and medium enterprises. On the other hand, we have also found a significant positive relationship between output growth and the capital–labour ratio, and between growth and production capacity of economic sectors. This indicates that productivity changes are factors associated with the long and short terms. Finally, the production function for the three sectors shows diminishing returns to scale for capital and labour, which is consistent with results from previous studies. However, the relationship between productivity growth and technological effort does not change very significantly when constant returns to scale condition are imposed. The estimated model may be affected by econometric limitations. Griliches and Mairesse (1995) suggest that an estimated production function from microdata presents some problems that also arise when the production function is transformed into logarithms. Accordingly, the empirical results obtained in this

262

J.C. Trejo-García et al.

work, although they are quite reasonable, must be contemplated with caution. Nevertheless, studies of this type are quite useful.

Notes 1 This procedure requires making assumptions about the value of the rate of depreciation of capital and taking an initial value of this capital 2 We use investment in R&D, and the estimate will focus on determining the rate of return of that capital, instead of its elasticity. Studies that attempt to estimate the rate of return on R&D expenditure from individual companies are common in other countries 3 g is the production elasticity related to R&D capital. It is given by:

References Bessen, J. (2000): ‘Adoption costs and the rate of return to research and development’, Working paper 1/00, Research on Innovation, Wallingford, PA. Online, available at: www.researchoninnovation.org (accessed 30 February 2009). Clark, B. and Griliches, Z. (1984): ‘Productivity growth and R&D at the business level: Results of the PIMS data base’, in Griliches, Z. (ed.): Patents and productivity, University of Chicago Press, Chicago, pp. 393–416. Denison, E.F. (1974): Accounting for United States economic growth 1929–1969, Brookings Institution, Washington, DC. Fecher, F. (1990): ‘Effects directs et indirects de la R&D sur la productivité: une analyse de l’industrie manufacturière belge’, Cahiers Économiques de Bruxelles, vol. 128, pp. 459–483. Griliches, Z. (1979): ‘Issues in assessing the contribution of research and development to productivity growth’, Bell Journal of Economics, vol. 10, pp. 92–116. Griliches, Z. and Mairesse, J. (1983): ‘Comparing productivity growth: An exploration of French and U.S. industrial and firm data’, European Economic Review, vol. 21, pp. 89–119. Griliches, Z. and Mairesse, J. (1995): ‘Production functions: The economic search for identification’, documento de trabajo nº 5067, NBER, Cambridge, MA. Hall, B.H. and Mairesse, J. (1995): ‘Exploring the relationship between R&D and productivity in French manufacturing firms’, Journal of Econometrics, vol. 65, pp. 263–293. Jasso, J. (1998): ‘Industrial organization, productivity and strategies business in Mexico’, Working Paper 162, Division of Economics, CIDE. Lichtenberg, F.R. and Siegel, D. (1991): ‘The impact of R&D investment on productivity: New evidence using linked R&D-LRD data’, Economic Inquiry, vol. 29, pp. 203–228. Lucas, R.E. Jr (1988): ‘On the mechanics of economic development’, Journal of Monetary Economics, vol. 22, pp. 3–42. Mankiw, G., Romer, D. and Weil, D. (1992): ‘A contribution to the empirics of economic growth’, Quarterly Journal of Economics, vol. 108, pp. 407–437. Odagiri, H. and Iwata, H. (1986): ‘The impact of R&D on productivity increase in Japanese manufacturing companies’, Research Policy, vol. 15, pp. 13–19. Parisi, M.L., Schiantarelli, F. and Sembenelli, A. (2002): ‘Productivity, innovation

Innovation and productivity in Mexico 263 creation and absoption, and R&D: Micro evidence for Italy’, Working Paper No. 526, Department of Economics, Boston College, Chestnut Hill, MA. Romer, Paul M. (1986): ‘Increasing returns and long run growth’, Journal of Political Economy, vol. 94, pp. 1002–37. Solow, R.M. (1957): ‘Technical change and the aggregate production function’, Review of Economics and Statistics, vol. 57, pp. 312–320. Unger, K. (1996): ‘International competitiveness and technological development: The Mexican manufacturing industry against trade liberalization’, Mexican Economy, New Era, Mexico, vol. V, no. 2. Wakelin, K. (2001): ‘Productivity growth and R&D expenditure in UK manufacturing firms’, Research Policy, vol. 30, pp. 1079–1090.

13 The robustness of Okun’s law – evidence from Mexico A quarterly validation, 1985.1–2006.4

1

Eduardo Loría and Leobardo de Jesús

13.1 Introduction In 1962 Arthur Okun found a statistical regularity of great relevance for the United States’ economy (1947.2–1960.4), which claimed that for each percentage point of reduction in the unemployment rate, the real GDP would grow 3.3 per cent (Okun, 1962); and inversely, for each percentage point of increase in output, unemployment would vary –0.3 points. This regularity is commonly known as 3:1, and also known as the Okun law, and since then has become a concept of great importance in modern macroeconomics. It is the consequence of relating output increase to the unemployment rate in a bi-directional way, coming out from three specifications: first differences, output gap and fitted trend and elasticity. Okun’s contribution enriches the modern macroeconomic analysis because (a) it allows to know variation of the unemployment variation in the long run, determined by structural factors such as demographics, institutions and technology, (b) it provides a proxy of the natural rate of unemployment and (c) it identifies that long run economic growth is the main factor that counteracts the reduction in employment creation capacity (see Loría and Ramos, 2007). Okun’s discovery has great importance because of its explicative capacity of economic development: The failure to use one year’s potential fully can influence future potential GNP: to the extent that low utilisation rates and accompanying low profits and personal incomes hold down investment in plant, equipment, research, housing, and education, the growth of potential GNP will be retarded. (Okun, 1962: 2) From the available literature, we found that since this contribution was made, several authors have estimated different variations of this law.2 In spite of the relevance of the subject and that in Mexico the problem of slow growth started since the early 1980s, it is surprising that we only found three references for the Mexican economy: Chavarín (2001), González (2002) and Loría & Ramos (2007). Our main purpose is to estimate the three Okun models (1962) for the Mexican economy using quarterly data (1985.1–2006.4), in order to prove that

Okun’s law and Mexico 265 unemployment constrains the long run growth, and compare our results with those obtained with annual series (1970–2004) by Loría & Ramos (2007). Therefore, we corroborate that in Mexico the Okun law is validated for data of different periodicity and length. Our results indicate that there is a bi-directional causality relationship between the unemployment rate and output growth – in its three variants – and that Okun’s coefficient is found in the interval of 2.3–2.5, which coincides with Loría and Ramos (2007). In the second section we present the original Okun results. Next, we analyse the statistical properties of quarterly data that we use and point out the differences with those used by Loría and Ramos. In the fourth section we estimate the three Okun models with the methodology of structural time series models, using the Kalman filter, and contrast them with the results obtained by those authors. Additionally, we confirm the bi-directional causality in Okun’s equations and test cointegration for model 3. Finally, we recover the main conclusion and outline some policy recommendations.

13.2 The Okun models Okun (1962) used three different econometric specifications to prove that there was a robust bi-directional statistical relationship between unemployment and economic growth for the economy of the United States (1947.2–1960.4), which are presented in Table 13.1. Okun’s empirical conclusion coming out from the first two estimations is that in the long run, unemployment reduction has a more than proportional effect on the dynamics of GDP (1/b2).3

13.3 Mexico: output and unemployment data, 1985.1–2006.4 We use quarterly GDP (Y) data (1993 prices) and the general rate of open unemployment (U) reported by INEGI (2007a, b), both without seasonal adjustments. The latter variable differs substantially from the macroeconomic unemployment Table13.1 Okun models Model

Estimation

First differences (1) Δ Ut = β1 + β2 yt + Єt Output gap (2) Ut = β1 + β2 Ytb + Єt Fitted trend and elasticity

Okun coefficient b2

1/b2

Δ Ut = 0.3 – 0.3 yt

0.3

3.3

Ut = 3.72 + 0.36 Ytb

0.36

2.8

In Et = 212 + 0.4 In Yt – 0.32t

0.4–0.35

2.5–2.8

(3) In Et = β1 + β2 In Yt +β3 + Єt t

Source: own estimations based on INEGI.

266

E. Loría and L. de Jesús

rate (MUR)4 of Loría and Ramos (2007), estimated for annual data (1970–2004). As a matter of example, towards 1970 the MUR was 2.03 per cent and since 1982 it began to grow until it reached 15.8 per cent in 1988, and 28.3 per cent towards the year 2004 (see Figure 13.1); while the general unemployment rate used here is stationary and currently has ranged from 2 to 5 per cent, although it reached high figures in 1995 and 1996. See Figure 13.2.

13.4 Analysis and discussion According to our main purpose, we estimated the three Okun models inversely, thus solving a serious econometric bias problem detected by Barreto and Howland (1993) in Okun’s seminal article. This problem consists in estimating the current regression and afterwards solving arithmetically for the exogenous, just by doing algebra. Therefore, it does not matter regressing U on Y or the other way around. By doing this Okun claims that it is possible to find economic sense in both directions. This procedure has been followed by many authors. Accordingly, when passing directly in estimations (1) and (2) from b2 to 1/b2 Okun was able to explain – at the same statistical level – either economic growth or unemployment. Nevertheless, in the original Okun’s models (1 and 2) there are two variables and the reading must be made as usual (from the right hand side to the left hand side), and the fact of reading inversely is not only related to the causality sense coming out from economic theory, but also – and not less important – has to do with the properties of a joint distribution function, which refers a conditional specification of random variables of the kind:

Figure 13.1 Mexico: macroeconomic rate of unemployment, 1970–2004 (source: Loría and Ramos (2007)).

Okun’s law and Mexico 267

Figure 13.2 GDP, unemployment, output gap and employment rate, 1985.1–2006.4 (source: INEGI).

Barreto and Howland outline that the correct specification depends on the specific question of interest. This task determines the regression direction: Thus Okun’s procedure [make the bi-directional reading as of b2, our aggregate] makes sense only if the underlying structure in the model is assumed to be stable, i.e., if the parameters of the model do not change between the sample period and the date on which the GNP gap is to be predicted. If any of the structural parameters have changed in the intervening time, then the sample relationship will produce biased estimates of the GNP gap. (1993: 4) Thus, in order to avoid the possible problem of referred bias and since our main purpose is to prove specifically that unemployment restricts economic growth, we choose the direct estimation for the three Okun estimations. That is, we proceeded by the inverse specification to that of Okun’s in the following way: Y = f(U), thus the reading is direct in terms of our hypothesis. Likewise in Loría and Ramos (2007), we estimated the three equations through the methodology of structural

268

E. Loría and L. de Jesús

time series models, using the Kalman filter (Kalman, 1960). See results in Table 13.a.2 and the Appendix to this chapter). One advantage of this procedure is that the estimated parameter ‘mt captures the long run movements of the series involved as well as the effects that b2 can’t explain’ (Loría and Ramos, 2007: 29). Empirical evidence reports that also for this data structure and with the inverse specification of Okun, this law was accomplished in Mexico. The value of the coefficients is congruent with the structure of the Mexican economy: labour intensive and low productivity. Model 3 depicts two results with high economic meaning. On the one hand, mt indicates the actual rate of potential output: 2.6 per cent, similar to the figure reported by Loría et al. (2008), 2.5 per cent for the span time 1980.1–2006.4. Likewise, from the parameter of the employment rate (E) we can calculate the output elasticity to employment (1/2.5892 = 0.386), that even with the anticipated methodological warnings, we can take it safely since it is congruent with the results obtained by Loría and Ramos and other applied works such as Loría (2006) and Hernández (1998) (see Table 13.a.2). Tests of unit roots (see Table 13.a.1 in the Appendix to this chapter) indicate that the GPD logarithm (ln y) and that of the employment rate (ln E) are series I(1), while the unemployment rate (U), GDP growth (y) and output gap (Yb) are I(0).5

13.5 Conclusion We empirically corroborated that for quarterly data and with the use of structural time series models (using the Kalman filter), Okun’s law applies in Mexico, and the coefficient varies in the interval 2.35–2.58, which is congruent with what Loría and Ramos (2007) estimated; and, furthermore, that these magnitudes are adequate for an economy that suffers from high structural unemployment and low productivity. Likewise, causality tests in the Granger sense run in a bidirectional way between unemployment and output. In order to avoid possible biases in the estimation of the slopes of the three models, we used inverse specifications to that of Okun’s, obtaining a direct Table 13.2 Mexico: Okun estimations Estimation

Average

Quarterly, 1985.1–2006.4

Annual, 1970–2004 (Loría and Quarterly Annual Ramos, 2007)

(1) yt = 1.1041 µt – 2.3538 ΔUt

(1) ΔUt = 2.349 µt – 0.403 yt

(2) YB = 9.5866 µt + 2.5383 Ut

(2) Ut = 14.65 µt + 0.456 YtB

2.49

2.25

(3) In Yt = 2.6115 µt + 2.5892 In Et (3) In Et = 0.481 In Yt – 2.661 µt Source: own estimations based on INEGI. Note All the models were estimated with GiveWin 2.3, module STAMP 6.0 (Koopman et al. 2000).

Okun’s law and Mexico 269 reading of the Okun coefficient and can prove in a reliable manner that unemployment constrains economic growth. This evidence supports the results of other authors for different series and periods, which allows us to use it as a good instrument of analysis and forecasting of the economic cycle; it also allows us to estimate the sacrifice rate of long run unemployment; and moreover, shows that economic policy must focus by all means on avoiding fluctuations in growth and at the same time reducing unemployment, because this way it will stimulate growth in the long run.

Appendix Table 13.a.1 Basic statistics and unit roots, 1985.1–2006.4 Mean 3.5046 Median 3.35 Std Dev. 1.0273 Skewness 1.3711 Kurtosis 5.2431 Jarque46.025 Bera 0 ADF –2.692 DF-GLS –1.7765 PP –3.785 KPSS 1.235

14.1067 14.0745 0.1899 0.044 1.7397 5.851

0.0067 0.7847 –0.0015 –0.1517 0.046 4.648 0.1151 0.1706 1.8257 1.8322 5.191 5.365

–0.053 –0.075 3.9131 –4.036* –2.618 –3.640** 3.6918 –34.907*** 202.862 0.155

4.5694 4.571 0.0107 –1.4092 5.383 49.953

–0.068 0 –3.1532 0.1003 –3.2306 –2.513 –4.7289 3.802 0.43211 1.214

0.00019 0.051 0.00055 0.0697 0.00532 2.8031 –0.3741 –0.3777 3.6816 3.2045 3.714 2.246 –0.156 –4.287 –2.258 –11.704 0.044

–0.325 –7.0484 –6.024 –16.3251 0.173

Source: own estimations based on INEGI. Notes Tests are non-significant at levels. ADF with four lags and intercept are valid at 90%; DF-GLS with four lags, trend and intercept; PP with four lags and intercept, valid at 99%. KPSS with four lags and intercept. * With intercept only; ** with 4 lags and intercept; *** with intercept; three lags

Table 13.a.2 Granger Causality Test, 1985.1–2006.4 for an unrestricted VAR(5) VAR model

Ho: not causality

x2 (5)

Probability

1

Δ Ut does not cause yt yt does not cause ΔUt

0.0007 0.0000

2

Ut does not cause YtB YtB does not cause Δ Ut

3

In Et does not cause In Yt In Yt does not cause In Et

21.39 28.42 x2 (5) 20.92 43.88 x2 (5) 27.79 17.47

0.0008 0.0000 0.0000 0.0037

Source: own estimations based on INEGI. Notes Models 1 and 3 include a GDP shock dummy and an inflation dummy for 1986.3 and 1995.2.

E. Loría and L. de Jesús

270

Figure 13.a.1 Model 1 first differences. Table 13.a.3 Mexico: Okun’s law, 1985.1–2006.4 R2 N DW r Q H

–1 –8 (8,6) –34

yt = 1.1041 µt – 2.354 ΔUt 0.882962 3.2632 (0.1956) 2.3062 –0.21520 (0.8296) 0.066376 (0.9471) 21.292 (0.0016) 0.74192 (0.8058)

Source: own estimations based on INEGI.

Figure 13.a.2 Model 1: first differences (source: estimates with information from INEGI). Note TGDPT: GDP growth.

Okun’s law and Mexico 271

Figure 13.a.3 Diagnostic tests: correlogram, density, QQplot, cusum residual (source: estimates with information from INEGI).

Figure 13.a.4 Model 2: output gap.

Table 13.a.4 Model 2 R N DW r 2

Q H

–1 –8 (8,6) –27

YB = 9.5866 µt + 2.5383 Ut 0.705865 4.2233 (0.1210) 1.749 0.086169 (0.9313) 0.076475 (0.9390) 10.342 (0.1110) 1.0967 (0.4061)

Source: own estimations based on INEGI.

Figure 13.a.5 Model 2: output gap (source: estimates with information from INEGI).

Figure 13.a.6 Diagnostic tests: correlogram, density, QQplot, cusum residual (source: estimates with information from INEGI).

Okun’s law and Mexico 273

Figure 13.a.7 Model 3: fitted trend and elasticity.

Table 13.a.5 Model 3 R2 N DW r Q H

–1 –8 (8,6) –34

In Yt = 2.6115 µt + 2.5892 In Et 0.993693 2.7109 (0.2578) 1.749 0.085145 (0.9321) 0.10788 (0.9141) 10.480 (0.1058) 1.1143 (0.3771)

Source: own estimations based on INEGI.

Figure 13.a.8 Model 3: fitted trend and elasticity (source: estimates with information from INEGI). Note LGDP: GDP logarithm.

274

E. Loría and L. de Jesús

Figure 13.a.9 Diagnostic tests: correlogram, density, QQplot, cusum residual (source: estimates with information from INEGI). Note LGDP: GDP logarithm.

Notes 1 We thank technical assistance from Manuel García R. and Jorge Ramírez, and comments from Armando Sánchez. As usual, the responsibility is only ours. 2 Upon reviewing the literature, available works can be classified in two main categories: (a) theoretical–empirical studies, in the sense that they review and discuss the estimation methods of Okun’s law (in this respect see: Barreto and Howland, 1993; Altig et al., 1997; Attfield and Silverstone, 1998; Sögner and Stiassny, 2000; Harris and Silverstone, 2001; Crespo, 2003; Friedman and Michael, 1974; Lang and de Peretti, 2002; Prachowny, 1993; Weber, 1995; Schorderet, 2001; Knoester, 1986; Paldam, 1987), and (b) empirical studies, whose main purpose is to estimate the Okun’s coefficients for some countries, even at the level of states or regions, in order to know the existing interrelations between different countries by identifying reciprocities between unemployment and output (see: Abril et al., 1996; Adanu, 2002; Arias et al., 2002; Garavito, 2002; Lemois, 2003; Murillo and Usabiaga, 2002; Freeman, 2001; Lee, 2001; Moosa, 1997 and Schnabel, 2002). 3 In Loría and Ramos (op. cit.) the economic implications of the results depicted in this table were carefully analysed. In addition, it is worth mentioning that (2) established the natural rate of unemployment (3.72 per cent) and (3) the output elasticity to employment. 4

Okun’s law and Mexico 275 where PEA = economically active population (Conapo, 2006), PO = employees (millions of persons) in the formal sector (INEGI, 2007a). 5 This way, the problem of spurious regression could only exist in model 3 and we followed the Johansen procedure (1988) to discard it. Accordingly, with a confidence level of 99 per cent we obtained a cointegrating vector with economic sense; statistic trace 27.78 (24.6), adjustment coefficient –0.6275 (standard error: 0.126).

References Abril, J.C., H.D. Ferullo and A. Gaínza (1996). ‘Estimación de la relación de Okun: Argentina 1980–1996’. Facultad de Ciencias Económicas, Universidad Nacional de Tucumán, Argentina. Adanu, K. (2002). ‘A Cross-Province Comparison of Okun’s Coefficient for Canada’. Department of Economics of Victoria, BC, Canada. Altig, D., T. Fitzgerald and P. Rupert (1997). ‘Okun’s Law Revisited: Should we Worry about Low Unemployment?’ Economic Commentary. Federal Reserve Bank of Cleveland. Arias, E., A.C. Kikut and J. Madrigal (2002). ‘Estimación de la Ley de Okun para Costa Rica’. Departamento de Investigación Económica, Banco Central de Costa Rica. Attfield, C. and B. Silverstone (1998). ‘Okun’s Law, Cointegration and Gap Variables’, Journal of Macroeconomics. Vol. 20, No. 3. Barreto, H. and F. Howland (1993). ‘There are Two Okun’s Law Relationships between Output and Unemployment’. Wabash College, Crawfordsville, USA. Chavarín, R. (2001). ‘El costo del desempleo medido en producto: Una revisión empírica de la Ley de Okun para México’, El Trimestre Económico. Vol. LXVIII, No. 270, FCE, México. Conapo (2006). Indicadores demográficos básicos. Consejo Nacional de Población. Online, available at: www.conapo.gob.mx/00cifras/00indicadores.htm (accessed September 2007). Crespo, J. (2003). ‘Okun’s Law Revisited’, Oxford Bulletin of Economics and Statistics. Vol. 65, No. 4. Freeman, D.G. (2001). ‘Panel Tests of Okun’s Law for Ten Industrial Countries’, Economic Inquiry. Vol. 39, No. 4. Friedman, B. and M.L. Wachter (1974). ‘Unemployment: Okun’s Law, Labor Force, and Productivity’, Review of Economics and Statistics. Vol. 56, No. 2. Garavito, C. (2002). ‘La Ley de Okun en el Perú: 1970–2000’. Pontificia Universidad Católica de Perú. González, J.A. (2002). ‘Labor Flexibility in Thirteen Latin American Countries and the United States: Revisiting and Expanding Okun Coefficients’, DREDPR. Working Paper No. 136. Stanford University. Granger, C.W.J. (1969). ‘Investigating Causal Relations by Econometric Models and Cross-Spectral Methods’, Econometrica. Vol. 37, No. 3. Harris, R. and B. Silverstone (2001). ‘Testing for Asymmetry in Okun’s Law: A CrossCountry Comparison’, Economics Bulletin. Vol. 5, No. 2. Hernández, E. (1998). ‘Apertura comercial, productividad, empleo y contratos de trabajo en México’, in V. Tokman and D. Martínez (ed.). Productividad y empleo en la apertura económica. OIT. Lima, Perú. INEGI (2007a). ‘Encuesta Nacional de Ocupación y Empleo’. Indicadores Económicos de Coyuntura. Instituto Nacional de Estadística, Geografía e Informática, México. Online,

276

E. Loría and L. de Jesús

available at: http://dgcnesyp.inegi.gob.mx/cgi-win/bdieintsi.exe/NIVA05#ARBOL (accessed May 2007). INEGI (2007b). ‘Sistema de Cuentas Nacionales de México’. Indicadores Económicos de Coyuntura. Instituto Nacional de Estadística, Geografía e Informática, México. Online, available at: http://dgcnesyp.inegi.gob.mx/cgi-win/bdieintsi.exe/NIVA050010#ARBOL (accessed May 2007). Johansen, S. (1988). ‘Statistical Analysis of Cointegrating Vectors’, Journal of Economic Dynamic and Control. Vol. 12, No. 2–3. Kalman, R. (1960). ‘A New Approach to Linear Filtering and Prediction Problems’, Journal of Basic Engineering. Vol. 82. Knoester, A. (1986). ‘Okun’s Law Revisited’, Weltwirtschaftliches Archiv. Vol. 122, No. 2. Koopman, S.J., A.C. Harvey, J.A. Doornik and N. Shephard (2000). ‘Structural Time Series Analyser, Modeller and Predictor’. Timberlake Consultants. Lang, D. and C. de Peretti (2002). ‘A Strong Hysteretic model for Okun’s Law: Theory and Preliminary investigation’. Université de la Méditerranée, France. Lee, J. (2001). ‘The Robustness of Okun’s Law: Evidence from OECD Countries’, Journal of Macroeconomics. Vol. 22, No. 20. Lemois, F.A. (2003). ‘Estimaciones de la Ley de Okun para Puerto Rico’. Junta de Planificación, Oficina de la Gobernadora, Estado Libre Asociado de Puerto Rico. Loría, E. (2006). Eudoxio: modelo macroeconométrico de la economía mexicana. UNAM. México. Loría, E. and M. Ramos (2007). ‘La ley de Okun. Una relectura para México, 1970–2004’, Estudios Económicos. Vol. 22, No. 1. Enero-Junio. El Colegio de México, México. Loría, E., M.G. Ramos and L. de Jesús (2008). ‘Producto potencial y ciclos económicos en México, 1980.1–2006.4’, Estudios Económicos. Vol. 23, No. 1. Enero-junio. El Colegio de México. México. Moosa, I.A. (1997). ‘A Cross-Country Comparison of Okun’s Law Coefficient’, Journal of Comparative Economics. Vol. 24, No. 3. Murillo, I.P. and C. Usabiaga (2002). ‘Estimaciones de la tasa de paro de equilibrio de la economía española a partir de la Ley de Okun’. Universidad de Extremadura y Universidad Pablo de Olavide, España. Okun, A. (1962). ‘Potential GNP: Its Measurement and Significance’, in J. Pechman (ed.) (1983), Economics for Policymaking. MIT Press. Cambridge, MA. Paldam, M. (1987). ‘How Much does One Percent of Growth Change the Unemployment Rate?’ European Economic Review. Vol. 31, No. 1–2. Prachowny, M. (1993). ‘Okun’s Law: Theoretical Foundations and Revised Estimates’, Review of Economics and Statistics. Vol. 75, No. 2. Schnabel, G. (2002). ‘Output Trends and Okun’s Law’, BIS Working Papers. Monetary and Economic Department, Bank for International Settlements, 111. Sögner, L. and A. Stiassny (2000). ‘A Cross-Country Study on Okun’s Law’, Growth and Employment in Europe: Sustainability and Competitiveness. University of Economics and Business Administration, Vienna. Schorderet, Y. (2001). ‘Revisiting Okun Law’s: An Hysteretic Perspective’, Discussion Papers. Department of Economics, University of California, San Diego. Weber, Christian E, (1995). ‘Cyclical Output, Cyclical Unemployment, and Okun’s Coefficient: A New Approach’, Journal of Applied Econometrics. Vol. 10, No. 4.

Index

Note: Page numbers in italics denote tables, those in bold denote figures. adverse effects 199 Andean Community (CAN) 111 development 113, 115–116 membership and representation 116–117 membership evolution/type of organisation 112 see also gravity model; regional integration agreements (RIAs) Andean Pact see Andean Community (CAN) Anderson, J.E. 119 Arbache, J.S. 198, 199, 205 Argentina business cycle 136 context of decline 133–134 debt 142–143 debt sustainability indicators 143 decomposition of elements of demand 138 development gap 41, 42, 44 economic performance 135–137 exports 143 external constraints 138–141 fiscal policy 140 foreign direct investment (FDI) 141–144 foreign exchange rate 139, 140 GDP 32 Gini coefficient 144 gross investment and savings 45 gross investment coefficient 44–45 growth and productivity 135 income elasticity of imports 48 income per capita 133 industrialisation 141–142 inequality 144–145 inflation rate 35 investment composition 142

ISI 136–137, 139, 142 manufacturing exports 143 model of post-war development 141 monetary policy 141 output decomposition 139 Peronism 136 public capital 142 real exchange rate 140 ‘Rodrigazo’ 137 role of exports 138–139 role of state 141–144 stabilisation 137 summary and conclusions 145–146 trade balance–GDP ratio 39 wage share 145 automatic balancing mechanism 45 balance of payments 17 balance of trade 17 Barreto, H. 267 Barro, R. 12 Bergstrand, J.H. 119, 155 Bertola, G. 16 Bessen, J. 251 Bhagwati, J. 7, 9 bias, in favour of rich countries 21 Bidlingmaier, T. 68 Bolivia, exports 117 Bougheas , S. 127 Bourguignon, F. 13 Brady Plan 233 Brazil devaluation 103 development gap 41, 42, 44 FDI inflows and exchange rate 104 GDP 32 gross investment and savings 46

278

Index

Brazil continued gross investment coefficient 45 income elasticity of imports 48 inflation rate 35 inward FDI 90–103 inward FDI by regions 99–100 inward FDI flows 97 inward FDI stock 97 sectoral changes to inward FDI 96–97 trade balance–GDP ratio 39 see also Mercosur; regional integration agreements (RIAs) Bretton Woods, effect of collapse 139–140 Bucheli, M. 171 business and finance income (B&F) 210–212 capital account liberalisation 48 capital flows, changed structure 89 capital goods, differentiated 61 capital market, liberalisation 45 ‘chance’ investments 21 Chen, S. 10 Chile bilateral agreements 150, 165 data and variables 159–161 development gap 41–42, 43 development strategy 151 economic growth 31 exports 150–151 FDI inflows 150–155, 151 FDI inflows by origin 154, 155 FDI inflows by sector 153 GDP 33 gross investment and savings 46 gross investment coefficient 45 income elasticity of imports 49 inflation rate 36 methodology 155–159 national economy, sectors 150 random effects estimation of the augmented equation 164 random effects estimation of the baseline equation 162 research study 155–165 results 161–165 sources of FDI 154–155 summary and conclusions 165 summary statistics 161 trade balance–GDP ratio 40 variables 160 Chilean Foreign Investment Committee 159, 161

China development gap 44, 44 economic growth 31 GDP 34 gross investment and savings 47 gross investment coefficient 45 income elasticity of imports 49 inflation rate 37 poverty 11 price stability 37–38 trade balance–GDP ratio 41 Clark, B. 251 classical economics 26–27 classical international trade theory 60 Cobb–Douglas production function 252 Colombia, exports 117 common external tariff (CET) Andean Community (CAN) 115–116 Mercosur 112, 113 comparative advantage 52 consumer’s decision problem 223–226 consumption dynamics 228–230 Convertibility 137, 141 Convertibility Plan 139 countercyclical policy 29 crisis-proofing 29–30 current account balance 17–18 Cyrus, T.L. 121 Deardoff, A.V. 120 debt and exchange rate depreciations 244 Mexico 240, 240 debt ratio 247 decline of labour market institutions argument 215 Denison, E.F. 255 devaluation, Brazil 103 development gap Argentina 41, 42, 44 Brazil 41, 42, 44 Chile 41–42, 43 China 44, 44 drivers 44–45 evolution of 38–50 income elasticity of demand for imports 48–49 Mexico 41, 43, 44 development, trade strategy for 20–23 developmental industrial policy 50–54 Developmentist State 52 Direction of Trade, IMF data set 159 Doha trade negotiations 20–21, 168 Dollar, D. 12, 20

Index Dow Jones Index composite average and FDI inflows 103 Dowrick, S. 13, 15 Dunning, J.H. 83 East Asia, inward FDI 88 economic growth 31 calculation 220 and trade liberalisation 7 Economic Partnership Agreements (EPAs) 21–22 economy, main characteristics 221–223 Ecuador, exports 118 education changing levels with changing income 202 and income 201, 216 and income bill share per sector 203 and income by sector 203 and income distribution 200–202 education investment, and human capital 63 Edwards, S. 20 Emmert, C.F. 88 European Union Economic Partnership Agreements (EPAs) 21–22 and Mercosur 113 exchange rate-based stabilisation 34 exchange rate depreciations, and debt 244 exchange rate fluctuations, pass-through effect 37 export performance 15 exports Argentina 114, 138–139, 143 Bolivia 117 Brazil 114 Chile 150–151 Colombia 117 as drivers of growth 19 Ecuador 118 Paraguay 115 Peru 118 Uruguay 115 Venezuela 119 extended neoclassical growth model countries in sample 68 data 67–69 derivation 62–66 empirical application 66–69 panel data regressions of per capita income 71, 76–77 results 69 summary and conclusions 75 extra-regional effects 85, 128

279

Faini, R. 16 Fecher, F. 251 Ffrench-Davis, R. 44 financial fragility hypothesis 233–237 see also Mexico: financial fragility financial liberalisation 48 firm size 142 firms, typology of growth rate 236 flexibility, labour market 30 Foley, D. 236 foreign debt crisis, 1982 38, 137 as catalyst for change 28 Chilean response 44 effects in Mexico 195 foreign direct investment as percentage of GDP 92 foreign direct investment (FDI) 61 Argentina 141–144 determinants 84 growth 149 inward stock as percentage of GDP 94 inward stocks 93 and regional integration agreements (RIAs) 82–89 see also inward FDI foreign direct investment, net inflows Mexico, Brazil and middle income countries 91 Mexico, Brazil and rest of world 90 foreign exchange rate Argentina 139 and stabilisation 139, 140, 141 Frankel, J. 121 free-market liberalism 220 see also Mexico: free-market liberalism study free trade, assumptions underlying 8 GDP Argentina 32 Brazil 32 Chile 33 China 34 Mexico 33 Gini coefficient Argentina 144, 144 decomposition by income source 210–212, 211, 213 for global inequality 13 international and global comparison 14 global inequality 13 global price variations, simulations 187–189

280

Index

globalisation, and income distribution 195–196 Goldberg, P. 11 Goldin, C. 199, 216 Golley, J. 13, 15 Gordon, J. 198 governments, role in reform 30 Gradin, C. 171 graphic evolution of gravity equation fixed effects for Mercosur and CAN 125 gravity equation for the panels of Mercosur and CAN: fixed effects 124 gravity equation for the panels of Mercosur and CAN: random effects 123 gravity model 110–112, 118–121, 155 distance as the only regressor in the gravity equation 127 exogeneity Hausman Test for distance in the gravity equation 126 graphic evolution of gravity equation fixed effects for Mercosur and CAN 125 gravity equation for the panels of Mercosur and CAN: fixed effects 124 gravity equation for the panels of Mercosur and CAN: random effects 123 hypotheses 119 method 122–123 research study 121–127 results 123–125 robustness and specification 126–127 summary and conclusions 127–128 theoretical support 119–120 Great Britain 10 Greenaway, D. 20 Griliches, Z. 250, 251, 252, 261 gross investment and savings Argentina 45 Brazil 46 Chile 46 China 47 Mexico 47 gross investment coefficient 44–45 Argentina 44–45 Brazil 45 Chile 45 China 45 Mexico 45 growth and the balance of payments, trade off 18–19, 18

growth models 9 growth performance 19–20 growth rate of firms, typology 236 Gupta, P. 198 Hall, B.H. 251–252 Hamilton, A. 10 Hausman Exogeneity Test 126 Hausmann, R. 21 Heckscher–Ohlin theorem 8, 11, 27, 121, 171–172 hedge financing 244, 247, 248 hegemonic tie 135 Helpman, E. 119, 121 household composition 216 Howland, F. 267 human capital accumulation 60–61, 63–64 human capital, and income 76–77 IADB 88 import growth 16–17 import-substitution industrialisation (ISI) Argentina 136–137, 139, 142 Mexico 195 Impossible Trinity proposition 34 income bill share by education and sector 203 change with changing education levels 202 and education 201 and education by sector 203 income distribution 13 additional forms 210–212 Argentina 144–145 as cyclical 215–216 and education 200–202 and globalisation 195–196 individual 200–205 and market openness 212–214, 214 and trade liberalisation 7 see also inequality income distribution targets 30 income elasticity of demand for imports development gap 48–49 effects of trade liberalisation 16–17 income elasticity of imports Argentina 48 Brazil 48 Chile 49 China 49 Mexico 49 income inequality within countries 11–12 Uruguay 185

Index industrialisation, Argentina 141–142 inequality Argentina 144–145 driving factors 215 effects of liberalisation 195–196 household vs. individual 210 international and global 13–15 Mexico see Mexico: inequality study Uruguay 171, 184, 185 see also income distribution infant country protection 22 inflation 31–38 inflation rate Argentina 35 Brazil 35 Chile 36 China 37 Mexico 36 innovation encouraging 21 Schumpeterian approach 50–51 institutions, building 30 international capital markets 34, 37 international inequality 13 international trade, theoretical aspects 60–62 Inverse Hyperbolic Sine Function 157 inward FDI Brazil, by regions 99–100, 99–100 changes by home country/region 97–100 changes by sector 95–97 changes in inflows and FDI stock 90–95 Chile 150–155, 151 Chile, by origin 154, 155 Chile, by sector 153 and Dow Jones Index composite average 103 empirical evaluation 90–103 flows, Brazil 97 flows, Mexico 96 and labour costs 103 Mexico, by regions 98–99, 98 possible explanations for increase 88–89, 100–103 stock, Brazil 97 stock, Mexico 95 see also foreign direct investment (FDI) inward potential index 101 Iwata, H. 251 Johnson, H. 8 Kaldoor-Verdoorn’s Law 135

281

Kalecki, M. 135 Katz, L.F. 199, 216 Keynes, J.M. 8, 54, 233 Kim, Y.-H. 87 knowledge, diffusion 61 knowledge spillover 63, 64 Kraay, A. 12, 20 Krugman, P. 119, 121 Krugman’s trade theory 9 Kuczynski, P.P. 29–30, 51 labour costs, and inward FDI 103 labour market, flexibility 30 law of comparative advantage 8 learning-by-doing, and human capital 64, 74, 78 Levy Yeyati, E. 88 liberalising reforms 30 Lichtenberg, F.R. 251 Lomé Convention 22 Loría, E. 265, 267–268 lost decade 29, 137 Lucas, R. 60–61, 255 macroeconomic performance, before and after structural reform 30–38 Mah, J.S. 16 Mairesse, J. 251–252, 261 Mandelson, P. 11 manufacturing exports, Argentina 143 market-incentive mechanisms 30 market openness and income distribution 212–214, 214 market-seeking FDI 85 Martinez-Zarzoso, I. 127 McKinnon’s theory 27 Medvedev, D. 83 Melo, O. 16 Mercosur 87–88, 91, 92, 93, 96–97, 104, 105 establishment 110–111 impact on trade flows 112–113 membership and representation 112–113 membership evolution/type of organisation 112 as model of integration 113 price changes, Uruguay 179 tariffs 176 Uruguay 170 see also gravity model; regional integration agreements (RIAs) mergers and acquisitions 90, 91, 149 Mexico average hourly income per level of education 201

282

Index

Mexico continued changes in income and education levels 202 debt 240 development gap 41, 43, 44 effects of trade liberalisation 11 FDI inflows and exchange rate 104 GDP 33 GDP, unemployment, output gap and employment rate, 1985.1–2006.4 267 gross investment and savings 47 gross investment coefficient 45 income elasticity of imports 49 inflation rate 36 inward FDI 90–103 inward FDI by regions 98 inward FDI flows 96 inward FDI stock 95 macroeconomic rate of unemployment, 1970–2004 266 observed growth rate from 1930 to 2002 230 Okun estimations 268 output and unemployment data 265–266 regional integration agreements (RIAs) 86–87 sectoral changes to inward FDI 95–96 trade balance–GDP ratio 40 see also North American Free Trade Organization (NAFTA); Okun models; regional integration agreements (RIAs) Mexico: financial fragility average values for sample of firms 239 dependent variables 243, 245, 246 empirical evidence 238–248 firms’ assets according to financial structure 242 firms’ composition according to financial structure 241, 242 growth and firms’ indebtedness 240 growth rates 239 growth, returns and interest rate: aggregate levels 238 investment 239 methodology 238 net wealth, growth and debt 240 overview 233 see also financial fragility hypothesis private debt 239 probability of hedge, speculative and Ponzi finance 247 summary and conclusions 248–249

Mexico: free-market liberalism study consumption dynamics 228–230 optimal consumption shares, parameters, and estimates 231 overview 220–221 simulation exercise 230–231 summary and conclusions 232 wealth dynamics 226–228 Mexico: inequality study additional forms of income distribution 210–212 data source 196 decomposition of household Gini by income source 213 decomposition of income by economic sectors 202–205 econometric analysis 205–210 factors driving inequality 215 Gini decomposition by income source 210–212, 211 hourly income and educational attainment by sector 203 household vs. individual inequality 210 income bill share by sector and education level 203 individual income distribution 200–205 market openness and income distribution 212–214, 214 methodology 200 overview 195–197 returns to education 200–202 returns to education: labour income 209 returns to labour by sectors 205–210 sector performance: labour income 206, 207 skill demand 204–205 summary and conclusions 214–216 theoretical debate 197–199 Mexico: innovation and productivity background to study 251–252 data 250, 254 econometric model 255–256 econometric model specification 256–257 estimate for rural municipalities weighted by product 259 estimate for rural municipalities weighted by registered units 260 estimate for urban municipalities weighted by product 258 estimate for urban municipalities weighted by registered units 260 estimation and model results 257–261 methodology 251 overview 250–251

Index summary and conclusions 261–262 theoretical aspects 252–255 variables 255 Milanovic, B. 12, 13 Miles, D. 171 Mill, J.S. 9 Mincerian earning functions 205 monetary consensus model 34–35 monetary policy, Argentina 141 Morrisson, C. 13 NAFTA 86–87, 91, 93, 95, 104, 105 see also regional integration agreements (RIAs) neoclassical theory of international trade 27, 59 new growth theory 81 new regionalism 81, 82–83 North American Free Trade Organization (NAFTA) 86–87, 91, 93, 95, 104, 105 see also regional integration agreements (RIAs) North–South integration 81–82 Nowak-Lehmann, F. 127 Ocampo, J.A. 28–29 Odagiri, H. 251 Okun, A. 264, 266 Okun models 265 analysis and discussion 266–268 basic statistics and unit roots, 1985.1–2006.4 269 diagnostic tests 271, 272, 274 econometric bias 266 Granger Causality Test 269 Mexico: Okun’s law, 1985.1–2006.4 270 model 1: first differences 270 Model 1 first differences 270 model 2 271 model 2: output gap 271, 272 model 3 273 model 3: fitted trend and elasticity 273 summary and conclusions 268–269 Okun’s law 264–265 Open Regionalism 111 openness 20 Argentina 141 and income distribution 212–214 and inequality 196 orthodox trade theory problems of 8–10 and wage inequality 11–12 Oxfam 22

283

Pacheco-López, P. 16–19, 48 Parikh, A. 17 Parisi, M.L. 252 Pavcnik, N. 11 per capita income (PCY) 13, 15 Perón, Juan Domingo 135 Peru, exports 118 Pissarides, C.A. 199, 216 Ponzi finance 236, 239, 244, 247, 248 Portantiero, J.C. 135, 136 poverty within countries 10–11 effects of trade liberalisation 11 and trade liberalisation 7 Uruguay 171, 184 Uruguay, before and after reform 184, 185 pragmatism 48 price stability 31–32 China 37–38 Mexico 239 primary-commodity dependence 21 privatisation 101 privatised firms, transaction values of cross-border M&As 101 protection arguments for 8 historical context 10 R&D sector, increasing human capital productivity 61 Ramirez, M.D. 88 Ramos, M. 265, 267–268 Ravallion, M. 10, 11 regional integration agreements (RIAs) Brazil 87–88 effect on FDI inflows 93 evaluation of inward FDI 90–103 extra-regional effects 128 FDI stock 93–94 influence 89 inter-regional effects 85 intra-regional effects 84–85, 128 and inward FDI 82–89 Mexico 86–87 North–South integration 81–82 overview 81–82 possible dynamic effects 86 possible static effects 82–84 summary and conclusions 103–106 see also Andean Community (CAN); Mercosur; North American Free Trade Organization (NAFTA) regional production networks 85

284

Index

regionalisation, Mercosur countries 112–113 rents, reduction in traded sector 199, 204, 214 Ricardo, D. 8, 26–27, 60 rise of service argument 196 rise of services 197–199, 204, 215 Robbins, D.J. 198 Rodriguez, F. 20 Rodrik, D. 8, 20–21, 22, 38, 51 Rojas-Suárez, L. 37 Romer, P. 61, 255 Rossi, M. 171 rules of origin (RoO) 85–86 Sachs, J. 20 Sala-i-Martin, X. 13 Samuelson, P. 8 Santos-Paulino, A. 15–16, 17 Schiff, M. 81, 82, 87 Schumpeter’s dynamic approach 50–51, 52, 54 self discovery 21 services sector 89 Shackle, G. 45 share of cross-border M&A sales in total FDI inflows in Mexico and Brazil 102 Siegel, D. 251 Sinha, A. 198–199 skill-biased technological change (SBTC) 195–196 skill demand 204–205 skill-enhancing trade hypothesis (SETH) 196, 198, 199, 204, 210, 215 Smith, Adam 10, 26 Solow model 9 Solow, R.M. 250, 255 specialisation 21, 60 speculative financing 239, 244, 247, 248 Spilimbergo, A. 12 stabilisation Argentina 137 and foreign exchange rate 139, 140, 141 stabilisation policy 29–30 standard theory 197–198 Stiglitz, J. 7, 9, 10, 19, 22 Stolper-Samuelson theorem (SST) 195, 197–198 structural change 38, 44–45 structural reform, effects on macroeconomic performance 30–38 structuralist economic theory 38

‘tariff-jumping’ FDI 84, 85, 87 tariff reduction, poverty and inequality effects 189–190 tariff reduction, poverty and inequality effects see also Uruguay tariffs impact on prices of traded goods 173 Uruguay 169 see also Uruguay technological innovation, effects of 52–53 technological progress 26 technological transfer, and human capital 75 technology, and income distribution 216 temporary adverse effects 199 tequila crises 92, 93 theory of technological innovation 255 Thirlwall, A.P. 15–19, 48 Tinbergen, J. 155 trade balance 17–18, 38 trade balance–GDP ratio Argentina 39 Brazil 39 Chile 40 China 41 Mexico 40 trade–development connection 26 trade flows gravity model 110–112 impact of Mercosur 112–113 trade integration, effect on income 63 trade liberalisation effects on inequality 195–196 overview 7–8 temporary adverse effects 199 trade performance 15–19 trade reforms 168–169 see also Uruguay trade, static and dynamic gains 9 trade strategy, for development 20–23 trade-to-GDP ratio 38 traded sector, rent reduction 199, 204, 214 traditional theory of international trade 60 transaction values of cross-border M&As privatised firms 101 transatlantic consensus 196 transfer income 212 trap of literacy 74, 78 Treaty of Asunción 110–111 Trujillo Protocol 115–116 Tuman, J.P. 88 UNCTAD 18, 83, 84, 88, 89, 91, 101

Index unemployment, as grounds for protection 8 Uruguay compensating variation 180, 182, 183 conclusions 189–190 consumption effect 177 data 190 domestic prices and labour income 175–176 Engle-Granger: cointegration test 187 estimation of effects of national trade reform 176–184 estimation of total effect 183 external trade reform, effects of 174–176, 184–189 income inequality 185 inequality 171, 184 international and domestic prices 174–175 intra and extra Mercosur trade flows 170 Mercosur 170 methodology 171–176 policy implications 189–190 poverty 171, 184 poverty and inequality effects of liberalisation 188 poverty, before and after reform 184, 185 price changes from Mercosur 179 price transmission 184–185 prices and income 174 prices co-integration 182 prices of traded goods and price of nontraded goods 173 probability of employment after Free Trade Agreement with USA 188 research study 171–189 selection models estimation 186 tariff structure 178 tariffs 169 tariffs and non-traded goods 177–180 tariffs and prices of traded goods 173 tariffs and traded goods 176–177 trade openness coefficient 170

285

trade reform, effects of 171–177 trade reforms 169–170 unit-root test: ADF 186 unit-root test: tradable and non-tradable prices 181 wage-price elasticities 180–183 USA 10 Venezuela, exports 119 vertical FDI 85 Vogt, M.G. 16 Wacziard, R. 20 wage inequality 11 wage-price elasticities, Uruguay 180–183 wage share, Argentina 145 Wakelin, K. 252 Warner, A. 20 Washington Consensus case for industrial policy 50–53 development gap 38–50 economic growth 31 framework 28–30 inflation 31–38 macroeconomic performance 30–38 overview 27–28 summary and conclusions 53–54 trade liberalisation and balance 38 wealth dynamics 226–228 Wealth of Nations 10 Welch, K. 20 Wicksell, K. 34 Williamson, J. 28, 29–30, 31, 50, 51, 53, 54 Wincoop, E.V. 120 Winters, A. 11 Winters, L.A. 81, 82, 87 World Trade Organization (WTO) 21, 22–23 Yacimientos Petroliferos Fiscales (YPF) 141