83837 Latin American Entrepreneurs Many Firms but Little Innovation Latin American Entrepreneurs Many Firms but Little Innovation Daniel Lederman, Julián Messina, Samuel Pienknagura, and Jamele Rigolini © 2014 International Bank for Reconstruction and Development / The World Bank 1818 H Street NW, Washington DC 20433 Telephone: 202-473-1000; Internet: www.worldbank.org Some rights reserved 1 2 3 4 16 15 14 13 This work is a product of the staff of The World Bank with external contributions. Note that The World Bank does not necessarily own each component of the content included in the work. The World Bank therefore does not warrant that the use of the content contained in the work will not infringe on the rights of third parties. The risk of claims resulting from such infringement rests solely with you. The fi ndings, interpretations, and conclusions expressed in this work do not necessarily reflect the views of The World Bank, its Board of Executive Directors, or the governments they represent. The World Bank does not guarantee the accuracy of the data included in this work. The boundaries, colors, denominations, and other information shown on any map in this work do not imply any judgment on the part of The World Bank concerning the legal status of any territory or the endorsement or acceptance of such boundaries. Nothing herein shall constitute or be considered to be a limitation upon or waiver of the privileges and immunities of The World Bank, all of which are specifically reserved. Rights and Permissions This work is available under the Creative Commons Attribution 3.0 Unported license (CC BY 3.0) http://creativecommons.org/licenses/by/3.0. Under the Creative Commons Attribution license, you are free to copy, distribute, transmit, and adapt this work, including for commercial purposes, under the following conditions: Attribution—Please cite the work as follows: Lederman, Daniel, Julián Messina, Samuel Pienknagura, and Jamele Rigolini. 2014. Latin American Entrepreneurs: Many Firms but Little Innovation. Wash- ington, DC: World Bank. doi:10.1596/978-1-4648-0012-2. License: Creative Commons Attribution CC BY 3.0 Translations—If you create a translation of this work, please add the following disclaimer along with the attribution: This translation was not created by The World Bank and should not be considered an offi cial World Bank translation. The World Bank shall not be liable for any content or error in this translation. All queries on rights and licenses should be addressed to the World Bank Publications, The World Bank Group, 1818 H Street NW, Washington, DC 20433, USA; fax: 202-522-2625; e-mail: pubrights@ worldbank.org. ISBN (paper): 978-1-4648-0012-2 ISBN (electronic): 978-1-4648-0013-9 DOI: 10.1596/978-1-4648-0012-2 Cover design: Critical Stages, Inc. Cover image: © Nicholas Wilton/Illustration Source; permission required for further reuse. Library of Congress Cataloging-in-Publication Data has been requested. Contents Foreword . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xi Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xiii Abbreviations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xv 1. Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1 Entrepreneurship is a driver of development . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1 Entrepreneurship is vibrant—but growth is weak . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4 The region has many entrepreneurs but little innovation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7 Few companies enter export markets . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10 Even large multinational corporations in the region are insufficiently innovative . . . . . . . . . 13 How can policy enable innovative entrepreneurs? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16 Structure of the report . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20 Notes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21 2. Entrepreneurship, Entry, and the Life Cycle of Firms in Latin America and the Caribbean: Are All Forms of Firm Creation Entrepreneurial? . . . . . . . . . . . . . . . . . 23 Low-level entrepreneurs, high-level entrepreneurs, and employees . . . . . . . . . . . . . . . . . . . . . 24 Theoretical framework . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26 Business creation in Latin America and the Caribbean . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38 Beyond entry: Firm dynamics in Latin America and the Caribbean . . . . . . . . . . . . . . . . . . . . 42 What is hindering high-growth entrepreneurship: Culture, institutions, or the environment? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49 Notes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56 3. Entrepreneurship by Incumbent Firms: What Explains the Innovation Gap? . . . . . . . . . . . . 61 What drives innovation? A conceptual framework . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63 How innovative are firms in Latin America and the Caribbean? . . . . . . . . . . . . . . . . . . . . . . 67 What explains the innovation gap? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73 Notes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 89 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 90 v vi CONTENTS 4. Export Entrepreneurship . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 95 Exporting as a transformative entrepreneurial act . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 96 Contribution of export entrepreneurship in the medium term . . . . . . . . . . . . . . . . . . . . . . . . 98 Descriptive benchmarking of export entrepreneurship . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 102 Econometric benchmarking of export entrepreneurship . . . . . . . . . . . . . . . . . . . . . . . . . . . 112 Export promotion policies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 116 Concluding remarks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 117 Notes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 118 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 119 5. Foreign Direct Investment, Multinational Corporations, and Innovation . . . . . . . . . . . . . . 121 Foreign multinational corporations in Latin America and the Caribbean . . . . . . . . . . . . . . 123 Multilatinas . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 131 Concluding remarks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 136 Notes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 136 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 138 6. Toward an Enabling Environment for Innovative Entrepreneurs . . . . . . . . . . . . . . . . . . . . . 141 What are the elements of an enabling environment for innovative entrepreneurs? . . . . . . . . 142 What explains the region’s innovation gap? The leading suspects . . . . . . . . . . . . . . . . . . . . 149 Notes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 150 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 151 Boxes 2.1 Main databases used in the study . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32 2.2 Do training programs for entrepreneurs work? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43 2.3 Comparing firm size across age cohorts in Colombia using Enterprise Surveys and administrative data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47 2.4 Importing entrepreneurs: Start-Up Chile . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51 3.1 Risk, laws, macroeconomics, and the innovation gap in Latin America and the Caribbean . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65 3.2 Management matters: How better practices could increase productivity in Latin America and the Caribbean . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70 3.3 Do’s and don’ts of entrepreneurship ecosystems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 88 5.1 Can a whale in a swimming pool create a splash? Intel and the upgrading of tertiary education in Costa Rica . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 129 Figures 1.1 Type of employment, by GDP per capita . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2 1.2 Innovation edge of medium and large firms over small firms in Latin America and the Caribbean, 2010 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3 1.3 Relationship between type of employment and GDP per capita, 2010 . . . . . . . . . . . . . . . . 4 1.4 Firm dynamics: entry, age, and size . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5 1.5 Employment growth in Colombia, by firm size and age . . . . . . . . . . . . . . . . . . . . . . . . . . 6 1.6 Percentage of firms in selected countries introducing a new product, 2006–10 . . . . . . . . . 8 1.7 Investment in research and development (R&D) in selected country groups, 2008–10 . . . 9 1.8 Number of patents per capita granted by U.S. Patent and Trademark Office, actual and benchmarked, by inventor’s country or place of residence . . . . . . . . . . . . . . . . 9 1.9 Management practices in selected economies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11 1.10 Average entry and one-year survival rates in selected countries (differences with respect to baseline) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12 1.11 Sources of export growth in selected countries, 2005– 07 and 2008– 09 . . . . . . . . . . . . . 13 CONTENTS vii 1.12 Innovation edge of foreign multinational corporations over local firms in Latin America and the Caribbean . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14 1.13 Predicted productivity gains from entry of new multinational corporations in selected country groups, countries, and economies . . . . . . . . . . . . . . . . . . . . . . . . . . . 14 1.14 Spending on research and development (R&D) in Latin America and the Caribbean . . . 15 1.15 Sectoral position of foreign subsidiaries relative to headquarters in selected country groups, countries, and economies, 2010–11 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15 1.16 Actual and benchmarked index of competition in 17 nontradable industries in selected countries or economies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17 1.17 Income and engineering density in selected economies, 1900 . . . . . . . . . . . . . . . . . . . . . 18 1.18 Number of engineers per million people in selected countries . . . . . . . . . . . . . . . . . . . . . 18 1.19 Actual and benchmarked index of intellectual property rights in selected countries or economies, 2005 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19 2.1 Model of entrepreneurship . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27 2.2 Occupational choice and GDP per capita, 2010 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29 2.3 Income distribution in Latin America and the Caribbean by type of occupation, circa 2011 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31 2.4 Rate of formal business ownership in selected country groups, countries, and economies, 2011 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33 2.5 Share of firms with no employees in selected country groups, countries, and economies, 2011 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35 2.6 Distribution of firm size in selected country groups, countries, and economies, 2011 . . . 35 2.7 Push versus pull entrepreneurship in selected country groups, countries, and economies, 2011 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37 2.8 New firm entry rates and GDP per capita in selected countries, 2004–11 . . . . . . . . . . . . 39 2.9 Time required to start a business in selected country groups, countries, and economies, 2004 and 2013 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40 2.10 Relationship between business formality and barriers to entry in selected countries, various years . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41 2.11 Employment growth in Colombia, by firm size and age cohort . . . . . . . . . . . . . . . . . . . . 44 2.12 Net employment growth rates by firms in Colombia, by establishment age and size, 1994–2009 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45 B2.3.1 Employment in establishments in Colombia, by age of establishment . . . . . . . . . . . . . . . 47 2.13 Firm size in Latin America and the Caribbean, by age of firm, 2006–10 . . . . . . . . . . . . 48 2.14 Age distribution of top 100 firms in selected country groups . . . . . . . . . . . . . . . . . . . . . 49 2.15 Share of business establishments in Argentina owned by foreigners, 1910 . . . . . . . . . . . 50 2.16 Entrepreneurship among immigrants and natives in the United States, by type of business and country . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53 2.17 Marginal effects of years in United States on entrepreneurship gap between migrant and nonmigrant white men, by cohort of arrival and region . . . . . . . . . . . . . . . . . . . . . . 54 3.1 Extensive and intensive margins of innovation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65 3.2 Percentage of firms in selected countries that introduced a new product in the past year . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67 3.3 Average investment in research and development, by region and level of GDP, 2008–10 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68 3.4 Number of patents per capita granted by U.S. Patent and Trademark Office, actual and benchmarked, by inventor’s country or place of residence . . . . . . . . . . . . . . . 69 B3.2.1 Correlation between management quality and productivity in selected countries and economies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70 3.5 Management practices in selected countries or economies . . . . . . . . . . . . . . . . . . . . . . . 71 3.6 Distribution of overall management scores in Brazil and the United States . . . . . . . . . . . 72 viii CONTENTS 3.7 Relationship between investor protection and time required to register property and innovation in Latin America and the Caribbean . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73 3.8 Doing Business in selected country groups and countries, circa 2004 versus 2013 . . . . . 75 3.9 Relationship between competition and various aspects of innovation in Latin America and the Caribbean . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76 3.10 Appointment of head of regulatory agency in Latin America and the Caribbean . . . . . . 77 3.11 Level of allowed government intervention in regulatory decisions . . . . . . . . . . . . . . . . . . 78 3.12 Scope of action of regulatory agencies in Latin America and the Caribbean . . . . . . . . . . 78 3.13 Transparency practices of regulatory agencies in Latin America and the Caribbean . . . 79 3.14 Maximum fines imposed by regulatory agencies in selected countries in Latin America and the Caribbean . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79 3.15 Credit, investment, and innovation in Latin America and the Caribbean . . . . . . . . . . . . 81 3.16 Depth of financial systems in selected country groups and countries, 1995 and 2005 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 82 3.17 Private equity and venture capital investments, by region, 2002–11 . . . . . . . . . . . . . . . . 83 3.18 Number and size of private equity and venture capital deals in Latin America and the Caribbean, by country, 2008–11 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 83 3.19 Access to credit, by region and age of firm . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 84 3.20 Likelihood of spin-off firm after five years, by type of entrant . . . . . . . . . . . . . . . . . . . . 85 3.21 Engineering density and GDP of selected countries, 1900 . . . . . . . . . . . . . . . . . . . . . . . . 85 3.22 Actual versus perceived management quality in Argentina, Mexico, Chile, and Brazil . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87 3.23 Domestic versus international spillovers in patenting activity . . . . . . . . . . . . . . . . . . . . . 89 4.1 Strategic objectives of export promotion agencies, by region . . . . . . . . . . . . . . . . . . . . . . 99 4.2 Share of total exports accounted for by new export entrants in seven countries in Latin America and the Caribbean, 2005– 09 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 100 4.3 Share of total exports accounted for by continuous exporters in seven countries in Latin America and the Caribbean, 2004– 09 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 101 4.4 Export growth and its components in selected countries, 2005– 07 and 2008– 09 . . . . 103 4.5 Product entry, exit, and first-year survival rates of incumbent exporters in selected countries, 2005– 07 and 2008– 09 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 105 4.6 Size of new product exports relative to incumbent products in selected countries, 2005– 07 and 2008– 09 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 106 4.7 Export growth of incumbent exporters in selected countries and its decomposition along the product dimension, 2005– 07 and 2008– 09 . . . . . . . . . . . . . . . . . . . . . . . . . . 107 4.8 Destination entry, exit, and first-year survival rates of incumbent exporters in selected countries, 2005– 07 and 2008– 09 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 109 4.9 Exports to new destinations as a share of total exports by incumbent exporters in selected countries, 2005– 07 and 2008– 09 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 110 4.10 Export growth of incumbent exporters in selected countries and its decomposition along the destination dimension, 2005– 07 and 2008– 09 . . . . . . . . . . . . . . . . . . . . . . . 111 4.11 Conditional benchmarking of export entry rates by sector, 2005– 09 . . . . . . . . . . . . . . 112 4.12 Conditional benchmarking of one-year export survival rates by sector, 2005– 09 . . . . 113 4.13 Conditional benchmarking of export entry rates in selected countries, 2005– 09 . . . . . 113 4.14 Conditional benchmarking of one-year export survival rate in selected countries, 2005– 09 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 114 4.15 Conditional benchmarking of export entry and one-year survival rate in selected countries after controlling for GDP per capita and comparative advantage, 2005– 09 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 115 4.16 Partial effects of 1 percent increase in index of revealed comparative advantage on export entrepreneurship indicators in seven countries in Latin America and the Caribbean, 2005– 09 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 115 CONTENTS ix 5.1 Inward foreign direct investment and multinational activity in Latin America and the Caribbean . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 124 5.2 Difference in number of patents held by multinational parent and local firms in home country in selected country groups, countries, and economies, 2010–11 . . . . 125 5.3 Difference in innovation between multinational affiliates and local firms in the host economy in Latin America and the Caribbean, 2010 . . . . . . . . . . . . . . . . . 126 5.4 Product innovation by foreign multinational affiliates in selected country groups and economies, 2010 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 127 5.5 Research and development by foreign affiliates of U.S. multinational corporations in selected regions, 1998 and 2008 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 127 5.6 Sources of predicted productivity gains associated with entry of multinational corporations, by country groups, countries, and economies, 2002–10 . . . . . . . . . . . . . 130 5.7 Share of total revenues of multilatinas by country or country group of origin, 2010–11 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 132 5.8 Actual and benchmarked revenue of multinational corporations in selected countries and economies, relative to given characteristics, 2010–11 . . . . . . . . . . . . . . . 133 5.9 Factors driving Brazilian firms to cross borders, 2010–11 . . . . . . . . . . . . . . . . . . . . . . . 134 5.10 Sectoral position of foreign subsidiaries relative to headquarters in selected country groups, countries, and economies, 2010–11 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 134 5.11 Origin of revenues of multilatinas, 2010 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 135 5.12 Management practices by firms in the United States and selected countries in Latin America and the Caribbean, by type of firm . . . . . . . . . . . . . . . . . . . . . . . . . . 135 5.13 Research and development by multinational corporations in selected country groups, countries, and economies, 2010–11 . . . . . . . . . . . . . . . . . . . . . . . . . . . 136 6.1 Actual and benchmarked access to credit by young firms in selected countries . . . . . . 143 6.2 Actual and benchmarked index of intellectual property rights in selected countries or economies, 2005 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 145 6.3 Actual and benchmarked contract certainty in selected countries or economies, 2012 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 146 6.4 Actual and benchmarked index of competition in 17 nontradable industries in selected countries or economies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 147 6.5 Actual and benchmarked index of openness to trade in selected countries, 2012 . . . . . 148 6.6 Actual and benchmarked share of engineers in selected countries, 2008–10 . . . . . . . . 149 Tables 2.1 Socioeconomic characteristics of business owners in selected country groups, countries, and economies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34 2.2 Dynamics of manufacturing firms in Colombia . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45 2.3 Five-year changes in size categories for establishments of different ages in Chile and Colombia . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46 3.1 Firm characteristics in Latin America, the United States, and China . . . . . . . . . . . . . . . 72 4.1 Number of new and incumbent exporters in seven countries in Latin America and the Caribbean, 2005– 09 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 100 4.2 Export growth by new entrants and incumbents in seven countries in Latin America and the Caribbean, 2004– 09 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 101 4.3 Treatment effects of export promotion agencies in seven countries in Latin America and the Caribbean . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 117 6.1 Factors that may account for innovation deficits in 13 countries in Latin American and the Caribbean, 2005 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 150 Foreword F or almost a decade, emerging market rate than the United States and, more impor- economies, including several countries tant, were unable to take advantage of their in Latin America and the Caribbean relative underdevelopment by catching up to (LAC), were regarded by analysts and inves- the United States and other developed econo- tors as new engines of growth. Their growth mies that became the sources of technologies before the global financial crisis sparked that are now commonplace around the globe. enthusiasm that, after a short pause during LAC did not need to invent, just to imitate the 2008 crisis, was cemented by vigorous and adopt technologies, as some economies recoveries in 2009 and 2010. A new story line in East Asia were able to do. seemed to dominate: thanks to deep struc- All this is not to say that the recent enthusi- tural changes, both domestic and global, the asm for LAC’s emerging markets was unwar- potential of emerging market economies had ranted. The enthusiasm was justified by the finally arrived. substantial and unprecedented social prog- In the past few months, enthusiasm for ress in the region during this recent growth emerging markets appears to have soured. spurt, as documented in a previous regional A notable slowdown has cast doubts on the flagship report, Economic Mobility and the sustainability of their high growth rates of the Rise of the Latin American Middle Class. past decade and revived old fears of macro- That report provided evidence of remarkable economic and fi nancial turbulence. Phrases progress: such as “submerging economies” have become common in financial periodicals. • Nearly 70 million people were lifted The truth is that major LAC economies out of poverty in the past decade. experienced lackluster growth for decades • Approximately 50 million people before the boom of the 2000s. At the begin- entered the ranks of the middle class ning of the 20th century, a simple aver- between 2003 and 2009. age of the region’s gross domestic product • Income inequality, as measured by the per capita was about 38 percent that of the Gini coefficient, fell steadily, dropping United States. By 2012, that ratio was about from its peak of 0.58 in 1996 to the 35 percent. lowest level ever recorded in the region, The change implies that over 110 years, 0.52, in 2011, a decline of more than the large economies of LAC grew at a slower 10 percent. xi xii FOREWORD • About one-third of the poverty reduc- • Improving logistics and infrastructure. tion was the result of social policies Modernizing ports, transport, and that transferred incomes to the poor, customs can add a competitive edge but labor market income during the to products from the region. The cur- boom years accounted for the remain- rent infrastructure deficit also needs to ing two-thirds. In other words, growth be addressed in order to end capacity is required to sustain poverty reduction constraints that become evident at low and middle class expansion. growth rates. • Enhancing competition. Although the What makes the productivity challenge press- region has globalized, many industries ing is precisely the fact that social progress has remain sheltered from competition. been tied to growth. Thanks to current poli- This protection has the dual negative cies, social programs can be maintained in the effects of reducing productivity growth short term. The risk is that these gains may be in those sectors and handicapping the lost if growth remains low for too long. export sector, which relies on their ser- With global tailwinds receding, the region vices and intermediate goods. will need to rely on its own devices to spur • Improving the contractual environ- growth. Those devices have only one name: ment. Although intellectual property productivity. With scant domestic sav- rights are not the only relevant aspect ings and receding external capital inflows, of domestic institutions that affect pro- income growth can be sustained only by pro- ductivity, innovation is unlikely to take ductivity gains. root without adequate protection. Leaders in the region are fully aware of the importance of boosting productivity. But what With LAC’s recent social gains, growing is this battle about? This report argues that it demands for access to good-quality services is about establishing an enabling environment have increased. Middle classes expect not in which entrepreneurs can emerge, compete, only income gains so that their children will and innovate. It is about building an innova- see even more progress in the future but also tive entrepreneurial class in which top-notch improved public services for the current gen- firms—firms that export goods, services, and eration. With increased productivity, private even capital—no longer look tepid in contrast incomes will rise, increasing public revenues to entrepreneurial superstars elsewhere. and the state’s capacity to invest in service Beyond generalities, the main elements of delivery. In time, if we win the productivity an enabling environment for entrepreneur- battle, we will enter into a virtuous cycle of ship and innovation include the following: stronger public sectors, higher growth, and opportunities for all. • Building human capital. The chal- lenge of raising the quality of education remains, but it goes well beyond test Augusto de la Torre, Chief Economist scores. For example, LAC has a historic Hasan Tuluy, Vice President deficit of engineers, dating at least to Latin America and the Caribbean Region the early 20th century. The World Bank Group Acknowledgments T his report was prepared by a team led Reis, Pablo Sanguinetti, and Antoinette by Daniel Lederman, Julián Messina, Schoar. While we are very grateful for the Samuel Pienknagura, and Jamele guidance received, these reviewers are not Rigolini. Important additional contributions responsible for any remaining errors, omis- were made by Paolo Benedetti, Claudio sions, or interpretations. Additional insights Bravo-Ortega, Maggie Chen, Paulo Correa, from Pablo Acosta, Tito Cordella, Leonardo Ana Paula Cusolito, Marcela Eslava, Ana M. Iacovone, Mariana Iootty de Paiva Dias, Fernandes, Mario Gutierrez-Rocha, Mary Esperanza Lasagabaster, Martha Martínez- Hallward-Driemeier, John Haltiwanger, Licetti, Marialisa Motta, Oltac Unsal, María Thomas Kenyon, Leora Klapper, William Pluvia Zuñiga, and other participants in M a l o n e y, Yo t a m M a r g a l i t , D a v i d a workshop that took place on November McKenzie, Camilo Mondragón, Marcelo 15–16, 2012 are gratefully acknowledged. Olarreaga, Aitor Ortiz, Caglar Ozden, We also want to thank Mauro Lopes Mendes Markus Poschke, Douglas Randall, Miguel de Azeredo, Marcela Sánchez-Bender, and Sarzosa, Murat Seker, Marco Vivarelli, and the World Bank’s Latin America and the Lucas Zavala. The team was ably assisted Caribbean Finance and Private Sector Devel- by Juan Manuel Puyana, Juan Pablo Uribe, opment and International Trade teams for and Cynthia van der Werf. The work was valuable comments. conducted under the general guidance Book design, editing, and production were of Augusto de la Torre, Chief Economist coordinated by the World Bank’s Publishing for the Latin America and the Caribbean and Knowledge department under the super- Region of the World Bank. vision of Patricia Katayama and Mark Inge- The team was fortunate to receive advice bretsen. Last but not least, we thank Ruth and guidance from four distinguished peer Delgado and Jacqueline Larrabure Rivero for reviewers: Caroline Freund, Jose Guillerme unfailing administrative support. xiii Abbreviations BEEPS Business Environment and Enterprise Performance Surveys CAFTA Central America Free Trade Agreement CRC Centro Regional de Competencia para América Latina EAP4 Indonesia, Malaysia, the Philippines, and Thailand ECA Europe and Central Asia EPA export promotion agency FTA free trade agreement GDP gross domestic product GIPBP Global Investment Promotion Best Practices HS Harmonized System ICRG International Country Risk Guide ICS Investment Climate Surveys IPA investment promotion agency IPRs intellectual property rights LAC Latin America and the Caribbean LAC5 Argentina, Brazil, Chile, Colombia, and Mexico MNC multinational corporation PEVC private equity and venture capital PPP purchasing power parity R&D research and development RCA revealed comparative advantage SME small and medium enterprise TPF total factor productivity USPTO U.S. Patent and Trademark Office xv Overview 1 Entrepreneurship is a and was greatly strengthened by the seminal driver of development work of Joseph Schumpeter—­ is that creative Successful entrepreneurs are individuals who entrepreneurs are not just byproducts of the transform ideas into profitable commercial development process but important drivers of enterprises. This process often requires special such a process. Entrepreneurs are key actors talents, including a capacity to innovate, to income societies in the transformation of low-­ introduce new products, and to explore new characterized by low productivity and often markets. It also requires an ability to manage subsistence self-­employment into dynamic others, to assign priorities to tasks to increase economies characterized by innovation and a the efficiency of production, and to make the rising number of well-­remunerated workers. best use of available resources. But these tal- To the extent that causal links from entre- ents are not enough. Successful entrepreneurs preneurship to productivity growth are at thrive in favorable economic and institutional work, there is room for using policy levers to environments that enhance the expected quicken the development process by improv- returns of innovation. When an enabling ing the incentives and supportive institutions environment exists, entrepreneurs take risks that facilitate innovation by entrepreneurs. and invest in innovation, spurring productiv- These analytical and policy issues motivate ity gains through the dynamics of firm entry this report, which explores the challenges and exit and innovation by incumbent firms, faced by potential high-­growth, transforma- thus fostering economic development. tional entrepreneurs in Latin America and Why should policy makers care about the Caribbean (LAC). entrepreneurs, who tend to be among the Figure 1.1 depicts the transition from better off in the population? The answer is employment toward wage employment self-­ simple: entrepreneurship is a fundamental that tends to go hand in hand with eco- driver of growth and development. Indeed, nomic development. It shows that up to a the basic premise of this report—­ one that is gross domestic product (GDP) per capita of shared by most economists since Adam Smith about $2,000 (adjusted for purchasing power 1 2   LATIN AMERIC AN ENTREPRENEURS FIGURE 1.1  Type of employment, by GDP per capita This stylized fact is shared across coun- tries, albeit with less intensity in the more 100 advanced economies. It is not attributable to observable differences in the distribution of 80 workers’ skills or education across firms of different sizes. Medium-­ size and large firms, which are Workers (%) 60 typically run by the most dynamic entrepre- 40 neurs, are also more likely to engage in vari- ous forms of innovation. They are more likely to export to foreign markets, obtain patents, 20 invest in research and development (R&D), introduce new products, improve produc- 0 tion processes, cooperate on innovation with 300 500 1,000 2,500 5,000 10,000 25,000 50,000 GDP per capita other firms, import new technologies, and All agricultural workers Nonagricultural wage and salaried export capital to establish affiliates in foreign Nonagricultural employer Nonagricultural own account markets (figure 1.2). Nonagricultural unpaid Research on entrepreneurship in LAC may deepen our understanding of the region’s Source: Gindling and Newhouse 2012. lagging productivity growth. Although LAC Note: Employment shares are calculated based on data from household surveys. GDP = gross domestic product. experienced remarkable growth in the first decade of the new millennium—­ especially parity), agricultural workers make up most compared with its own past and growth in of the labor force, followed by the nonagri- the advanced economies—­ there are reasons cultural self-­employed; wage employment to doubt the long-­ term sustainability of such outside agriculture comes only third. The high growth rates. A significant part of the incidence of wage employment rises gradually recent growth spurt appears to be related to thereafter, becoming the most important type the commodity boom. Productivity growth of employment at a GDP per capita of about remains modest (Busso, Madrigal, and $5,000. In countries such as Canada and the Pagés-Serra 2012), particularly in the non- United Kingdom, more than 85 percent of tradable services sector (Pagés-Serra 2010), employment consists of salaried employees which through the natural process of struc- (Gindling and Newhouse 2012). tural transformation is attracting a growing The transition from self-­ employment to share of the LAC urban workforce. wage employment is part and parcel of the Measuring entrepreneurship is not an easy development process, in which entrepreneurs task, however, because it is related to the play a crucial role. Creative entrepreneurs individual talents and characteristics of a few are typically behind the most dynamic and elite businesspeople. Following Schumpeter productive firms—­ t he ones that innovate, (1911), this report adopts a broad definition expand production, and generate jobs at a of entrepreneurship that focuses on what is comparatively rapid pace. These firms not new for the market. 2 Entrepreneurship thus only create employment opportunities, they includes firm entry into new or existing also create better employment. For a given markets (both domestic and foreign), the set of skills, across the world, more pro- introduction of new products to the mar- ductive firms, which tend to be the larger ket, and organizational improvements that ones, pay higher wages. In LAC, for exam- enable firms to improve the quality or price ple, medium firms (with 5–­ 2 5 employees) of their products or achieve more efficient pay 20–­ 40 percent higher wages than small modes of production. The report adopts var- firms, and large firms (with more than 25 ious terms to refer to this type of innovative employees) pay 30–­ 60 percent higher wages.1 entrepreneurship, including “high-­ growth,” O v er v ie w   3 “high-­ end,” and Lerner’s and Schoar’s (2010) FIGURE 1.2  Innovation edge of medium and large firms over small “transformational” entrepreneurship. The firms in Latin America and the Caribbean, 2010 important point is to differentiate entrepre- neurs with high growth potential from small Labor productivity firms and self-­employed individuals with low growth potential. Exporter The report uncovers some bright spots. It finds that LAC is a region of entrepreneurs, Exports share as evidenced by the large number of busi- ness owners per capita relative to countries Invested in R&D with similar incomes per capita. Moreover, Patent abroad the large number of entrepreneurs is not—­ as often believed—­ mainly a reflection of a large Patent, trademark, or copyright informal sector in which low-­ productivity New or signi cantly firms are constantly emerging and dying. The improved process share of business owners with formally reg- istered firms is also relatively high in several Patent in country LAC economies. New products introduced At the top end of the entrepreneurial Medium firms spectrum, LAC experienced impressive Technology from a foreign- owned company Large firms export entrepreneurship activity during 95% confidence 2004–­ 09. Stimulated by global tail winds Cooperates on innovation interval and augmented by comparative advantage, 0 10 20 30 40 50 recently implemented trade agreements, and Marginal e ect (%) well-­t argeted export promotion policies, the region saw impressive survival rates by Source: World Bank, based on data from 2010 Enterprise Surveys. exporters. It also witnessed the emergence Note: Bars represent the marginal effect of a medium and large dummy variable in a regression controlling for firm, sector, and country characteristics. Small firms have 0–­50, medium firms of multinational enterprises—­ multilatinas—­ 51–­100, and large firms more than 100 employees. Robust standard errors were calculated. Each which are increasingly extending their country has the same weight in the regional average. R&D = research and development. influence beyond their countries’ borders, particularly into neighboring countries. productive than similar multinationals from These bright spots notwithstanding, the other regions. report identifies a glaring weakness in LAC’s The rest of this overview is structured as entrepreneurship landscape—­ n amely, the follows. The next section documents the sur- low level of innovation. Firms in the region prising vibrancy of entrepreneurship in the suffer from a chronic and substantial inno- region, as measured by the large number of vation gap relative to comparator countries enterprises. It highlights the crucial distinc- and regions. This gap exists not only in terms tion between “small” and “young” firms. of R&D and patenting but also in terms of Businesses that grow rapidly and become product and process innovation. Innovation employment poles are more likely to be young gaps are found among small and large firms firms, but they are not necessarily small. The alike. Indeed, even the region’s superstar third section documents the acute shortfall entrepreneurs—­exporters and multilatinas—­ in innovation that characterizes LAC entre- lag in important dimensions of innovation. preneurship—­ i n product innovation, pat- Entry rates into exporting activities by LAC ents, R&D, and managerial practices. The firms have been particularly low, although fourth section examines various stylized facts incumbent exporters did become more about export entrepreneurship in the region, innovative under duress during the global including low entry rates coupled with solid financial crisis of 2008–­ 09. Multilatinas are survival rates and strong responsiveness to less innovative, less well managed, and less adverse circumstances. The fifth section 4   LATIN AMERIC AN ENTREPRENEURS examines the performance of multilatinas in fundamental policy question. Addressing it the broader context of multinational corpora- requires a change in policy paradigm from tions in LAC, with a focus on their low level the current emphasis on supporting small of innovation. The last section discusses pos- firms toward an emphasis on supporting sible links between entrepreneurship, innova- start-­ups and young firms. tion, and structural features of the enabling Figure 1.3 captures both the vibrancy of environment in LAC. the entrepreneurial environment and some of its deficits. It shows that in many countries in the region, the share of (nonagricultural) Entrepreneurship is vibrant—­ employers in the population is much larger but growth is weak than in countries at similar levels of eco- In contrast to commonly held views, LAC nomic development (panel a). However, these is characterized by vibrant entrepreneur- employers do not generate sufficient wage ship, as measured by the number of firms employment, as the share of own-­ account per capita. The share of entrepreneurs in workers in the population is also above the the population is higher than in compar- expected levels (panel b). This characteristic ator countries and regions. Perhaps more is linked to the large informal sectors that surprisingly, the incidence of formal busi- constitute a developing country hallmark. nesses is also high. This fact suggests that Entry into the higher end of the formal the enterprise sector is much more than a sector, measured by registration of new lim- large informal sector. However, the region ited liability firms, remains low in many LAC lags in the nature of the businesses created. countries3 relative to their level of economic Firms in LAC tend to be smaller (in terms of development. Figure 1.4 (panel a) displays the number of employees) at birth than firms the relationship between firm entry (mea- in other regions at similar levels of develop- sured by the average annual number of new ment, and the growth process fails to com- limited liability firms registered per 1,000 pensate for the initial gap in employment. working-­ age people during 2004–­ 11) and Even the largest firms in LAC create fewer the level of economic development (mea- jobs than the largest firms in other regions. sured by the average per capita income for How to address the gap in firm growth is a the same period) across 129 countries. Entry FIGURE 1.3  Relationship between type of employment and GDP per capita, 2010 a. Nonagriculture, employer b. Nonagriculture, own account 8 40 HTI 6 CRI DOM 30 COL VEN ECU VEN SLV PRY Percent HND 4 Percent MEX PER GTM SLV URY BOL URY BOL PER 20 PRY ECU JAM HND COL CHL CHL CRI JAM GTM 2 DOM MEX 10 HTI 0 0 6 7 8 9 10 11 6 7 8 9 10 11 Log of GDP (PPP) per capita Log of GDP (PPP) per capita LAC countries Non-LAC countries Source: World Bank, based on data from Gindling and Newhouse 2012 and World Development Indicators. Note: Curves shows quadratic fitted values. GDP = gross domestic product. LAC = Latin America and the Caribbean. PPP = purchasing power parity. O v er v ie w   5 FIGURE 1.4  Firm dynamics: entry, age, and size a. Entry rates and GDP per capita b. Average workers by rm age 20 300 250 15 Number of employees 200 Entry density 10 150 100 5 50 0 0 6 7 8 9 10 11 1–4 5–9 10–19 20–29 30–39 40+ Log of GDP per capita LAC countries Non-LAC countries LAC High-income ECA EAP4 countries Sources: Panel a: World Bank, based on data from World Development Indicators and World Bank Group Entrepreneurship Snapshots (WBGES). Panel b: World Bank, based on data from 2006–­10 Enterprise Surveys. Note: Panel a: Each point represents the average between 2004 and 2011. Curve shows quadratic fitted values. GDP = gross domestic product. LAC = Latin America and the Carib- bean. Panel b: ECA (Eastern Europe and Central Asia): Albania, Armenia, Azerbaijan, Belarus, Bosnia and Herzegovina, Bulgaria, Georgia, Kazakhstan, Latvia, Lithuania, FYR Macedonia, Moldova, Romania, Russian Federation, Serbia, Turkey, Ukraine, and Uzbekistan. EAP4: Indonesia, Malaysia, the Philippines, and Thailand. High income: Croatia, the Czech Republic, Hungary, Poland, the Slovak Republic, Slovenia, and Spain; LAC: Latin America and the Caribbean. The most recent survey available for each country was used. Each country has the same weight in the regional averages. Size at birth above 10,000 was replaced by “missing.” is positively associated with GDP per capita, business declined steadily in the 2000s. LAC and in many LAC countries entry rates are was no exception, exhibiting stronger dereg- below the expected level. However, there is ulation among countries that started with the substantial heterogeneity within the region, highest levels at the turn of the millennium. with some countries located above the bench- However, the significant reduction in entry mark. The most salient example is Costa barriers has not made a visible dent in the Rica, with an entry rate of almost 16 new region’s entry rates of limited liability firms, firms per 1,000 working-­ age people—­ four which lie at the high end of the formal sector. times the international benchmark. Argen- This failure could be interpreted as an indica- tina and Mexico, by contrast, exhibit rates of tion that the effects of changes in entry bar- entry substantially below those suggested by riers come with a considerable lag. A more their GDP per capita. plausible interpretation may be that either The fact that on average LAC displays entry barriers are not the most binding con- uninspiring rates of entry of formal limited straint to formal entry in LAC or that reduc- liability firms has led many observers to sin- ing entry barriers alone, without achieving a gle out entry barriers as the main culprit. critical mass of complementary reforms, is In the last decade, however, LAC countries insufficient to spur entry. made significant progress in reducing such Another salient feature of LAC entrepre- barriers. The burden imposed by red-­ t ape neurship is that new firms do not grow as entry-­related regulations is still higher in LAC much as firms in other regions and thus tend than in comparator regions. But the time to to remain small. Panel b of figure 1.4 plots set up a business, for instance, was halved in the average age of firms against the average less than a decade (World Bank 2013). number of employees for different regions. Moreover, the variance across countries It shows that LAC has the smallest new in the number of procedures, length of time, firms (in terms of number of employees) of and costs associated with setting up a new any region.4 Even the largest new firms (the 6   LATIN AMERIC AN ENTREPRENEURS 90th percentile of the size distribution of new importance for the design and effectiveness firms) are about half the size of new firms in of SME support programs. other regions.5 Moreover, differences in size The empirical basis for emphasizing this widen as firms age: LAC firms that are 40 or distinction is illustrated by a detailed analysis more years old are on average half the size of of the dynamics of (formal) manufacturing firms the same age from high-­ income coun- firms in Colombia by Eslava and Haltiwan- tries and Eastern Europe and Central Asia ger (2013), as well as by research on firm (ECA) and one-­ third the size of firms in the dynamics in the United States. Figure 1.5 middle-­ i ncome countries of East Asia and presents some of the results on the impor- Pacific (EAP4)—­ I ndonesia, Malaysia, the tance of firm size versus firm age for the gen- Philippines, and Thailand. eration of employment in Colombia. Panel a Policy makers in LAC have typically tried focuses on “continuers” (that is, firms that to address the lackluster growth of firms by remained alive throughout the sample period) focusing on smallness per se. Together with and therefore abstracts from firm entry and a concern about employment, this focus has exit. Growth increases with size and declines taken the form of a myriad of government-­ with age, as stands to reason (that a firm that sponsored programs that support small and did not expand quickly during its youth or medium enterprises (SMEs). Eligibility for middle-­ age years is arguably less likely to accessing support depends largely on size, enjoy a growth spurt in old age). However, typically measured by the number of employ- differences in growth rates are much more ees. The evidence in this report casts doubt marked along the age dimension than along on this overemphasis on smallness and points the size dimension. Firms of all sizes grow to the need to shift the focus toward young fastest in their early years, especially their (rather than small) firms. Most young firms first four years. are small, but a relatively large share of small Even more interesting is the fact that the firms are not young—­ a distinction this report average growth rates of firms in their early highlights as having potentially critical years increase rapidly with size—­ that is, firms FIGURE 1.5  Employment growth in Colombia, by firm size and age a. Continuers b. All establishments 0.6 0.10 0.4 0.05 Growth rate Growth rate 0.2 0 0 –0.2 –0.05 Small Medium Large Small Medium Large Size of rm Size of rm Age of rm: All 0–4 years 5–9 years 10–14 years 15+ years Source: Eslava and Haltiwanger 2013. Note: Small: fewer than 50 employees; medium: 51–­200 employees; large: more than 200 employees. Growth rates are defined as in Davis, Haltiwanger, and Schuh (1996). They are the change in employment between two consecutive periods divided by the average employment between the two periods. O v er v ie w   7 that are young and large grow the most, mak- abilities of private agents, and using risk-­ ing the largest contribution to job creation. sharing arrangements to align incentives This fact contradicts the popular belief that could help governments try to pinpoint firms most employment generation occurs among worthy of public sector support. small firms. The confusion stems from the failure to distinguish between the stock of firms and their growth dynamics. Even if at The region has many any point in time small firms were to account entrepreneurs but little for most of the jobs in the economy, it does innovation not follow that all small firms (independent There are many potential reasons why LAC of age) are equally responsible for employ- firms grow as slowly as they do. One is the ment generation over time. Rather, it appears lack of innovation. Entry is just the beginning that job creation comes from young firms, of the story. In order to grow, or even survive, regardless of their size. firms need to continuously innovate. When all firms in the Eslava-­Haltiwanger It is in this domain of entrepreneurship sample (not just firms that stayed alive that businesses in LAC score relatively badly. during the sample period but also firms that LAC firms introduce new products less fre- were created or died during that period) are quently than firms in otherwise similar econ- examined, the picture changes in an import- omies, high-­ end entrepreneurs tend to be far ant respect (panel b of figure 1.5). Although away from global best practices in the man- young firms continue to be the main contrib- agement of their enterprises, firms’ invest- utors to employment growth, the role of size ment in R&D is low, and patent activity is is reversed, with small firms dominating. The well below benchmark levels. average employment growth rate of small Some of the most successful LAC firms firms up to four years old jumps from 4 per- have managed to grow out of their national cent for continuers to 53 percent for all firms. boundaries during the last decade and are This result stems from the fact that the vast now competing on world markets. The suc- majority of entrants are small, and by con- cess of high-­ end companies such as Vale, struction the growth rates of newly created Embraer, and CEMEX notwithstanding, firms are highest. innovation in LAC is limited, with even some Hence, the evidence on firm dynamics in of the giant multilatinas underperforming Colombia suggests that young rather than their peers from other countries. Many for- small firms are the main employment cre- mal firms in the region are engaged in some ators. This evidence is consistent with recent form of innovation, but the intensity of findings for the United States (Haltiwanger, innovation tends to be low or poorly suited Jarmin, and Miranda 2013). Further research to raise productivity. Figure 1.6 shows the could determine the role of young firms in percentage of firms that developed or intro- employment generation across LAC. duced a new product (product innovation) in However, increasing the effectiveness selected countries between 2006 and 2010. of programs aimed at supporting firm (and The LAC countries are bunched toward the employment) growth may call not just for a low end of the scale.6 On average, firms in shift of emphasis from small to young firms. the region are 20 percent less likely to have A deeper understanding of the characteristics introduced a new product than the middle-­ of young firms of all sizes that enable them income countries in ECA—­ and the picture to survive and thrive in market economies appears even grimmer for most of the Carib- is also necessary. Unfortunately these char- bean, where the likelihood of introducing acteristics of young dynamic firms remain a new product drops to half that of firms unknown, thus making policy making in this in ECA. area complicated. Coordinating efforts with Figure 1.6 measures the share of firms the private sector, leveraging the screening involved in innovation activities, which is 8   LATIN AMERIC AN ENTREPRENEURS FIGURE 1.6  Percentage of firms in selected uninformative about the quality and intensity countries introducing a new product, 2006–10 of innovation, two factors strongly associ- ated with high-­ productivity firms. Datasets St. Lucia exploring these fundamental factors in a Dominica comparable way across countries are of poor Jamaica Antigua and Barbuda quality. The few available indicators suggest Nicaragua that the quality of innovation in LAC may be Venezuela, RB Mexico as much of an obstacle to firms’ growth and Guyana productivity as the quantity. Trinidad and Tobago St. Kitts and Nevis Figure 1.7 shows aggregate investment in Ecuador R&D. Panel a compares regional averages St. Vincent and the Grenadines Malaysia as a percentage of value added in manufac- Dominican Republic turing (the sector where most R&D takes El Salvador Uzbekistan place). Panel b benchmarks R&D against Uruguay the average of countries at similar stages of Guatemala Romania development.7 Average R&D investment in Spain the five largest LAC economies is two-­ thirds Bolivia Honduras that of China when expressed as a percentage Costa Rica of manufacturing value added and one-­ third Chile Paraguay when expressed as a percentage of GDP. For Colombia the remaining LAC countries, R&D invest- Greece Peru ment is about a third that in China when Turkey expressed as a percentage of manufacturing Argentina Suriname value added and a tenth that of China when Bulgaria expressed as a percentage of GDP. These Korea, Rep. Azerbaijan innovation gaps are worrisome. Grenada A second feature that distinguishes LAC Kazakhstan Germany from China and high-­ income countries is the Croatia preponderant role the public sector plays in Ireland Ukraine R&D (the public sector also accounts for a Hungary large share of R&D in ECA) (Pagés-Serra Georgia Poland 2010).8 This is not to say that the public sec- Moldova tor in LAC invests excessively in R&D: as a Slovak Republic Macedonia, FYR percentage of GDP, it invests much less than Russian Federation China or high-­ income countries. The finding Thailand Serbia rather reflects how little private LAC firms Czech Republic invest in innovation. Albania Latvia The extent to which lower levels of R&D Armenia are likely to translate into lower produc- Bosnia and Herzegovina Slovenia tivity and economic growth is, of course, Lithuania influenced by many factors. But panel b of Belarus figure 1.7 indicates that economies that expe- 0 20 40 60 80 100 Percent rienced periods of sustained growth often LAC countries Other countries had bursts of R&D investments that placed them well above their peers (relative to the Source: World Bank, based on data from Seker 2013 and 2006–­10 Enter- blue line). LAC’s low levels of R&D, and the prise Surveys. fact that little of it is conducted by the private Note: LAC = Latin America and the Caribbean. sector, appears to be one of the main culprits behind the region’s well-­ documented history of low productivity growth. O v er v ie w   9 FIGURE 1.7  Investment in research and FIGURE 1.8  Number of patents per capita granted by U.S. Patent development (R&D) in selected country groups, and Trademark Office, actual and benchmarked, by inventor’s 2008–­10 country or place of residence a. R&D by region Uzbekistan LAC countries Haiti Bolivia Other countries Other LAC Paraguay or economies Albania El Salvador Benchmark Indonesia ECA Honduras Kazakhstan Bosnia and Herzegovina LAC5 Guatemala Peru Azerbaijan Colombia China Ecuador Macedonia, FYR Dominican Republic High-income Philippines countries Ukraine Turkey 0 2 4 6 8 10 12 14 16 Venezuela, RB Serbia R&D/manufacturing value added (%) Belarus Georgia Business enterprise Government Oman India Higher education Private nonpro t Thailand Armenia Foreign Mexico Jamaica b. R&D by level of GDP Brazil Uruguay 5 United Arab Emirates Argentina Israel Saudi Arabia Latvia 4 Poland Chile Russian Federation Finland Korea, China R&D (% of GDP) 3 Rep. Trinidad and Tobago Portugal Slovak Republic Lithuania 2 St. Kitts and Nevis Greece China Croatia Brazil Costa Rica Costa Bulgaria 1 Rica India Kuwait Uruguay Antigua and Barbuda Colombia Argentina Czech Republic Mexico Malaysia 0 Guatemala Hungary 100 1,000 10,000 100,000 Spain Slovenia GDP per capita (PPP) Italy New Zealand Ireland France Sources: Panel a: World Bank, based on data from World Development Norway Indicators (WDI) and UNESCO. Panel b: Updated from Lederman and Belgium Maloney 2003 using WDI. United Kingdom Note: For countries and economies included in each group, see note 4. Australia Austria GDP = gross domestic product. PPP = purchase power parity. R&D = Hong Kong SAR, China research and development. The blue line is a regression-fitted line esti- Singapore mated with data from 1996 to 2011 covering 119 countries. Netherlands Denmark Canada Germany Sweden A similar picture emerges from data on Finland Korea, Rep. patents. Figure 1.8 shows the number of pat- Switzerland Israel ents per million people that inventors from Japan different countries received from the U.S. Pat- 1 10 100 1,000 10,000 ent and Trademark Office (USPTO) between Patents per 1 million people 2006 and 2010. No LAC country exhibits a level of patents that approaches that of high-­ Source: World Bank, based on data from USPTO 2012 and World Development Indicators. Note: Dots represent predictions from a multivariate regression analysis that includes the log of patents income countries, and most LAC countries per million people on the log of gross domestic product (GDP) (adjusted for purchasing power parity), the log of population, and the log of merchandise exports to the United States. They indicate where each country stands with respect to countries with similar levels of GDP, population, and merchandise exports to the United States. The regression used all countries and economies for which data were available; the figure presents only comparator countries. Data are averages for 2006–10. LAC = Latin America and the Caribbean. 10   LATIN AMERIC AN ENTREPRENEURS received fewer patents than their middle-­ managers and entrepreneurs. Factors exter- income country peers. Brazil, for instance, nal to the firms, such as the business envi- registered only 5 patents per million people ronment and other country characteristics, between 2006 and 2010, half the number per are also likely to explain the region’s deficit capita of China (10) and slightly less than a in managerial practices and hence process quarter the number per capita of Bulgaria innovation. (22). To be sure, part of these differences can be explained by lower levels of economic development and lower exports to the United Few companies enter States (which imply fewer incentives to apply export markets for patents from the USPTO). But even after Accessing new markets through trade is controlling for per capita income, population arguably a salient manifestation of transfor- size, and exports to the United States, the pat- mational entrepreneurship. Barring firms that ent intensity in most countries in the region benefit from high rents, only firms with supe- remains below their benchmark, including rior performance can thrive in export mar- Brazil (figure 1.8). kets. In fact, most new entrants into export R&D and patenting are proxy measures of markets do not survive beyond one year. the intensity and quality of innovation. They This report documents a number of styl- indicate only indirectly how firms perform in ized facts that characterize LAC exporting terms of process innovation. An additional firms. In particular, although entry rates dimension is the quality of management into exporting activities remain significantly practices, which can be assessed following below those in (poorer) comparator coun- the methodology developed by Bloom and tries, the survival rates of the few firms that Van Reenen (2007). attempt to export tend to be at or slightly Figure 1.9 compares management prac- above benchmark levels. Moreover, analysis tices of manufacturing firms across different of the contraction of foreign demand during dimensions for a number of high-­ income and 2008– ­ 0 9 suggests that exporting entre- LAC countries as well as China and India (the preneurs respond well to pressure: in the sample of comparator countries is dictated face of the crisis, they nimbly opened new by countries in which management surveys exporting firms and developed new export were conducted). LAC countries other than products, in the process penetrating new Mexico score toward the bottom of the dis- export markets. Thus, it seems that the old tribution, with management practices closer adage “necessity is the mother of invention” to those of Chinese and Indian firms than to applies to export entrepreneurship. The high-­income countries. Given that LAC firms report also provides evidence that export face higher labor costs than firms in China promotion policies that help entrepreneurs and India, poor management practices in the surmount certain barriers to entry by pro- region pose a more severe competitive disad- viding information about global markets. vantage for them. Research conducted for this report bench- Part of the LAC “management gap” can marked entry and survival rates in the be explained by firm characteristics. Firms region using a new firm-­ level database, the in high-­income countries have a larger share World Bank’s Exporter Dynamics Database of employees with college degrees, are larger, (figure 1.10).9 and are more likely to be multinationals The results are striking: virtually all than firms in LAC. These firm characteris- LAC countries in the sample show export tics explain at most a third of the manage- entry rates that are below the benchmark. ment gap between the median firm in LAC In contrast, in Asia, the Middle East, and and the United States, however. Part of the even Africa, entry rates of firms into export- remaining two-­ t hirds of the gap could be ing activities are above the benchmark. explained by the training and ability of LAC LAC countries fare better in the survival O v er v ie w   11 FIGURE 1.9  Management practices in selected economies a. Overall management practices b. Operation management c. Performance monitoring United States United States Sweden Japan Germany United States Germany Sweden Germany Sweden Japan Canada Canada Canada Japan United Kingdom Australia France Italy New Zealand United Kingdom France Italy Mexico Australia France Italy Mexico United Kingdom Australia Poland Greece Portugal Northern Ireland Argentina Poland New Zealand Portugal New Zealand Portugal Mexico Argentina Republic of Ireland Northern Ireland Chile Argentina Chile Brazil Chile Republic of Ireland Northern Ireland Greece China Republic of Ireland China Poland Greece Brazil Brazil China India India India 2.0 2.5 3.0 3.5 2.0 2.5 3.0 3.5 2.0 2.5 3.0 3.5 Mean score Mean score Mean score d. Target management e. Talent management Japan United States Germany Canada United States Japan Sweden Germany Italy United Kingdom Canada Poland France Sweden Australia Northern Ireland United Kingdom Italy Poland Mexico Mexico Australia New Zealand Republic of Ireland Northern Ireland China Portugal France Republic of Ireland Chile Argentina New Zealand Brazil Greece Greece Argentina China Portugal Chile India India Brazil 2.0 2.5 3.0 3.5 2.0 2.5 3.0 3.5 LAC countries Other countries Mean score Mean score or economies Source: Maloney and Sarrias 2012. Note: Surveys sampled manufacturing firms with 100–­5,000 employees recorded in Orbis. LAC = Latin America and the Caribbean. 12   LATIN AMERIC AN ENTREPRENEURS FIGURE 1.10  Average entry and one-­year survival rates in selected However, exporting entrepreneurs tend countries (differences with respect to baseline) to display a significant capacity to adapt to and cope with adverse circumstances, which suggests that greater competitive pressures Chile could be an antidote to the dearth of inno- El Salvador vation among high-­ end export entrepreneurs Costa Rica in LAC. The agility of incumbent exporters Colombia Bulgaria is illustrated by their reactions to the drop Mexico in foreign demand in 2008– ­ 0 9. During Guatemala this period, average LAC export growth Macedonia, FYR by incumbent exporters was negative. But Peru their sales of new products raised exports Ecuador Jordan by 3 percent on average, and their sales to Morocco new destinations raised exports by 4 percent Mauritius (Fernandes, Lederman, and Gutierrez-­ Rocha South Africa 2013). Furthermore, the contribution of Dominican Republic new exporters (entrants) to national export Mali Senegal growth increased when the global crisis hit Nicaragua in 2008, even though entry rates did not rise. Bangladesh During the steady growth period (2005–­ 07), Kenya incumbent exporters played a dominant role Burkina Faso in explaining export growth in both LAC Tanzania Pakistan and non–­ L AC countries, among all types Cameroon of exporters (natural resource based, simple Iran processing, and diversified manufactures) Cambodia (panel a of figure 1.11). In contrast, new Malawi exporting firms were an important contrib- Niger Uganda utor to exports in LAC during 2008– ­ 09. Export growth in LAC during the global –0.3 –0.2 –0.1 0 0.1 0.2 0.3 0.4 Log of GDP per capita crisis would have declined more sharply Average entry rate Average 1-year survival rate of new entrants than it did if exports by new entrants had not compensated for the exit of incumbent firms (panel b of figure 1.11) and incumbent Source: Estimations by Ana M. Fernandes and Daniel Lederman (World Bank), based on data from the World Bank’s Exporter Dynamics Database, World Development Indicators, and World Inte- exporters had not found new markets. grated Trade Solution (WITS) database. Export promotion services also appear to Note: Figure shows estimates of each country’s dummy variable from an econometric model that also includes (the log of) GDP per capita (adjusted for purchasing power parity), the Vollrath (1991) increase entry and survival rates and there- index of revealed comparative advantage at the six-­digit level of the Harmonized System (HS) clas- fore overall export activity. The economic sification, industry dummies, and year dummies. The industry dummies are defined at the two-­digit level of the HS. The excluded benchmark country is Albania. Data are for 2005–­09. justification for export promotion is often based on some form of information failure, related to the public good nature of infor- mation that leads to its underproduction by dimension, with survival rates of the (rela- private firms. For instance, existing exporters tively small number of) firms that enter into have no incentives to share information about exporting markets above the benchmark. foreign market conditions and opportunities However, no LAC country appears to be with potential competitors after incurring the an overachiever on the survival front when costs of discovering how to export profitably compared to most of the other developing (Hausmann and Rodrik 2003). countries included in the database, as shown In research conducted for this report, in figure 1.10, after controlling for GDP Lederman, Olarreaga, and Zavala (2013) use per capita. firm surveys from seven LAC countries from O v er v ie w   13 FIGURE 1.11  Sources of export growth in selected countries, 2005–­07 and 2008–­09 a. Decomposition, 2005–07 b. Decomposition, 2008–09 Chile Ecuador resources Peru Costa Rica Natural resources Natural Ecuador Peru South Africa Colombia Costa Rica Chile Nicaragua South Africa Average Manufacturing Simple processing Bangladesh Average Manufacturing Simple processing Cambodia Cambodia Bangladesh Guatemala Nicaragua Dominican Republic Guatemala El Salvador Dominican Republic El Salvador Egypt, Arab Rep. Brazil Egypt, Arab Rep. Mexico Brazil Mexico countries of LAC countries LAC countries LAC countries of LAC –0.1 0 0.1 0.2 0.3 0.4 –0.2 –0.1 0 0.1 0.2 0.3 Export growth rate Export growth rate Incumbents Exiters Entrants Source: Fernandes, Lederman, and Gutierrez-Rocha 2013, based on data from the World Bank’s Exporter Dynamics Database. Note: Figures for Ecuador in panel a are for 2006–­07. LAC = Latin America and the Caribbean. 2006 and 2010 to analyze the effectiveness transitional dislocations. Although the elim- of export promotion services. They find that ination of inefficient local firms may not ulti- firms that used export support services have mately be bad for a country’s economy, in the a significantly higher probability of entering short term it may adversely affect workers and surviving in export markets. and create social and political tensions. This report provides evidence that multi­ national corporations have had significant Even large multinational net positive impact in LAC economies in corporations in the region are recent years: the positive impacts from tech- insufficiently innovative nology transfers, knowledge spillovers, and Under the right business environment and linkages have overwhelmingly dominated contractual conditions, multinational corpo- the negative impacts from greater competi- rations can be good for the local economy. tion in product and factor markets. The full They tend to be more productive and to use potential of multinational corporations has the latest technologies; through their engage- not been fully realized, however, because ment with and support of local suppliers, they multinational affiliates in LAC behave like can transfer knowledge and better technol- local firms, investing very little in innovation. ogies to the local economy, which raise the Thus, either LAC is not attracting the most quality of inputs and the productivity of firms innovative multinationals or the obstacles (Moran 2001; Javorcik and Spatareanu 2005). that local firms face to innovate also act as At the same time, they can have nega- barriers to innovation for foreign firms oper- tive impacts: by competing in local product ating in the region. and factor markets, they can drive less effi- The recent emergence of multilatinas cient local firms to exit, thereby generating has not changed this picture. On average, 14   LATIN AMERIC AN ENTREPRENEURS multilatinas conduct less research than their Multinationals are also more likely than peers from other regions. The large major- local firms to apply for a patent, trademark, ity of their business is concentrated in Bra- or copyright; collaborate for innovation pur- zil, Mexico, and Chile. They therefore miss poses with other institutions; invest in R&D; the opportunities presented by greater inte- and adopt foreign technologies. The differ- gration, both regionally and globally. When ences are even larger for efforts to improve multilatinas expand abroad, typically to the quality of products. Multinational corpo- neighboring countries, their affiliates often rations are 21 percentage points more likely operate in the same sector as the parent com- to engage in quality-­improving investments pany, suggesting that these firms are driven and 25 percentage points more likely to have by the search for larger markets and the international quality certifications than local desire to diversify country risk rather than firms, perhaps because they are more likely the desire to establish linkages and clusters, to export. thereby deepening their involvement in pro- Figure 1.13 quantifies the relative impor- ductive networks and global value chains. tance of the competition and knowledge The higher productivity and more innova- transfer channels, in order to assess the tive behavior of multinational corporations impact of the entry of multinational cor- relative to local firms in LAC are reflected porations on firm-­ level and aggregate pro- in many dimensions. Everything else equal, ductivity. The estimations use a sample of the probability that a firm introduces a new manufacturing firms from 60 countries, product is about 11 percentage points higher 5 of which are in LAC (Argentina, Brazil, for a foreign-­owned firm operating in LAC Chile, Colombia, and Mexico). The results than for domestic firms, and the probabil- are striking: other things equal, doubling ity of introducing a new process is about the number of multinational corporations in 5 percentage points higher (figure 1.12). LAC would increase aggregate productivity FIGURE 1.12  Innovation edge of foreign multinational corporations over local firms in Latin America and the Caribbean FIGURE 1.13  Predicted productivity gains from entry of new multinational corporations in selected country groups, countries, and Has an international quality certi cation economies Invested to improve quality control or obtain certi cation 4.0 Uses foreign technology 3.5 New or signi cantly improved product 3.0 Percentage points Invested in research and development 2.5 New or signi cantly improved process 2.0 1.5 Cooperates on innovation with others 1.0 Filed for patent, trademark, or copyright 0.5 0 5 10 15 20 25 0 Additional likelihood by MNC a liates High-income China ECA LAC5 (percentage points) economies Signi cant at 10% Not signi cant at 10% Knowledge spillover Market reallocation Source: World Bank, based on data from 2010 Enterprise Surveys. Source: Alfaro and Chen 2013. Note: Figures are for the manufacturing sector only. Bars are the coefficients of a dummy variable tak- Note: Figures are for the manufacturing sector only. Bars represent total ing the value 1 if the firm is foreign owned in a regression of innovation variables. Additional controls productivity gains from doubling the probability of multinational cor- include country and industry fixed effects. Standard errors are clustered at the industry level. MNC = poration entry, estimated though a structural model. For countries and multinational corporations. economies included in each group, see note 4. O v er v ie w   15 by 3.8 percent. This number is six times FIGURE 1.14  Spending on research and development (R&D) higher than in ECA or high-­ income econo- in Latin America and the Caribbean mies and seven times higher than in China. Moreover, in contrast with other regions, a. R&D spending by foreign b. R&D spending by multinational knowledge spillovers run the entire show in a liates of U.S. multinationals, corporations, by home region 1998 and 2008 LAC: they explain almost all the estimated US$ of R&D per ten thousand US$ of revenue aggregate productivity gains from entry of 100 30 multinational corporations. 90 80 25 Alas, the full potential of productivity gains from knowledge spillovers from mul- 70 20 tinational corporations in LAC is not being 60 Percent fully realized, in part because of very low 50 15 levels of R&D by foreign companies oper- 40 10 ating in LAC and multilatinas. The share of 30 R&D in LAC accounted for by U.S. multi- 20 5 national corporations, for instance, is only 10 about one-­ fi fth the share of R&D done by 0 0 1998 2008 mie e ina P4 ECA C5 ia eco incom Ind the same companies operating in Asia. More- s EA LA Ch Europe and Canada Africa over, trends are not encouraging: the share of no h- Middle East Asia Hig R&D performed by U.S. multinational cor- LAC porations in LAC fell 1.2 percentage points, to just 3.9 percent of total R&D, between Sources: Panel a: National Science Board 2012; panel b: World Bank, based on data from Orbis. 1998 and 2008 (panel a of figure 1.14). Note: Panel a covers only the manufacturing sector. For countries and economies included in each group in panel b, see note 4. The emergence of multilatinas, welcome as it is, has not fundamentally changed the innovation picture. To be sure, the num- FIGURE 1.15  Sectoral position of foreign subsidiaries relative to ber of multilatinas is still small, and they headquarters in selected country groups, countries, and economies, are concentrated in three countries (Bra- 2010–11 zil, Chile, and Mexico). But despite tower- ing over other LAC companies in size, they 60 are not sufficiently innovative. On average, Share of foreign subsidiaries (%) multilatinas from the manufacturing sec- 50 tor invest only $0.6 per $10,000 of reve- 40 nue on R&D (panel b of figure 1.14). This figure stands in sharp contrast with R&D 30 intensity in high-­ income economies and even China and the four economies of EAP4. For 20 example, multinationals from EAP4 invest 10 $17 in R&D for every $10,000 of revenue—­ almost 30 times the R&D investment of the 0 average multilatina. LAC China and EAP4 High-income ECA India economies A partial explanation for the low level of Location of headquarters innovation of multilatinas may be found in their motives for sending capital abroad. Mul- Upstream (relative to parent) Horizontal tilatinas appear to set up operations abroad Downstream (relative to parent) mainly to expand the markets in which they Source: World Bank, based on data from Orbis. sell and to diversify country risk rather than Note: The sectoral position was calculated using the input-­output matrix for the United States. to integrate into global value chains. A subsidiary is defined as downstream if the parent company’s sector is a net supplier of the sub- sidiary’s sector. A subsidiary is defined as upstream if the subsidiary’s sector is a net supplier of the Figure 1.15 divides the subsidiaries of mul- parent company’s sector. For countries and economies included in each group, see note 4. LAC = tinational corporations from different regions Latin America and the Caribbean. 16   LATIN AMERIC AN ENTREPRENEURS into three groups: companies operating in more important for growth as LAC contin- the same sectors as headquarters (horizon- ues to consolidate their hard-­ earned achieve- tal activity), companies providing inputs to ments on the macroeconomic and financial headquarters (upstream activity), and com- stabilization fronts. panies obtaining inputs from headquarters Pinpointing the enablers of innovative (downstream activity). Almost half of foreign entrepreneurship is complex, however, subsidiaries of multilatinas operate in the because of the intricate interactions and inter- same sector as their headquarters compared dependencies between the various dimen- with 30–­ 40 percent for other regions. Sub- sions of the enabling environment that matter sidiaries of multinational corporations from for innovation. These components include the other regions are thus more likely to establish clarity and reliability of legal rights (includ- vertical (upstream and downstream) link- ing intellectual property rights) and the ages with their headquarters. The implica- judicial process, the quality of information tion is that many multilatinas fail to transfer disclosure and accounting standards, regula- knowledge to the home economy through tions and policies (including procompetition their involvement in global value chains. This policy) that affect industry and commerce, lack of integration may be exacerbated by the access to suitable financial services, the qual- border activity of fact that most of the cross-­ ity of human capital (education and skills), multilatinas takes place in large countries in and programs and policies that promote the region (Brazil, Chile, and Mexico jointly or support business development or R&D. account for 70 percent of total multilatinas’ Complexity also arises because both entre- revenues); less than 15 percent of multilati- preneurial innovation and its possible deter- nas’ revenues comes from outside LAC. minants may be affected by common factors and hence jointly determined. For instance, an economy’s contractual environment may How can policy enable simultaneously affect both access to credit innovative entrepreneurs? and innovation. In a tribute to innovation as the key to Some areas where policy action may be growth, Yale University’s Robert Shiller most fruitful can nevertheless be identified (2013) recently asserted that “capitalism is by highlighting some of the dimensions of the culture. To sustain it, laws and institutions enabling environment that are vital to inno- are important, but the most fundamental role vation and on which LAC countries signifi- is played by the basic human spirit of inde- cantly underperform. pendence and initiative.” But where should Competition is a first and highly plausi- policy makers look for remedies to cure the ble candidate. To be sure, the relationship low growth and low innovation of LAC enter- between competition and innovation may prises if not in the laws and institutions that follow an inverted U-­ shape, as Aghion and shape the enabling environment for entrepre- others (2005) compellingly argue: too much neurs? The answer surely lies well beyond the competition may weaken the incentives to traditional concern with laws and regulations innovate for firms that lack basic capabilities that impose barriers to entry per se. and are far from the technological frontier, The main policy challenges seem to be whereas too little competition may not pro- related to deeper structural features of the vide sufficient incentives to invest in innova- enabling environment for innovative entre- tion. The evidence suggests, however, that preneurship, including not only laws and LAC suffers from too little rather than too institutions but also endowments such as much competition, particularly in the mar- infrastructure and the quantity and qual- kets for inputs and nontradable services. This ity of human capital. These elements of the lack of competition undermines the incen- enabling environment are likely to be even tives to innovate, as enterprises can remain O v er v ie w   17 profitable by dint of their market power FIGURE 1.16  Actual and benchmarked index of competition in rather than their innovative efforts. Without 17 nontradable industries in selected countries or economies a perceived necessity to innovate, the private sector may not give birth to invention. United States LAC countries Bulgaria Figure 1.16 benchmarks LAC countries Romania Other countries Poland or economies in terms of revealed market concentration Canada in industries that are arguably not subject Hungary Benchmark Russian Federation to international competition.10 Most LAC Lithuania Czech Republic countries appear at the upper end of the dis- Norway Colombia tribution of the (nontradable) market con- Latvia Korea, Rep. centration index, and all but two (Colombia Japan and Brazil) exhibit average levels of market Portugal United Kingdom concentration well above their international Switzerland Macedonia, FYR benchmarks. Hence, competition should China Italy remain at the top of the policy agenda in Ireland Germany most LAC economies. Croatia A second fundamental factor behind the Serbia Belarus lack of innovation in LAC seems to be its Thailand Spain human capital gap, particularly in the edu- Sweden Austria cation quality dimension. The region lacks Finland Netherlands the type of human capital—­ engineers and Argentina scientists—­ t hat is likely to produce inno- Greece Denmark vative entrepreneurs. A country’s stock of Brazil Bosnia and Herzegovina human capital is often measured by average Singapore Australia years of schooling of the labor force and by Belgium Turkey the quality of education, assessed through Mexico standardized scholastic test scores. LAC France Philippines countries underperform international com- Malaysia Moldova parators on both measures, especially quality Israel New Zealand (Ferreira and others 2013). However, human Kuwait Hong Kong SAR, China capital for entrepreneurship and innovation Ecuador only partially overlaps with general curric- Kazakhstan Indonesia ula and is probably badly captured by gen- Chile Peru eral schooling attainment or achievements. Albania Saudi Arabia Hence, it is worth also examining the region’s India Uruguay chronic shortage of scientific and engineering Dominican Republic training. Oman Guatemala LAC has long suffered from a dearth of Bolivia Jamaica engineers: despite higher income per capita, United Arab Emirates Paraguay Argentina, Chile, and Mexico all had lower Venezuela, RB densities of engineers than Spain and Portu- Trinidad and Tobago Nicaragua gal in 1900 (figure 1.17). Such historical gaps El Salvador Costa Rica appear to be important. Maloney and Valen- Honduras cia Caicedo (2012) find a positive association 0 0.2 0.4 0.6 0.8 between engineering density in the 1900s Her ndahl index and per capita income in the 2000s. Source: World Bank, based on data from World Development Indicators and firm-­level data from Orbis. LAC countries still have fewer engineers Note: Bars show the average Herfindahl index of concentration of revenues across a selection of two-­digit nonfinancial services sectors for which data were available for more than 80 countries. than the median country and fewer than A value of 1 represents a market captured entirely by a single firm (the highest level of concen- would be expected given their current level tration); lower values indicate less concentration. Revenues were averaged across 2007–­10. Dots represent a benchmark predicted value from a regression for each sector with (log of) population and GDP (adjusted for purchasing power parity) as explanatory variables. The regression model was estimated for each of 17 sectors separately; the dots are the averages of all sectors. The regression used all available countries. The figure presents only comparator countries. LAC = Latin America and the Caribbean. 18   LATIN AMERIC AN ENTREPRENEURS FIGURE 1.17  Income and engineering density in FIGURE 1.18  Number of engineers per thousand selected economies, 1900 people in selected countries 180 Honduras LAC countries United States, north Other Guyana Engineers per 100,000 male workers 150 Uruguay countries El Salvador Benchmark 120 Brazil Argentina 90 United States Indonesia Colombia 60 United States, south Mexico Venezuela, RB Serbia 30 Portugal Spain Brazil Chile Saudi Arabia Colombia Argentina Peru Mexico Chile 0 6.0 6.5 7.0 7.5 8.0 8.5 9.0 Turkey Log of GDP per capita (US$ in 1900) Hungary Netherlands Source: Maloney and Valencia Caicedo 2012. Armenia Note: GDP = gross domestic product. Norway United States Latvia Croatia of development (figure 1.18). Even the larger Sweden and more advanced countries in the region Bulgaria (Brazil, Chile, Colombia, and Mexico) have Belgium relatively few engineers. Greece LAC students may be inclined toward Germany nonscientific studies for at least two poten- Denmark tial reasons. First, for historical reasons, Japan LAC universities have long emphasized the Malaysia humanities; law; and social, economic, and Portugal political fields of study, possibly constraining Austria their ability to educate more engineers and Lithuania scientists. Switching their emphasis would Poland require very aggressive public policy, such as Slovenia Switzerland the United States adopted when it developed Spain mining and engineering studies in the early Czech Republic 20th century. Second, young people may be New Zealand attracted to fields of studies that are relevant Slovak Republic to pressing problems faced by their societ- Ireland ies, which may explain why LAC may have Finland formed many sociologists and more macro Ukraine than micro economists. Given the progress Thailand the region has made in taming macro insta- 0 10 20 30 bility, there may be more incentives for stu- Engineering graduates per thousand dents to embark on scientific careers. That inhabitants, ages 15–24 said, a big push to expand engineering and Source: World Bank, based on data from World Development Indicators scientific education at the secondary and ter- and UNESCO 2013. tiary levels may be required to accompany Note: Bars show average number of engineering graduates per thousand people ages 15–­24. Dots are a benchmark predicted by a regression with rising demand for such careers. (the log of) population and GDP (adjusted for purchasing power parity) as the explanatory variables. The regression uses all the available countries. The figure presents only comparator countries. Data are averages for 2008–­ 10. LAC = Latin America and the Caribbean. O v er v ie w   19 Factors that affect firms’ economic perfor- FIGURE 1.19  Actual and benchmarked index of intellectual mance may also adversely affect innovative property rights in selected countries or economies, 2005 entrepreneurship, although the nexus may not be as straightforward as often believed. Guyana LAC countries Thailand Despite substantial reform, business regu- Indonesia Other countries Dominican Republic or economies lations may still hamper innovative behav- Costa Rica Benchmark ior. Which specific regulations bite and how Paraguay much damage they cause, however, remain Haiti Nicaragua questions for future research. Saudi Arabia Honduras Although the region underperforms in Grenada terms of financial services, such as long-­ Guatemala Venezuela, RB term credit and venture capital, young firms Peru Jamaica in LAC are not necessarily more credit con- Uruguay strained than young firms in other regions. Bolivia El Salvador This potential link requires careful research, Malaysia but prima facie, it is difficult to categorically Brazil Ukraine state that lack of access to finance is a sig- Russian Federation Colombia nificant cause of the region’s innovation gap. Ecuador To be sure, as documented in the report on Trinidad and Tobago Hong Kong SAR, China Financial Development in LAC (de la Torre, India Mexico Ize, and Schmukler 2012), the region’s gap in Argentina bank credit is significant and has been grow- Lithuania Turkey ing over the past 15 years. However, much New Zealand of this gap appears to be explained by LAC’s China Israel turbulent macro and financial history and by Australia Norway the shortage of promising productive projects Romania (that is, a shortage of innovation) rather than Philippines Poland by credit rationing and credit supply-side Slovak Republic constraints per se. Moreover, the constraint Singapore Chile that seems to be most relevant for bank credit Greece Switzerland supply in LAC is weaknesses in the contrac- Spain tual (rather than the informational) envi- Austria Czech Republic ronment, which can undermine both credit Korea, Rep. Portugal supply and entrepreneurial innovation. Germany The role of the contract enforcement envi- Hungary Bulgaria ronment in the region’s innovation deficit is Sweden also nuanced. Insufficient intellectual prop- United Kingdom Belgium erty rights may be an issue (figure 1.19), and France Canada other weaknesses in the contractual environ- Italy ment may also hinder innovation. But indexes Japan Finland of contract viability and the risk of expro- Ireland priation do not indicate that LAC countries Denmark Netherlands systematically underperform relative to com- United States parators in other regions. More research is 2 3 4 5 therefore needed to understand the subtleties Park index (as of 2005) of, and complex interactions and interdepen- Source: World Bank, based on data from World Development Indicators and Park 2008. dencies between, the fundamental underpin- Note: The Park index is the sum of five components: coverage of patents in eight industries; partici- nings of LAC’s peculiar combination of many pation in five international property rights (IPR) treaties; duration of protection (relative to a global standard, such as 15–­20 years for patents); the existence of up to three enforcement mechanisms; entrepreneurs and little innovation. and the existence of up to three types of restrictions on patent rights. Bars show the 2005 Park index for each country. Dots show the predicted percentage of firms from a regression that includes (the log of) population and GDP (adjusted for purchasing power parity) as explanatory variables. The regression used all available countries. The figure presents only comparator countries. LAC = Latin America and the Caribbean. 20   LATIN AMERIC AN ENTREPRENEURS Structure of the report to increase when conditions are favorable. LAC countries appear to underperform The report uses a fictional story to illustrate poorer countries in terms of both entry and the characteristics of entrepreneurs and the survival rates. complex tradeoffs they face. Javier Vizzi, a Chapter 5 studies the role of foreign direct young man from Mendoza, Argentina, had investment and multinational corporations in a comfortable middle-­ class upbringing. His fostering a more entrepreneurial LAC. It first parents provided him with a decent educa- analyzes how foreign-­ owned firms operating tion, and he did not waste the opportunity. in LAC generate positive aggregate and firm-­ After graduating from a local university, he level spillovers. It then turns to the emergence found his first job with a local winemaker of multinational corporations from LAC and rapidly moved up the ranks. After direct- (multilatinas) and their impact on LAC’s ing the Buenos Aires branch of the winery entrepreneurial potential. for a few years, Javier started his own com- Chapter 6 concludes by mapping the ele- pany. He wanted to produce higher-­ quality ments of an enabling environment in LAC, in wines with potentially higher profit margins. an attempt to explain the region’s innovation This endeavor required extensive experi- gap. Its brief review of empirical benchmark- mentation, which his previous employer was ing exercises indicates a few priority policy unwilling to undertake. Javier took risks and areas. engaged in activities that were uncommon in his region. He hired international consul- tants to teach him the latest techniques in Notes wine-­ making and marketing experts to find 1. World Bank calculations based on data the best ways to sell his wines. A few years from 2010 household surveys from 15 LAC after opening, the winery had 50 employees countries. and exported a small selection of bottles to 2. Schumpeter (1911) defines entrepreneurship the United States. as “(1) The introduction of a new good … or The rest of the report comprises five chap- of a new quality good. (2) The introduction ters that track the difficult choices that entre- of a new method of production…. (3) The preneurs like Javier typically face at home opening of a new market…. (4) The conquest and abroad. Chapter 2 discusses the creation of a new source of supply of raw materials of new firms and firm dynamics in LAC. It manufactured goods…. (5) The car- or half-­ rying out of the new organization of any pays particular attention to the nature of industry….” the business being created, distinguishing 3. The LAC region comprises the following between formal and informal and small and countries: Antigua and Barbuda, Argentina, large enterprises. The chapter sets the scene Bolivia, Brazil, Chile, Colombia, Costa Rica, for the rest of the report by elaborating on Dominica, Dominican Republic, Ecuador, the key distinction between transformational El Salvador, Grenada, Guatemala, Guyana, and low-­ growth entrepreneurs. Honduras, Jamaica, Mexico, Nicaragua, Par- Chapter 3 focuses on barriers to innova- aguay, Peru, St. Kitts and Nevis, St. Lucia, St. tion faced by survivors (incumbent firms) Vincent and the Grenadines, Suriname, Trin- along both the product and process dimen- idad and Tobago, Uruguay, and República sions. It also discusses the policy areas gov- Bolivariana de Venezuela. 4. Throughout this chapter we use the following ernments can explore to enhance innovative groups of economies unless otherwise noted. entrepreneurship. LAC5 includes Argentina, Brazil, Chile, Chapter 4 examines a different form of Colombia, and Mexico. Other LAC includes entrepreneurship, namely, the exploration Bolivia, Costa Rica, the Dominican Repub- of new markets through exports. The mes- lic, Ecuador, El Salvador, Guatemala, Hon- sage is loud and clear: the key to success duras, Nicaragua, Paraguay, Peru, Uruguay, in export markets is survival, which tends and República Bolivariana de Venezuela. O v er v ie w   21 Caribbean includes Antigua and Barbuda, References Cuba, Dominica, Grenada, Guyana, Haiti, Jamaica, St. Kitts and Nevis, St. Lucia, St. Aghion, P., N. Bloom, R. Blundell, R. Griffith, Vincent and the Grenadines, Suriname, and and P. Howitt. 2005. “Competition and Inno- Trinidad and Tobago. ECA (Eastern Europe vation: An Inverted U Relationship.” Quar- and Central Asia) includes Albania, Armenia, terly Journal of Economics 120 (2): 701–­ 28. Azerbaijan, Belarus, Bosnia and Herzegovina, Alfaro, L., and M. X. Chen. 2013. “Market Bulgaria, Georgia, Kazakhstan, Latvia, Lith- Reallocation and Knowledge Spillovers: The uania, FYR Macedonia, Moldova, Romania, Gains from Multinational Production.” NBER the Russian Federation, Serbia, Turkey, Turk- Working Paper 18207, National Bureau of menistan, Ukraine, and Uzbekistan. EAP4 Economic Research, Cambridge, MA. includes Indonesia, Malaysia, the Philip- Bloom, N., and J. Van Reenen. 2007. “Measuring pines, and Thailand. High-income economies and Explaining Management Practices across include Australia; Canada; Hong Kong SAR, Firms and Countries.” Quarterly Journal of China; Israel; Japan; the Republic of Korea; Economics 122 (4): 1351–­ 408. Kuwait; New Zealand; Oman; Saudi Ara- Busso, M., L. Madrigal, and C. Pagés-Serra. bia, Singapore; Switzerland; the United Arab 2012. “Productivity and Resource Misalloca- Emirates; the United States; and all countries tion in Latin America.” BE Journal of Macro- in the European Union not included in ECA. economics 13 (1): 1–­30. The set of economies from each group used in Davis, S. J., J. Haltiwanger, and S. Schuh. 1996. figures throughout this chapter varies accord- Job Creation and Destruction. Cambridge, ing to data availability. MA: MIT Press. 5. The typical LAC firm at the 90th percentile de la Torre, A., A. Ize, and S. L. Schmuk- has fewer than 25 employees, as opposed to ler. 2012. Financial Development in Latin 40 in ECA and high-­ income countries and America and the Caribbean: The Road almost 55 in East Asia and the Pacific (EAP4). A head . Washington, DC: World Bank. 6. Grenada is a regional outlier. Its performance ht t p s: //op e n k nowle d g e .worldb a n k .or g reflects the small number of firms rather than /handle/10986/2380. the high incidence of new products. Eslava, M., and J. Haltiwanger. 2013. “Young 7. The OECD (2002) Frascati Manual on R&D Businesses, Entrepreneurship, and the Dynam- statistics, which is used around the world, ics of Employment and Output in Colombia’s excludes investments in soil analysis and min- Manufacturing Industry.” Working paper, eral exploration from R&D activities. Conse- CAF, Caracas, Venezuela. quently investments in innovation in agriculture Enterprise Surveys (database). World Bank Group, and mining tend to be underreported. Washington, DC. http://www.enterprisesurveys 8. R&D data are classified as “productive-­ .org/. sector” R&D when financing comes from Exporter Dynamics Database. World Bank, a company that participates in the market. Washington, DC. http://data.worldbank.org Companies can be publicly owned, blurring /data-catalog/exporter-dynamics-database. the distinction between “private” and “pub- Fernandes, A. M., D. Lederman, and M. Gutierrez-­ lic” R&D. In this report, as in others, such as Rocha. 2013. “Export Entrepreneurship and Pagés-Serra (2010), the term private is used to Trade Structure in Latin America during Good characterize “productive-­ sector” R&D. and Bad Times.” World Bank Policy Research 9. This exercise took into consideration cross-­ Working Paper 6413, Washington, DC. country differences in GDP per capita, sec- Ferreira, F. H. G., J. Messina, J. Rigolini, L.-F. toral composition, and year-­ specific effects López-­ Calva, M. A. Lugo, and R. Vakis. 2013. (such as the global recession of 2008–­ 09). Economic Mobility and the Rise of the LAC 10. The distinction between tradables and non- Middle Class. Washington, DC: World Bank. tradables is important. Domestic market Gindling, T. H., and D. L. Newhouse. 2012. concentration could be high in the sense that “Self-­ Employment in the Developing World.” few domestic firms participate in an industry, Policy Research Working Paper 6201, World but if domestic firms compete with imports, Bank, Washington, DC. domestic market concentration would be a Haltiwanger, J., R. Jarmin, and J. Miranda. poor proxy for competition. 2013. “Who Creates Jobs? Small versus Large 22   LATIN AMERIC AN ENTREPRENEURS versus Young.” NBER Working Paper 16300, Orbis (database). Bureau van Dijk, Amsterdam, the National Bureau of Economic Research, Cam- Netherlands. http://www.bvdinfo.com/en-us bridge, MA. /products/company-information/international Hausmann, R., and D. Rodrik. 2003. “Develop- /orbis-(1). ment as Self-­Discovery.” Journal of Develop- Pagés-­ Serra, C., ed. 2010. The Age of Productiv- ment Economics 72 (2): 603–­ 33. ity: Transforming Economies from the Bottom Javorcik, B., and M. Spatareanu. 2005. “Disen- Up. Washington, DC: Inter-­ A merican Devel- tangling FDI Spillover Effects: What Do Firm opment Bank. Perceptions Tell Us?” In Does Foreign Invest- Park, W. 2008. “International Patent Protection, ment Promote Development?, edited by The- 1960–­2005.” Research Policy 37: 761–­ 66. odore Moran, Edward Graham, and Magnus Schumpeter, J. 1911. Theorie der Wirtschaftlichen Blomström, 45–­ 72. Washington, DC: Center Entwicklung. Leipzig: Duncker & Humblot. for Global Development. Seker, M. 2013. “Innovation Performance in Lederman, D., and W. Maloney. 2003. “R&D Latin America and Caribbean Region.” Back- and Development.” Policy Research Working ground paper for this report. Paper 3024, World Bank, Washington, DC. Shiller, R. S. 2013. “Why Innovation Is Still Capi- Lederman, D., M. Olarreaga, and L. Zavala. talism’s Star.” New York Times, August 17. 2013. “Export Promotion and Firm Entry UNESCO (United Nations Educational, Scien- and Survival: Evidence from a Panel of LAC tific and Cultural Organization). 2013. Data Firms.” World Bank, Washington, DC, and Center, Institute for Statistics. http://stats.uis University of Geneva, Geneva. .unesco.org/unesco/ReportFolders/Report Lerner, J., and A. Schoar. 2010. “Introduction Folders.aspx. to International Differences in Entrepreneur- USPTO (U.S. Patent and Trademark Office). ship.” In International Differences in Entre- 2012. “Patents by Country, State, and Year: preneurship, 1–­13. Cambridge, MA: National All Patent Types.” Washington, DC. http:// Bureau of Economic Research. www.uspto.gov/web/offices/ac/ido/oeip/taf Maloney, W. F., and M. Sarrias. 2012. “What /cst_all.htm. Makes LAC Managers So Bad?” World Bank, Vollrath, T. 1991. “A Theoretical Evaluation Washington, DC. of Alternative Trade Intensity Measures of Maloney, W. F., and F. Valencia Caicedo. 2012. Revealed Comparative Advantage.” Review of Part II: Engineers, Innovative Capacity and World Economics 2: 265–­80. Development in the Americas. World Bank, World Bank. 2013. Doing Business 2013: Smarter Washington, DC. Regulations for Small and Medium- ­ S ize Moran, T. 2001. Parental Supervision: The New Enterprises. Washington, DC: World Bank. Paradigm for Foreign Direct Investment and World Bank Group Entrepreneurship Snap- Development. Washington, DC: Institute for shots (WBGES). World Bank, Washington, International Economics. DC . http://w w w.doingbusiness.org /data National Science Board. 2012. “Science and Engi- /exploretopics/entrepreneurship. neering Indicators 2012.” National Science World Development Indicators (database). World Foundation, Arlington, VA. Bank, Washington, DC. http://data.worldbank OECD (Organisation for Economic Co-­ operation . o r g /d a t a - ­c a t a l o g / w o r l d -­d e v e l o p m e n t and Development). 2002. Frascati Manual: -­indicators. Proposed Standard Practice for Surveys on World Integrated Trade Solution (WITS) (data- Research and Experimental Development. base). World Bank, Washington, DC. http:// Paris: OECD. wits.worldbank.org/wits/. Entrepreneurship, Entry, and the Life Cycle of Firms in Latin America 2 and the Caribbean: Are All Forms of Firm Creation Entrepreneurial? Contrary to popular perception, Latin America and the Caribbean has a vibrant entrepre- neurial sector. Indeed, the share of entrepreneurs, employers, and formal businesses is larger than in other middle-income regions. Firms are smaller than in other regions at similar levels of development, however, with even the largest firms creating fewer jobs than their coun- terparts in other regions. These patterns are reproduced in other environments. After long periods in the United States, migrants from the region are about as likely as natives to own small businesses, but few of them own large, employment-generating firms. E ntrepreneurship is multifaceted; sim- new market. The two extremes are inversely ple defi nitions fail to capture the het- related. If there are many dynamic entrepre- erogeneity of innovative acts included neurs in the economy, there will be an abun- under this umbrella. Most defi nitions view dance of good jobs, reducing the incentives the creation of new fi rms as a critical dimen- for start- ups with low growth potential. sion of the entrepreneurial process. Indeed, Conversely, too few innovative entrepreneurs entry has been considered central to the will generate few employment opportunities, complex process of entrepreneurship since pushing some workers who may not have the seminal work of Schumpeter (1934). an innate ability or interest in running their However, not all entry is the same: simply own business to accept employment oppor- working for oneself or creating an enterprise tunities with low growth potential. Although is not the same as engaging in job- creating both types of entrepreneurs are found in all entrepreneurship. countries, the lack of good jobs in developing Some business owners create their fi rms countries suggests that low-growth entrepre- very much in the Schumpeterian tradition, neurship may be more prevalent at low levels with the goal of creating something new of development. to bring to the market, revolutionizing the Not surprisingly, this tremendous hetero- economy, and creating jobs. Others create geneity in entry motives translates into a no their fi rms in response to grim employment less heterogeneous picture in the distribution prospects, as a mean of subsisting rather of incumbent firms. As in a forest, where than creating a new product or entering a small and large trees coexist, large and small 23 24 LATIN AMERICAN ENTREPRENEURS firms compete even within very narrowly bias toward smaller firms generates insuf- defined sectors. Large firms are the larg- ficient formal employment opportunities. In est employers in every sector. In the United the absence of better employment prospects, States, for instance, the largest 5 percent of many people end up working for themselves, firms accounted for more than 75 percent fueling a vicious cycle of small size and few of employment by the end of the 2000s; good jobs for future job seekers. in Mexico, the largest 10 percent of firms The chapter ends by investigating whether accounted for 70 percent of employment the behavior of entrepreneurs in LAC is (Bartelsman, Haltiwanger, and Scarpetta linked to the environment in which they 2009). At the same time, the vast majority operate or has deeper causes, perhaps linked of firms are small. In high-income countries, to cultural roots or human capital character- about 70 percent of firms had fewer than five istics. It fi nds that historically, people in the employees in 2010 (Klapper and Randall region have not been predisposed to become 2012). In Argentina, Brazil, Chile, Colombia, entrepreneurs who transform the business and Mexico (LAC5), 9 of every 10 firms have environment. Most large fi rms in the region fewer than five employees. Indeed, slightly at the beginning of the 20th century were more than 60 percent of business owners in foreign owned. Even in the United States—in the Latin American region report having no an environment that is more conducive to paid employees (Klapper and Randall 2012). entrepreneurship—people from LAC are less This chapter has a double purpose. First, entrepreneurial than migrants from other it examines the process of business creation regions of the world. But there is some room in Latin America and the Caribbean (LAC), for optimism. Migrants from LAC slowly benchmarking its performance against that adapt to the new business environment. After of other regions and characterizing the long periods in the United States, they catch nature of entry across countries. Second, it up with natives and migrants from other studies the life cycle of fi rms in the region, regions in ownership of small- scale firms. the frequency with which they grow, and dif- They continue to lag in ownership of large ferences in the process across countries and companies, however. type of firm. The chapter does not provide an in-depth analysis of the behavior of incum- bent firms, the subject underlying most of the Low-level entrepreneurs, following chapters. Instead, it provides an high-level entrepreneurs, overview of business dynamics in the region, and employees leaving the discussion of the determinants of The story of a fictional family of Italian these dynamics for the rest of the report. immigrants illustrates the different motiva- The chapter shows that there is substan- tions that may trigger the creation of new tial creation of new fi rms in LAC countries, firms. The Vizzis settled in Mendoza, Argen- at both the low and high ends of the entre- tina, at the beginning of the 20th century to preneurial spectrum. Indeed, in the formal work in the fields. The eldest of their three sector, the process of creation (and destruc- children, Maria, had no opportunity to go to tion) does not differ much from that found in school. As soon as she learned to read and other regions at similar levels of development write, her parents asked her to help with the and even shares some characteristics found crops. Her brothers, Lucio and Javier, were in more advanced countries. However, the luckier. By the time they reached school age, vast majority of new businesses in LAC are Maria was 15 and contributing to the house- microfirms that will remain tiny throughout hold income, so their parents could pay for their life span. Even fi rms that grow rapidly their studies. The boys did not waste their never catch up in size with firms the same opportunities, finishing high school and age in other parts of the world. This strong enrolling in college. Their parents were very E N T R E P R E N E U R S H I P, E N T R Y , A N D T H E L I F E C Y C L E O F F I R M S I N L A T I N A M E R I C A A N D T H E C A R I B B E A N 25 proud to see their sons become economists loan, bought a trolley, and in a matter of (although they never quite understood what weeks was selling alfajores on the streets of economists exactly did). Mendoza. She had no intentions to grow her The three siblings saw few opportuni- business, preferring to stay below the radar ties for progress in the fields; as soon as they screen of the government. Hence, she never could, they went to the province’s capital to formally registered her business. look for jobs. Being hard workers, they soon Lucio and Javier were luckier with the found themselves with their fi rst paid jobs. recession. Lucio worked in the public sec- Maria worked in a restaurant at the train tor, where wages and jobs were relatively station. Lucio and Javier started as clerks. insulated from business cycle fluctuations. Lucio found a job as an accountant for He was happy with his job, which was not the local government. Javier found a job at very challenging but paid a good salary and Vinos Torreón, a family fi rm that produced provided him with a lot of free time. He was and sold one of the soon to be well-known not ambitious. He accepted a promotion as wines from the area. Life was good. Wages a manager, but after a few stressful months were not spectacular, but they were enough returned to his old job. All he wanted was to to allow all three siblings to save some money enjoy a quiet life, perhaps raising a couple of and send some cash to their parents every children one day. other month. The 1981 recession hit the winery, but Life was soon to change for them all. In Javier had no trouble keeping his job. He 1979, passenger service between Mendoza had the rare ability to create an affable and and Chile was terminated. Business at the relaxed work environment where co-workers restaurant fell steadily, and by the end of were happy, becoming more reliable and 1980, Maria found herself searching for a more committed to their job. Javier enjoyed new job. the process and knew he was good at it. His At first Maria was optimistic. During bosses also recognized his talent. Hence, it her time at the restaurant, she learned how came as no surprise when his company asked to cook, which she thought would give her him to run the small office the company was plenty of opportunities in the many restau- planning to open in Buenos Aires. rants in the city. But the 1981 recession had The Buenos Aires office was a success. hit hard. As a result, she was able to find only Javier learned a lot about the wine busi- temporary jobs, which provided no job stabil- ness and developed a wide network of cli- ity and did not allow her to use her recently ents. Buenos Aires exposed him to the great acquired cooking skills. French, Italian, and Spanish wines, which After six months, money was starting to he learned to love. He had talked with wine become a serious concern. One day Maria experts around the world and had the con- was walking down the street after another viction that conditions in Mendoza were failed attempt to see a chef when she stopped right for producing top- quality wines, not at the gate of a school to watch a group of just decent table wines like Vinos Torreón. children playing football. When the bell rang, Prospects were promising, but every time he the kids ran to the exit, briefly greeting the tried to convince the company’s owners of parents waiting at the gates before scram- the need to move upward in the value-added bling toward a little trolley from which a man chain, they looked at him with incredulity. was selling alfajores de maizena, Argentina’s Feeling increasingly frustrated, he decided sweet biscuits. In that second Maria saw the the time had come to move on, perhaps to light: “I could do this!” she thought out loud. start his own business. Soon enough, she put her passion and the Javier quit his job in 1987 and went back skills acquired at the restaurant into a new to Mendoza. After talking to more than 100 venture. She asked her brothers for a small local farmers, he found the right hill on which 26 LATIN AMERICAN ENTREPRENEURS to plant his grapes. The soil was perfect and similar teachers and peers, but they have dif- the orientation ideal. Now he needed money. ferent abilities and ambitions. Javier has the Obtaining it proved more difficult than Javier rare ability to motivate others. He can make had expected. Local bankers in Mendoza did everyone more productive by identifying peo- not understand his business plan. Why waste ple’s relative strengths and combining them all that money bringing oak barrels from effectively. Lucio does not have this ability. France when local barrels worked just fi ne? Although he is a good worker, he is not moti- They thought he was a visionary enthusiast. vated by the challenge of change, and he finds In Buenos Aires it was not much easier, but management stressful. He will never be inter- in the end he managed to convince a banker ested in running his own business. (These of the merits of his project. Once money was entrepreneurial skills are studied in detail in secured, he bought the land, registered the CAF 2013.) new business with the relevant authorities, Entrepreneurs like Maria are abundant. hired a small group of laborers, and planted On average, 28.8 percent of income earners the first grapes. In 1990, he produced his first in LAC are self-employed or small employ- wine; by 1993, he employed 50 workers. ers.1 Few of them ever hire workers. In This story highlights some of the features Colombia, for example, only 0.3 percent of entrepreneurism in every economy in the of the self- employed became employers in world. All three siblings started as wage earn- a three-year period (Mondragón and Peña ers; two of them decided to become entrepre- 2010). In contrast, high- growth entrepre- neurs, albeit for different reasons and with neurs like Javier are extremely rare: less than different sets of skills. Maria and Javier have 0.4 percent of income earners in LAC own different education levels, which are typically a business employing 50 workers or more. 2 associated with different levels of ability to However, their contribution to employment transform a raw idea into a business project. is huge: some 45 percent of employees in On average, transformational entrepreneurs LAC work for medium-size and large fi rms like Javier have more years of schooling than (firms employing 50 workers or more). Wage low- growth- potential entrepreneurs like employees like Lucio represent the most com- Maria. The nature of the business they cre- mon type of worker in the region, accounting ated is also different. Maria adopted a well- for the remaining 70.8 percent. established business model. Her prospects for growth are probably low. In contrast, Javier created something new by introducing high- Theoretical framework quality wines in an area specialized in table Most of this report focuses on the different wines. entrepreneurial activities that innovators like Education is clearly related to the type of Javier put in place. Before focusing on Javier- business created. As shown later in this chap- type entrepreneurs, this chapter starts by ter, formal business owners are much more comparing the differences between the three likely to have attended college than informal siblings highlighted in the story in a more business owners. On average, across coun- formal setting. tries in LAC, 21 percent of people with a The main ingredients of this fictional story tertiary degree own a business. About 15 per- can be built into a simple theoretical frame- cent registered their business; 6 percent did work following Poschke (2013a). Consider not. Among people with only primary educa- a population that is heterogeneous in “abil- tion the pattern is reversed: about 9 percent ity,” as proxied, for example, by educational own an informal business and 5 percent own attainment. Workers with different abilities a formal enterprise. Education is not the only can choose to work for themselves (that is, determinant of entrepreneurship, as the con- to become entrepreneurs) or for someone trast between Javier and Lucio shows. Javier else (that is, to work as wage employees). and Lucio went to the same schools, had The value of dependent employment can be E N T R E P R E N E U R S H I P, E N T R Y , A N D T H E L I F E C Y C L E O F F I R M S I N L A T I N A M E R I C A A N D T H E C A R I B B E A N 27 FIGURE 2.1 Model of entrepreneurship a. Two types of entrepreneurs b. Taxes and evasion B’ Value of employment (wages) Value of employment (wages) B C A D Employees A’ aL aH Ability Ability Subsistence Transformational entrepreneurs entrepreneurs Employment Entrepreneurship Source: Poschke 2013a. thought of as a linear function of ability: the some cases, entrepreneurship serves as a buf- more able an individual is, the higher the fer before workers find better employment value of becoming an employee (for simplic- opportunities. In others, where the income ity, measured by wages), as shown by the red obtained from the small business is higher lines in figure 2.1. than the wage offered by the market, low- What is the value of entrepreneurship? In end entrepreneurship becomes permanent. contrast with wage employment, the value These low-ability entrepreneurs are labeled of entrepreneurship does not need to be lin- low- growth-potential entrepreneurs or, for ear in ability. At the high end of the ability simplicity, low-growth entrepreneurs. spectrum, the value of entrepreneurship is Hence, the value of entrepreneurship is high, because the value of a great entrepre- a convex function of individual talent (and neurial idea or great management skills can education), as depicted by the orange curves be spread across inputs used in the firm, in figure 2.1.3 The payoff functions for wage augmenting these inputs. The idea that an employment and entrepreneurship intersect entrepreneur’s ability multiplies the value of twice, at aL and aH . Individuals with a < aL inputs in the production process goes back or a > aH become entrepreneurs; individuals to the seminal work of Lucas (1978). Hence, with intermediate skills (for example, Lucio for very high-ability individuals, becoming in our fictional story) become employees. an entrepreneur is more lucrative than being High-ability individuals like Javier become an employee. These individuals are “transfor- entrepreneurs because their ability allows mational entrepreneurs” (following Schoar them to expand the marginal product of the 2010), high-ability entrepreneurs, or high- firm’s inputs. Low- ability individuals like growth-potential entrepreneurs. Maria become entrepreneurs because the Entrepreneurship is also more valuable expected payoff from entrepreneurship is than employment at the low end of the abil- higher than the payoff from wage employ- ity distribution, for a variety of reasons. Self- ment, a finding that is very much in line with employment may serve as an alternative to the evidence reported by Maloney (2004). dependent employment after job loss, when Thus, there is no market segmentation in the finding a job takes a long time. This motiva- model. High-ability entrepreneurs are likely tion is important in developing countries. In to run larger, more complex firms than their 28 LATIN AMERICAN ENTREPRENEURS low-ability counterparts— a prediction con- entrepreneurship moves to the left, from A firmed by the data analyzed in this chapter. to Aʹ. However, the reduction in the num- To be sure, ability and the relative pay- ber of high- ability entrepreneurs reduces offs of entrepreneurship versus paid employ- the number of large firms hiring workers, ment are not the only factors determining reducing the wage rate, as illustrated by the an individual’s occupational choice. Some movement to the right of the wage schedule. individuals with a natural talent for entre- Low-ability workers who previously found it preneurship may dislike the risk involved in advantageous to work as employees now find entrepreneurism. Alternatively, employees dependent employment less valuable, opting with no particular talent for entrepreneurism for entrepreneurship. In the new equilibrium, may want to be their own bosses. They may the number of low- ability entrepreneurs prefer opening their own business even if the increases, as depicted by point D. Thus, when monetary value of dependent employment is taxes fall more than proportionally on large higher. If such preferences are uncorrelated firms, rising taxation reduces employment in with ability, the main insights of this simpli- larger firms and, through lower wages, stim- fied theoretical framework remain valid. In ulates the creation of small enterprises. the presence of preference heterogeneity, it In the real world, the distinction between would still be true that the average ability of low- and high- ability entrepreneurs is high-end entrepreneurs is higher than that of blurred; it is probably better approximated employees and low-end entrepreneurs. by a bimodal distribution with a concentra- One of the important insights of the the- tion of entrepreneurs at the low and high ends oretical discussion here is that it helps us of the ability distribution but a continuum understand that, in general equilibrium, fac- across the ability spectrum.5 The heterogene- tors affecting fi rm profits also affect wages, ity of real world experiences is captured in a and thus may alter both the value of entrepre- very rough manner in the data. neurship and that of employment. An excel- This chapter uses several proxies for the lent example is the effect of changes in fi rm two types of entrepreneurs that help approxi- size– related taxes. Many rules and regula- mate the heterogeneous nature of entrepre- tions apply only to firms above a certain size neurship around the world in general and in or are enforced more strictly for larger firms. LAC in particular. Depending on the data, In many high-income countries, for example, low-ability entrepreneurs are self- employed, small fi rms are exempted from employment own unregistered businesses, have no protection and severance payments.4 In low- employees, and were pushed into entrepre- and middle-income countries, where tax eva- neurship by lack of opportunity in the for- sion is pervasive, taxation is expected to fall mal sector. High- ability entrepreneurs are more than proportionally on larger firms. employers, own registered businesses, and Larger firms also face stricter enforcement of were pulled into entrepreneurship because payment of nonwage benefits to workers, as they had a great idea or saw a good business Almeida and Carneiro (2011) show for Brazil. opportunity. Panel b of figure 2.1 shows the impact of an increase in fi rm size– contingent taxes Employers, employees, on entrepreneurship. The increase in taxes and the self-employed reduces firms’ profits, muting the incen- tives to become a high-ability entrepreneur. One fundamental distinction is the ability or The threshold for high-ability entrepreneurs willingness of the entrepreneur to hire work- shifts to the right, from point B to C. At the ers. Our framework predicts that transfor- low end of the ability distribution, the gen- mational entrepreneurs will run larger fi rms eral equilibrium effects become fundamental than low-growth entrepreneurs, because their for occupational choice. If wages were fixed, comparative advantage lies precisely in orga- the share of low-ability entrepreneurs would nizing the working environment. A rough also fall, as the threshold for low- ability empirical counterpart of this distinction that E N T R E P R E N E U R S H I P, E N T R Y , A N D T H E L I F E C Y C L E O F F I R M S I N L A T I N A M E R I C A A N D T H E C A R I B B E A N 29 is easily found in household surveys around of occupation shares—including agricultural the world is the separation between employ- workers, self- employed, employers, wage ers and self-employed.6 employees, and nonpaid employees (in gen- Following Gindling and Newhouse eral family members who work in the family (2012), figure 2.2 examines the correlation business without a regular wage)—with gross FIGURE 2.2 Occupational choice and GDP per capita, 2010 a. Agricultural workers b. Nonagriculture, wage and salaried 100 100 NonAg, wage and salaried (%) Agricultural workers (%) 80 80 CHL URY 60 60 CRT JAM VEN GTM SLV SLV DOM COL BOL PRY ECH 40 HTI HND 40 HND PER PER BOL GTM PRY ECU HTI 20 COL JAM VEN 20 SLV MEX DOM CRI CHL URY 0 0 6 7 8 9 10 11 6 7 8 9 10 11 Log of GDP (PPP) per capita Log of GDP (PPP) per capita c. Nonagriculture, unpaid d. Nonagriculture, employer 6 8 SLY NonAg, employer (%) 4 6 NonAg, unpaid (%) PER CRI HND ECU BOL MEX ECU GTM PRY VEN VEN PRY 4 SLV MEX 2 DOM GTM PER URY URY BOL COL HND COL CRI CHL CHL JAM HTI JAM 2 DOM 0 HTI 0 –2 6 7 8 9 10 11 6 7 8 9 10 11 Log of GDP (PPP) per capita Log of GDP (PPP) per capita e. Nonagriculture, own account LAC countries Non-LAC countries 40 HTI NonAg, own account (%) DOM 30 COL VEN HND SLV PER BOL URY 20 PRY ECU JAM CHL GTM CRI MEX 10 0 6 7 8 9 10 11 Log of GDP (PPP) per capita Source: World Bank based on Gindling and Newhouse 2012 and data from World Development Indicators. Note: Curves show quadratic fitted values in each panel. GDP = gross domestic product. LAC = Latin America and the Caribbean. NonAg = nonagricultural. PPP = purchasing power parity. 30 LATIN AMERICAN ENTREPRENEURS domestic product (GDP) per capita across 74 The simple theoretical framework also countries.7 At very low levels of GDP per cap- generates an implicit distribution of income, ita, the vast majority of workers are involved in which low-growth entrepreneurs earn less in primary activities or, if they work outside than wage employees, who in turn earn less agriculture, unpaid labor. As GDP per cap- than transformational entrepreneurs. Indi- ita rises, the share of workers in agriculture vidual heterogeneity with regard to prefer- declines and self-employment increases. This ences for entrepreneurship would widen the pattern is consistent with a move to cities, distribution of income within each group, where a new form of informal employment, but it should still be the case that on aver- nonagricultural self- employment, is com- age the income of the transformational entre- mon. As GDP per capita continues rising, the preneurs exceeds that of employees, which shares of self- employed and unpaid family exceeds the income of the self- employed. In workers decline, hand in hand with mono- seeking empirical support for this prediction tonic increases in the shares of employers and of the framework, we continue with the par- paid employees.8 Although these patterns are allel between self-employed and low-growth- obtained from a cross-section of countries, potential entrepreneurs, as well as employers they are very consistent with the evolution of and high- growth entrepreneurs, studying self-employment over time that takes place as their distribution of income together with countries develop.9 that of wage employees in LAC countries They are also consistent with our sim- using data from household surveys.10 ple theoretical framework if technological Figure 2.3 shows the distribution of change more than proportionally benefits annual income across the three groups for the high-ability individuals. In this scenario, a LAC5 countries as a group and for 11 other pattern that is the reverse of that shown in countries in the region for which comparable figure 2.1 (panel b) emerges. Technologi- data is available (“Other LAC”).11 In both cal advances push high- ability individuals groupings, the differences in the distributions into entrepreneurship. At the other end of confirm the predictions of the theory. On the spectrum, better technologies provide average, employers dominate the income dis- incentives for low-ability individuals to enter tribution of employees, and the lowest paid entrepreneurship, but higher wages induced workers are own-account workers. by technical change more than outweigh the There is an important exception to this direct effect of technology on occupational rule. Panels c and d present the cumulative choice. Thus, technical change reduces the distribution functions of the three groups. share of low- ability workers moving into The horizontal differences are informative self-employment. about the income distances between each Considering its level of development, LAC group at each percentile of the income dis- stands out as a fairly entrepreneurial region tribution. In both LAC5 and Other LAC, when benchmarked against the rest of the employers are better off than paid employees world. Its share of employers is well above only after the 20th percentile; the bottom the share predicted by GDP per capita (panel 20 percent of paid employees do better than d in figure 2.2). However, these employers the bottom 20 percent of employers. do not generate sufficient wage or salaried A second important stylized fact is that employment, as the share of own- account the distributions of entrepreneurs (both low- workers is also above expected levels (panel e end and high-end) have higher variances than in figure 2.2). These data suggest that there is the distribution for employees. This pattern something in the nature of the fi rms created may indicate the ex post realization of one in LAC that prevents them from generating important dimension of entrepreneurship not sufficient paid employment for the working- discussed so far: risk.12 High- and low-ability age population. This report tries to provide entrepreneurs appear to face higher ex ante some answers as to why this is the case. risk than paid employees. E N T R E P R E N E U R S H I P, E N T R Y , A N D T H E L I F E C Y C L E O F F I R M S I N L A T I N A M E R I C A A N D T H E C A R I B B E A N 31 FIGURE 2.3 Income distribution in Latin America and the Caribbean by type of occupation, circa 2011 a. Distribution in LAC5 b. Distribution in Other LAC 0.8 0.6 0.6 0.4 Density Density 0.4 0.2 0.2 0 0 0 2 4 6 8 10 0 2 4 6 8 10 Log of labor income Log of labor income c. Cumulative distribution in LAC5 d. Cumulative distribution in Other LAC 1.0 1.0 Cumulative population Cumulative probability 0.8 0.8 0.6 0.6 0.4 0.4 0.2 0.2 0 0 0 2 4 6 8 10 0 2 4 6 8 10 Log of labor income Log of labor income Employer Wage employee Self-employed Source: Socio-Economic Database for Latin America and the Caribbean (SEDLAC). Note: Distribution includes people ages 25– 65 years with positive income. Distribution is weighted so that each country has the same importance. Outliers (points in the top or bottom 0.5 percent for each country and category) are excluded. For countries included in each group, see note 11. Low-growth and high-growth work, or for their own or their family’s busi- entrepreneurs ness. Respondents who answered “yes” to Self-employment may provide a rough proxy being self-employed but “no” to being a busi- for low-growth entrepreneurs, but the group ness owner were classified as self- employed is highly heterogeneous. Professionals, and not considered as business owners in this including doctors, architects, lawyers, and analysis (Klapper and Randall 2012). journalists, often work as freelancers and On average, 15 percent of adults in LAC are consequently self- employed, even if they report owning a business. This figure is do not necessarily own a business. The dis- exactly in line with the average in the rest tinction between self-employed workers and of the developing world. It hides substantial business owners is blurred and not easily within-group heterogeneity, however: owner- identifiable from household surveys. How- ship ranges from less than 10 percent in Uru- ever, the Gallup World Poll Survey (described guay and Panama to more than 20 percent in box 2.1) separates the two. People who in Bolivia, Colombia, Ecuador, and Haiti. responded affirmatively to the question “Do An additional 12 percent of adults in LAC you currently own a business?” were clas- report being self-employed but not owning a sified as business owners. Adults were clas- business. sified as self- employed if they worked even A follow-up question was added in the minimally in the last seven days for them- Gallup World Poll Survey to investigate selves, as a freelancer, performing contract the importance of formal versus informal 32 LATIN AMERICAN ENTREPRENEURS BOX 2.1 Main databases used in the study Exporter Dynamics Database SEDLAC The Exporter Dynamics Database covers measures The Socio- Economic Database for Latin America of exports growth and selected characteristics of and the Caribbean (SEDLAC) compiles the micro- exporters in 38 developing and 7 developed coun- data from the main household surveys carried out tries. The fi rm-level information, collected directly in LAC countries. Great effort is made to standard- from customs information, is available primar- ize the data to allow cross- country comparability. ily for the period 2003–10. Information includes The database includes information from more than basic characteristics of exporters (numbers, size, 200 household surveys carried out in 25 countries growth); their concentration and degree of diver- (Argentina, the Bahamas, Belize, Bolivia, Brazil, sification in products and markets; their dynam- Colombia, Costa Rica, Chile, Dominica, the Domin- ics (entry, exit, and survival); and the average unit ican Republic, Ecuador, El Salvador, Guatemala, prices of the products they trade. More informa- Guyana, Haiti, Honduras, Jamaica, Mexico, Nica- tion is available at http://econ.worldbank.org ragua, Panama, Paraguay, Peru, Suriname, Uruguay, /exporter-dynamics-database. and República Bolivariana de Venezuela). In each period, the sample of countries represents more than Gallup World Poll Survey 97 percent of the total population. The database The Gallup World Poll Survey surveyed more than mainly covers the 1990s and 2000s, although it also 150,000 adults in 148 economies in 2011. The sur- has information for previous decades on a few coun- vey is representative of the adult population in each tries. More information is available at http://sedlac country. The core questionnaire includes detailed .econo.unlp.edu.ar/eng/. information on demographics (gender, age, marital status, education); income; well-being and life/job World Bank Enterprise Surveys satisfaction; trust in institutions, family, and strang- The World Bank’s Enterprise Surveys database ers; and jobs. The World Bank and the Bill & Melinda includes fi rm-level information on a representative Gates Foundation recently partnered to include sample of registered fi rms in the nonagricultural for- information on the use of formal and informal pay- mal private sector. The surveys cover a broad range ments, savings, credit, and insurance. More infor- of business environment topics, including corruption, mation is available at http://www.worldbank.org infrastructure, crime, competition, access to fi nance, /globalfi ndex and http://www.gallup.com/strategic and performance measures. The World Bank collects consulting/en-us/worldpoll.aspx. these data through face-to-face interviews with top managers at and owners of more than 130,000 com- Orbis panies in more than 135 economies. The database Orbis is a commercial database compiled by the includes about 187 surveys from about 100 countries. Bureau Van Dijk. It contains standardized finan- Typically, 1,200–1,800 interviews are conducted cial accounting information on companies world- in larger economies, 360 interviews in medium-size wide, with an emphasis on private sector fi rms. It economies, and 150 in smaller economies (for exam- contains information on more than 100 million ple, the survey for The Gambia in 2006 included 33 listed and unlisted companies, including 50 million firms, the one in Ecuador in 2010 included 366 fi rms, in Europe, 30 million in the Americas, and 15 mil- and the one Brazil in 2009 included 1,802 fi rms). In lion in the Asia- Pacific region. The database covers LAC, fi rms with five or more employees are included, 2002–11, but the availability of information var- and firms owned 100 percent by the state are ies greatly depending on the country. Orbis does excluded. Two different surveys are conducted, one not follow a particular sampling strategy, which for service industry fi rms and another for the manu- poses serious questions on the extent of the data’s facturing sector; both surveys contain a common representativeness. Listed firms present the most core set of questions. The last and most complete complete level of information. More information wave of data are for 2009–10. More information is is available at http://www.bvdinfo.com/Products available at http://data.worldbank.org/data- catalog /Company-Information/International/Orbis-(1). /enterprise-surveys. E N T R E P R E N E U R S H I P, E N T R Y , A N D T H E L I F E C Y C L E O F F I R M S I N L A T I N A M E R I C A A N D T H E C A R I B B E A N 33 business ownerships. People identified as FIGURE 2.4 Rate of formal business ownership in selected country business owners were asked: “Have you for- groups, countries, and economies, 2011 mally registered your business?”13 Respon- dents who answered “yes” were classified as 100 formal business owners; respondents who 90 answered “no” were classified as informal 80 business owners. 70 More than half of the businesses owned in LAC are not formally registered, a share 60 Owners (%) similar to China (56 percent) and the EAP4 50 countries (Indonesia, Malaysia, the Philip- 89.0 40 pines, and Thailand) in East Asia and Pacific 70.3 (60 percent) but much higher than in Eastern 30 45.4 44.0 43.0 40.3 Europe and Central Asia (ECA) (30 percent) 20 40.0 and high-income economies, where only 1 in 10 19.0 every 10 businesses is not formally registered 0 (figure 2.4). Within LAC, Chile, Ecuador, High- ECA LAC5 China Caribbean Other EAP4 India Panama, Paraguay, and Uruguay stand out income LAC economies as the countries with the highest rate of busi- Formal firm Informal firm ness formality. More than half of business Source: Klapper and Randall 2012, based on data from Gallup World Poll Survey. owners in these countries formally registered Note: The sample includes only the working-age population. Each country or economy has the their businesses. In contrast, less than 30 per- same weight in the regional average. For countries and economies included in each group, see note 11. cent of business owners in the Dominican Republic, El Salvador, Guatemala, and Haiti reported doing so. have only elementary education. Similar pat- Do business owners of formal and infor- terns are found in the other country groups. mal firms look alike? The propensity of The gender gap is consistent across the two people to engage in different forms of entre- types of business owners but differs greatly in preneurial activities— and job creation— magnitude and across country groups. The naturally depends on social and individual difference between men and women is larger characteristics, such as gender, culture, and for formal business ownership. In LAC5, religion (Iyer and Schoar 2010; Ardagna and for instance, 8.8 percent of men own a for- Lusardi 2010; Djankov and others 2005). mal business, compared with 5 percent of The typical formal and informal business women; 9.4 percent of men and 7.2 percent owner has very different socioeconomic char- of women own informal firms. Similar differ- acteristics (table 2.1). Compared with the ences are found in the other middle-income general population, formal business own- regions, with China and India standing out ers tend to be older, male, urban, and well at the two extremes of the distribution. Chi- educated, with an income in the upper level nese women are almost as likely as men to be of the within- country income distributions. formal entrepreneurs, and India has the larg- Informal business owners are also older and est gender entrepreneurship gap. more likely to be men, but there is no clear One of the striking features of businesses relationship with income, and they are likely in LAC is their small size. Indeed, the most to be less educated than the average worker. common firm in LAC has no employees, In LAC5, for example, about 9 percent of according to survey respondents who own an adults with elementary education as their establishment. Such bias toward micro estab- highest educational credential own an infor- lishments is evident among formal firms; it is mal business, as opposed to some 6 percent even more marked among informal ones. In of adults with tertiary education. In contrast, Caribbean countries, more than 70 percent of among formal business owners, 15 percent unregistered business establishments have no have tertiary education and just 5 percent employees, according to business owners. This 34 TABLE 2.1 Socioeconomic characteristics of business owners in selected country groups, countries, and economies (business owners as percentage of working-age population) LAC5 Other LAC Caribbean High Income ECA EAP4 China India Formal Informal Formal Informal Formal Informal Formal Informal Formal Informal Formal Informal Formal Informal Formal Informal % % % % % % % % % % % % % % % % % of business owners 7.0 8.0 7.0 9.8 6.3 9.3 8.0 0.9 3.7 2.8 7.0 12.5 7.0 9.0 3.0 11.0 Gender Male 8.8 9.4 8.5 10.6 7.0 9.7 10.7 1.1 4.9 3.5 8.3 12.8 8.0 11.0 5.0 18.0 Female 5.0 7.2 5.8 9.3 6.0 9.3 5.5 0.6 2.4 2.0 5.5 12.5 7.0 7.0 1.0 4.0 Education Elementary 5.6 9.2 4.7 11.2 4.7 9.3 4.9 0.5 1.0 3.2 4.5 13.0 7.0 10.0 2.0 11.0 Secondary 6.4 7.8 7.0 9.2 6.7 9.7 8.0 1.0 3.5 2.5 7.3 13.8 10.0 8.0 5.0 11.0 Tertiary 15.0 5.8 16.3 7.7 23.3 8.7 11.0 0.9 7.1 1.6 12.0 10.0 6.0 3.0 5.0 10.0 Age 15–24 2.4 3.4 2.6 3.6 3.7 4.3 1.8 0.5 1.2 2.1 2.5 5.0 6.0 6.0 1.0 7.0 25– 64 8.2 10.4 9.3 12.9 8.7 12.3 10.3 1.0 4.9 3.2 8.5 15.5 8.0 11.0 4.0 13.0 65+ 7.6 3.8 5.2 9.2 2.3 6.0 4.8 0.7 0.8 1.7 6.5 11.0 2.0 1.0 1.0 6.0 Setting Urban 7.6 8.0 8.0 10.0 7.7 10.3 7.3 0.8 4.0 2.0 7.3 14.0 9.0 9.0 3.0 11.0 Rural 6.0 8.4 6.2 9.5 6.0 9.3 9.4 1.0 3.2 3.4 6.3 12.0 7.0 9.0 3.0 11.0 Income quintile 1 (top) 2.2 8.6 3.8 9.9 4.0 9.0 4.6 0.9 1.3 2.9 3.0 8.0 3.0 5.0 1.0 8.0 2 5.2 9.0 4.8 9.3 7.7 9.0 6.3 0.8 2.2 2.3 3.8 13.0 6.0 6.0 2.0 12.0 3 5.2 6.4 5.5 9.8 6.3 11.7 6.6 0.9 2.7 2.9 6.8 13.5 7.0 11.0 3.0 12.0 4 8.2 8.6 8.8 10.2 4.7 9.0 9.7 0.8 4.3 2.6 9.3 13.5 7.0 10.0 5.0 10.0 5 (bottom) 14.0 8.2 13.3 10.2 10.3 8.0 13.8 1.1 7.5 3.1 13.3 15.3 13.0 14.0 5.0 14.0 Source: Klapper and Randall 2012, based on data from Gallup World Poll Survey. Note: Each country or economy has the same weight in the regional average. For countries and economies included in each group, see note 11. E N T R E P R E N E U R S H I P, E N T R Y , A N D T H E L I F E C Y C L E O F F I R M S I N L A T I N A M E R I C A A N D T H E C A R I B B E A N 35 percentage is very similar to that observed FIGURE 2.5 Share of firms with no employees in selected country in LAC5 and Other LAC. In contrast, the groups, countries, and economies, 2011 figures are 56 percent in EAP4 and 35 per- cent in Europe and Central Asia (figure 2.5). 80 As expected, informal businesses tend to be 70 smaller, but even among formal businesses in LAC, almost half have no employees. Group- 60 ing formal and informal businesses together 50 Firms (%) makes the bias toward microfirms in LAC 40 even more striking (figure 2.6). Although the share of firms with more than five employees 30 reaches 20 percent in high-income economies, 20 it barely reaches 3 percent in LAC. 10 0 India Caribbean Other LAC5 China High- EAP4 ECA Push versus pull factors LAC income economies Economists have long tried to understand Informal firm Formal firm the determinants of business creation. Fol- lowing Vivarelli (2013), the traditional view Source: Klapper and Randall 2012, based on data from Gallup World Poll Survey. gives the expected level of profit a prominent Note: Each country or economy has the same weight in the regional average. For countries and economies included in each group, see note 11. role (Orr 1974; Khemani and Shapiro 1986). More recent studies in this stream of litera- ture highlight not only profit expectations FIGURE 2.6 Distribution of firm size in selected country groups, but also other pull factors, such as economic countries, and economies, 2011 growth and high innovative potential (see Acs and Audretsch 1989a, 1989b; Geroski 1995). 100 Authors such as Knight (1921); Schum- 90 peter (1934, 1939); and Oxenfeldt (1943) 80 29.8 drew attention to the characteristics of the 33.7 34.2 38.8 founder of a new firm, highlighting the 70 44.7 47.9 importance of individual heterogeneity and 60 50.7 43.3 Firms (%) the desire to innovate and put new ideas into practice as drivers of entrepreneurial spirit. 50 Potential entrepreneurs seem to be strongly 40 influenced by specific psychological attitudes, 68.1 63.1 30 62.4 58.2 such as the desire to be independent, the need 49.8 45.7 20 38.3 37.1 for autonomy in the workplace, an aspiration to fully exploit previous job experience and 10 acquired ability, and the desire to be socially 0 useful and to acquire social status (see Creedy Caribbean LAC5 Other India EAP4 China High- ECA LAC income and Johnson 1983; Evans and Leighton economies 1990; Vivarelli 1991, 2004; Blanchflower Employees: 0 1–5 6–50 50+ and Meyer 1994; Blanchflower and Oswald 1998; Zacharakis, Bygrave, and Shepherd Source: Klapper and Randall 2012, based on data from Gallup World Poll Survey. 2000). Pursuit of these goals—in the form of Note: Each country or economy has the same weight in the regional average. For countries and profitability, growth, or simply the desire to economies included in each group, see note 11. put in place original ideas—is associated with a view of entrepreneurship in which the entre- foundation of a new fi rm is not fostered by preneur is pulled into business creation. absolute profitability but by the difference In the occupational choice model out- between expected profits and current local lined at the beginning of this chapter, the wages in the same sector, taking into account 36 LATIN AMERICAN ENTREPRENEURS the surrounding environmental conditions. A third explanation for entrepreneurship The introduction of relative considerations may simply be “entry mistakes” (Cabral 1997; opens the door to examine entrepreneurs Geroski and Mazzucato 2001). Such mistakes who are pushed into entrepreneurship rather are likely to result in early failure, turbulence, than pulled by absolute profits. One can eas- and churning. Mistakes may occur if poten- ily extend the relative approach to consider tial entrepreneurs are overconfident (Dosi and the risk differential between the two occu- Lovallo 1998; for an experimental econom- pational alternatives (Kihlstrom and Laffont ics exercise, see Camerer and Lovallo 1999). 1979; Parker 1997; Cressy 2006). Parker (2006) discusses both the psychology The implication of comparing the relative literature, which gives reasons for expecting virtues of self-employment versus dependent entrepreneurs to be especially prone to over- employment means that entry may have a optimism, and previous empirical evidence countercyclical component. It is in periods of showing that optimism is significantly and slow growth and profit prospects that many positively associated with the propensity to small firms are created, simply as an alterna- be an entrepreneur (de Meza 2002; Åstebro tive to the prospects of dependent employ- 2003; Coelho, de Meza, and Reyniers 2004). ment, which become less attractive (see A set of questions in the Gallup World Poll Highfield and Smiley 1987; Hamilton 2000). Survey sheds some light on the importance of Pushing this argument further, founding a pull versus push factors in LAC (see Klapper new fi rm may be an alternative to uncertain and Randall 2012 for details). The poll asked future career prospects or represent an escape business owners if each of the following rea- from unemployment (see Oxenfeldt 1943; sons was a very important reason why they Evans and Leighton 1990; Storey 1991, started their business, allowing for multiple 1994). The empirical evidence suggesting the responses: important role of job losses in fostering entry A. You could not find a suitable job. is indeed quite robust (see Storey and Jones B. You were afraid of losing your job. 1987; Santarelli, Carree, and Verheul 2009). C. You saw an opportunity to make more A complication in identifying the push fac- money. tor is related to the fact that in general, times D. You wanted to be your own boss. of low job finding rates are usually recessions, E. You had a great idea for a business. which may also imply lower expected profits from self- employment. The contemporane- Factors A and B are clearly associated with ous presence of these two channels has made push motives for entrepreneurship; factors it hard to identify unemployment push entre- D and E are more likely to be pull motives, preneurship in the data. The use of micro although they may also constitute an addi- data has helped researchers overcome this tional incentive to entrepreneurship among problem. Using data from the National Lon- people who were pushed. Factor C can be gitudinal Survey of Youth, Rissman (2007) associated with both push and pull factors; documents the presence of push entrepreneur- it is harder to interpret. For instance, in our ship among young men in the United States. framework, all potential entrepreneurs com- Millán (2012) finds a similar pattern in sev- pare expected profits with expected wages; eral European countries, using data from the making money is thus always a motive for European Community Household panel. starting a business. For this reason, response Thus, in some situations, a business owner C (to which more than 70 percent of respon- may be pushed rather than pulled into entre- dents answered affirmatively) was disre- preneurship. When unemployment and the garded in the analysis. risk of failure of entrepreneurial projects are Entrepreneurs who were pushed into incorporated, push factors will more than entrepreneurship are those who answered proportionally discourage entrepreneurship positively to questions A or B. Pulled entre- by high-ability individuals (Poschke 2013c). preneurs are defi ned as survey respondents E N T R E P R E N E U R S H I P, E N T R Y , A N D T H E L I F E C Y C L E O F F I R M S I N L A T I N A M E R I C A A N D T H E C A R I B B E A N 37 who answered “no” to A and B and “yes” LAC stands out as the region where push to D or E. A residual category, which covers factors are most clearly linked to informal some 8 percent of the business owners, was entrepreneurship and pull factors with for- neither pulled nor pushed; it is ignored in the mality. More than 50 percent of pull entrepre- analysis. neurs in LAC register their business, against One in every two business owners in LAC just 30 percent of pushed entrepreneurs (fig- is pushed into entrepreneurship, the same ure 2.7, panel b). Differences are also posi- proportion as in ECA and about twice as tive but smaller in other regions dominated many as in high-income economies. About by middle-income countries, such as EAP4, 60 percent of entrepreneurs in high-income where there is a difference of about 10 per- countries are pulled by great ideas for a busi- centage points in the share of formal business ness or a desire to be their own bosses against owners that were pulled versus pushed into 45 percent in LAC and 40 percent in ECA entrepreneurship. In regions where infor- (figure 2.7, panel a). Guatemala, Honduras, mality is almost nonexistent, such as high- Brazil, El Salvador, Peru, and Haiti lie above income countries and ECA, or informality is the regional average, with more than 50 per- the predominant form of ownership (India), cent of business owners pushed into entre- differences between pull and push factors are preneurship. In the Caribbean, almost 70 not fundamental for formality status. percent of business owners declared having The differences between push and pull opened a business out of fear of losing their factors are also significant for the capacity job or because jobs were not available. Uru- (or willingness) of the entrepreneur to gen- guay, Chile, and Mexico, the countries with erate employment. Pull entrepreneurs are the lowest share of push entrepreneurs in the more likely to have employees. This differ- region, still lie above the 28 percent push ence is largest in India, where the likelihood entrepreneurship that characterizes high- that an entrepreneur generates at least one income countries. job is more than 25 percentage points higher FIGURE 2.7 Push versus pull entrepreneurship in selected country groups, countries, and economies, 2011 a. Pull and push entrepreneurship b. Difference between push and pull factors 70 30 Percentage of push and pull entrepreneurs % of firms pulled – % of firms pushed 25 60 20 50 15 40 10 30 5 0 20 –5 10 –10 0 –15 no me P4 no me 5 ia AC ean ECA 4 a 5 ean AC ina ECA ia in LAC P LAC Ind Ind s s EA EA eco -inco eco co mie mie er L er L Ch Ch ibb ibb n h-i Oth Oth Car Car h Hig Hig Pull Push Formal + 1 employee Source: Klapper and Randall 2012, based on data from Gallup World Poll Survey. Note: The sample consists of the working-age population only. Each country or economy has the same weight in the regional average. For countries and economies included in each group, see note 11. 38 LATIN AMERICAN ENTREPRENEURS if he or she was pulled into entrepreneur- market growth (Malchow-Møller, Schjern- ship as opposed to pushed. In LAC, 43 per- ing, and Sørensen 2011). Indeed, industrial cent of pull entrepreneurs have at least one dynamics (that is, the entry and exit of firms) employee, against 30 percent of those who accounts for 20– 40 percent of total produc- were pushed. The largest difference between tivity growth in eight selected Organisation pull and push entrepreneurs is observed in for Economic Co-operation and Develop- the Caribbean (15 percentage points), closely ment (OECD) countries, according to OECD followed by EAP4 (13 percentage points) (2003), supporting the idea that entrepre- and LAC5 (12 percentage points). Excluding neurs represent one of the driving forces of China, a clear outlier in the sample, this dif- economic growth and structural change ference is smallest in Other LAC (5 percent- (Audretsch and Keilbach 2005; Foster, Halti- age points). wanger, and Syverson 2005). The reasoning These fi ndings suggest that entrepreneurs is that new entrants can displace obsolescent set up businesses for a large variety of rea- fi rms in a process of “creative destruction” sons and that such differences are important (see Schumpeter 1939, 1943), which may be determinants of the type of business activity an important micro determinant of produc- that will be developed. Entrepreneurs who tivity dynamics that eventually results in eco- are pulled into entrepreneurship are more nomic growth.14 likely to end up registering their business and Recent studies based on data from the hiring more workers than entrepreneurs who Global Entrepreneurship Monitor have are pushed into entrepreneurship because of identified a U - shaped relationship between the fear of losing their jobs or lack of better a country’s rate of entrepreneurial activ- employment opportunities. The proportion ity as measured by net entry and its level of of businesses that are created because of push economic development (Reynolds and oth- factors is much larger in developing regions, ers 2001; Wennekers and others 2005). The including LAC, than in high-income coun- creation of new fi rms is very active in both tries. However, push and pull factors are two highly developed and extremely poor coun- sides of the same coin. If there are insufficient tries, a fact that emphasizes the multifaceted high-end entrepreneurs, or the entrepreneurs phenomenon of entrepreneurship and demys- that exist generate little employment, there tifies simplistic mechanical links between will be fewer good jobs for jobseekers and firm creation and innovation, productivity some of them will be pushed into low-growth growth, and economic development. Indeed, potential forms of entrepreneurship. The rest only when transformational entrepreneurs of this chapter investigates the creation and are distinguished from low-growth entrepre- dynamics of formal firms. neurs is a positive linear relationship between economic development and entrepreneur- ship restored (Carree and others 2007; Acs Business creation in Latin 2008; Acs, Desai, and Klapper 2008). In America and the Caribbean developing countries, a positive relationship New fi rm formation may play a crucial role between entrepreneurship and job creation is in fostering competition, inducing innova- detectable only when self-employment with- tion, and boosting the emergence of new sec- out employees and informal companies are tors, as discussed by Wennekers and Thurik excluded from the analysis (Ghani, Kerr, and (1999) and Dejardin (2011). Entrepreneurs O’Connell 2011). leading the new small firms may compensate Identifying transformational entrepre- for the restructuring of mature sectors and neurs and distinguishing them from low- the downsizing of larger incumbent firms. growth entrepreneurs is a hard task. The Ultimately, new firms may contribute sub- previous analysis of push versus pull factors stantially to job creation, provided that the offers some hints, however. For instance, net effect of new entrants brings about overall registered business owners in developing E N T R E P R E N E U R S H I P, E N T R Y , A N D T H E L I F E C Y C L E O F F I R M S I N L A T I N A M E R I C A A N D T H E C A R I B B E A N 39 countries are more likely to be pulled into and Dominica (4 new fi rms). Argentina and entrepreneurship than business owners who Mexico stand at the opposite end of the did not register their business, suggesting distribution, with rates of entry substan- that registered business owners have a higher tially below those suggested by their GDP likelihood of becoming transformational per capita. entrepreneurs. As the focus of this study is on The relatively weak performance of LAC transformational entrepreneurs, we start the countries in formal business creation raises discussion by examining the creation of for- the question of what determines entry and mal businesses. Is LAC lagging behind in net the even more important question of how the formal entry? process of business creation can be enhanced Figure 2.8 displays the relationship in the region. In an attempt to answer these between firm entry and the level of devel- questions, the literature has attributed a opment across 129 countries. Perhaps as prominent role to regulatory barriers. The expected, entry, as measured by firm reg- importance of entry costs as an obstacle to istration per 1,000 working-age people, is the creation of new businesses— and conse- weakly positively associated with GDP per quently a healthy reallocation of productive capita. Formal entry rates in LAC tend to factors in the economy— has been docu- be below the level predicted by their income mented as a limiting factor in fi rms’ invest- per capita, although differences with respect ments (Nicoletti and Scarpetta 2003) and an to the predicted values are not always large obstacle to productivity and growth (Alesina and there is substantial heterogeneity within and others 2005) and the creation of new the region. By far the most dynamic economy firms (Klapper, Laeven, and Rajan 2006). in the region is Costa Rica, with an average Within a sample that includes 85 developed entry rate of almost 16 new fi rms per 1,000 and developing countries, entry regulation working-age people between 2004 and 2011, has also been found to promote corruption followed by St. Kitts and Nevis (6 new firms) and larger unofficial economies (Djankov and others 2002). Could the removal of reg- ulatory barriers spur the creation of formal FIGURE 2.8 New firm entry rates and GDP per businesses in the region? capita in selected countries, 2004–11 We examine the association between entry and the share of formal business in the total 20 business population on the one hand and two different indicators of administrative barri- ers to entrepreneurship on the other. The first 15 indicator of entry barriers is the total cost of setting up a business, obtained from Doing Entry density 10 Business data. This indicator includes an imputation of the monetary costs associated with the numbers of days needed to set up a 5 business and the direct monetary cost related to fees and other taxes for a sample of 132 0 economies.15 6 7 8 9 10 11 The second indicator is a summary mea- Log of GDP per capita sure of barriers to entrepreneurship that is LAC countries Non-LAC countries calculated as a weighted average of three subindexes: an indicator of regulatory and Source: World Bank, based on data from World Development Indicators and administrative opacity, an indicator of World Bank Group Entrepreneurship Snapshots (WBGES). administrative barriers to start-ups, and an Note: Each point represents the average between 2004 and 2011. Curve shows quadratic fitted values. GDP = gross domestic product. LAC = Latin indicator of barriers to competition. Each America and the Caribbean. subindex ranks countries on a scale from 0 to 40 LATIN AMERICAN ENTREPRENEURS 6, with 0 representing lax and 6 representing worldwide deregulation in product markets. strict regulations. The cross-country variance in the number of There is great heterogeneity across LAC procedures, the time, and the cost associated countries in these red-tape barriers to entry with setting up a business declined steadily (figure 2.9). In Guatemala, it takes 37 days, between 2004 and 2013, as countries have 12 procedures, and the equivalent of 52 per- become more aware of the need to create cent of GDP per capita to open a business; in a more favorable business environment.17 Chile it takes 7 days, 7 procedures, and the Despite progress, however, the burden equivalent of 5 percent of GDP per capita (in imposed by red-tape regulations in the region contrast, in Canada, it takes just one admin- is still higher than in other regions of similar istrative procedure to set up a new business). income per capita, such as ECA or EAP4. Great heterogeneity across countries is We examine next the association between also present in the broader summary measure entry and formal business ownership and of barriers to entrepreneurship. The average barriers to entrepreneurship around the score for OECD countries excluding Mexico world. We present partial correlations that is 1.36. This score is lower than in the best- control for differences in GDP and popula- scoring country in LAC (Colombia at 1.79), tion across countries because entry barriers where the framework is much more business tend to be concentrated among developing friendly than in the worst-scoring countries countries. The analysis yields very similar (Honduras 3.65, Argentina 3.28, and Nica- results for barriers to entry and barriers to ragua 3.18).16 entrepreneurship (panels a– d in figure 2.10). LAC countries have made significant The partial correlations are weakly negative progress in reducing such barriers in the and not uniformly statistically significant. last few years. The time to set up a busi- For similar levels of entry barriers or barri- ness was halved in less than a decade (see ers to entrepreneurship, there is tremendous figure 2.9), reflecting a general trend of variability in the degree of formalization and entry. Moreover, in countries such as Peru and Brazil, where regulation was expected FIGURE 2.9 Time required to start a business in selected country to be a major obstacle to business formaliza- groups, countries, and economies, 2004 and 2013 tion, levels of entry are in broad accordance with the level of development. 140 These findings do not imply that barri- ers to entry need not be detrimental to fi rm 120 performance and resource reallocation. The 100 literature documents the obstacles imposed by entry barriers for a variety of outcomes, 80 including employment in dynamic service sec- Days 60 tors (Messina 2006), firm investment (Alesina and others 2005), and the creation of new 40 firms (Klapper, Laeven, and Rajan 2006). The 20 weak associations of the data here may reflect the fact that other counteracting forces blur 0 the cross-country correlations under study. It no me ECA a 4 5 AC ia n in P LAC ea Ind s EA er L Ch eco -inco mie should also be noted that most of the litera- ibb Oth Car ture cited relied on OECD data or included a h Hig 2004 2013 very limited set of middle-income countries. The analysis raises a question about the Source: World Bank 2011. relevance of regulatory barriers for the for- Note: Each bar represents the average for each region. The thin lines show the standard deviation for the countries within each region. For countries and economies included in each group, see mation of new businesses in developing coun- note 11. tries. It may be that in developing countries E N T R E P R E N E U R S H I P, E N T R Y , A N D T H E L I F E C Y C L E O F F I R M S I N L A T I N A M E R I C A A N D T H E C A R I B B E A N 41 FIGURE 2.10 Relationship between business formality and barriers to entry in selected countries, various years a. Entry density vs. cost of starting a business b. Formal business owners vs. cost of starting a business 15 Slope: −0.00 0.4 Slope: −0.00 Formality (% of business owners) CRI p-value: 0.76 p-value: 0.46 10 0.2 Entry density PRY HTI ECU 5 0 NICBOL HND CHL PER COL URY JAM CRI MEX BRA ARG TTO GTM PER BRA 0 CHL GTMKNAURY COL MEX BOL −0.2 ARG SLV VEN JAM DOM DOM SLV LCA DMA SUR −5 −0.4 −100 0 100 200 −200 0 200 400 600 Cost, income per capita (%) Income per capita (%) c. Entry density vs. barriers to entrepreneurship d. Formal business owners vs. barriers to entrepreneurship 15 Slope: −0.41 0.3 Slope: −0.12 Formality (% of business owners) p-value: 0.83 p-value: 0.03 0.2 10 NIC Entry density 0.1 HND CRI 5 0 COLPER PER MEX 0 COL −0.1 CRI ARG MEX ARG SLV SLV −0.2 DOM −5 −0.5 0 0.5 1 1.5 −0.5 0 0.5 1 1.5 Barriers to entrepreneurship (index) Barriers to entrepreneurship (index) e. Entry density vs. governance f. Formal business owners vs. governance 15 Slope: 1.75 0.4 Slope: 0.08 Formality (% of business owners) p-value: 0.00 CRI p-value: 0.00 10 0.2 Entry density PRY HTI ECU 5 0 BOL NIC HND CHL COLPER URY BRA MEX CRI ARG TTO 0 PER COLGTM URY CHL −0.2 GTM BRA MEX BOL ARG KNA VEN DOM JAM SLV DOM SLV DMA LCA SUR −0.4 −5 −1.5 −1 −0.5 0 0.5 1 −1.5 –1 −0.5 0 0.5 1 Governance (index) Governance (index) LAC countries Non-LAC countries Source: World Bank, based on data from World Development Indicators, 2012; World Bank Entrepreneurship Database; Klapper and Randall 2012; Wölfl, Koz- luk, and Nicoletti 2009; and Kaufman, Kraay, and Mastruzzi 2010; and Worldwide Governance Indicators. Note: Each point represents the residuals of the regression between each variable, gross domestic product (GDP) adjusted for purchasing power parity, and population. Panels a and e use the average for 2004–2011; panels b, c, d, and f use the average for 2004– 09. The slope is the coefficient of a regression between the two variables. The p -value shows the significance of the coefficient. LAC = Latin America and the Caribbean. 42 LATIN AMERICAN ENTREPRENEURS what matters for the business environment is businesses because of more limited access to the quality of broader institutional and policy finance or capacity to deal with burdensome arrangements, which may range from respect regulations. However, even in the United of the rule of law (Botero and others 2004) to States, the typical small business is engaged the development of other markets, including in low-growth entrepreneurship, does not finance and insurance. necessarily represent an engine of employ- To shed some light on this hypothesis, ment creation, and has no intention to grow we correlate our measures of entry and for- (Hurst and Pugsley 2011). Thus, targeting mality with a broader index, the Worldwide small business as the sole criteria of entrepre- Governance Indicator (WGI), which sum- neurship programs may involve substantial marizes information on the degree of voice inefficiencies. Moreover, even if some mar- and accountability in the economy, political ket failures are concentrated among small stability and absence of violence, government businesses, little is known about the type of effectiveness, regulatory quality, the rule of policies that may be successful in promot- law, and control of corruption (Kaufman, ing entrepreneurship; the few studies that Kraay, and Mastruzzi 2010). The index esti- attempt to scrutinize these types of interven- mates the quality of governance on a scale tions are inconclusive (box 2.2). of –2.5 to 2.5 that increases in the quality of Recent studies have tried to distinguish governance. Because this indicator is collected between low-growth and high-growth entre- annually, all of the information on entry rates preneurs (see Gindling and Newhouse 2012; for the period 2004–11 can be exploited, Fafchamps, Woodruff, and Yin 2013). In which adds some precision to the estimates.18 parallel, some studies have emphasized the The cross- correlations of entry and the important distinction between young and share of formality with the quality of gov- small firms as sources of growth. Most of the ernance are highly consistent and suggest job-creating process among small businesses that in countries with better governance, the in the United States is accounted for by new share of business owners and the creation of entrants and young businesses (Haltiwan- new formal firms are higher, even after con- ger, Jarmin, and Miranda 2013). In contrast, trolling for GDP and population. The impli- small mature businesses have on average neg- cation, which deserves further scrutiny, is ative net job creation. There is also consid- that to stimulate a better business climate in erable heterogeneity in terms of job creation the quest for a vigorous and vibrant entrepre- within any definition of firm class. Haltiwan- neurial sector, governments should examine ger (2011) shows that the typical small or the overall business environment rather than median young business in the United States specific aspects of it. displays very low growth, but average growth is high for this group because a small fraction of firms are growing very rapidly. Beyond entry: Firm dynamics in LAC does not seem to be lagging tre- Latin America and the Caribbean mendously behind in the creation of new The importance of entry in the process of businesses. However, the productivity perfor- structural change and productivity dynam- mance of the region during the last decades ics is hard to dispute. It is probably for this has been very disappointing. Total factor reason that all governments have specific productivity in the manufacturing sector has programs to support entry and the perfor- not increased since the 1970s, and it actually mance of small fi rms. Indeed, if small busi- declined in some countries (Busso, Madrigal, nesses are the engines of net job creation (as and Pagés-Serra 2012). The combination of suggested by Neumark, Wall, and Zhang these two facts suggests that the problems 2011 for the United States), there is possi- of resource misallocation and inefficiencies bly a role for public support because of the may lie either in the nature (rather than the presence of market failures. Small businesses number) of the businesses created or in the are likely to face greater barriers than larger postentry performance of fi rms. Chapter 3 E N T R E P R E N E U R S H I P, E N T R Y , A N D T H E L I F E C Y C L E O F F I R M S I N L A T I N A M E R I C A A N D T H E C A R I B B E A N 43 BOX 2.2 Do training programs for entrepreneurs work? The typical micro or small business in a develop- cant effects in three out of six estimations, ranging ing country does not implement many of the busi- from 3 percent to 57 percent. However, it is unclear ness practices considered standard in the devel- whether training merely speeds up or permanently oped world. For this reason, entrepreneur training increases the rate of entry, as the control group programs have become increasingly popular in the of one program seems to catch up after two years developing world. There has been little rigorous evi- (de Mel and others 2012). dence of the impacts of these programs; when these An important channel through which training evaluations have taken place they have encountered may improve business outcomes is better business serious methodological challenges. practices; almost all studies fi nd a positive effect. Several issues arise in assessing the studies. First, Seven out of 9 estimations fi nd signifi cant effects, the impact of training is likely to depend on who ranging from 3 percent to 203 percent. Although receives the training. Second, the business training some of these effects seem large, however, in abso- offered varies across studies substantially. McKen- lute terms they are low, because business practices at zie and Woodruff (2013) distinguish four strategies baseline are very weak (less than 30 percent of fi rms for participant selection among existing studies: keep records in most cases). classroom-based training offered by microfi nance Another relevant outcome is the increase in prof- organizations or banks to their clients, training its after training. Only two of seven studies find sig- offered to fi rms in a particular industry or cluster, nificant effects on profits, with effects of 24 percent individual application to training as part of a compe- and 43 percent on female participants. There is also tition, and training offered to a random subsample little evidence of employment creation, with just of a representative population of microenterprises. 1 in 20 trained entrepreneurs hiring an additional It is hard to draw general conclusions from exist- worker. ing studies because of four fundamental challenges. Five issues need better answers before govern- First, the studies often lack statistical power: only ments start implementing large-scale interventions: 2 of the 15 studies have enough statistical power to • Studies have not been able to say who benefits safely yield conclusions. Second, most of the studies most from these programs. In theory, it would include only one follow-up interview and are very be optimal to target firms where management short run, looking at impacts after one year or less skills represent a constraint on growth, but after the training. Third, attrition rates range from identifying those firms is a complicated en- 6 percent to 28 percent. Their relation to business deavor. It is still unclear through which chan- failure, disappointment with training effects, and nels training affects business outcomes. location movements complicates inference. Fourth, • How do markets and the competition react to sales and profits reported may not be true indicators newly trained firms? of impact: training may simply reduce bookkeeping • The short-run scope of existing studies pre- mistakes rather than improve actual outcomes. vents extracting lessons in the long-run. Are Keeping these caveats in mind, a few tentative there any market constraints preventing firms conclusions can be drawn. On the one hand, there from accessing helpful training programs by is little evidence of a relationship between training themselves? and survival, with signifi cant effects for three out • What is the effect of attitudes and personalities of seven estimations and estimated impacts ranging of business owners on performance? from 6 percent to 9 percent for men. On the other hand, the evidence suggests that business training • Can people be turned into entrepreneurs? generates short-run impacts on start-up, with signifi- characterizes the behavior of incumbent Eslava and Haltiwanger (2013) analyze fi rms and their degree of entrepreneurship. business dynamics in the formal manufac- The next few paragraphs describe some of turing sector of Colombia, contrasting them the most salient stylized facts of business when possible with similar data for the dynamics in the region. United States.19 They find similar patterns 44 LATIN AMERICAN ENTREPRENEURS in the two countries. Figure 2.11, panel a, 53 percent.21 Once entry and exit are consid- reveals the importance of the size versus ered, the fastest-growing establishments are the age of the plant for the generation of small and young. Compared with the United employment.20 It separates plants into small, States, younger Colombian firms appear medium, and large based on their average “healthier”: they exhibit stronger growth and employment levels in two consecutive peri- are less likely to die. This evidence suggests ods and shows net employment creation for that selection dynamics are stronger in the establishments of different ages. United States, where only the fittest firms are Growth clearly increases with size and able to survive. declines with age. However, the differ- Even within relatively homogeneous size ences are much more marked along the age and age classes there is tremendous heteroge- dimension. On average, young fi rms are net neity in growth rates across firms. Excluding employment generators for all size classes, entry and exit, it is always the case that the and average growth rates increase rapidly fastest-growing fi rms are among the young- with size. The fastest-growing young estab- est (figure 2.12). Employment by young lishments are the largest, a fact that appears gazelles— fi rms at the 90th percentile of the to contradict the idea that most employ- growth distribution— can increase by almost ment generation occurs among small fi rms. 50 percent in one year, even if they are already Moreover, and in line with evidence from large. Across age classes, there is much more the United States, small plants older than five variation among top performers than among years contract rather than grow. contracting firms (firms in the 10th percentile This evidence does not imply that small of the growth distribution). Smaller fi rms in firms are not important for growth. When the 10th percentile tend to decline faster than the contribution of all firms (including new larger ones, with very little differences across firms and firms that die) is considered, the pic- age classes. When the additional impact of ture changes dramatically (figure 2.11, panel exits on employment growth is considered, b). The average growth rate of small plants it is also true that youngest fi rms are more up to four years old jumps from 4 percent to likely to decline fastest. FIGURE 2.11 Employment growth in Colombia, by firm size and age cohort a. Continuers b. All establishments 0.6 0.10 0.4 0.05 Growth rate Growth rate 0.2 0 0 –0.2 –0.05 Small Medium Large Small Medium Large Size of firm Size of firm Age of firm: All 0–4 years 5–9 years 10–14 years 15+ years Source: Eslava and Haltiwanger 2013. Note: Small: fewer than 50 employees; medium: 51–200 employees; large: more than 200 employees. Growth rates are defined as in Davis, Haltiwanger, and Schuh (1996); they are the change in employment between two consecutive periods divided by the average employment between the two periods. E N T R E P R E N E U R S H I P, E N T R Y , A N D T H E L I F E C Y C L E O F F I R M S I N L A T I N A M E R I C A A N D T H E C A R I B B E A N 45 Young firms grow faster than mature firms FIGURE 2.12 Net employment growth rates by firms in in Colombia, even among the small fraction Colombia, by establishment age and size, 1994–2009 of gazelles. However, young establishments tend to be smaller. Does the growth of young 0.40 establishments matter in terms of aggregate employment? The answer is a resounding 0.20 DHS growth rates yes. Table 2.2 shows the dynamics of Colom- 90th bian manufacturing firms by birth cohort percentile for 1994–2009. Firms that were 10 years or 0 older in 1994 had shrunk in size by 2009. 10th percentile Indeed, if it were not for the creation of new –0.20 fi rms, the Colombian manufacturing sector would have contracted considerably during –0.40 the sample period. Small Medium Large In contrast to commonly held views, there Firm age (years) is substantial mobility across establishments 10–14 15+ 0–4 5–9 in the few LAC countries for which data 10–14 15+ 0–4 5–9 are available. Table 2.3 shows transition matrixes for five years in Chile and Colom- Source: Eslava and Haltiwanger 2013. bia across three establishment size classes: Note: Small: fewer than 50 employees; medium: 51–200 employees; large: more than 200 employees. Growth rates are defined as in Davis, Haltiwanger, and Schuh (1996); small (10– 49 employees), medium (50–249 they are the change in employment between two consecutive periods divided by the employees), and large (250 or more employ- average employment between the two periods. ees). There is substantial upward mobility in both countries: about a third of the medium in Colombia), but the analysis does not con- and large firms in Chile (a fifth in Colombia) sider the death and birth of fi rms. Upward belonged to a smaller size class five years mobility is even greater if only young firms in earlier. Downward mobility is very small in the base year are considered. Restricting the Chile (somewhat greater, at about 7 percent sample to establishments that were less than TABLE 2.2 Dynamics of manufacturing firms in Colombia Plant’s initial Before year of operation 1970 1970–79 1980– 84 1985– 89 1990–94 1995–97 1998–2000 2001– 03 2004– 06 2007– 09 Total Total employment 1994 316,612 139,428 80,396 73,248 26,377 — — — — — 636,061 1997 279,372 124,205 75,739 74,119 44,811 17,114 — — — — 615,360 2000 222,464 102,478 63,371 64,540 43,868 20,669 8,297 — — — 525,687 2003 201,227 97,512 64,491 67,379 57,669 26,381 18,559 4,423 — — 537,641 2006 215,886 106,163 69,771 78,947 68,357 37,073 25,226 12,544 3,182 — 617,149 2009 203,989 98,969 67,484 73,960 72,750 39,525 33,305 23,703 17,268 12,545 643,498 1994–2009 –112,623 – 40,459 –12,912 712 46,373 39,525 33,305 23,703 17,268 12,545 7,437 Number of establishments 1994 1,756 1,931 1,500 1,484 593 — — — — — 7,264 1997 1,643 1,891 1,511 1,585 1,032 375 — — — — 8,037 2000 1,374 1,524 1,243 1,329 975 426 196 — — — 7,067 2003 1,212 1,375 1,104 1,271 1,051 521 388 138 — — 7,060 2006 1,112 1,247 1,031 1,228 1,110 594 493 315 86 — 7,216 2009 1,029 1,114 968 1,235 1,286 740 794 693 596 373 8,828 1994–2009 –727 –817 –532 –249 693 740 794 693 596 373 1,564 Source: Eslava and Haltiwanger 2013. 46 LATIN AMERICAN ENTREPRENEURS TABLE 2.3 Five-year changes in size categories for establishments of different ages in Chile and Colombia Size in year t + 4 Establishments less than Establishment of all ages four years old in year t Country/size in year t Small Medium Large Small Medium Large Chile Small 96.0 28.0 2.0 96.0 31.0 1.0 Medium 4.0 71.0 33.0 4.0 68.0 38.0 Large 0 1.0 65.0 0 1.0 61.0 Colombia Small 92.6 19.3 0.6 94.8 38.5 6.4 Medium 7.1 74.4 18.8 5.1 58.1 35.0 Large 0.2 6.3 80.5 0.2 3.4 58.7 Sources: World Bank data for Chile; Eslava and Haltiwanger 2013 for Colombia. Note: Small: 10– 49 employees. Medium: 50–249 employees. Large: More than 250 employees. four years old at baseline, some 38 percent of fi rms that have formalized their businesses. medium-size establishments in Chile (35 per- Moreover, the question regarding initial size cent in Colombia) were large establishments is ambiguous. If the firm started as an infor- five years later. This evidence shows dyna- mal establishment, some managers may refer mism across younger establishments in LAC, to the initial size as the size when the first a feature that is consistent with the evidence full-time employee was hired, whereas others for the United States reviewed above. may refer to the size of the firm when it was The detailed analysis presented so far formally registered. However, administrative leaves several questions open. How much data featuring long panels of firms yields sim- of the observed patterns of firm dynamics ilar results, providing some confi rmation of can be generalized to the region? Ideally, the analysis of firm dynamics using this data one would like to trace firms in all countries set (box 2.3). over time to observe the contribution to total The relationship between age cohort and employment across birth cohorts. Unfortu- initial firm size is strong in LAC. Companies nately, such long panels of firms in the region that were 30–39 years old in 2009 increased are not easily available. their initial size by a factor of eight over the Firm dynamics can be examined across 30- to 40-year period (figure 2.13). This different birth cohorts in a large number of performance is less impressive than that countries using Enterprise Surveys, however observed in high-income countries (where the (see box 2.1 for a description of this data set). multiplying factor exceeds 14) but better than Although Enterprise Surveys are representa- in regions of similar levels of development. tive of a cross-section of firms at one point in In EAP4, the relative size is close to seven; time, they contain a key question that makes in ECA it barely exceeds 4. Note, however, the analysis possible. Managers of the fi rms that in ECA the fi rms created in the 1970s are asked “how many permanent full-time and 1980s were state firms, which underwent employees did this establishment employ massive transformation in the 1990s during when it started operations?” The question the transition to a market economy. But even allows their current size to be compared with if only the youngest cohorts are examined, the size at the time of setting up the business. LAC has a relative advantage in firm growth Enterprise Surveys poll only formally reg- with respect to initial size. istered firms. This sample is thus highly selec- The impressive employment growth per- tive, including only those relatively successful formance of LAC fi rms hides a fundamental E N T R E P R E N E U R S H I P, E N T R Y , A N D T H E L I F E C Y C L E O F F I R M S I N L A T I N A M E R I C A A N D T H E C A R I B B E A N 47 BOX 2.3 Comparing firm size across age cohorts in Colombia using Enterprise Surveys and administrative data In an attempt to provide some external validity to 15 or more than at age 1–5. In contrast, Eslava and the Enterprise Surveys, we compared the differences Haltiwanger fi nd that the size of these fi rms barely in fi rm size across age cohorts in Colombia from the doubled. However, if the sample in the Enterprise Enterprise Surveys with the differences Eslava and Surveys is restricted to eliminate fi rms with 5–10 Haltiwanger (2013) document based on a universal employees, a virtually identical picture of the size establishment registry. The data in Eslava and Halti- structure across ages emerges from the two sources. wanger cover only the manufacturing sector; refer to The impact of eliminating very small fi rms from plants, not fi rms; and have a different size threshold the sample is clearly observed in panel b of fig- from the Enterprise Surveys (more than 10 employees ure B2.3.1. The average fi rm size for age category rather than more than 4 employees). The picture that 1– 4 is below 20 when the smallest fi rms are included emerges from the Enterprise Surveys restricted to the in the sample; it more than doubles when fi rms in manufacturing sector is significantly different. Firms the 5–9 category are dropped. in the Enterprise Surveys are five times larger at age FIGURE B2.3.1 Employment in establishments in Colombia, by age of establishment a. Average of old and young firms b. Average number of employees per firm size 5 120 100 Relative size with respect 4 Number of employers to firms 1–4 years old 80 3 60 2 40 1 20 1–4 5–9 10–14 15+ 1–4 5–9 10–14 15+ Firm establishment age (years) Firm establishment age (years) Manufacturing Manufacturing Manufacturing (excluding micro firms) (as measured by Eslava-Haltiwanger) Source: World Bank, based on Eslava and Haltiwanger 2013. weakness, however. At the time of creation the typical LAC firm at the 90 percent percen- LAC firms are smaller than in any other tile barely reaches 25 employees, as opposed country group. The gap in initial firm size to 40 in ECA and high-income countries and is not obvious for the average firm; rather, almost 55 in EAP4 (panel b of figure 2.13). LAC seems to be lacking top performers. The The imbalance in initial fi rm size is such median firm size at the start of operations in that LAC fi rms never catch up in size with LAC is five employees, very much in line with firms from other regions. LAC firms that ECA and high-income countries. However, are 40 years old or older are on average half 48 LATIN AMERICAN ENTREPRENEURS FIGURE 2.13 Firm size in Latin America and the Caribbean, by age of firm, 2006–10 a. Average size of firm relative to its initial size b. Average initial size of firm 16 60 Ratio between current and initial firm size 14 50 12 Number of employees 40 10 8 30 6 20 4 10 2 0 0 1–4 5–9 10–19 20–29 30–39 LAC ECA High-income EAP4 economies LAC High-income ECA EAP4 Average 10th percentile economies 50th percentile 90th percentile c. Average employment by firm age 300 250 Number of employees 200 150 100 50 0 1–4 5–9 10–19 20–29 30–39 40+ LAC High-income ECA EAP4 economies Source: World Bank, based on data from World Bank Enterprise Surveys. Note: The last survey available for each country or economy was used. Each country or economy has the same weight in the regional average. Size at birth above 10,000 employees was replaced by “missing.” For countries and economies included in each group, see note 11. the size of fi rms in similar age cohorts from times the initial size after 40 years of opera- high-income countries or countries in ECA tion. Differences in sampling frames between countries (panel b of figure 2.12). The size the Enterprise Surveys and the Mexican data gap is also notable in comparison with EAP4. used by these authors are likely to be the main Even within the cohort of firms 10–19 years factor behind these differences. The Hsieh old, firms in EAP4 are twice the size of LAC and Klenow data set includes all fi rms, for- firms, at 100 employees on average versus 50. mal or informal, except street vendors. These The finding of rapid growth of LAC firms micro firms were included in Hsieh and Kle- may seem to contradict a recent study by now but excluded from the Enterprise Sur- Hsieh and Klenow (2012) for Mexico. They veys. Most of them have no employees and report an average firm size of merely two are very unlikely to grow. E N T R E P R E N E U R S H I P, E N T R Y , A N D T H E L I F E C Y C L E O F F I R M S I N L A T I N A M E R I C A A N D T H E C A R I B B E A N 49 The high selectivity of the Enterprise FIGURE 2.14 Age distribution of top 100 firms in selected country Surveys sample is clearly observed in the groups reported initial fi rm size of five employees. Analysis of the Gallup data suggest that even Top 100 firms in developing vs developed countries among formal business owners, the median 0.025 number of employees is 0. The implication is that there is a subset of fi rms in LAC that 0.02 are highly dynamic. They start small in com- parison with similar fi rms in other parts of 0.015 Density the world but grow relatively quickly. How- ever, these fi rms represent a small subset of 0.01 the economy. The vast majority of LAC firms start small and never cross the size threshold 0.005 of five employees to be considered in the sam- pling frame of the Enterprise Survey. 0 A final piece of evidence supporting the rel- 0 50 100 150 ative dynamism of good markets in the region Age of firm (years) is obtained by examining the age distributions LAC5 (developing) Other LAC (developing) of fi rms in different parts of the world (fig- EAP (developing) Continental Europe United States ure 2.14). The Orbis data (see box 2.1) were (developed) (developed) used to plot the size distribution of the 100 Source: World Bank, based on data from Orbis. largest firms in terms of revenue in different Note: The distribution includes all firms within a region for which data were available. EAP (East Asia regions, including LAC5, Other LAC, EAP4, and Pacific): Indonesia, Malaysia, the Philippines, and Thailand. Continental Europe: Belgium, France, Germany, Italy, the Netherlands, and Spain. For countries included in LAC5 and Other LAC, see note 11. Continental Europe, and the United States. If the largest fi rms in LAC are public sector companies that later privatized but still benefit is somewhat lower than expected, but the from a position of dominance in the market, share of formal businesses is larger than in one would expect to observe that the larg- other regions of similar levels of develop- est firms are relatively old. In contrast, if the ment. However, the share of informal busi- privatization of the 1990s resulted in a cleans- ness owners is also relatively large. These two ing effect, killing unproductive firms and giv- set of facts mesh when one observes that for- ing birth to a new entrepreneurial class, the mal sector fi rms in LAC tend to be smaller age distribution should be tilted toward the than fi rms in other parts of the world. Even relatively young. The problem is determining fi rms that manage to grow and generate sig- the right benchmark for comparison. Some nificant numbers of jobs are substantially top U.S. firms have had a remarkable ability smaller than in EAP4 or ECA. The last sec- to reinvent themselves: some companies date tion examines whether this bias toward small as far back as the early 1800s (Siegel 2007). firms is dictated by the environment in which Something similar is likely to have happened LAC firms operate or is instead more deeply in Europe. Perhaps for this reason, the most rooted in cultural and historical factors. interesting comparison is with EAP4. LAC firms are on average younger than firms in EAP4. In particular, both Other LAC and What is hindering high-growth LAC5 have relatively large numbers of large entrepreneurship: Culture, fi rms that are very young (for example, less institutions, or the environment? than 30 years old) and have a long tail, includ- Is LAC missing truly innovative entrepre- ing some firms that are 100 years old or older. neurs? Firms in the region are small given the The analysis so far provides a mixed level of development, limiting employment picture. The formal sector in LAC is rela- opportunities, creating too few good-paying tively dynamic. The rate of firm creation jobs, and contributing to the flourishing 50 LATIN AMERICAN ENTREPRENEURS of low- growth firms and self- employment. FIGURE 2.15 Share of business establishments in These facts may point to an environment Argentina owned by foreigners, 1910 that is not business friendly; they may also be signs of insufficient entrepreneurial zeal. Sugar refining The two hypotheses may be connected, as an Sugar mills environment that is less favorable to innova- Gas tion and high- growth entrepreneurship is Wool fabric likely to push potential employees into less Printing dynamic forms of entrepreneurship or even Flour mills outside the market (through migration, for Beer instance). Fruit vending One way to shed light on these questions Meat processing plant is to look back at history. At the beginning Electricity of the 20th century, insufficient entrepre- neurial spirit reflected the institutions and Sack cloth attitudes toward entrepreneurship inherited Tobacco from Spain on the one hand, and the lack of Saddlery techno-literacy and knowledge among the Lumber mills native population on the other (Maloney Glass 2012). Foreign-born entrepreneurs were in Foundries charge of the vast majority of businesses in Meat preserves the Americas. Indeed, census data for 1910 Brickmaking in Argentina show that 7 out of 10 busi- Tanneries nesses registered in Argentina were owned by All foreigners. Designers The influence of foreign-born ownership Mechanic shops in Argentina was heavily tilted toward sectors Dressmaking that were more technologically advanced, Carpentry including trolleys (100 percent foreign pres- Iron works ence), carriages and other vehicles (79 per- cent), iron works (71 percent), mechanic Baking shops (70 percent), and lumber mills (58 per- Tin work cent) (figure 2.15). Although foreign presence Footwear was also very important in some sectors that Shoe making are arguably less technologically advanced Carriages/Vehicles (for example, baking and cloth vending), in Tailoring general it was less important in some of the Clothes vending more traditional sectors, such as sugar mill- Espadrilles ing (30 percent), wool production (39 per- Furniture making cent), flour milling (43 percent), and beer Shirt making production (46 percent). Trolleys The tremendous presence of foreign-born Wool cleaning individuals in the productive network of Argentina is naturally influenced by the large 0 10 20 30 40 50 60 70 80 90 100 Foreign ownership (%) influx of migrants there. However, foreign- born residents own a large share of businesses Source: Maloney 2012. even in countries with smaller shares of migrants. In 1888 in Barranquilla, the major center of economic activity in Colombia at the time, some 64 percent of establishments E N T R E P R E N E U R S H I P, E N T R Y , A N D T H E L I F E C Y C L E O F F I R M S I N L A T I N A M E R I C A A N D T H E C A R I B B E A N 51 were owned by people born abroad (Maloney entrepreneurial drive among locals is behind 2012). In Mexico in 1935, the share of estab- innovative programs that try to attract for- lishments directed by people born abroad eign entrepreneurs (box 2.4). was 35 percent, while some 90 percent of the Cross-country heterogeneity in local con- workers in the same sectors were Mexican. ditions such as access to credit, barriers to Spaniards dominate the foreign-born pres- entry, and attitudes of institutions toward ence among the number of directors (16 per- entrepreneurship make it difficult to draw cent), followed by people born in the United causal relationships about why on average States (3 percent) and in Germany, France, people from some countries are more likely Poland, and the Russian Federation (2 per- than people from other countries to become cent each). entrepreneurs. It is almost impossible to iso- This evidence suggests that people from late the role of the environment from the role LAC were not particularly prone to entre- of innate entrepreneurial ability or predispo- preneurial activities at the turn of the cen- sition from cross-country comparisons. One tury. If this tendency reflected cultural possible albeit imperfect way to do so is to traits or deficits in human capital that were compare immigrants from different countries strongly persistent, it could explain the bias in a particular country. toward low-growth entrepreneurial fi rms in Messina, Özden, and Sarzosa (2013) study the region. 22 The perception of insufficient differences across countries of origin in the BOX 2.4 Importing entrepreneurs: Start-Up Chile In August 2010, the Chilean economic development hosted; more than $8 million had been raised from agency, CORFO, launched an innovative initiative investors in the United States, Argentina, and Mex- with the aim of enhancing the country’s competitive- ico; and projects had achieved sales of $550,000 and ness through technology, innovation, and entrepre- employed 228 people. The goal is to reach 1,000 neurship. The program, Start-Up Chile, aims to cre- projects by 2014. ate a new entrepreneurial environment by enhancing The program also created a network of entrepre- international connections and removing the barriers neurs. Through the online platform Meetups, Start- faced by entrepreneurs: limited access to credit, low Up Chile entrepreneurs and local interested parties adoption of new technologies, and the lack of inter- can meet to share experiences and challenges. This national customers. part of the program is intended to promote entre- Early-stage, high-potential entrepreneurs received preneurial activity and contribute to changing the seed capital of $40,000, which they had to match culture in the local environment. with at least $4,000 of their own resources. Critics of the program complain about the Entrepreneurs were approved in an admission bureaucracy of the reimbursement process and the process conducted by Silicon Valley experts and a lack of commitment of certain participants, high- Chilean innovation board. lighting problems with the selection process. The The truly innovative aspect of the initiative is program needs to defi ne its long-term goals and a that the program targets foreign entrepreneurs or method for measuring results. Although it is too Chileans developing projects abroad. Entrepreneurs early to assess the impact on economic activity, this are required to spend at least six months in Chile, initiative reveals the increasing interest of govern- where a variety of facilities, including a one-year ments in attracting the most promising entrepre- visa, social security, a bank account, and a work- neurs to their countries. place with wireless Internet, are provided. Source: Applegate and others 2012. The pilot launched a modest 22 start-ups from 14 countries. By June 2012, 323 start-ups had been 52 LATIN AMERICAN ENTREPRENEURS entrepreneurial experiences of migrants to the the Current Population Survey (CPS), con- United States. The main advantage of looking ducted by the U.S. Bureau of Labor Statistics, at migrants in one country is that they share about 30 percent of self- employed people the same economic environment. Perhaps the with incorporated businesses had firms with most important limitation is that migration more than 10 employees; in contrast, only is not a random phenomenon. A combina- 1 percent of unincorporated businesses did tion of factors, including the socioeconomic so. Hence, the proxy for transformational situation at home and expected prospects entrepreneurship used in the analysis is the after migration, determines the decision to incorporation of the business. migrate. If such selection were similar across Figure 2.16 shows differences across coun- countries, one could compare differences in tries of origin in the self-employment rates of entrepreneurship across migrants from dif- migrants in the United States after control- ferent birth countries and draw conclusions ling for differences in education, age, and about differences in entrepreneurial drive. year of arrival. Results for men are exam- In fact, these factors differ across migrant ined, in order to avoid dealing with problems groups. Indian migrants in the United States associated with self-selection into participa- tend to be highly educated, even more so tion in the labor market. As expected, most than natives, whereas migrants from Mexico migrants have a lower likelihood of being and El Salvador have, on average, less educa- self- employed than non- Hispanic U.S. tion than U.S. natives. The year of migration natives, but very interesting differences across also differs across groups, and the moment of region of origin emerge.24 arrival is likely to influence the entrepreneur- With the exception of Spanish- speaking ial experience. The costs of migration are Caribbeans, people from LAC appear to likely to be different as well, with geographi- be less entrepreneurial than migrants from cal proximity reducing such costs. Some of other regions. The least entrepreneurial these differences can be accounted for by among migrants are Mexicans, closely fol- controlling for observable characteristics of lowed by migrants from Central America and migrants and the year of migration. non-Spanish-speaking Caribbean islands. Another complication is that the destina- Migrants from South America do some- tion of migrants within the recipient country what better, about as well as migrants from is not random. Migrants from different ori- East Asia and the Anglo- Saxon countries gins tend to cluster in geographical enclaves, (Australia, Ireland, New Zealand, and the and the characteristics of each of these geo- United Kingdom). The estimated effects are graphical areas, including the entrepreneur- large. Being of Mexican origin reduces the ial environment, are likely to differ. Indeed, likelihood of being self- employed by almost in a pioneering study, Borjas (1986) finds 70 percent with respect to being a U.S. native that part of the migrant/native gap in self- (from 14 percent to 4 percent). employment rates reflects “enclave” effects.23 T he gap bet ween L at i n A meric a n The fi nal challenge to studying entrepre- migrants and U.S. natives is much larger neurship among migrants in the United States among self- employed people with incorpo- is related to the difficulties in separating high- rated businesses, although differences by and low-growth-potential entrepreneurship. region of origin are stable across classifica- Messina, Özden, and Sarzosa (2013) employ tions. Among Mexican immigrants, the gap a fundamental dimension that distinguishes with respect to non-Hispanic U.S. natives is the two forms of entrepreneurship: their almost 4 percentage points, which suggests capacity to generate employment. Trans- the virtual nonexistence of Mexicans within formational entrepreneurs by and large run this type of entrepreneurial activity, as the larger fi rms than low-growth entrepreneurs share of white U.S. natives that have incorpo- and hence generate more jobs. According to rated businesses is about 4 percent. E N T R E P R E N E U R S H I P, E N T R Y , A N D T H E L I F E C Y C L E O F F I R M S I N L A T I N A M E R I C A A N D T H E C A R I B B E A N 53 FIGURE 2.16 Entrepreneurship among immigrants and natives in the United States, by type of business and country a. Self-employed b. Not incorporated Mexico Central America Central America Mexico Caribbean Caribbean South Asia South Asia East Asia East Asia Anglo-Saxon Anglo-Saxon South America South America Eastern Europe Eastern Europe Africa Africa Middle East and North Africa Spanish-speaking Caribbean Spanish-speaking Caribbean Middle East and North Africa Spanish-speaking United States Western Europe Western Europe Spanish-speaking United States –0.1 –0.05 0 0.05 –0.06 –0.04 –0.02 0 Marginal effect Marginal effect c. Incorporated LAC countries Non-LAC countries Mexico Caribbean Central America South Asia East Asia South America Africa Anglo-Saxon Eastern Europe Middle East and North Africa Spanish-speaking Caribbean Spanish-speaking United States Western Europe –0.04 –0.02 0 –0.02 0.04 Marginal effect Source: Adapted from Messina, Özden, and Sarzosa 2013. Note: Figure shows the marginal effects of country of origin grouped by region for a 35-year- old male migrant with secondary education who immigrated between 1992 and 1996 and is observed in the period 2001– 04 (that is, 5–12 years after arrival in the United States). The baseline category is U.S. non-Hispanic natives. These estimates are obtained from logit regressions (panel a) and multinomial logits (panels b and c) in specifications that control for age, age squared, educational attainment, the log of number of years since arrival in the United States, sector dummies, a citizenship dummy, and country of origin dummies. Points represent the marginal effect for each group. Bars represent the 95 confidence interval. Red points are migrants from Latin America and the Caribbean (LAC). The large gap among LAC migrants in the the gap to shrink as the immigrant spends share of incorporated businesses in the United time in the host country and assimilates. This States may reflect difficulties in accessing hypothesis is examined by looking at the gap credit and other market imperfections that in businesses owned by immigrants from dif- are more likely to affect migrants than U.S. ferent regions over time. natives (although such difficulties are likely Across most regions of origin, the gap to be similar across migrants of different ori- in not incorporated self- employment with gin). If this were the case, one would expect respect to non- Hispanic natives dissipates 54 LATIN AMERICAN ENTREPRENEURS FIGURE 2.17 Marginal effects of years in United States on entrepreneurship gap between migrant and nonmigrant white men, by cohort of arrival and region a. Mexico b. South Asia Nonincorporated firms Incorporated firms Nonincorporated firms Incorporated firms 0.02 0.02 0.1 0.1 Marginal effects 0 0 0.05 0.05 –0.02 –0.02 0 0 –0.04 –0.04 –0.06 –0.06 –0.04 –0.04 3 5 7 9 11 13 15 17 19 3 5 7 9 11 13 15 17 19 3 5 7 9 11 13 15 17 19 3 5 7 9 11 13 15 17 19 Number of years in Number of years in Number of years in Number of years in United States United States United States United States c. Caribbean d. Central America Nonincorporated firms Incorporated firms Nonincorporated firms Incorporated firms 0.02 0.02 0.02 0.02 Marginal effects 0 0 0 0 –0.02 –0.02 –0.02 –0.02 –0.04 –0.04 –0.04 –0.04 –0.06 –0.06 –0.06 –0.06 3 5 7 9 11 13 15 17 19 3 5 7 9 11 13 15 17 19 3 5 7 9 11 13 15 17 19 3 5 7 9 11 13 15 17 19 Number of years in Number of years in Number of years in Number of years in United States United States United States United States Year of arrival in United States: 1987–91 cohort 1992–97 cohort 1998–2001 cohort Source: Messina, Özden, and Sarzosa 2013. Note: Each line represents the marginal effects of years in the United States for each cohort of immigrants. Estimates are obtained from multinomial logits in specifications that con- trol for age, age squared, educational attainment, the log of number of years since arrival in the United States, sector dummy, a citizenship dummy, and countries-of-origin dummies. The marginal effects are obtained for a male migrant with secondary education. after 15–17 years in the United States (fig- Notes ure 2.17). In contrast, the gap in incorporated 1. These statistics are averages of household businesses persists among migrants from surveys from Argentina, Chile, Colombia, many regions, although there is some conver- the Dominican Republic, Ecuador, El Salva- gence after 15 years for some regions outside dor, Mexico, Panama, Peru, and Paraguay LAC. Convergence occurs for migrants from in 2009–10. The samples were restricted to South Asia (panel b) but not among Mexicans workers ages 25–60 who worked more than (panel a), non-Spanish-speaking Caribbeans 30 hours in the reference week (SEDLAC). (panel c), or Central Americans (panel d). 2. Some of the most successful entrepreneurs in L AC m ig rants thus catch up w ith the region are probably not captured by the non- Hispanic natives in the type of self- household surveys, which tend to underrepre- employment they are used to at home— sent the upper tail of the income distribution. namely, small-scale self-employment. But just 3. Poschke (2013a) argues that a firm’s produc- tivity is an increasing and convex function of like at home, they have a harder time engag- the ability of the entrepreneur who runs it— ing in the dynamic activities that have high that is, both the level of productivity and the employment-generation potential. E N T R E P R E N E U R S H I P, E N T R Y , A N D T H E L I F E C Y C L E O F F I R M S I N L A T I N A M E R I C A A N D T H E C A R I B B E A N 55 rate at which productivity increases are posi- Asia) includes Albania, Armenia, Azerbaijan, tively associated with ability. Belarus, Bosnia and Herzegovina, Bulgaria, 4. See, for instance, Boeri and Jimeno (2005) for Georgia, Kazakhstan, Latvia, Lithuania, FYR a rationale of size-related employment protec- Macedonia, Moldova, Romania, the Russian tion in Italy and an analysis of its impact on Federation, Serbia, Turkey, Turkmenistan, the labor market. Ukraine, and Uzbekistan. EAP4 includes 5. Jovanovic (1994) provides a very rich treat- Indonesia, Malaysia, the Philippines, and ment of the occupational choice between Thailand. High-income economies include entrepreneurship and wage work. In this Australia; Canada; Hong Kong SAR, China; model, entrepreneurs are concentrated among Israel; Japan; the Republic of Korea; Kuwait; high-ability individuals. In contrast, the simple New Zealand; Oman; Saudi Arabia; Singa- framework presented here highlights the coex- pore; Switzerland; the United Arab Emirates; istence of two critical masses of entrepreneurs the United States; and all countries in the at the extremes of the ability distribution. European Union not included in ECA. The set 6. For simplicity and comparability across sur- of economies from each group used in figures veys, and unless otherwise stated in the text, throughout this chapter varies according to the self-employed are assumed not to have data availability. paid employees, although in the data some 12. A perhaps more obvious explanation would self-employed run small firms. On average be that entrepreneurs are a more heteroge- in Latin America and the Caribbean, about neous group than salaried workers. Repli- 14 percent of self-employed people had at least cation of the analysis with the residuals of one employee in 2009. a flexible Mincer regression that includes a 7. These 73 countries represent two-thirds of second-order polynomial in age and educa- the population of the developing world. Data tion, a gender dummy, and their interactions sources vary by country. They are harmonized results in residual wage distributions that still microlevel household surveys collected by the display substantially higher variance for both Development Economics Group (DEC) of the groups of entrepreneurs. It is not possible to World Bank in the International Income Dis- rule out the possibility that such higher vari- tribution Database (I2D2). See Gindling and ance in earnings is the result of greater het- Newhouse (2012) for details. erogeneity in unobservable characteristics (for 8. Loayza and Rigolini (2011) provide a detailed example, ability). treatment of the relationship between self- 13. Interviewers were instructed to ask if the employment and GDP per capita. businesses were registered with the relevant 9. For example, the share of self-employment in authorities and had a license or certificate. the United States fell by half between 1910 14. For an account in an endogenous growth and 1990, from 16 percent to 8 percent of framework, see Aghion and Howitt (1992). workers in the adult population (Fairlie and 15. The procedures, time, and costs to start up Meyer 1999). a business refer to the requirements to reg- 10. In order to mitigate differences in the length of ister formally a limited liability company of the schooling period across countries, we con- small to medium size (10–50). Our entry data centrate on the population in the age bracket include all formal registrations. The costs to 25–65. register smaller firms may be different from 11. Throughout this chapter we use the following those captured by Doing Business. In addi- groups of economies unless otherwise noted. tion, local authorities, such as authorities in LAC5 includes Argentina, Brazil, Chile, charge of zoning laws and building permits, Colombia, and Mexico. Other LAC includes impose some potentially important restric- Bolivia, Costa Rica, the Dominican Republic, tions on small entrepreneurs that the Doing Ecuador, El Salvador, Guatemala, Honduras, Business indicators do not include. However, Nicaragua, Paraguay, Peru, Uruguay, and the costs reported by Doing Business are likely República Bolivariana de Venezuela. Carib- to be correlated with the overall costs to set up bean includes Antigua and Barbuda, Cuba, a business. Dominica, Grenada, Guyana, Haiti, Jamaica, 16. Administrative barriers to entrepreneurship in St. Kitts and Nevis, St. Lucia, St. Vincent and OECD countries refer to the regulatory frame- the Grenadines, Suriname, and Trinidad and work in 2008, whereas in LAC the indicator Tobago. ECA (Eastern Europe and Central was constructed in 2013. See Wölfl, Kozluk, 56 LATIN AMERICAN ENTREPRENEURS and Nicoletti (2009) for a description of the Acs, Z. J., S. Desai, and L. F. Klapper. 2008. methodology. “What Does ‘Entrepreneurship’ Data Really 17. The standard deviation of the costs of red tape Show?” Small Business Economics 31 (3): declined by a factor of more than four—from 265– 81. 218 in 2004 to 48 in 2013—for the 151 coun- Aghion, P., and P. Howitt. 1992. “A Model tries for which data were available throughout of Growth through Creative Destruction.” the period. Econometrica 60 (2): 325–51. 18. Results are very similar if the yearly observa- Alesina, A., S. Ardagna, G. Nicoletti, and F. Schi- tions or period averages are used. antarelli. 2005. “Regulation and Invest- 19. The data set covers Colombian manufacturing ment.” Journal of the European Economic firms with more than 10 employees or annual Association 3 (4): 791– 825. production of more than about $100,000. Almeida, R., and P. Carneiro. 2011. “Enforce- The 10-employee threshold has implications ment of Labor Regulation and Informality.” for the study of business dynamics, as dis- IZA Discussion Paper 5902, Institute for the cussed in box 3.3 in chapter 3. Study of Labor, Bonn. 20. Although all of the analysis is based on estab- Applegate, L. M., R. K. William, J. Lerner, D. D. lishments, the main patterns remain when Pomeranz, G. A. Herrero, and C. Scott. 2012. firms are considered (see Eslava and Halti- “Start- Up Chile: April 2012.” Harvard Busi- wanger 2013 for details). ness School Case 812–158. 21. Growth rates in this section are defined as the Ardagna, S., and A. M. Lusardi. 2010. “Explain- difference in firm size between two consecutive ing International Differences in Entrepreneur- periods divided by the average employment in ship: The Role of Individual Characteristics the two periods. These growth rates, popular- and Regulatory Constraints.” In International ized by the work of Davis, Haltiwanger, and Differences in Entrepreneurship, edited by Schuh (1996), present two main advantages. J. Lerner and A. Schoar, 17– 62. Chicago: Uni- First, they are symmetric for expansions and versity of Chicago Press. contractions, ranging in the interval [–2, +2]. Åstebro, T. 2003. “The Return to Independent Second, they allow the treatment of firm birth Invention: Evidence of Unrealistic Optimism, and death in the computation of the growth Risk Seeking or Skewness Loving?” Economic rate. Journal 113 (484): 226–39. 22. For an interesting discussion of the main per- Audretsch, D. B., and M. Keilbach. 2005. “Entre- sonality traits entrepreneurs typically have, preneurship Capital: Determinants and see CAF (2013). Impact.” CEPR Discussion Paper 4905, Centre 23. There is little we could do to tackle this for Economic Policy Research, London. problem with the data we have. Hence, the Bartelsman, E., J. Haltiwanger, and S. Scarpetta. results remain informative but present some 2009. “Measuring and Analyzing Cross- limitations. Country Differences in Firm Dynamics.” In 24. Migrants other than Mexicans were pooled by Producer Dynamics: New Evidence from region of origin because the number of entre- Micro Data, edited by T. Dunne, J. B. Jensen, preneurial migrants in the American Commu- and M. J. Roberts, 15–76. Chicago: University nity Survey (U.S. Census) samples were not of Chicago Press for the National Bureau of large enough for some countries in particular Economic Research. cohorts of arrival. Mexicans constitute a suffi- Blanchflower, D. G., and B. D. Meyer. 1994. ciently large group to be considered separately. “A Longitudinal Analysis of the Young Self- Employed in Australia and the United States.” Small Business Economics 6 (1): 1–19. References Blanchflower, D. G., and A. J. Oswald. 1998. Acs, Z. J. 2008. “Foundations of High Impact “What Makes an Entrepreneur?” Journal of Entrepreneurship.” Foundations and Trends Labor Economics 16 (1): 26– 60. in Entrepreneurship 4 (6): 535– 620. Boeri, T., and J. F. Jimeno. 2005. “The Effects Acs, Z. J., and D. B. Audretsch. 1989a. “Small- of Employment Protection: Learning from Firm Entry in U.S. Manufacturing.” Econom- Variable Enforcement.” European Economic ica 56 (222): 255– 65. Review 49 (8): 2057–77. ———. 1989b. “Small Firms in U.S. Manufactur- Borjas, G. L. 1986. “The Self- Employment ing: A First report.” Economics Letters 31 (4): Experience of Migrants.” Journal of Human 399– 402. Resources 21 (4): 485–506. E N T R E P R E N E U R S H I P, E N T R Y , A N D T H E L I F E C Y C L E O F F I R M S I N L A T I N A M E R I C A A N D T H E C A R I B B E A N 57 Botero, J., S. Djankov, R. La Porta, F. Lopez-de- Doing Business (database). World Bank Group, Silanes, and A. Shleifer. 2004. “The Regula- Washington, DC. http://www.doingbusiness tion of Labor.” Quarterly Journal of Econom- .org. ics 119: 1339– 82. Dosi, G., and D. Lovallo. 1998. “Rational Entre- Busso, M., L. Madrigal, and C. Pagés-Serra. preneurs or Optimistic Martyrs? Some Consid- 2012. “Productivity and Resource Misalloca- erations on Technological Regimes, Corporate tion in Latin America.” Inter-American Devel- Entries, and the Evolutionary Role of Deci- opment Bank, Research Department, Wash- sion Biases.” In Foresights and Oversights in ington, DC. Technological Change, edited by R. Garud, Cabral, L. 1997. ”Entry Mistakes.” CEPR Dis- P. Nayyar, and Z. Shapiro, 41– 68. Cambridge: cussion Paper, Centre for Economic Policy Cambridge University Press. Research, London. Enterprise Surveys (database). World Bank Group, CAF (Corporación Andina de Fomento). 2013. Washington, DC. http://www.enterprise Emprendimientos en América Latina: Desde surveys.org/. la subsistencia hacia la transformación pro- Eslava, M., and J. Haltiwanger. 2013. “Young ductiva. Caracas, República Bolivariana de Businesses, Entrepreneurship, and the Dynam- Venezuela: Publicaciones CAF. ics of Employment and Output in Colombia’s Camerer, C., and D. Lovallo. 1999. “Overcon- Manufacturing Industry.” Working paper, fidence and Excess Entry: An Experimental Corporación Andina de Fomento, Caracas, Approach.” American Economic Review 89 República Bolivariana de Venezuela. (1): 306–18. Evans, D. S., and L. S. Leighton. 1990. “Small Carree, M., A. J. van Stel, A. R. Thurik, and Business Formation by Unemployed and S. Wennekers. 2007. “The Relationship Employed Workers.” Small Business Econom- between Economic Development and Business ics 2 (4): 319–30. Ownership Revisited.” Entrepreneurship and Fafchamps, M., C. Woodruff, and W. Yin. 2013. Regional Development 19 (3): 281–91. Identifying Gazelles among Micro and Small Coelho, M., D. de Meza, and D. J. Reyniers. Enterprises in Ghana. Ongoing research 2004. “Irrational Exuberance, Entrepreneurial project, Innovations for Poverty Action, New Finance and Public Policy.” International Tax Haven, CT. and Public Finance 11 (4): 391– 417. Fairlie, R. W., and B. D. Meyer. 1999. “Trends Creedy, J., and P. S. Johnson. 1983. “Firm For- in Self-Employment among White and Black mation in Manufacturing Industry.” Applied Men: 1910 –1990.” NBER Working Paper Economics 15: 177– 85. 7182, National Bureau of Economic Research, Cressy, R. 2006. “Why Do Most Firms Die Cambridge, MA. Young?” Small Business Economics 26 (2): Foster, L., J. Haltiwanger, and C. Syverson. 103–16. 2005. “Reallocation, Firm Turnover, and Effi- Davis, S. J., J. Haltiwanger, and S. Schuh. 1996. ciency: Selection on Productivity or Profitabil- Job Creation and Destruction. Cambridge, ity?” NBER Working Paper 11555, National MA: MIT Press. Bureau of Economic Research, Cambridge, Dejardin, M. 2011. “Linking Net Entry to MA. Regional Economic Growth.” Small Business Geroski, P. A. 1995. “What Do We Know about Economics 36: 443– 60. Entry?” International Journal of Industrial De Mel, S., D. McKenzie, and C. Woodruff. 2012. Organization 13 (4): 421– 40. “Business Training and Female Enterprise Geroski, P. A., and M. Mazzucato. 2001. “Mod- Start-Up, Growth, and Dynamics: Experimen- elling the Dynamics of Industry Populations.” tal Evidence from Sri Lanka.” Working Paper, International Journal of Industrial Organiza- World Bank, Washington, DC. tion 19 (7): 1003–22. De Meza, D. 2002. “Overlending?” Economic Ghani, E., W. R. Kerr, and S. D. O’Connell. Journal 112 (477): F17– F31. 2011. ”Who Creates Jobs?” World Bank, Pov- Djankov, S., R. La Porta, F. Lopez-de-Silanes, and erty Reduction and Economic Management A. Shleifer. 2002. “The Regulation of Entry.” Network, Washington, DC. Quarterly Journal of Economics 117 (1) 1–37. Gindling, T. H., and D. Newhouse. 2012. “Self- Djankov, S., E. Miguel, Y. Qian, G. Roland, and Employment in the Developing World.” Policy E. Zhuravskaya. 2005. “Who Are Russia’s Research Working Paper 6201, World Bank, Entrepreneurs?” Journal of the European Washington, DC. Economic Association 3 (2–3): 1–11. 58 LATIN AMERICAN ENTREPRENEURS Haltiwanger, J. 2011. “Job Creation and Firm Loayza, N. V., and J. Rigolini. 2011. “Informal Dynamics in the United States.” Innovation Employment: Safety Net or Growth Engine?” Policy and the Economy, National Bureau of World Development 39 (9): 1503–15. Economic Research, Cambridge, MA. Lucas, R. 1978. “On the Size Distribution of Busi- Haltiwanger, J., R. Jarmin, and J. Miranda. 2013. ness Firms.” Bell Journal of Economics, 9 (2): “Who Creates Jobs? Small versus Large ver- 508–523. sus Young.” NBER Working Paper 16300, Malchow- M øller, N., B . S chjerning, and National Bureau of Economic Research, Cam- A. Sørensen. 2011. “Entrepreneurship, Job bridge, MA. Creation and Wage Growth.” Small Business Hamilton, B. H. 2000. “Does Entrepreneurship Economics 36 (1): 15–32. Pay? An Empirical Analysis of the Returns to Maloney, W. F. 2004. “Informality Revisited.” Self-Employment.” Journal of Political Econ- World Development 32 (7): 1159–78. omy 108 (3): 604–31. ———. 2012. “Part I: What If E.T. Had Landed in Highfield, R., and R. Smiley. 1987. “New Busi- Latin America in 1850? Techno-literate Entre- ness Starts and Economic Activity: An Empiri- preneurship and Development.” World Bank, cal Investigation.” International Journal of Washington, DC. Industrial Organization 5 (1): 51– 66. McKenzie, D., and C. Woodruff. 2013. “What Hsieh, C. T., and P. J. Klenow. 2012. “The Life Are We Learning from Business Training Cycle of Plants in India and Mexico.” National and Entrepreneurship Evaluations around the Bureau of Economic Research, Cambridge, Developing World?” CAGE Online Work- MA. ing Paper 115, Competitive Advantage in the Hurst, E., and B. Pugsley. 2011. “What Do Small Global Economy (CAGE). Businesses Do?” Brookings Papers on Eco- Messina, J. 2006. “The Role of Product Mar- nomic Activity, Washington, DC. ket Regulations in the Process of Structural Iyer, R., and A. Schoar. 2010. “Are There Cul- Change.” European Economic Review 50 (7): tural Determinants of Entrepreneurship?” 1863–90. In International Differences in Entrepreneur- Messina, J., C. Özden, and M. Sarzosa. 2013 ship, edited by J. Lerner and A. Schoar, 209– “Entrepreneurship Choices among Latin- 40. Chicago: University of Chicago Press for American Immigrants to the U.S.” World the National Bureau of Economic Research. Bank, Washington, DC. Jovanovic, B. 1994 “Firm Formation with Hetero- Millán, A. 2012. “The Choice to Become Self- geneous Management and Labor Skills.” Small Employed: Acknowledging Frictions.” Univer- Business Economics 6 (3): 185–91. sity Carlos III, Madrid. Kaufman, D., A. Kraay, and M. Mastruzzi. 2010. Mondragón, C., and X. Peña. 2010. “Business “Response to ‘What Do the Worldwide Gover- Ownership and Self- Employment in Devel- nance Indicators Measure?’” European Jour- oping Economies: The Colombian Case.” In nal of Development Research 22 (1): 55–58. International Differences in Entrepreneur- Khemani, R. S., and D. M. Shapiro. 1986. “The ship, 89– 127. Cambridge, MA: National Determinants of New Plant Entry in Canada.” Bureau of Economic Research. Applied Economics 18 (11): 1243–57. Neumark, D., B. Wall, and J. Zhang. 2011. “Do Kihlstrom, R. E., and J.-J. Laffont. 1979. “A Gen- Small Businesses Create More Jobs? New Evi- eral Equilibrium Entrepreneurial Theory of dence for the United States from the National Firm Formation Based on Risk Aversion.” Establishment Time Series.” Review of Eco- Journal of Political Economy 87 (4): 719– 48. nomics and Statistics 93 (1): 16–29. Klapper, L., L. Laeven, and R. Rajan. 2006. Nicoletti, G., and S. Scarpetta. 2003. “Regula- “Entry Regulation as a Barrier to Entrepre- tion, Productivity, and Growth: OECD Evi- neurship.” Journal of Financial Economics dence.” Policy Research Working Paper 2944, 82 (3): 591– 629. World Bank, Washington, DC. Klapper, L., and D. Randall. 2012. “Entrepre- OECD (Organisation for Economic Co-operation neurship in LAC: Evidence from the Gallup and Development). 2003. The Sources of Eco- World Poll Survey.” Background paper for this nomic Growth in OECD Countries. Paris: report. OECD. Knight, F. 1921. Risk, Uncertainty, and Profi t. Orr, D. 1974. “The Determinants of Entry: Boston: Houghton Mifflin. A Study of the Canadian Manufacturing E N T R E P R E N E U R S H I P, E N T R Y , A N D T H E L I F E C Y C L E O F F I R M S I N L A T I N A M E R I C A A N D T H E C A R I B B E A N 59 Industries.” Review of Economics and Statis- National University of La Plata, La Plata, tics 56 (1): 58– 66. Argentina, and World Bank, Washington, DC. Oxenfeldt, A. R. 1943. “New Firms and Free Siegel, J. 2007. Stocks in the Long Run: The Enterprise: Pre-War and Post-War Aspects.” Definitive Guide to Financial Market Returns American Council on Public Affairs, Washing- and Long Term Investment Strategies. New ton, DC. York: McGraw-Hill. Parker, S. 2006. “A Selection-Based Theory of the Storey, D. J. 1991. “The Birth of New Firms: Transition from Employment to Entrepreneur- Does Unemployment Matter? A Review of the ship: The Role of Employer Size.” IZA Dis- Evidence.” Small Business Economics 3 (3): cussion Paper 2071, Institute for the Study of 167–78. Labor, Bonn. ———. 1994. Understanding the Small Business Parker, S. C. 1997. “The Effects of Risk on Self- Sector. London: Routledge. Employment.” Small Business Economics Storey, D. J., and A. M. Jones. 1987. “New Firm 9 (4): 515–22. Formation: A Labour Market Approach to Poschke, M. 2013a. “The Decision to Become an Industrial Entry.” Scottish Journal of Political Entrepreneur and the Firm Size Distribution: Economy 34 (1): 37–51. A Unifying Framework.” Background paper Vivarelli, M. 1991. “The Birth of New Enter- for this report. prises.” Small Business Economics 3 (3): ———. 2013b. “Entrepreneurs out of Necessity: 215–23. A Snapshot.” Applied Economics Letters ———. 2004. “Are All the Potential Entrepre- 20 (7): 658– 63. neurs So Good?” Small Business Econom- ———. 2013c. “Who Becomes an Entrepreneur? ics 23 (1): 41– 49. Labor Market Prospects and Occupational ———. 2013. “Entrepreneurship in Advanced and Choice.” Journal of Economic Dynamics and Developing Countries: A Microeconomic Per- Control 37 (3): 693–710. spective.” Background paper for this report. Reynolds, P. D., S. Michael Camp, W. D. Bygrave, Wennekers, S., and A. R. Thurik. 1999. “Link- E. Autio, and M. Hay. 2001. Global Entre- ing Entrepreneurship and Economic Growth.” preneurship Monitor. 2001 Summary Report, Small Business Economics 13 (1): 27–55. London Business School, London, and Babson Wennekers, S., A. J. van Stel, A.R. Thurik, and College, Babson Park, MA. P.D. Reynolds. 2005. “Nascent Entrepreneur- Rissman, E. R. 2007. “Labor Market Transitions ship and the Level of Economic Development.” and Self-Employment.” Federal Reserve Bank Small Business Economics 24 (3): 293–309. of Chicago Working Paper 2007–14, Chicago. Wölfl, I., T. Kozluk, and G. Nicoletti. 2009. “Ten Santarelli, E., M. Carree, and I. Verheul. 2009. Years of Product Market Reform in OECD “Unemployment and Firm Entry and Exit: Countries: Insights from a Revised PMR Indi- An Update on a Controversial Relationship.” cator.” OECD Economic Department Work- Regional Studies 43 (8): 1061–73. ing Paper 695, Organisation for Economic Co- Schoar, A. 2010. “The Divide between Subsis- operation and Development, Paris. tence and Transformational Entrepreneur- World Bank. 2011. Doing Business. Washington, ship.” In Innovation Policy and the Economy, DC: World Bank. vol. 10, edited by J. Lerner and S. Stern, 57– 81. World Bank Entrepreneurship Database. Wash- Chicago: University of Chicago Press. ington, DC. http://www.doingbusiness.org Schumpeter, J. A. 1934. The Theory of Economic /data/exploretopics/entrepreneurship. Development. Cambridge, MA: Harvard Uni- World Development Indicators (database). World versity Press. Bank, Washington, DC. http://data.worldbank ———. 1939. Business Cycles: A Theoretical, . o r g /d a t a - c a t a l o g / w o r l d - d e v e l o p m e n t Historical and Statistical Analysis of the -indicators. Capitalist Process. New York: McGraw-Hill. Worldwide Governance Indicators (database). ———. 1943. Capitalism, Socialism and Democ- World Bank, Washington, DC. http://info racy. London: Allen and Unwin (originally .worldbank.org/governance/wgi/index.asp. published in the United States in 1942; Zacharakis, A. L., W. D. Bygrave, and D. A. Shep- reprinted by Routledge, London in 1994). herd. 2000. Global Entrepreneurship Moni- SEDLAC (Socio- Economic Database for Latin tor 2012 United States Report. Babson Col- America and the Caribbean). Center for lege, Babson Park, MA. Distributive, Labor and Social (CEDLAS), Entrepreneurship by Incumbent Firms: What Explains the 3 Innovation Gap? Latin America and the Caribbean suffers from an innovation gap. On average, its entrepre- neurs introduce new products less frequently, invest less in research and development, and hold fewer patents than entrepreneurs in other regions; moreover, their management practices are far from global best practices. A deficit in human capital for innovation, lack of competi- tion, and inadequate intellectual property rights may explain the region’s underperformance. E ntry is only one dimension of entre- high-tech sectors (although it briefly reviews preneurship. To survive, firms must this area as well). It shows that, with the innovate. Incumbent fi rms can inno- exception of a few top- performing firms vate by bringing new products into the mar- (examined in chapters 4 and 5), the types of ket or by exploring new markets at home innovations that drive productivity growth and abroad (product innovation). They can in low- and middle-income countries differ also improve their internal processes and from those that drive growth in high-income management practices (process innovation). economies. Most fi rms in emerging markets These innovations are usually invisible to engage in activities that lie far from the tech- the final consumer but may be even more nological frontier; they innovate by adopt- important for surviving the test of the mar- ing and adapting products and production ket than other innovations. processes that have already been tested in Innovations by incumbent firms are likely countries that are at the technology frontier to be at least as important as entry rates for (Grossman and Helpman 1991; Segerstrom long-term economic growth and employment 1991; Dutz 2007; Ayyagari, Demirgüc- generation. This chapter reviews the innova- Kunt, and Maksimovic 2011). Cutting- edge tive performance of firms in Latin Ameri- innovation tends to gradually become more can and the Caribbean (LAC) in terms of important when firms in a country approach their propensity to introduce new products, the world technology frontier (Acemoglu, change their internal processes in search of Aghion, and Zilibotti 2006). efficiency, invest in research and development Overall, the evidence suggests that LAC (R&D), and receive patents. fi rms tend to score toward the bottom end The chapter moves away from tradi- of the spectrum in product innovation. tional analyses of cutting-edge innovation in Firms in East Asia and Eastern Europe tend 61 62 LATIN AMERICAN ENTREPRENEURS to introduce new products more frequently, addressing to some extent the vulnerability conduct more R&D, and obtain patents in concern caused by inflexible labor markets. the United States more often than do fi rms This progress notwithstanding, LAC in LAC. countries still underperform their peers in Assessing process innovation across coun- some aspects, and regulation may still ham- tries is difficult because of data constraints. per fi rms’ ability to innovate. Nevertheless, Fortunately, thanks to the emergence of a although some benefits may still be extracted wave of management surveys that are com- from improving the regulatory framework, parable across countries, it is now possible regulation may no longer be the most severe to compare the quality of management prac- obstacle to unleashing the private sector’s tices across countries and assess their rela- innovative potential. tionship with fi rm productivity. The picture The toughest challenge ahead may be to that emerges suggests that much remains to address other aspects of the regulatory envi- be done in LAC on the process front as well. ronment, such as governance and uncom- With a few exceptions, management pro- petitive practices. To be sure, the relationship cesses (even by the relatively large firms that between competition and innovation is com- are included in the surveys) remain weak and plex. In sectors with increasing returns to are comparable to those of firms in China or scale, for instance, there is a strong rationale India. Weak processes may not be the sole for allowing a single (well-regulated) monop- determinant of long-term fi rm productivity, olist to operate. And new research indicates but given LAC’s labor costs, which are signif- that even in sectors with low returns to scale, icantly higher than Asia’s, poor management “excess” competition can, at times, reduce processes can hamper productivity. firms’ incentives to innovate (Aghion, Bloom, After comparing innovation performance and others 2005). LAC is far from the tipping across regions, the chapter reviews factors point at which excess competition may hurt that can potentially affect fi rms’ innovative innovation, however. Although there are seri- potential. The focus is on four factors that ous technical challenges in measuring de facto have been shown to affect innovation: regu- competition, LAC shows a pattern consistent lations, competition, access to fi nance, and with a few actors grabbing a large share of the entrepreneurial skills. It also briefly reviews market and having little incentive to innovate. the extent to which policy makers can affect Overall, as discussed in chapter 6, the region entrepreneurship and innovation by exploit- exhibits high concentration both across and ing agglomerations and spatial spillovers. within markets: production remains less Designing regulations requires balancing diversified than in other countries, and within the protection of workers and consumers sectors, especially in nontradable industries, a against the ability of firms to operate without few firms dominate the market. unnecessary obstacles. The consensus on the An important stream of research has historical evolution of the region’s regulatory documented the channels through which environment is that in the 1980s, policy dis- underdeveloped financial markets and insuf- tortions were excessively tilted against fi rms ficient or inefficient fi nancial intermediation and protecting only a minority of formal may hurt productivity and innovation (see workers. Some have argued that the regula- de la Torre, Ize, and Schmukler 2012). This tory environment was so onerous for private chapter focuses on early-stage fi nancing, an sector firms that ultimately workers and con- aspect of financial intermediation supposedly sumers were negatively affected rather than designed for young, innovative fi rms. Using protected (de Soto 1989). Over the past 20 a new database that surveys the region’s pri- years, however, the region made regulatory vate equity deals, it shows that most deals are reforms that improved the business envi- large and involve mature firms. Venture capi- ronment substantially. It also substantially tal targets large fi rms in traditional sectors, expanded social assistance for the poor, and angel investors are missing in action. ENTREPRENEURSHIP BY INCUMBENT FIRMS: WHAT EXPLAINS THE INNOVATION GAP? 63 Enterprise Survey data indicate that for both for policy action, which would one choose? young and old fi rms, the region is probably Overall, the competition and skills fronts seem not lagging other emerging markets in terms most important in LAC. Greater competition of firms’ access to credit, however. generates pressure to innovate, but without Regarding the human capital aspect of the human capital to do so, the momentum entrepreneurship, often referred to as “entre- will probably be lost. There is therefore a preneurial skills,” entrepreneurs’ technical need to produce skilled managers and tech- and managerial background has been found nicians who, by promoting innovation, may to increase fi rms’ chances of success (Viva- increase competition. By acting on these two relli 2012). Historical studies also find a fronts, governments may be able to sustain a positive association between densities of engi- virtuous cycle of innovation and competition. neers at the beginning of the 20th century These issues are revisited in chapter 6. and long-term economic growth (Maloney Finally, we would like to make a pitch and Valencia- Caicedo 2012). Our analysis for more rigorous evaluations of entrepre- suggests that the region is still lagging along neurship and innovation programs. These this dimension. Although cross-country com- programs are plentiful in the region, as parisons remain a challenge in this field, most comprehensively reviewed by recent studies LAC countries have smaller percentages of by the Inter-American Development Bank science and engineering graduates than do (Pagés-Serra 2010) and the Corporación countries in Eastern Europe or East Asia. To Andina de Fomento (CAF 2013). Despite some extent, many Latin American countries the substantial resources invested in these remain “rent-seeking societies,” in the words programs, there is a sense that few deliver of Murphy, Shleifer, and Vishny (1991). economic benefits. This failure can be a The region also appears to have surpris- consequence of an environment that may be ingly few well-managed modern firms, as hindering innovation, but it may also reflect reflected by the poor quality of management poor program design. In addition to working practices in relatively large formal sector on the broader constraints reviewed in this firms. The correlation between how good chapter, there is a need to better understand managers are, and how good they think they which programs work and which do not. are, is also very weak. To the extent that Few rigorous evaluations of entrepreneur- management practices can be taught— and ship and innovation programs have been evidence suggests that they can— these fi nd- conducted in the region. Only by improving ings leave room for public action. the understanding of which programs are The chapter ends by discussing an aspect effective through more rigorous evaluations of innovation policies that has always fasci- will it be possible to act effectively on both nated policy makers: the (alleged) ability to the micro- and macroeconomic fronts. foster self-sustaining innovation clusters by exploiting geographic spillovers. Geographic spillovers do indeed substantially affect What drives innovation? firms’ incentives and ability to innovate. A conceptual framework However, it is extremely difficult to generate Multiple factors drive innovation, often inter- self-sustaining innovation clusters ex novo. In acting in a complex manner. The fictional experimenting with these risky (and costly) story of Javier, the winemaker introduced in strategies, it is therefore extremely important chapter 1, can shed some light on the process. to build on natural advantages, partner with His story illustrates a conceptual framework the private sector, and ensure that exit strat- about the difficult choices faced by entrepre- egies are well defined to avoid subsidizing neurs around the world— namely, whether failed attempts. and how much to invest in innovation. All these factors, and many others, affect Javier’s dreams have no boundaries. innovation. But if one were to pick priorities He wants to take advantage of Mendoza’s 64 LATIN AMERICAN ENTREPRENEURS high- altitude Malbec grape varietal and they are the main source of innovations that export his product to the ever-growing wine improve existing products (Bartelsman and market in New York. He knows, however, Doms 2000; Foster, Haltiwanger, and Krizan that in order to indulge the refi ned palates 2001; Akcigit and Kerr 2010). of New York’s wine connoisseurs he needs The fi rst lesson from Javier’s story is that to improve the quality of the wine from his there are two parts of the decision to inno- vineyards. His fi rst wines were good for the vate: whether to invest in innovation (the price, but many competitors produce similar extensive margin of innovation) and, if so, wines. If he wants to conquer the competitive how much (the intensive margin). Javier can U.S. market, he knows he needs to go one continue to produce at the local scale or he step further. can improve his production processes in After talking to some of his friends in the order to export. Panel a of figure 3.1 analyzes wine business, Javier realizes that achiev- this type of decision. The entrepreneur com- ing his goal of producing a top-quality wine pares preinnovation profits with the value of requires investments. The first is a fixed innovating— expected profits less any initial investment that will be spent regardless of fi xed costs of engaging in innovation activi- whether the business plan works. Javier ties. Any factor influencing either preinnova- needs to conduct a detailed analysis of the tion profits or the value of innovation may vineyard’s soil, the genetic characteristics of affect an entrepreneur’s decisions to innovate. the fruit, and the blending process in order For instance, if the United States imposed to have a better sense of the scope for quality import quotas on Argentine wines, the value improvements. Before producing a single bot- of innovation would drop, and Javier might tle of his longed-for wine, Javier has to invest choose not to innovate. a significant sum of money. The second aspect of the innovation deci- Once he determines the feasibility of pro- sion—how much to spend on innovation— ducing a higher- quality wine, his next step captures the amount of resources spent by is to hire an experienced winemaker. Javier a firm to improve its internal processes or understands that hiring an internationally products. The amount of the investment is renowned expert would increase his chances likely associated with the probability of suc- of reaching his goal but would also signifi- cess. Panel b of figure 3.1 summarizes this cantly increase his costs. And there is a risk choice. It illustrates that the entrepreneur has that he fails to achieve the desired quality incentives to invest in innovation up to the of wine, which would mean not only that point at which a small additional investment the wine will not be good enough to export in innovation (that is, the marginal cost) but that the local market may also be lost, equals the small additional gains that result because of the increased costs of production. from it (that is, the marginal benefit). Any Failure would also mean that Javier loses his factor affecting the marginal costs or benefits sunk-costs investment without a payoff. from innovation may change fi rms’ innova- After evaluating these options, Javier pon- tion investments. ders his options. Should he settle for the quiet Thus far we have discussed the forces driv- life of a small local wine producer, or should ing firms’ innovation decisions without delv- he incur the costs and run the risks required ing into the factors affecting these forces. to pursue his dreams of becoming an interna- What affects fi rms’ profitability? How does tional winemaker? a country’s legal framework affect firms’ Javier’s story illustrates the decision to incentives to innovate? Many factors affect innovate of almost any incumbent firm. Such the joint choices of whether to invest in inno- firms are a major force behind the growth vation and how much to invest. The rest of and innovation processes. In the United this chapter focuses on a few of them, such as States, for instance, incumbent establish- regulation, competition, or access to finance. ments account for almost 75 percent of aver- Box 3.1 discusses others like risk, laws, or age total factor productivity growth, and macroeconomic stability. ENTREPRENEURSHIP BY INCUMBENT FIRMS: WHAT EXPLAINS THE INNOVATION GAP? 65 FIGURE 3.1 Extensive and intensive margins of innovation a. Extensive margin: Whether to innovate b. Intensive margin: Whether to innovate Preinnovation profits Optimal innovation intensity Marginal costs and benefits Equality of preinnovation profits and value of Marginal costs Do not innovating of innovating innovate Profits Value of innovating: Marginal benefits Innovate Expected profits minus of innovating any fixed costs of innovating Preinnovation profits Investment in innovation Source: Adapted from Pienknagura 2013. BOX 3.1 Risk, laws, macroeconomics, and the innovation gap in Latin America and the Caribbean Our framework appears to be silent on the role of the literature. One of the few theoretical papers to risk as an important constraint for innovative entre- argue that badly designed bankruptcy laws could preneurship, because it assumes that fi rms are risk hurt the intensive margin of innovation is Manso neutral. However, uncertainty does play a role; it is (2011). Bankruptcy laws can have two opposing hidden behind expected profits, which depend on effects on the intensive margin. On the one hand, the probability of being successful in the R&D pro- lenient bankruptcy laws that limit the losses of a cess. To the extent that fi rms and entrepreneurs fear failed innovation could decrease incentives to invest losses more than they value gains, risk can play an in innovation, as the “punishment” factor is reduced even larger role in driving the decision to innovate, (a downward shift of the marginal benefits curve in as it reduces the value of innovation directly. figure 3.1). On the other hand, limited liability may Risk is so embedded in everyday business prac- reduce the marginal cost of innovation, providing tices that it is difficult to measure the full extent to fi rms with an incentive to increase investments (this which it affects innovation (chapter 6 briefly exam- could be the case, for instance, if fi rms take loans ines the risk of contracts being broken and the role to fi nance innovation, which will be repaid only if of intellectual property rights). A few studies look- they are successful). The net effect therefore remains ing at the impacts of bankruptcy laws on innova- ambiguous a priori, which emphasizes the need to tion provide a glimpse into the ways in which risk conduct empirical analyses. can affect innovation. These studies tend to point The contractual environment in which entrepre- toward a negative impact of harsh bankruptcy laws neurs operate also affects the incentives to invest in on the extensive margin of innovation (Armour and innovation. Intellectual property rights, for instance, Cumming 2008; Acharya and Subramanian 2009; increase the expected profits from innovation invest- Primo and Green 2011). Lenient bankruptcy laws ments. Risk caused by unexpected macroeconomic are associated with lower penalties if investments in fluctuations can also hamper private sector invest- innovation fail to bear fruit; they therefore reduce ment (Servén 1998). LAC, however, enjoyed a period the expected fi xed costs of innovation under the sce- of relative macroeconomic calm during the years nario in which the profits associated with an innova- covered by the analyses in this chapter. It is therefore tion do not materialize. unlikely that the innovation gap documented in this The relation between bankruptcy laws and inno- chapter reflects macroeconomic volatility. vation intensity has received much less attention in 66 LATIN AMERICAN ENTREPRENEURS Regulations Several studies have challenged this idea, providing evidence of a positive correla- Chapter 2 documented how poorly designed tion between innovation and competition regulations may affect the entry of private (Nickell 1996; Blundell, Griffith, and van sector firms into the marketplace. Regula- Reenen 1999). The logic is that innovation tions can also affect both the intensive and may serve as a vehicle for escaping competi- extensive margins of innovation investments. tion by providing the innovator with an edge Regulations that increase the fi xed costs of over competitors (Aghion, Harris, and oth- innovation, for instance, reduce the value of ers 2001; Aghion, Bloom, and others 2005). innovation, limiting the number of innova- In the example of Javier, an increase in the tors and an economy’s aggregate investments number of wine producers serving Mendoza’s in innovation. Similarly, regulations that local market would reduce preinnovation increase the cost of either capital or labor used profits, possibly putting pressure on Javier to for R&D, such as laboratory equipment or improve the quality of his wine in order to researchers, affect the marginal costs of inno- export to New York or compete in a higher- vation and thus the size of the optimal invest- quality niche market in Argentina. ment in innovation. This is not to say that These two views represent the extreme all regulations are bad; regulation is needed cases. Reality is probably more complex. to solve market failures, prevent unfair prac- Whether competition is good or bad for inno- tices, and protect workers and consumers. But vation needs to be assessed empirically, as regulations may have impacts on innovation done below. that need to be taken into account. How regulation is implemented and enforced also matters.1 Risk- averse firms Access to finance may be better off knowing the rules with The empirical literature has uncovered a certainty— even if they are costly—than fac- strong association between financial develop- ing uncertainty. Higher uncertainty in the ment and growth (see King and Levine 1993; returns from innovation reduces the value Beck, Levine, and Loayza 2000; and Rajan of innovation, discouraging some firms from and Zingales 1998, to mention a few studies). innovating. From a theoretical standpoint, this positive correlation is consistent with the role played by financial intermediation in allowing firms Competition to invest in growth-enhancing activities such Innovation decisions are tightly linked to as R&D (Aghion, Angeletos, and others profits, which depend crucially on the level of 2010). competition. The link between competition Financial development can positively and innovation (or more generally produc- affect innovation by decreasing the fixed tivity) has been central in the policy debate and variable costs of innovation by reduc- in developing countries for many decades. ing financing costs. Financial development There are two opposing views regarding can thus have a positive effect on innovation. this relation. Less competition can provide In particular, deepening fi nancial develop- incentives for entrepreneurs to invest in inno- ment may have large effects on innovation if vation. This view has led many to argue in it expands available credit to firms that were favor of stronger patent protection as a way previously financially constrained. to boost incentives to innovate (see, for exam- ple, Romer 1990; Aghion and Bolton 1992). Entrepreneurial skills Such protection is equivalent to an increase in the value of innovation relative to the value Entrepreneurial skills can affect both the of the status quo, which leads more fi rms to number of innovations and the intensity of innovate. investment in innovation. To the extent that ENTREPRENEURSHIP BY INCUMBENT FIRMS: WHAT EXPLAINS THE INNOVATION GAP? 67 entrepreneurial skills increase the ability to FIGURE 3.2 Percentage of firms in selected introduce new products or improve existing countries that introduced a new product in the technologies, they raise the value of innova- past year tion and thus the likelihood of engaging in innovation activities. St. Lucia Dominica Jamaica Antigua and Barbuda Agglomeration and spatial spillovers Nicaragua Venezuela, RB It is theoretically plausible that fi rms could Mexico upgrade their product mix and manage- Guyana Trinidad and Tobago ment practices without investing much of St. Kitts and Nevis their own resources in innovation. Such Ecuador St. Vincent and the Grenadines an outcome could be possible if there are Malaysia strong knowledge spillovers across firms, Dominican Republic El Salvador both within countries (through, for instance, Uzbekistan agglomeration effects) and across countries Uruguay Guatemala (through international knowledge spillovers). Romania The last section of this chapter briefly studies Spain Bolivia the potential of exploiting both domestic and Honduras Costa Rica international knowledge spillovers. Chile Paraguay Colombia How innovative are firms in Latin Greece Peru America and the Caribbean? Turkey Argentina LAC underwent substantial regulatory Suriname Bulgaria reforms in the past decade. These reforms Korea, Rep. improved firms’ investment and employ- Azerbaijan Grenada ment decisions. As the next chapters docu- Kazakhstan ment, the most successful fi rms managed to Germany Croatia grow beyond their national boundaries to Ireland compete on the world scene. Nevertheless, Ukraine Hungary although the success of companies such as Georgia Vale, Embraer, and CEMEX has been her- Poland Moldova alded, going beyond these top performers the Slovak Republic picture is more nuanced. In fact, even some Macedonia, FYR Russian Federation of these LAC giants may be underperforming Thailand relative to their peers (see chapters 4 and 5). Serbia Czech Republic Many formal firms in the region are engaged Albania in some form of innovation, but in many Latvia Armenia cases the intensity and type of innovation Bosnia and Herzegovina may not increase much productivity. Slovenia Lithuania Figure 3.2 shows the percentage of fi rms Belarus that developed or introduced a product that 0 20 40 60 80 100 is new to the market (product innovation). Percent With a few exceptions, Latin American firms LAC countries Other countries tend to engage less in innovation than fi rms in other parts of the world. On average, firms Source: World Bank, based on data from Seker 2013 and 2006–10 Enter- prise Surveys. in the region are 20 percent less likely to have Note: LAC = Latin America and the Caribbean. introduced a new product than their coun- terparts in Eastern Europe and Central Asia 68 LATIN AMERICAN ENTREPRENEURS (ECA) or in high-income countries.2 And the FIGURE 3.3 Average investment in research and picture is even grimmer in the Caribbean, development, by region and level of GDP, 2008–10 where the likelihood of introducing a new product is half that observed in ECA or high- a. R&D by region income countries. Other LAC These raw numbers indicate the percent- age of fi rms involved in innovation activities ECA each year; they are uninformative about the quality and intensity of innovation, two fac- LAC5 tors strongly associated with fi rms’ growth and productivity. Datasets exploring these China fundamental factors at the level of the fi rm that are comparable across countries are of High-income countries poor quality, but the few indicators avail- 0 2 4 6 8 10 12 14 16 able suggest that the quality of innovation in R&D/manufacturing value added (%) LAC may be as much of an obstacle to firms’ Business enterprise Government growth and productivity as the quantity. Higher education Private nonprofit Figure 3.3 shows aggregate investment in Foreign R&D. Panel a compares average R&D as a percentage of value added in manufactur- b. R&D by level of GDP ing (the sector where most R&D takes place) 5 across regions. Panel b benchmarks R&D in Israel LAC against the average of countries at simi- 4 lar stages of development.3 On average, R&D Finland investment in the five largest Latin American Korea, R&D (% of GDP) 3 Rep. economies other than República Bolivariana de Venezuela is two-thirds the level of China 2 when expressed as a percentage of manu- China facturing value added, and one-third when Costa Brazil expressed as a percentage of gross domestic 1 Rica India Uruguay product (GDP).4 For the remaining LAC coun- Colombia Argentina Mexico tries, R&D investment is about one-third the 0 Guatemala Chinese level when expressed as a percentage 100 1,000 10,000 100,000 GDP per capita (PPP) of manufacturing value added and one-tenth when expressed as a percentage of GDP. Sources: Panel a: World Bank, based on data from World Development A second feature that distinguishes LAC Indicators and UIS. Panel b: Adapted from Lederman and Maloney 2003. from China and high-income countries is the Note: For countries and economies included in each group, see note 2. GDP = gross domestic product. LAC = Latin America and the Caribbean. preponderant role played by the public sec- PPP = purchasing power parity. R&D = research and development. tor in R&D (Pagés- Serra 2010). 5 (This fea- ture is also observed in Eastern Europe.) This is not to say that the public sector invests that economies that experienced periods of excessively in R&D: as a percentage of GDP, sustained growth often had levels of R&D public sector R&D is lower than in China or investments well above their peers. The low high-income countries. Instead, it reflects the levels of R&D in LAC, and the fact that little low level of private investment in innovation. R&D is conducted by the private sector, may Many factors influence the extent to which represent a drag on productivity and growth lower levels of R&D are likely to trans- in the medium term. late into lower productivity and economic A similar picture emerges by looking growth. But panel b of figure 3.3 shows at patents granted by the U.S. Patent and ENTREPRENEURSHIP BY INCUMBENT FIRMS: WHAT EXPLAINS THE INNOVATION GAP? 69 Trademark Office (USPTO). Figure 3.4 FIGURE 3.4 Number of patents per capita granted by U.S. Patent shows the number of patents per million and Trademark Office, actual and benchmarked, by inventor’s people that inventors from each country country or place of residence received from the USPTO between 2006 and Uzbekistan LAC countries 2010. It also displays the results of a multi- Haiti Other countries Bolivia variate regression analysis that shows where Paraguay or economies Albania each country stands with respect to countries El Salvador Benchmark Indonesia with similar levels of GDP, population, and Honduras exports to the United States.6 Both compari- Kazakhstan Bosnia and Herzegovina sons suggest that patenting activity of most Guatemala Peru LAC countries remains low: no LAC country Azerbaijan Colombia exhibits a level of patenting that equals that Ecuador Macedonia, FYR of a high-income country, and most coun- Dominican Republic Philippines tries have lower levels than their peers. Bra- Ukraine Turkey zil, for instance, registered just 5 patents per Venezuela, RB Serbia million people between 2006 and 2010, half Belarus Georgia the number per capita of China (10) and only Oman slightly less than a quarter the number per India Thailand capita of Bulgaria (22). Armenia Mexico Part of these differences can be explained Jamaica Brazil by lower levels of income per capita and Uruguay United Arab Emirates lower intensities of exports to the United Argentina Saudi Arabia States (which implies fewer incentives to Latvia Poland apply for patents with the USPTO). As the Chile Russian Federation benchmarking exercise shows, however, China these structural factors do not fully account Trinidad and Tobago Portugal for the low patenting intensity of LAC firms. Slovak Republic Lithuania With very few exceptions, patenting intensity St. Kitts and Nevis Greece in most LAC countries falls well below the Croatia Costa Rica benchmark numbers. For example, given its Bulgaria Kuwait GDP, population, and level of exports to the Antigua and Barbuda Czech Republic United States, Brazil is expected to register Malaysia Hungary 1.5 times as many patents as were granted Spain during 2006–10. For many countries in the Slovenia Italy region, the difference is even larger. New Zealand Ireland R&D and patenting are indicators of the France Norway intensity and quality of innovation, but they Belgium United Kingdom indicate only indirectly how firms perform in Australia Austria terms of process innovation. Until recently, Hong Kong SAR, China Singapore comparable data across countries were Netherlands Denmark scarce, which is unfortunate given the strong Canada link between process innovation and pro- Germany Sweden ductivity. In 2007, however, Bloom and van Finland Korea, Rep. Reenen published a methodology that has Switzerland Israel since been applied to a large number of devel- Japan oped and developing countries. It assesses 1 10 100 1,000 10,000 the quality of management practices, which Patents per 1 million people are both an input and an outcome of process Source: World Bank, based on data from USPTO 2012 and World Development Indicators. innovation (box 3.2). Note: Dots represent predictions from a multivariate regression analysis that includes the log of patents per million people on the log of gross domestic product (GDP) adjusted for purchasing power parity, the log of population, and the log of merchandise exports to the United States. They indicate where each country stands with respect to countries with similar levels of GDP, population, and merchandise exports to the United States. The regression used all countries and economies for which data were available; the figure presents only comparator countries. Data are averages for 2006–10. LAC = Latin America and the Caribbean. 70 LATIN AMERICAN ENTREPRENEURS BOX 3.2 Management matters: How better practices could increase productivity in Latin America and the Caribbean About half of per capita income and productivity Until recently, few studies focused on low- or differences across countries cannot be explained middle-income countries. One reason for the dearth by the accumulation of factors of production such of work on management as a development issue as labor and capital. How effectively these factors is the absence of comparable cross- country data. are combined accounts in part for the gap: the uti- Bloom and van Reenen (2007) surveyed manufac- lization and combination of factors of production turing firms about management practices in the requires a particular type of organizational capital United States and Europe, a methodology that has management quality— something the literature has been extended to four Latin American countries: overlooked until recently. Argentina, Brazil, Chile, and Mexico. Historically, the literature stressed the impor- As expected, management quality appears sig- tance of management practices. Chandler (1990) nifi cantly correlated with (average) labor produc- and Lazonic (1990), for instance, argue that differ- tivity across countries (figure B3.2.1). But even ences in management and organizational practices within countries, management scores are correlated account for the United States overtaking the United with fi rm-level productivity, growth, and survival. Kingdom by the turn of the 20th century. Wom- Although causality cannot be confidently assigned, ack, Jones, and Roos (1990) see the organization of the extreme heterogeneity of management quality Japanese fi rms as critical to their growth miracle. suggests that significant gains in efficiency could Bertrand and Schoar (2003) focus on the impact of result from increasing managerial quality, a dimen- changing chief executive officers and chief fi nancial sion of human capital formation and fi rm behavior officers in very large publicly traded U.S. fi rms (see that should not be ignored. also Bloom and van Reenen 2007). FIGURE B3.2.1 Correlation between management quality and productivity in selected countries and economies 6 Sweden Republic of Ireland Japan Italy Germany Northern Ireland France Greece United Kingdom United States Portugal 5 Canada Log of sales per employee Argentina Chile Poland Australia Brazil 4 India Mexico China 3 New Zealand 2 2.6 2.8 3 3.2 3.4 Overall management score LAC countries Other countries and economies Source: Maloney and Sarrias 2012, based on Bloom and van Reenen 2007. Note: Samples are drawn from manufacturing firms with 100–5,000 employees. LAC = Latin America and the Caribbean. ENTREPRENEURSHIP BY INCUMBENT FIRMS: WHAT EXPLAINS THE INNOVATION GAP? 71 Figure 3.5 compares management prac- which management surveys were conducted). tices of manufacturing firms across dif- With the exception of Mexico, Latin Ameri- ferent dimensions in LAC, high- income can countries score toward the bottom of the countries, China, and India (countries in distribution: management practices remain FIGURE 3.5 Management practices in selected countries or economies a. Overall management practices b. Operation management c. Performance monitoring United States United States Sweden Japan Germany United States Germany Sweden Germany Sweden Japan Canada Canada Canada Japan United Kingdom Australia France Italy New Zealand United Kingdom France Italy Mexico Australia France Italy Mexico United Kingdom Australia Poland Greece Portugal Northern Ireland Argentina Poland New Zealand Portugal New Zealand Portugal Mexico Argentina Republic of Ireland Northern Ireland Chile Argentina Chile Brazil Chile Republic of Ireland Northern Ireland Greece China Republic of Ireland China Poland Greece Brazil Brazil China India India India 2.0 2.5 3.0 3.5 2.0 2.5 3.0 3.5 2.0 2.5 3.0 3.5 Mean score Mean score Mean score d. Target management e. Talent management Japan United States Germany Canada United States Japan Sweden Germany Italy United Kingdom Canada Poland France Sweden Australia Northern Ireland United Kingdom Italy Poland Mexico Mexico Australia New Zealand Republic of Ireland Northern Ireland China Portugal France Republic of Ireland Chile Argentina New Zealand Brazil Greece Greece Argentina China Portugal Chile India India Brazil 2.0 2.5 3.0 3.5 2.0 2.5 3.0 3.5 LAC countries Other countries Mean score Mean score or economies Source: Maloney and Sarrias 2012. Note: Samples are drawn from manufacturing firms with 100–5,000 employees. LAC = Latin America and the Caribbean. 72 LATIN AMERICAN ENTREPRENEURS closer to the practices of Chinese and Indian FIGURE 3.6 Distribution of overall management firms than to firms in Germany, Japan, or scores in Brazil and the United States the United States. Given that Latin American fi rms have higher labor costs than fi rms in 0.6 China and India, management practices may pose a more severe constraint for labor pro- ductivity for them. 0.4 Maloney and Sarrias (2012) investigate Density the factors associated with the poor manage- ment practices of Latin American firms. Low average scores could be driven by a long, fat 0.2 tail of underperforming fi rms that conceals good practices by top firms, but this does not appear to be the case. Figure 3.6 presents 0 the distribution of management scores for 1 2 3 4 5 Brazilian and U.S. firms. It shows that the Management score whole distribution of Brazilian firms tends to Brazil United States underperform the distribution of their U.S. counterparts at all levels; very few Brazilian Source: Maloney and Sarrias 2012. Note: Samples are drawn from manufacturing firms with 100–5,000 employees. firms reach the management scores of top U.S. firms. Moreover, the actual distribution of management practices in Latin America is firms (which, on average, tend to be less well likely to be even weaker than the survey data managed than publicly traded companies) is indicate, because the survey covers firms with almost twice as large in Latin America (about 100–5,000 employees, which should be bet- 20 percent) as in the United States (about 10 ter managed than the average firm. percent). Furthermore, a large proportion Part of the “management gap” between of firms in Latin America are run by their the United States and Latin America can founder (up to a third in Brazil). be explained by fi rm characteristics. In the These differences in firm characteristics United States, midsized fi rms have a larger account for a share of the management gap share of employees with a college degree, are but not all of it. A decomposition exercise larger, and are more likely to be multinational following the Machado-Mata (2005) meth- corporations than firms in Latin America odology shows that median firm charac- (table 3.1). The proportion of family-owned teristics can explain at most a third of the TABLE 3.1 Firm characteristics in Latin America, the United States, and China (percent, except where indicated otherwise) Firm characteristic Argentina Brazil Chile Mexico United States China Number of employees 518 581 487 594 1,254 960 Employees with college degree 5 8 12 15 15 4 Output exported 22 14 30 38 17 48 Multinational 36 21 38 39 48 52 Family firms 21 23 19 19 12 5 Run by founder 21 32 14 19 8 29 Share of management gap with respect to United States explained by firm characteristics 28 37 13 3 n.a. 15 Share of management gap left unexplained 72 63 87 97 n.a. 85 Source: Maloney and Sarrias 2012. Note: Samples are drawn from manufacturing firms with 100–5,000 employees. n.a. = Not applicable. ENTREPRENEURSHIP BY INCUMBENT FIRMS: WHAT EXPLAINS THE INNOVATION GAP? 73 management gap between median firms in development, entrepreneurial skills, and Latin America and the United States (bottom agglomerations/externalities have repeatedly two rows of table 3.1). Moreover, in Mexico, been shown to affect innovation. LAC is not where firm characteristics and the quality of likely to boost innovation without addressing management are closer to the United States’, these factors. characteristics of the median firm do not appear to account at all for the management Regulation still matters— gap. To be sure, the survey probably missed but less than in past decades important firm characteristics associated Early analyses— such as the work of de with the quality of management, increasing Soto (1989), Rauch (1991), Loayza (1996), the unexplained component of the manage- and Johnson, Kaufmann, and Shleifer ment gap. But given the results, it is unlikely (1997)— spurred an important stream of that firm characteristics fully account for dif- research that aimed to understand and ferences in the quality of management. document the many ways in which poorly designed regulations affect entrepreneurial What explains the innovation gap? activity. Overall, most of these studies fi nd Many factors affect innovation, both directly a positive association between the quality of and through their interaction with one regulation, innovation, and economic growth another. There is probably no universal recipe (examples include Bassanini and Ernst 2002 for enhancing innovation, just as there is no and Djankov, McLiesh, and Ramalho 2006). single recipe for growth. But certain factors Figure 3.7 shows the association between including regulations, competition, financial the strength of investor protection and the FIGURE 3.7 Relationship between investor protection and time required to register property and innovation in Latin America and the Caribbean a. Protecting investors: Strength b. Registering property of investor protection index (1–10) (time in days) New or significantly New or significantly improved product improved product New or significantly New or significantly improved process improved process Invested in R&D Invested in R&D Cooperates on Cooperates on innovation with others innovation with others Adopted foreign Adopted foreign technology technology Filed for a patent Filed for a patent –0.02 0 0.02 0.04 –0.06 –0.04 –0.02 0 Additional likelihood to innovate Additional likelihood to innovate Significant at 10% Not significant Source: World Bank, based on data from 2006–10 Enterprise Surveys. Note: Bars show the impact on the dependent variable of one standard deviation of the explanatory variable. The regression includes other country-level Doing Business indicators; sectoral concentration (Herfindahl) indexes; the number of competitors; size; age; the number of establishments; legal organization; the manager’s years of experience; the percentage of workers with complete university education; and whether the firm was registered at start-up, has a foreign owner, is an exporter, and has taken a loan. Robust standard errors are clustered at the country level. Countries include Argentina, Chile, Colombia, Costa Rica, the Dominican Republic, Ecuador, El Salvador, Guatemala, Honduras, Jamaica, Mexico, Nicaragua, Paraguay, Peru, Trinidad and Tobago, and Uruguay. R&D = research and development. 74 LATIN AMERICAN ENTREPRENEURS time required to register a property (from the aspect of the regulatory environment seems World Bank’s Doing Business indicators) on particularly weak in Caribbean countries, the one hand, and the propensity of firms to where the number of days required to regis- engage in innovative activities on the other. ter property is four times the number in East The underlying econometric analysis controls Asia. Caribbean countries also score poorly for fi rms’ characteristics, other Doing Busi- on the cost of registering property, which ness indicators, the number of competitors, averages almost 10 percent of the property and sectoral concentration. Even after con- value, against an average of less than 4 per- trolling for competition, regulation appears cent for other LAC countries. Interestingly, to matter: a one standard deviation improve- as in the rest of the world except some East ment in either investor protection or the Asian economies, export costs (imposed by time required to register property is associ- government regulations, excluding trade ated with 2– 4 percent more fi rms engaging taxes) have soared in LAC and remain second in innovative activities, such as cooperating only to Eastern European countries.7 on innovation with others, adopting foreign These numbers should be taken with a technologies, filing for patents, and investing grain of salt, as they focus on very selective in R&D. aspects of the regulatory environment. But The significance of these correlations they suggest that there is room for further should be interpreted with caution. Some improvement of the regulatory environment, could stem from unobserved characteristics which could stimulate firms’ propensity to of the regulatory environment that are also innovate. How much they would boost inno- correlated with Doing Business indicators. vation remains an open question, however: But taken as a whole, they remain sugges- thanks to recent waves of regulatory reforms tive of the role that the quality of regulation in the region, regulation may no longer be the plays in promoting or preventing innovation. main bottleneck to innovation. As discussed Although the magnitudes of the association below, new challenges are emerging. may at first sight seem small, one should not forget that the regressions consider only Competition: An unfinished agenda variation within LAC, a fairly homogeneous group of countries in which innovation is rel- The ability to foster innovation through com- atively low to start with. petition is of particular relevance for low- and The good news is that, just as in the middle-income countries, as governments may case of the entry regulations described in fi nd it easier and more effective to level the chapter 2, many LAC countries have made playing field than to use more intervention- substantial regulatory reforms. However, ist policies with strong governance challenges because these reforms are part of a global (Allen and Gale 2000; Ayyagari, Demirgüc- deregulation process taking place in a broad Kunt, and Maksimovic 2011). However, the international context, LAC countries still lag empirical association between competition, their peers along important dimensions. In productivity, and innovation is complex. addition, although the largest countries in Although there is little doubt that in many cir- the region (Argentina, Brazil, Chile, Colom- cumstances competition can have a positive bia, and Mexico [LAC5]) have for the most impact on growth and innovation (Blundell, part improved investor protection, many of Griffith, and van Reenen 1995; Nickell 1996; the smaller countries have not made much Galdon- Sanchez and Schmitz 2002; Ayya- progress (figure 3.8). The number of days gari, Demirgüc-Kunt, and Maksimovic 2011), required to register property, for instance, there are many instances in which its impact remains high throughout the region (on aver- may be limited or even negative. age, at the same level as in India), well above In sectors with strong returns to scale, Eastern European and East Asian peers. This for instance, well-regulated monopolies may ENTREPRENEURSHIP BY INCUMBENT FIRMS: WHAT EXPLAINS THE INNOVATION GAP? 75 FIGURE 3.8 Doing Business in selected country groups and countries, circa 2004 versus 2013 a. Investor protection b. Cost to export 10 2,000 Strength of investor protection index 1,800 8 1,600 US$ per container 1,400 6 1,200 1,000 4 800 600 2 400 200 0 0 AC ina n ECA ia 5 mie e P4 P4 ina no ome ia ean AC 5 ECA eco incom LAC LAC bea Ind Ind s s EA EA mie er L er L Ch Ch ibb c ib n Oth Oth no h-i Car Car h- eco Hig Hig c. Registering property d. Costs to register property 140 12 120 10 100 Property value (%) 8 Time (days) 80 6 60 4 40 20 2 0 0 P4 ECA mie e ina AC 5 ia n ECA AC ina Hig LAC5 mie e P4 ia n eco incom eco incom LAC bea bea Ind Ind s s EA EA er L er L Ch Ch ib ib Oth Oth no no Car Car h- h- Hig LAC countries, 2013 LAC countries, 2004 Other country groups, Other country groups, countries, and economies, 2013 countries, and economies, 2004 Source: Authors, based on Doing Business indicators. Note: For countries and economies in each group, see note 2. be the way to go. In some fi nancial services, analysis, low levels of competition lead to a excessive competition can also be harmful, as few oligopolistic fi rms sharing similar pro- it may prevent the building of sound reserves, duction costs and laggard technologies. In raising the overall vulnerability of the system. such an environment, competition is good, And recent studies have uncovered that com- as it drives rents down; firms react to higher petition may hurt innovation (and productiv- competition by “escaping” it through innova- ity) even in more traditional sectors, such as tion (along both the products and the process manufacturing. A seminal study by Aghion, dimensions). In contrast, in highly com- Bloom, and others (2005) suggests that the petitive sectors, a (short-lived) technological relationship between competition and inno- leader captures the market. In these settings, vation may have an inverted U-shape. In their further increases in competition may reduce 76 LATIN AMERICAN ENTREPRENEURS fi rms’ incentives to innovate, as competition innovative plants and minor products and reduces any rent from innovation.8 increased sales of larger plants and main Competition may also generate short-term products. Although the long-term increase distributional effects that have to be taken in productivity may well compensate for the into account. Using fi rm-level data from the short-term losses associated with the exit of United Kingdom, Aghion, Blundell, and oth- the least productive fi rms, any competition ers (2009) fi nd that the intensity of compe- reform should include measures to protect tition, measured by entry into an industry, workers who may suffer from a potential fosters innovation and productivity growth short-term surge in plant closures. among the more technologically advanced These caveats notwithstanding, LAC may incumbents but slows it among less efficient not have reached the tipping point at which incumbents. Iacovone, Rauch, and Win- an increase in competition hurts innovation. ters (2013) fi nd similar results for Mexico, On the contrary, lack of competition may where a surge in Chinese exports from 1994 inhibit firms’ incentives to innovate. to 2004 reduced sales of smaller and less Figure 3.9 shows the association between various dimensions of innovation and the number of competitors. The regressions con- FIGURE 3.9 Relationship between competition and various trol for other fi rm-level characteristics and aspects of innovation in Latin America and the Caribbean country- level Doing Business indicators. Although there seems to be little association between competition and the generic ques- New or significantly tion about whether firms have developed new improved product products or processes, competition seems to New or significantly be associated with the quality of innovation improved process activities that firms claim to conduct: a more competitive environment is associated with a Invested in R&D higher likelihood of firms collaborating with others on innovation, adopting foreign tech- nologies, and filing patent applications.9 Cooperates on The magnitude of the correlations is at innovation with others times substantial: all else being equal, having Adopted foreign more than one competitor is associated with a technology 4 percent increase in the likelihood of cooper- ating on innovation and an 8 percent increase Filed for a patent in the likelihood of applying for a patent. To be sure, unobserved factors (such as higher 0 0.04 0.08 0.12 0.16 0.20 profit opportunities driving both entry and Additional likelihood to innovate innovation) may be behind these associations. 2–5 competitors, >5 competitors, But these fi ndings and the causal impact of significant at 10% significant at 10% competition found in most of the literature 2–5 competitors, >5 competitors, suggest that in many sectors, gains from not significant at 10% not significant at 10% increased competition could be substantial. How could LAC governments foster a Source: World Bank, based on data from 2006–10 Enterprise Surveys. more competitive environment? Until now, Note: The regression includes other country-level Doing Business indicators; sectoral concentration efforts have focused on improving the regula- (Herfindahl) indexes; number of competitors; size; age; number of establishments; legal organiza- tion; manager’s years of experience; percentage of workers with complete university education; and tory environment. However, many countries whether firm was registered at start-up, has a foreign owner, is an exporter, and has taken a loan. have made substantial progress along the Robust standard errors are clustered at the country level. Countries include Argentina, Chile, Colom- bia, Costa Rica, the Dominican Republic, Ecuador, El Salvador, Guatemala, Honduras, Jamaica, Mexico, regulatory front. Although there may still be Nicaragua, Paraguay, Peru, Trinidad and Tobago, and Uruguay. R&D = research and development. room for improvement, additional regulatory ENTREPRENEURSHIP BY INCUMBENT FIRMS: WHAT EXPLAINS THE INNOVATION GAP? 77 reform is not likely to bring countries in the ministry, and is responsible directly to the region to the competitive frontier. parliament or legislature for its budget is LAC countries are in a situation in which structurally independent. Operational inde- more active aspects of competition policies pendency refers to the freedom to use the start to matter. These initiatives are mul- budget, organize the agency, and carry out tifaceted and fall under the umbrella of the enforcement activities and advocacy func- governance structure and the effectiveness of tions, without having a ministry supersede competition and consumer protection poli- decisions (Clark 2005). cies and authorities. Data collected by the CRC suggest that in Many countries in the region enacted only about half of the surveyed agencies is the competition laws in recent years (Honduras head appointed or cleared by Congress (fig- in 2005, Nicaragua in 2006, the Domini- ure 3.10). Mexico and Brazil have the stron- can Republic in 2008, and Ecuador in 2011) gest checks and balances to avoid political or saw major legislative reforms (Mexico in interference; their agencies are structurally 2011, Brazil in 2012). These countries based and operationally independent, and there is a their legal frameworks on international best low risk that their decisions will be overruled practices. Many of their competition laws by the executive branch. Although in Chile, grant significant power to competition agen- the head of the Fiscalía Nacional Económica cies to carry out investigations and impose is not cleared by Congress, strong check and sanctions.10 Many countries also included balances are in place. The institutional design provisions in their laws shielding competition of the competition agencies in the Domini- agencies from political interferences to guar- can Republic, Ecuador, and Honduras also antee their independence. favors, at least de jure, the independency of The challenges, however, lie in imple- the president of the regulatory agency and its menting these policies and laws. To varying commissioners (who are appointed by Con- extents, and with a few exceptions, countries gress, or jointly by Congress and the govern- in the region are characterized by limited cul- ment, for a relatively long period of time), tures of competition, concentrated markets, and only a court can overrule the agencies’ vested interests, scarce human and economic decisions. In contrast, Argentina, Colombia, resources, lack of cooperation among regula- Costa Rica, and Uruguay may potentially tors, opposition from large corporations, and judiciary systems with little or no experience FIGURE 3.10 Appointment of head of regulatory in competition matters. All these elements agency in Latin America and the Caribbean may reduce the capabilities of the competition agencies to enforce the law (Ortiz 2013). The extent to which they will be successful in pro- One minister moting competition will depend on whether or more they have the means, independence, and pow- Government ers to operate effectively. New data collected by the Centro Regional de Competencia para América Latina (CRC) allow some of these Congress aspects to be explored in greater detail. Government and Congress Independence of competition agencies T he Organisation for E conom ic Co - 0 1 2 3 4 5 6 operation and Development (OECD) consid- Number of agencies ers two aspects of independence: operational Source: Centro Regional de Competencia para América Latina 2013. and structural. An agency that is created as Note: Covered countries are Argentina, Brazil, Chile, Colombia, Costa Rica, the a separate entity, rather than as part of a Dominican Republic, Ecuador, El Salvador, Honduras, Mexico, and Uruguay. 78 LATIN AMERICAN ENTREPRENEURS be more exposed to interference: their legal criminal sanctions, affect their effectiveness. frameworks allow the government to inter- Much heterogeneity is evident along these vene in the agencies by selecting its president two dimensions. Competition agencies in and commissioners (including participating Brazil, Chile, and Mexico appear to be well in their reelection), giving instructions to endowed, operating with budgets ranging the agencies about current investigations, or from $9 million to almost $20 million. At determining the internal organization of the the other end of the spectrum, some agencies, agency. such as the ones in in Costa Rica, Honduras, Another source of vulnerability may and Uruguay, operate with much more lim- stem from the ability of the government to ited budgets.11 interfere in the agencies’ decisions: agencies The scope of action also varies across in only half of the surveyed countries are agencies. Although all agencies are entitled shielded from explicit government interven- to conduct “dawn raids” (on-site investiga- tions (figure 3.11). In the other half, the gov- tions) at the premises of the companies inves- ernment can overrule the agencies’ decisions tigated, only 6 out of 11 did so between 2010 or interfere in an ongoing investigation. and 2013 (figure 3.12). All agencies can also Given their limited experience in enforce- request information from the companies ment, it is too early to assess whether some of investigated, but only 5 out of 11 can record these recently created agencies are not only de or ask the police to record conversations jure but also de facto independent. To date, between employees to collect evidence. Fur- none of the agencies has seen its decision thermore, in some countries, such as Brazil, overruled, and none reports having received Costa Rica, and Mexico, some sectors are instructions from the executive branch. Nev- exempt from competition law enforcement. ertheless, lack of legislation makes agencies more vulnerable to potential interference. FIGURE 3.12 Scope of action of regulatory Budget and scope of action agencies in Latin America and the Caribbean The budget of competition agencies, as well as the legal means they have been given to Sanctions to conduct investigations and impose fi nes and individuals Recording conversations FIGURE 3.11 Level of allowed government Interim intervention in regulatory decisions measures Leniency programs Request No intervention information Overrule the Settlements agency’s decision Dawn raids Instructions in an ongoing investigation Dawn raids (effectively Appointment of head conducted) commissioners 0 2 4 6 8 10 12 Legal actions available 0 1 2 3 4 5 6 7 (from antitrust agencies) Number of agencies Source: Centro Regional de Competencia para América Latina 2013. Source: Centro Regional de Competencia para América Latina 2013. Note: Covered countries are Argentina, Brazil, Chile, Colombia, Costa Rica, Note: Covered countries are Argentina, Brazil, Chile, Colombia, Costa Rica, the Dominican Republic, Ecuador, El Salvador, Honduras, Mexico, and the Dominican Republic, Ecuador, El Salvador, Honduras, Mexico, and Uruguay. Uruguay. Effectively conducted dawn raids cover 2010–13. ENTREPRENEURSHIP BY INCUMBENT FIRMS: WHAT EXPLAINS THE INNOVATION GAP? 79 Transparency mergers. The remaining nine agencies do Good transparency practices include publish- not have guidelines in some of these areas, ing decisions, disclosing the facts and theo- although some are in the process of draft- ries of harm under consideration, providing ing them. Although the absence of guidelines access to evidence, providing opportunities need not reduce transparency, it is desirable to to meet with the agency, responding to con- make available to the public the methodology cerns, and guaranteeing the confidentiality of used by the agency when assessing conduct. third parties. Most agencies display a relatively high Anticompetitive conduct and fines standard of transparency (figure 3.13). All The data on anticompetitive conduct and 11 agencies publish their decisions and the fines are far from exhaustive. The surveys col- underlying legal and economic reasoning, lected information only about the existence of and all 11 provide access to the file once certain anticompetitive conduct and the pos- the decision has been made (sometimes this sibility of investigating and sanctioning it. information is publicly available, but some- All agencies share similar capacities to times it is necessary to formally request it). investigate horizontal agreements, vertical All 11 agencies also release annual reports agreements, abuse of dominant positions, and other types of reports to inform the and mergers. There are, however, important public about their activities, and all of them differences in various dimensions, reflecting guarantee confidentiality when parties sub- the different roles agencies play in each coun- mit information. try. Countries differ in how anticompetitive An area where there is still room for conducts are assessed. Brazil, Chile, Colom- improvement is the adoption of guidelines bia, and Ecuador, for instance, do not con- for how specific anticompetitive practices sider all cartels as anticompetitive per se. are treated. Brazil and Mexico are the only The maximum fines that have been countries that have published guidelines imposed also differ substantially across coun- explaining horizontal agreements, vertical tries (figure 3.14). The toughest agencies are agreements, abuse of dominant position, and FIGURE 3.14 Maximum fines imposed by regulatory agencies in FIGURE 3.13 Transparency practices of selected countries in Latin America and the Caribbean regulatory agencies in Latin America and the Caribbean Ecuador Dominican Publish Republic guidelines Costa Rica Uruguay Oral hearings Colombia Access to files El Salvador Honduras Accounts reviewed by Chile third party Argentina Publish decisions Mexico 0 2 4 6 8 10 12 Brazil Number of agencies 1 10 100 1,000 10,000 100,000 1,000,000 Source: Centro Regional de Competencia para América Latina 2013. US$ (thousands) Note: Covered countries are Argentina, Brazil, Chile, Colombia, Costa Rica, the Dominican Republic, Ecuador, El Salvador, Honduras, Mexico, and Uruguay. Source: Centro Regional de Competencia para América Latina 2013. 80 LATIN AMERICAN ENTREPRENEURS CADE (Brazil), which imposed a fine of $1.1 Colombia and Mexico are the only two billion on White Martins, and CFC (Mex- countries in the region that conduct a “com- ico), which fined Telcel $1 billion. The lowest petition impact assessment” of new pro- maximum fines were imposed by Uruguay posed regulation, implement sector studies, ($10,000) and Costa Rica ($4,600). Agen- and carry out ex post evaluations of their cies in the Dominican Republic and Ecuador activities— and the competition agency in have not yet imposed fines. Of course, not Mexico is the only one entitled to issue bind- all of this heterogeneity can be attributed ing opinions under certain circumstances. In to the agencies’ effectiveness: many of these all countries, the competition agency can issue agencies were established or reformed only (nonbinding) opinions to prevent the adoption recently and are in the process of defining of regulation with negative effects in the mar- their scope and scaling up operations. ket and publish market and sector studies. Training and advocacy need to be custom- Mergers ized to the local context. Instruments used in Good merger evaluation practices should one country (conferences, training courses, include a comprehensive framework to opinions, market studies, media appear- address mergers that are likely to harm com- ances) may not be effective in another. Lim- petition significantly. Factors other than mar- ited economic resources and a still nascent ket share or increase in market share, such as competition culture make training and advo- entry barriers, should also be evaluated. cacy costly and challenging. Nonetheless, it Mergers represent the area with possi- is important to keep investing in this area, bly the greatest divergence across agencies. because it affects the likelihood of enforce- Agencies in four countries (Brazil Colombia, ment of competition policies. Costa Rica, and El Salvador) have to approve all mergers and acquisitions before the par- Does financial underdevelopment ties can close the deal. Agencies in three explain the innovation gap? countries (Ecuador, Honduras, and Mexico) have to approve only some operations before Innovation is a risky activity; if markets fail the parties can proceed. In Chile, there is no to share some of these risks, entrepreneurs need to approve a merger or acquisition. The may fi nd it difficult to innovate. The asso- Dominican Republic and Uruguay do not yet ciation between financial intermediation and have a merger notification system in place. growth and innovation has been documented Disparities can also be found in the analy- extensively in the literature. Early works by sis of mergers. Argentina, Chile, Honduras, King and Levine (1993) and Beck, Levine, and Mexico take into account only criteria and Loayza (2000) find a positive association based on competition grounds (that is, effi- between fi nancial development and growth ciencies) in determining whether to approve a (see Levine 2005 for a review). merger. In contrast, Brazil, Colombia, Ecua- In addition to the depth of financial mar- dor, and El Salvador also take into consider- kets, it also appears that regulations and the ation other issues, such as public interest or type of fi nancial intermediation tools avail- impacts on the labor market. able to fi rms affect economic performance. According to Bekaert, Harvey, and Lundblad Training and advocacy (2005), for example, liberalization of the Training and advocacy should be central equity market led to an average increase in activities of any competition agency. The annual real economic growth of 1 percent. more judges, policy makers, and the pri- Most of these studies suffer from reverse vate sector are trained on the benefits of causality biases, as better-performing econo- good competitive practices and regulation, mies may foster the development of finan- the easier it will be to sustain a competitive cial markets. In an attempt to draw causal environment. relationships, Rajan and Zingales (1998) ENTREPRENEURSHIP BY INCUMBENT FIRMS: WHAT EXPLAINS THE INNOVATION GAP? 81 construct a measure of “financial depen- FIGURE 3.15 Credit, investment, and innovation in Latin America dency” of each sector. They show that indus- and the Caribbean trial sectors that need more external finance grow faster in countries with more developed New or significantly financial markets. improved product Financial development, in terms of both New or significantly the depth and the diversification of financing improved process instruments, appears to exert a dispropor- tionately positive effect on small firms, which Invested in R&D tend to fi nd it more difficult to raise funds (Guiso, Sapienza, and Zingales 2004; Beck Cooperates on and others 2008). Because small firms tend to innovation with others operate locally, local fi nancial development Filed for patent, trademark, has also been found to be an important deter- or copyright minant of the economic success of an area Invested to improve quality control or obtain (Guiso, Sapienza, and Zingales 2004). certification Ayyagari, Demirgüc- Kunt, and Maksi- –0.04 0 0.04 0.08 0.12 0.16 movic (2011) explore the association between Marginal effect financial development and innovation in Loan, significant at 10% emerging markets. They define innovation to Loan, not significant at 10% include the introduction of new products and Purchased fixed assets, significant at 10% technologies, knowledge transfers, and new Purchased fixed assets, not significant at 10% production processes. They find that access to Source: World Bank, based on data from 2006–10 Enterprise Surveys. external fi nancing is associated with greater Note: Robust standard errors are clustered at the country level. As additional controls, the regres- fi rm innovation. Although data constraints sions include firm size, age, legal organization, the number of establishments, whether the firm was registered at start-up, whether it is foreign owned, the percentage of full-time workers with limit the causal interpretation of the associa- university degrees, and country and sector fixed effects. Countries include Antigua and Barbuda, tion, the study highlights an important chan- Argentina, the Bahamas, Barbados, Belize, Chile, Colombia, Costa Rica, Dominica, the Dominican Republic, Ecuador, El Salvador, Grenada, Guatemala, Guyana, Honduras, Jamaica, Mexico, Nicaragua, nel through which financial development can Panama, Paraguay, Peru, St. Kitts and Nevis, St. Lucia, St. Vincent and the Grenadines, Trinidad and affect productivity. Tobago, and Uruguay. R&D = Research and development. These analyses are not specific to LAC, but figure 3.15 shows that the region is no exception. Based on a multivariate regression that controls for other firm-level characteris- a 9 percent higher probability of investing in tics, sector, and country effects, it shows the quality control and certification. association between having taken a loan and Investment in fixed assets also appears various forms of innovation. It also shows the to be an important channel through which association between investment in fixed assets innovation takes place. The fact that the fixed and innovation (controlling for taking a loan), assets indicator remains significant even after as these types of investments are an impor- controlling for borrowing suggests that many tant avenue through which firms innovate. fi rms invest using other means, such as self- The results show that at least one of the financing and (for large firms) equity financ- two variables is significantly associated with ing. For some fi rms, however, these types of any form of innovation captured in Enter- financing may be second-best choices dictated prise Surveys. The magnitudes of these effects by the lack of good financial intermediation. are also relatively large: everything else being LAC financial markets developed substan- equal, having taken a loan is associated with tially in the last two decades, as de la Torre, a 9 percent higher probability of introduc- Ize, and Schmukler (2012) document. Bond ing a new product, a 5 percent higher prob- and equity markets have gained ground, ability of improving processes, an 11 percent institutional investors now play a central role, higher probability of conducting R&D, and new markets and instruments have sprung 82 LATIN AMERICAN ENTREPRENEURS up, maturities have lengthened, and dollar- FIGURE 3.16 Depth of financial systems in selected ization has been reduced. However, many of country groups and countries, 1995 and 2005 these gains are benefitting only larger fi rms; significant gaps remain in the financing of 1995 smaller ones, as manifested by the depth China and efficiency (as measured by interest rate 2005 margins) of banking intermediation and the 1995 EAP4 liquidity of domestic equity markets. 2005 These gaps are of concern because they 1995 coincide with some of the financial indicators Turkey that have been shown to be the best predictors 2005 of future output growth and because, except High-income 1995 for bank margins, there is little evidence of economies 2005 convergence toward benchmark levels con- sistent with the economic development of the 1995 India region and its basic structural characteristics 2005 (figure 3.16). The lack of depth of the banking 1995 sector, an important financing avenue of small LAC5 fi rms, may in particular hurt the innovation 2005 potential of emerging firms. Another area of Other 1995 concern is the limited capacity of institutional LAC 2005 investors to expand their portfolios beyond 0.5 1.0 1.5 2.0 2.5 3.0 3.5 0 the safest and most liquid investments. Percentage of GDP These features are not identical across Banks, LAC countries Banks, other countries countries. There is substantial heterogeneity Bonds, LAC countries and economies in fi nancial development within the region, Equities, LAC countries Bonds, other countries with smaller, lower-income countries gener- and economies ally lagging behind. Equities, other countries and economies The innovation potential of firms in LAC—in particular small ones—may also be Source: Adapted from Didier and Schmukler 2011. hindered by the dearth of private equity and Note: For countries and economies included in each group, see note 2. venture capital financing options. Mondragón (2012) benchmarks the private equity and venture capital (PEVC) industry in LAC rela- were minimal in all emerging economies). It tive to other low- and middle-income regions dramatically picked up in Asia but remains using a new dataset that merges various relatively modest in LAC and Eastern Europe sources of information on PEVC fi nancing. and Central Asia. Although the dataset may miss smaller deals, In addition to being in its infancy, the it is one of the most comprehensive efforts to PEVC industry in the region focuses on large measure PEVC intensity at the regional level. deals, in mature industries, in a few coun- Despite unprecedented growth since tries (figure 3.18). More than 90 percent of the mid-2000s, PEVC investments in LAC activity takes place in Mexico or the major remains relatively low, below the region’s economies of South America, with Bra- share of world GDP and capital inflows (fig- zil accounting for half of reported deals in ure 3.17). In 2011, PEVC investments in LAC 2008–11 and two-thirds of investments. The totaled $3.2 billion. This figure was close larger number of deals in Brazil suggests that to the figure in Eastern Europe and Central market size and liquidity may be important Asia ($3.6 billion) but well below PEVC factors driving the expansion of the PEVC investments in emerging Asia ($18.7 billion). industry; without liquidity and a constant The industry is quite new in middle-income stream of potential deals, it may be difficult countries (before 2005, PEVC investments for the industry to expand. ENTREPRENEURSHIP BY INCUMBENT FIRMS: WHAT EXPLAINS THE INNOVATION GAP? 83 It may thus not come as a surprise that in FIGURE 3.17 Private equity and venture capital investments, the smaller countries, very few PEVC deals by region, 2002–11 have been reported. More surprising is the size of the deals and the type of company the 40 PEVC industry is targeting. Only 20 percent of deals are smaller than $5 million (see fig- 30 ure 3.18), and the average deal is $30 mil- lion. Moreover, companies that benefit from US$ billions PEVC fi nancing do not appear to be young, 20 innovative start-ups: the average company is about 18 years old. After controlling for 10 other company characteristics, each addi- tional year in operation implies $0.9 million of additional PEVC investments. Investments 0 also have a strong bias toward energy and 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 natural resources, which account for about a Emerging Asia Eastern Europe and Central Asia third of PEVC investments. Latin America and the Caribbean Middle East and North Africa Overall, start-up financing remains a chal- Sub-Saharan Africa lenge in the region: venture capital accounts for less than 10 percent of total PEVC invest- Source: Mondragón 2012, based on data from the Emerging Markets Private Equity Association (EMPEA). Note: Emerging Asia: All countries; excludes funds whose primary mandate is investments in Japan, ments. Moreover, venture capital typically Australia, or New Zealand. Eastern Europe and Central Asia: Countries in Central and Eastern Europe, does not finance young, innovative firms. the Commonwealth of Independent States, the Baltics, and the Balkans; the European Union acces- sion countries; and Turkey. Latin America and the Caribbean: countries in Central America, South The dearth of financing opportunities for America, and the Caribbean excluding Puerto Rico; Middle East and North Africa: Afghanistan, Alge- young, innovative firms may be hindering ria, the Arab Republic of Egypt, the Gulf Cooperation Council countries, the Islamic Republic of Iran, Iraq, Libya, Jordan, Lebanon, Morocco, Pakistan, the Palestinian Territories, Sudan, Syria, Tunisia, and the region’s entrepreneurial potential. But the Republic of Yemen. Sub-Saharan Africa: All countries; excludes funds whose primary mandate is the problem may not necessarily be solved by investments in Algeria, Egypt, Libya, Morocco, Sudan, or Tunisia. FIGURE 3.18 Number and size of private equity and venture capital deals in Latin America and the Caribbean, by country, 2008–11 a. Number of deals, 2008–11 b. Deal size distribution Belize 0–1 Cayman Islands Nicaragua Guyana 1–5 Jamaica Panama 5–10 Paraguay Dominican Republic 10–20 US$ (millions) Honduras Trinidad and Tobago 20–50 Guatemala Uruguay 50–100 Costa Rica Peru 100–250 Colombia Chile 250–500 Argentina Mexico >500 Brazil 0 50 100 150 200 250 0 5 10 15 20 25 Deals % of total deals Source: Mondragón 2012. Note: Information on the size of the investment was not available on all deals. 84 LATIN AMERICAN ENTREPRENEURS simple supply-side interventions. The region FIGURE 3.19 Access to credit, by region and age may be trapped in a vicious cycle of low of firm innovation leading to too little demand for a healthy PEVC industry to flourish. It may be 70 possible to break such a cycle with govern- ment-led supply side interventions—but with- 60 out a larger mass of young, innovative firms, supply-only interventions are likely to fail. 50 Firms (%) In addition, the facts that LAC has under- developed capital markets and that venture 40 capital appears to chase big deals in tradi- tional industries do not by themselves imply 30 that the region’s innovation gap is caused by lack of access to fi nance. Young rather than 20 small firms drive growth and employment 1–4 5–9 10–19 20–29 30–39 generation in the long run (see chapter 2). Age of firm (years) Hence, it is worthwhile to explore the link LAC High-income countries between firm age and access to finance. ECA EAP4 Figure 3.19 shows the share of firms across regions and age groups covered by the Source: World Bank, based on data from 2006–10 Enterprise Surveys. Note: The last survey available is used for each country. Each country has Enterprise Survey database that report hav- the same weight in the regional average. Initial size of more than 10,000 ing access to credit. LAC appears not to have was replaced by missing. Data do not take into account the initial year for firms more than 39 years old. High income: Croatia, the Czech Republic, a meaningful gap. Consequently, it seems dif- Hungary, Poland, the Slovak Republic, Slovenia and Spain. For countries ficult to conclude that lack of fi nance is the included in other groups, see note 2. main explanation for the region’s innovation gap. Much remains to be done to continue particularly for high-tech start-ups, where transforming the region’s capital markets the founder’s human capital is a key driver of into engines of growth without hampering growth (Colombo and Grilli 2010; Arvanitis recently achieved stability. Yet lack of finance and Stucki 2012). is unlikely to be the main driver of the inno- Education in technical or scientific fields vation gap. Chapter 6 revisits this issue. and work experience in technical and com- mercial functions within the same industry also matter for success, especially for new Entrepreneurial skills: technology– based firms (Almus and Ner- A key missing link? linger 1999; Colombo and Grilli 2005; Bal- Entrepreneurial skills are personal traits, coni and Fontana 2011; Ganotakis 2012). experience, and human capital that favor Entrepreneurs who are well endowed in a experimentation, risk taking, and ultimately variety of fields—“jacks-of-all trades”—may the growth of incumbent firms. For the pur- also have higher probabilities of success, poses of this section, entrepreneurs include because entrepreneurs have to manage dif- all individuals in a position to make impor- ferent people and tasks and be well versed tant strategic decisions in a fi rm, including in a variety of management skills (Lazear managers, chief executive officers, founders, 2004, 2005). Using cross- section analyses, and engineers. Lazear (2005) and Wagner (2003) fi nd that Some personal traits have been associated accumulation of a balanced skills mix (that with entrepreneurial success. Education, for is, general human capital) is associated with instance, increases the likelihood of survival above- average postentry performance (see of new fi rms and subsequent economic per- also Vivarelli 2012). formance (Bates 1990; Gimeno and others These aspects of entrepreneurship appear 1997; Acs, Armington, and Zhang 2007), to be relevant in LAC. In Brazil, for instance, ENTREPRENEURSHIP BY INCUMBENT FIRMS: WHAT EXPLAINS THE INNOVATION GAP? 85 employee spin-offs from existing enterprises spin-offs have a lower likelihood of exit (fig- account for one- sixth to one-third of new ure 3.20): employee spin- offs, for instance, formal sector fi rms, and these fi rms exhibit are 6 percent more likely than ordinary new a higher likelihood of survival. Muendler, firms to survive. Rauch, and Tocoian (2012) analyze post- These results point to the importance of entry performance of new firms using a knowing the sector and the country con- unique employee- employer database from text for the success of new firms, but they Brazil that offers comprehensive individ- remain silent about the particular skills ual employee information on occupations, that are needed. Lessons from high-income demographic characteristics, and earnings. countries— derived largely from analyses of Exploiting these data, they compare differ- high-tech sectors— may be of limited rele- ences in likelihood of exit after five years vance for most countries in LAC, where firms between “ordinary” new firms and spin- do not necessarily operate at the technology offs, distinguishing between spin- offs by frontier. employees, spin- offs generated by the fi rms New studies of LAC show that people with themselves out of diversification purposes, good technical and managerial skills may and “divestitures,” where a new firm absorbs improve a firm’s innovative potential and, plants from an existing firm. They also break ultimately, productivity. Drawing on gradu- out spin- offs that included high-level man- ation records, membership in professional agers from the mother firm from spin- offs societies, and historical census data, Maloney that included only employees. Their results and Valencia- Caicedo (2012) generated new suggest that the founders’ profi le influences data on the stock of engineers at the end of postentry performance: in almost all cases, the 19th century at the subnational level for a panel of five countries (the United States, Argentina, Chile, Mexico, and the República FIGURE 3.20 Likelihood of spin- off firm after five Bolivariana de Venezuela). Two findings years, by type of entrant stand out from these data. First, Argentina, 4 Chile, and Mexico had lower densities of engineers in 1900 than Spain and Portugal, Additional likelihood of exit (%) despite higher income per capita (figure 3.21). 0 FIGURE 3.21 Engineering density and GDP of –4 selected countries, 1900 180 –8 United States, north Engineers per 100,000 male workers 150 –12 Director/manager Five or more 120 employees 90 United States Employee spinoff, significant at 5% Divestiture, significant at 5% 60 United States, south Diversification venture, significant at 5% Venezuela, RB Diversification venture, not significant 30 Portugal Spain Brazil Chile Colombia Argentina Peru Mexico Source: Muendler, Rauch, and Tocoian 2012. 0 Note: Bars show coefficients of a multivariate regression on the likelihood 6.0 6.5 7.0 7.5 8.0 8.5 9.0 of exit. “Director/manager” means that a director or manager from the Log of GDP per capita (US$ in 1900) former company was involved in the new company; “five or more employ- ees” means that five or more employees of the former company joined the new company. The omitted category is all other new formal sector Source: Maloney and Valencia-Caicedo 2012. firms. Regressions include cohort and sector fixed effects. Note: GDP = gross domestic product. 86 LATIN AMERICAN ENTREPRENEURS This gap in the number of engineers per cap- with efforts to improve management perfor- ita appears to have persisted over the course mance itself, could potentially be an effective of history. Second, at both the national and avenue for policy action. subnational levels, there is a significant asso- ciation between engineering density in the Agglomerations and spillovers: 1900s and per capita income in the 2000s. Can policy makers affect them? A low prevalence of people with good techni- cal skills in LAC may thus have hurt fi rms’ The holy grail of entrepreneurship policies innovative potential and affected long-term is scale effects: once a critical mass of entre- growth prospects. preneurial firms is achieved, the ball may Why do engineers matter? The roman- keep rolling almost by itself. Scale effects tic vision of engineers conducting high-tech can be driven by technological factors, such research and pushing the technology fron- as increasing returns to scale in production, tier may represent only part of the story, as well as by factors beyond technology that especially in middle-income countries. Most can, in principle, be steered through policy engineers play a less fashionable but equally actions. Such factors include positive exter- important role in continuously improving nalities generated by geographic agglomera- basis products and production processes tions of similar industries. and adapting foreign products and technolo- The idea of designing policies that promote gies to local conditions. There is no need agglomeration is tempting, and the evidence for a degree from Harvard or MIT to equip supporting the presence of externality effects engineers to perform such functions: a solid is strong. Natural advantages can explain only curriculum that combines good analytical about 20 percent of geographic concentration; thinking with practice— and teaches what although correcting for omitted geographic the industry is demanding— may suffice. characteristics might raise the share they Chapter 6 provides further evidence on the account for slightly, agglomeration effects association between the region’s innovation would not exist without localized intraindus- gap and its low density of engineers. try spillovers (Ellison and Glaeser 1999). But technology is only one part of the In the United States, for instance, employ- equation. How people and technologies are ment growth is strongly predicted by the managed also matters: a potentially great presence of small establishments, across both new product may never see the light of day cities and industries within cities. The pres- if engineering efforts are poorly coordi- ence of entrepreneurs may thus attract more nated. And, as discussed, the region’s larg- entrepreneurs, by lowering the cost of entry est fi rms seem to have subpar management through the growth of suppliers, venture cap- practices. italists, and by developing an entrepreneur- An additional fi nding is that few manag- ial culture (Chinitz 1961; Glaeser and Kerr ers may be aware of how they are running 2009; Glaeser, Kerr, and Ponzetto 2010). their company. Maloney and Sarrias (2012) Greenstone, Hornbeck, and Moretti (2010) present survey evidence on managers’ rat- also fi nd that five years after the opening of ings of how well they think they are running a large plant, the total factor productivity of the company (figure 3.22) The correlation incumbent plants is 12 percent higher than in between how good managers are (as mea- counties without large plant openings. More- sured by the management score) and how over, this productivity spillover is larger for good they think they are is very low (about plants that share similar labor and technol- 0.2). This low correlation is also evident in ogy pools with the new plant. other countries, even high-income countries. Human capital spillovers can also be sig- But given the lower average management per- nificant. Moretti (2004) finds that the produc- formance of LAC firms, it suggests that inter- tivity of plants in cities that experienced large ventions aimed at improving awareness of the increases in the share of college graduates rises importance of management practices, along more than the productivity of similar plants ENTREPRENEURSHIP BY INCUMBENT FIRMS: WHAT EXPLAINS THE INNOVATION GAP? 87 FIGURE 3.22 Actual versus perceived management quality in Argentina, Mexico, Chile, and Brazil a. Argentina b. Mexico 10 Self-perception, index 8 6 4 2 1 2 3 4 5 1 2 3 4 5 Average of all management questions c. Chile d. Brazil 10 Self-perception, index 8 6 4 2 1 2 3 4 5 1 2 3 4 5 Average of all management questions Source: Maloney and Sarrias 2012. in cities that experienced small increases. For Given the large impacts of externalities, it human capital, too, “technological proxim- is tempting for policy makers to invest large ity” seems to matter: within a city, spillovers sums to attract firms and promote agglomer- between industries that are economically ations. And indeed, when there is competition close are larger than spillovers between indus- to attract investments, providing (reasonable) tries that are economically distant. fi nancial incentives to make the balance tilt Local multipliers also play a role in fos- in a region’s favor may be money well spent tering agglomerations. Moretti and Thulin (Moretti 2010). But there is very little evi- (2012), for instance, distinguish between dence that it is possible to artificially gener- locally tradable and nontradable jobs. They ate self-sustaining agglomerations. Successful fi nd that every time a local economy gener- industrial parks that were promoted by the ates a new job by attracting a new business government are hard to find, and there is lit- in the traded sector, a significant number of tle evidence that large tax incentives aimed at additional jobs is created in the nontraded attracting foreign firms and investments have sector. The type of job generated matters: the systematically provided good returns. local multiplier varies from a third to three. Success stories often result out of the com- It is particularly large for high-tech indus- bination of natural advantages (including tries and other jobs that require high levels of “first-mover” advantages), luck, and good human capital. policies that nudged incentives in already 88 LATIN AMERICAN ENTREPRENEURS favorable local conditions. Silicon Valley, the international spillovers (mostly generated gold standard of entrepreneurship ecosys- by the transfer of knowledge) affect innova- tems, was never “founded.” It evolved from a tion. The policy implications may change unique set of circumstances: Stanford Univer- dramatically depending on the source of the sity’s interest in hooking up with the indus- spillovers. If spillovers are domestic, there try, a strong aerospace industry, a handful of may be space for government intervention industries making breakthrough progresses to solve free-riding problems (the so- called in the semiconductor industry, a liberal immi- “appropriability problem”): because firms do gration policy toward doctoral students, and not internalize the positive impacts of their pure luck, among others (Isenberg 2010). Sili- production of knowledge on other firms, they con Valley also benefitted from a first-mover may underinvest in knowledge production. If, advantage. It is not clear that if the same con- in contrast, knowledge spillovers are gener- ditions appeared today they would lead to the ated internationally—by technology transfers same success. through trade, for example—then the appro- Overall, there is not enough evidence to priate policy may be one of laissez-faire and say with certainty what works and what trade liberalization, especially for small open does not. But some do’s and don’ts are start- economies that do not affect prices in inter- ing to emerge (box 3.3). And one fact seems national markets. to be repeatedly confirmed: it is very hard to Bravo- Ortega, Causalito, and Leder- get it right. man (2013) attempt to identify the origins Geography is only one channel through of spillovers. Using a cross- country panel which spillovers may affect firms’ productiv- of investment in R&D and USPTO patents, ity and innovative behavior. Another is trade. they look at the extent to which domestic There have been few attempts to understand patenting activity is influenced by domes- the extent to which domestic as opposed to tic investments in R&D, the domestic stock BOX 3.3 Do’s and don’ts of entrepreneurship ecosystems In a 2010 article in the Harvard Business Review, cultural factors in fostering entrepreneurship Daniel Isenberg, a business school professor, sum- has been understudied. But there is a growing marizes years of experience in analyzing conditions sense that culture matters significantly. under which entrepreneurs thrive (“entrepreneurship • Experiment holistically, but do not overengi- ecosystems”). His list of dos and don’ts includes the neer. For ecosystems to thrive, several elements following: must work well together—hence the need to experiment holistically along several fronts. • Shape the ecosystem around local conditions. Overengineering, however, may also lead to Natural advantages cannot fully account for failure. The more there is a need to intervene, the higher productivity of agglomerations, but the less likely it is that ecosystems will eventu- they can give a good head start. The less natural ally thrive alone. advantage there is a, the more difficult it will be • Think about an exit strategy. Not all experi- to foster a self-sustaining agglomeration. mentation will deliver the desired results—if it • Engage the private sector from the start. Profit- did it would not be an experiment. But there is driven motives may lead firms to miss the ben- a danger, often for political economy reasons, eficial effect of spillovers, but they provide the to keep subsidizing pilots that do not work. All best perspective to judge whether a venture will experiments should have clearly defined time work. horizons and exit strategies. • Tackle cultural change head-on. Because cul- ture is at times intangible, the importance of Source: Adapted from Isenberg 2010. ENTREPRENEURSHIP BY INCUMBENT FIRMS: WHAT EXPLAINS THE INNOVATION GAP? 89 of patents, the stock of patents from other technology leaders is large and they prefer to countries in the same region, and the stock import technology. of patents in the rest of the world. The results The ambiguous interpretations about the suggest some scope for government interven- role of international movements of goods, tion (figure 3.23). Net of R&D expenditures, services, and capital on domestic innovation the domestic stock of USPTO patents does highlights the need to understand the roles affect the number (that is, flow) of patents that trade and multinational activity play in by a country, a result that is consistent with fostering innovation and entrepreneurship. the presence of domestic knowledge spill- The following chapters examine the perfor- overs. The elasticity is less than one, how- mance of some of the top entrepreneurs in ever, suggesting that “big push” interventions LAC and the fi rms that enter and survive in to generate a self-sustaining virtuous cycle of highly competitive export markets and seek innovation may be difficult to achieve. out foreign markets by investing abroad. On the international front, the stock of These firms are led by the highest end of the patents of countries in the same region does region’s entrepreneurs. But how innovative are seem to positively affect domestic patenting they? Chapters 4 and 5 examine these issues. activity, suggesting the presence of positive regional knowledge spillovers. In contrast, the stock of patents from the rest of the world Notes appears to negatively affect domestic patent- 1. Some observers argue that it may be optimal ing activity. This negative coefficient can be not to enforce some poorly conceived regu- interpreted in two ways. It could simply be lations that may hurt firms or people exces- an outcome of “patent races” between high- sively, but weak enforcement is a second-best option that often creates more complications and low-income countries, where low-income than it solves. countries do not see a need to patent innova- 2. Throughout this chapter we use the following tions because of lower engagement at a global groups of economies unless otherwise noted. scale or because of lower returns from pat- LAC5 includes Argentina, Brazil, Chile, enting in sectors where firms in high-income Colombia, and Mexico. Other LAC includes countries are also active. It could also be a Bolivia, Costa Rica, the Dominican Republic, genuine negative externality, where firms Ecuador, El Salvador, Guatemala, Honduras, in low-income countries do not see a need Nicaragua, Paraguay, Peru, Uruguay, and to innovate because their distance from the República Bolivariana de Venezuela. Carib- bean includes Antigua and Barbuda, Cuba, Dominica, Grenada, Guyana, Haiti, Jamaica, FIGURE 3.23 Domestic versus international St. Kitts and Nevis, St. Lucia, St. Vincent and spillovers in patenting activity the Grenadines, Suriname, and Trinidad and Tobago. ECA (Eastern Europe and Central Asia) includes Albania, Armenia, Azerbaijan, Belarus, Bosnia and Herzegovina, Bulgaria, R&D expenditures Georgia, Kazakhstan, Latvia, Lithuania, FYR Macedonia, Moldova, Romania, the Russian Domestic stock Federation, Serbia, Turkey, Turkmenistan, of patents Ukraine, and Uzbekistan. EAP4 includes Regional stock Indonesia, Malaysia, the Philippines, and of patents Thailand. High-income economies include Stock of patents Australia; Canada; Hong Kong SAR, China; in the rest Israel; Japan; the Republic of Korea; Kuwait; of the world –0.4 –0.2 0 0.2 0.4 0.6 0.8 New Zealand; Oman; Saudi Arabia, Singa- Elasticity of patenting activity pore; Switzerland; the United Arab Emirates; the United States; and all countries in the Source: Bravo- Ortega, Cusolito, and Lederman 2013. European Union not included in ECA. The set Note: R&D = research and development. of economies from each group used in figures 90 LATIN AMERICAN ENTREPRENEURS throughout this chapter varies according to 11. Budget information is not provided because of data availability. the challenges in obtaining comparative data. 3. The OECD (2002) Frascati Manual on R&D Some agencies are incorporated into minis- statistics, which is used around the world, tries, making it difficult to estimate the budget excludes investments in soil analysis and min- available to the agency. eral exploration from R&D activities. Conse- quently investments in innovation in agriculture and mining tend to be underreported. References 4. It is likely that multinational corporations Acemoglu, D., P. Aghion, and F. Zilibotti. 2006. rather than domestic firms conduct a large “Distance to Frontier, Selection, and Economic share of R&D in China. A similar bias in Growth.” Journal of the European Economic reported R&D statistics may affect some LAC Association 4 (1): 37–74. economies, such as Mexico. It is unlikely that Acharya, V. V., and K. V. Subramanian. 2009. LAC economies would overcome China’s “Bankruptcy Codes and Innovation.” Review measured R&D effort even if the data were of Financial Studies 22 (12): 4949– 88. corrected for this bias, however. (The authors Acs, Z. J., C. Armington, and T. Zhang. 2007. would like to thank William F. Maloney for “The Determinants of New- Firm Survival pointing this out.) across Regional Economies: The Role of 5. R&D data are classified as “productive- Human Capital Stock and Knowledge Spill- sector” R&D when the source of the financ- over.” Papers in Regional Science 86 (3): ing comes from a company that participates 367–91. in the market. However, these companies can Aghion, P., G.-M. Angeletos, A. Banerjee, and be publicly owned, thus blurring the distinc- K. Manova. 2010. “Volatility and Growth: tion between private and public R&D. In Credit Constraints and the Composition of this report, as in others such as Pagés-Serra Investment.” Journal of Monetary Economics (2010), the term private is used to character- 57 (3): 246– 65. ize productive-sector R&D. Aghion, P., N. Bloom, R. Blundell, R. Griffith, 6. Specifically, the log of patents per million peo- and P. Howitt. 2005. “Competition and Inno- ple was regressed on the log of GDP (expressed vation: An Inverted U Relationship.” Quar- in purchasing power parity terms), the log terly Journal of Economics 120 (2): 701–28. of population, and the log of merchandise Aghion, P., R. Blundell, R. Griffith, P. Howitt, exports to the United States. The dots in fig- and S. Prantl. 2009. “The Effects of Entry ure 3.4 represent the estimated intensity of pat- on Incumbent Innovation and Productivity.” enting in the United States given each country’s Review of Economics and Statistics 91 (1): characteristics along these three dimensions. 20–32. 7. Official export costs include costs for docu- Aghion, P., and P. Bolton. 1992. “An Incomplete ments, administrative fees for customs clear- Contracts Approach to Financial Contracting.” ance and technical control, customs broker Review of Economic Studies 59 (3): 473–94. fees, terminal handling charges, and inland Aghion, P., C. Harris, P. Howitt, and J. Vickers. transport. The cost measure does not include 2001. “Competition, Imitation and Growth tariffs or trade taxes. with Step- by- Step Innovation.” Review of 8. In the aggregate, productivity may still Economic Studies 68 (3): 467–92. increase, because more firms innovating raises Akcigit, U., and W. R. Kerr. 2010. “Growth the chances of a technological breakthrough through Heterogeneous Innovations.” NBER (Bento, forthcoming). Working Paper 16443, National Bureau of 9. The measured effects are in addition to the Economic Research, Cambridge, MA. impact of regulation on innovation as mea- Allen, F., and D. Gale. 2000. “Corporate Gov- sured by the Doing Business indicators (see ernance and Competition.” In Corporate figure 3.7), which, by affecting competition Governance: The Theoretical and Empirical themselves, also capture part of the effects of Perspectives, edited by X. Vives, 23–94. Cam- working though competition. bridge, U.K.: Cambridge University Press. 10. This section benefitted immensely from dis- Almus, M., and E. A. Nerlinger. 1999. “Growth cussions with Paolo Benedetti and Aitor Ortiz, of New Technology-Based Firms: Which Fac- of the Centro Regional de Competencia para tors Matter?” Small Business Economics América Latina (CRC), as well as from infor- 13 (2): 141–54. mation provided by the CRC. ENTREPRENEURSHIP BY INCUMBENT FIRMS: WHAT EXPLAINS THE INNOVATION GAP? 91 Armour, J., and D. Cumming. 2008. “Bank- Technological Innovation.” Economic Journal ruptcy Law and Entrepreneurship.” American 105 (429): 333– 44. Law and Economics Review 10 (2): 303–50. ———. 1999. “Market Share, Market Value, and Arvanitis, S., and T. Stucki. 2012. “What Deter- Innovation in a Panel of British Manufacturing mines the Innovation Capability of Firm Firms.” Review of Economic Studies 66 (3): Founders?” Industrial and Corporate Change 529–54. 21 (4): 1049– 84. Bravo-Ortega, C., A. P. Cusolito, and D. Leder- Ayyagari, M., A. Demirgüc-Kunt, and V. Maksi- man. 2013. “Faraway or Nearby? Domestic movic. 2011. “Firm Innovation in Emerging and International Spillovers in Patenting and Markets: The Role of Finance, Governance, Product Innovation.” Unpublished manuscript. and Competition.” Journal of Financial and CAF (Corporación Andina de Fomento). 2013. Quantitative Analysis 46 (6): 1545– 80. “Entrepreneurship in Latin America: From Balconi, M., and R. Fontana. 2011. “Entry and Subsistence to Productive Change.” Caracas, Innovation: An Analysis of the Fabless Semi- Venezuela, RB. conductor Business.” Small Business Econom- Centro Regional de Competencia para América ics 37 (1): 87–106. Latina (CRC) (database). 2013. http://www Bartelsman, E. J., and M. Doms. 2000. “Under- .crcal.org. Available upon request. standing Productivity: Lessons from Longitu- Chandler, A. D., Jr. 1990. Scale and Scope: The dinal Microdata.” Journal of Economic Lit- Dynamics of Industrial Capitalism. Cam- erature 38 (3): 569–94. bridge, MA: Belknap Press. Bassanini, A., and E. Ernst. 2002. “Labour Mar- Chinitz, B. 1961. “Contrasts in Agglomeration: ket Institutions, Product Market Regulation, New York and Pittsburgh.” American Eco- and Innovation: Cross- Country Evidence.” nomic Review 51 (2): 279– 89. Working Paper 316, OEC D E conomics Clark, J. 2005. “Competition Advocacy: Chal- Department, Organisation for Economic Co- lenges for Developing Countries.” OECD operation and Development, Paris. Journal: Competition Law and Policy 6 (4): Bates, T. 1990. “Entrepreneur Human Capital 69– 80. Inputs and Small Business Longevity.” Review Colombo, M. G., and L. Grilli. 2005. “Found- of Economics and Statistics 72 (4): 551–59. ers’ Human Capital and the Growth of New Beck, T., A. Demirgüc- Kunt, L. Laeven, and Technology- Based Firms: A Competence- R. Levine. 2008. “Finance, Firm Size, and Based View.” Research Policy 34 (6): 795– 816. Growth.” Journal of Money, Credit and Bank- ———. 2010. “On Growth Drivers of High-Tech ing 40 (7): 1379– 405. Start- Ups: Exploring the Role of Founders’ Beck, T., R. Levine, and N. V. Loayza. 2000. Human Capital and Venture Capital.” Journal “Financial Intermediation and Growth: Cau- of Business Venturing 25 (6): 610–26. sality and Causes.” Journal of Monetary Eco- de la Torre, A., A. Ize, and S. L. Schmukler. 2012. nomics 46: 31–77. Financial Development in Latin America and Bekaert, G., C. R. Harvey, and C. Lundblad. the Caribbean: The Road Ahead. Washington, 2005. “Does Financial Liberalization Spur DC: World Bank. Growth?” Journal of Financial Economics de Soto, H. 1989. The Other Path: The Invisible 77 (1): 3–55. Revolution in the Third World. New York: Bento, P. Forthcoming. “Competition as a Discov- Basic Books. ery Procedure: Schumpeter Meets Hayek in a Didier, T., and S. L. Schmukler. 2011. Financial Model of Innovation.” American Economic Development in Latin America: Stylized Facts Journal; Macroeconomics. and the Road Ahead. LAC Finance Develop- Bertrand, M., and A. Schoar. 2003. “Managing ment Flagship. Washington, DC: World Bank. with Style: The Effect of Managers on Corpo- Djankov, S., C. McLiesh, and R. Ramalho. 2006. rate Policy.” Quarterly Journal of Economics “Regulation and Growth.” Economics Letters 18 (4): 1169–20. 92 (3): 395– 401. Bloom, N., and J. van Reenen. 2007. “Measuring Doing Business (database). World Bank Group, and Explaining Management Practices across Washington, DC. http://www.doingbusiness Firms and Countries.” Quarterly Journal of .org. Economics 122 (4): 1351– 408. Dutz, M. 2007. Unleashing India’s Innovation: Blundell, R., R. Griffith, and J. van Reenen. Towards Sustainable and Inclusive Growth. 1995. “Dynamic Count Data Models of Washington, DC: World Bank. 92 LATIN AMERICAN ENTREPRENEURS Ellison, G., and E. L. Glaeser. 1999. “The Geo- Johnson S., D. Kaufmann, and A. Shleifer. 1997. graphic Concentration of Industry: Does “The Unofficial Economy in Transition.” Natural Advantage Explain Agglomeration?” Brookings Papers on Economic Activity 2: American Economic Review 89 (2): 311–16. 159–239. Enterprise Surveys (database). World Bank Group, King, R. G., and R. Levine. 1993. “Finance and Washington, DC. http://www.enterprise Growth: Schumpeter Might be Right.” Quar- surveys.org/. terly Journal of Economics 108 (3): 717–37. Foster, L., J. C. Haltiwanger, and C. J. Krizan. Lazear, E. P. 2004. “Balanced Skills and Entre- 2001. “Aggregate Productivity Growth; Les- preneurship.” American Economic Review sons from Microeconomic Evidence.” In New 94 (2): 208–11. Developments in Productivity Analysis, edited ———. 2005. “Entrepreneurship.” Journal of by C. R. Hulten, E. R. Dean, and M. J. Harper, Labor Economics 23 (4): 649– 80. 303–72. Chicago: University of Chicago Press. Lazonic, W. 1990. Competitive Advantage on the Galdon- Sanchez, J. E., and J. A. Schmitz. 2002. Shop Floor. Cambridge, MA: Harvard Univer- “Competitive Pressure and Labor Productiv- sity Press. ity: World Iron- Ore Markets in the 1980’s.” Lederman, D., and W. Maloney. 2003. “R&D American Economic Review 92 (2): 9–23. and Development.” Policy Research Working Ganotakis, P. 2012. “Founders’ Human Capital Paper Series 3024, World Bank, Washington, and the Performance of UK New Technol- DC. ogy Based Firms.” Small Business Economics Levine, R. 2005. “Finance and Growth: Theory 39 (2): 1–21 and Evidence.” In Handbook of Economic Gimeno, J., T. B. Folta, A. C. Cooper, and C. Y. Growth, vol. 1, edited by P. Aghion and Woo. 1997. “Survival of the Fittest? Entrepre- S. Durlauf, 865–934. Amsterdam: Elsevier. neurial Human Capital and the Persistence of Loayza N. 1996. “The Economics of the Informal Underperforming Firms.” Administrative Sci- Sector: A Simple Model and Some Empirical ence Quarterly 42 (4): 750– 83. Evidence from Latin America.” Carnegie- Glaeser, E. L., and W. R. Kerr. 2009. “Local Rochester Conference Series on Public Policy Industrial Conditions and Entrepreneurship: 45 (1): 129– 62. How Much of the Spatial Distribution Can We Machado, J., and J. Mata. 2005. “Counterfactual Explain?” Journal of Economics & Manage- Decomposition of Changes in Wage Distribu- ment Strategy 18 (3): 623– 63. tions Using Quantile Regression.” Journal of Glaeser, E. L., W. R. Kerr, and G. A.M. Ponzetto. Applied Econometrics 20 (4): 445– 65. 2010. “Clusters of Entrepreneurship.” Journal Maloney, W. F., and M. Sarrias. 2012. “What of Urban Economics 67 (1): 150– 68. Makes Latin American Managers So Bad?” Greenstone, M., R. Hornbeck, and E. Moretti. World Bank, Washington, DC. 2010. “Identifying Agglomeration Spillovers: Maloney, W. F., and F. Valencia- Caicedo. 2012. Evidence from Winners and Losers of Large “Part II: Engineers, Innovative Capacity and Plant Openings.” Journal of Political Econ- Development in the Americas.” Background omy 118 (3): 536–98. paper for this report. World Bank, Washing- Grossman, G. M., and E. Helpman. 1991. “Qual- ton, DC. ity Ladders in the Theory of Growth.” Review Manso, G. 2011. “Motivating Innovation.” Jour- of Economic Studies 58 (1): 43– 61. nal of Finance 66: 1823– 69. Guiso, L., P. Sapienza, and L. Zingales. 2004. Mondragón, C. 2012. “Private Equity and Ven- “Does Local Financial Development Mat- ture Capital in Latin America.” World Bank, ter?” Quarterly Journal of Economics 119 (3): Washington, DC. 929– 69. Moretti, E. 2004. “Human Capital Externalities in Iacovone, L., F. Rauch, and L. A. Winters. 2013. Cities.” In Handbook of Regional and Urban “Trade as an Engine of Creative Destruction: Economics, edited by J. V. Henderson and J. F. Mexican Experience with Chinese Competi- Thisse, 2243–91. Amsterdam: Elsevier. tion.” Journal of International Economics ———. 2010. “Local Multipliers.” American 89 (2): 379–92. Economic Review 100 (2): 373–77. Isenberg, D. J. 2010. “The Big Idea: How to Start Moretti, E., and P. Thulin. 2012. “Local Multi- an Entrepreneurial Revolution.” Harvard pliers and Human Capital in the United States Business Review June: 2–11. and Sweden.” Working Paper 914, Research Institute of Industrial Economics, Stockholm. ENTREPRENEURSHIP BY INCUMBENT FIRMS: WHAT EXPLAINS THE INNOVATION GAP? 93 Muendler, M., J. E. Rauch, and O. Tocoian. 2012. Segerstrom, P. S. 1991. “Innovation, Imitation, “Employee Spinoffs and Other Entrants: Styl- and Economic Growth.” Journal of Political ized Facts from Brazil.” International Journal Economy 99 (4): 807–27. of Industrial Organization 30 (5): 447–58. Seker, M. 2013. “Innovation Performance in Murphy, K. M., A. Shleifer, and R. W. Vishny. Latin America and Caribbean Region.” Back- 1991. “The Allocation of Talent: Implications ground paper for this report. for Growth.” Quarterly Journal of Economics Servén, L. 1998. “Macroeconomic Uncertainty 106 (2): 503–30. and Private Investment in Developing Coun- Nickell, S. J. 1996. “Competition and Corporate tries: An Empirical Investigation.” World Bank Performance.” Journal of Political Economy Policy Research Working Paper 2035, Wash- 104 (4): 724– 46. ington, DC. OECD (Organisation for Economic Co-operation UIS (UNESCO [United Nations Education, Sci- and Development). 2002. Frascati Manual: entific and Cultural Organization] Institute for Proposed Standard Practice for Surveys on Statistics). Montreal, Quebec, Canada. http:// Research and Experimental Development. www.uis.unesco.org/. Paris: OECD. USPTO (U.S. Patent and Trademark Office). Ortiz, A. 2013. Honduras and Nicaragua’s Study 2012. “Patents by Country, State, and Year: on Recommendations for Competition Advo- All Patent Types.” Washington, DC. http:// cacy Policy. World Bank, Washington, DC. www.uspto.gov/web/offices/ac/ido/oeip/taf Pagés-Serra, C., ed. 2010. The Age of Productiv- /cst_all.htm. ity: Transforming Economies from the Bottom Vivarelli, M. 2012. “Drivers of Entrepreneurship Up. Washington, DC: Inter-American Devel- and Post-Entry Performance: Microeconomic opment Bank. Evidence from Advanced and Developing Pienknagura, S. 2013. “The Extensive and Inten- Countries.” Policy Research Working Paper sive Margin of Innovation by Incumbent 6245, World Bank, Washington, DC. Firms: A Conceptual Framework.” Unpub- Wagner, J. 2003. “Testing Lazear’s Jack- of-All- lished manuscript. Trades View of Entrepreneurship with Ger- Primo, D. M., and W. S. Green. 2011. “Bank- man Micro Data.” Applied Economics Letters ruptcy Law and Entrepreneurship.” Entrepre- 10 (11): 687– 89. neurship Research Journal 1 (2): 1–22. Womack, J. P., D. T. Jones, and D. Roos. 1990. Rajan, R. G., and L. Zingales. 1998. “Financial The Machine that Changed the World. New Dependence and Growth.” American Eco- York: Harper- Collins. nomic Review 88 (3): 559– 86. World Development Indicators (database). World Rauch, J. E. 1991. “Modelling the Informal Sec- Bank, Washington, DC. http://data.worldbank tor Formally.” Journal of Development Eco- . o r g /d a t a - c a t a l o g / w o r l d - d e v e l o p m e n t nomics 35 (1): 33– 47. -indicators. Romer, P. 1990. “Endogenous Technological Change.” Journal of Political Economy 98 (5): S71– S102. Export Entrepreneurship 4 Exporting is difficult: only highly successful firms ever try to do so, and most of them exit exporting within a year. Export entrepreneurship by the small number of superstar firms that survive drives long-term export growth. New evidence indicates that the economies of Latin America and the Caribbean underperform poorer economies in terms of export entry rates. However, companies in the region proved resilient in the face of the contraction of foreign demand in 2008–09, experimenting with new export products and attempting to penetrate new markets. These findings suggest that competitive pressures can spur innovation by high- end entrepreneurs. If the costs of entry into export markets partly reflect the need to gather information about the characteristics of foreign markets, export promotion policies that pro- vide such information could increase export entry and enhance the likelihood of survival in global markets. Preliminary evidence suggests that export promotion policies in the region are having these effects, although even the region’s superstars enter exporting at lower rates than comparators in other regions. T his chapter focuses on one type of Export entrepreneurship has several high- end innovative entrepreneur- dimensions. This chapter studies the rate of ship: the act of entering and surviv- entry and exit of exporting fi rms, the likeli- ing in global markets.1 Like the forms of hood of survival of fi rms in export markets innovation discussed in chapter 3, exporting after entry, and the size of their exports at is costly. Firms thus carefully analyze the entry. (Chapter 6 revisits the issue of whether potential costs and benefits of entering new LAC is lagging similar economies in its level markets. of openness to international trade.) The chapter begins with an assessment of Partly because of LAC’s comparative the contribution of export entrepreneurship advantage in mining and agriculture com- to overall export growth in Latin America modities, observers have long been pessimistic and the Caribbean (LAC) in the medium about the region’s growth and entrepreneur- term. It benchmarks LAC countries relative ial potential (see, for example, the literature to other developing economies in terms of reviewed by Lederman and Maloney 2007). export entrepreneurship. However, the literature has remained silent 95 96 LATIN AMERICAN ENTREPRENEURS with respect to the role of entrepreneurship fi rms— enter and thrive in export markets. 2 as a driver of export growth in developing In an influential review of the literature on countries with diverse trade structures. This firms in international trade, Bernard and oth- chapter therefore benchmarks broad sectors ers (2007) document that in 2000 only about of economic activity in terms of their poten- 4 percent of the 5.5 million fi rms operating tial for export entrepreneurship and discusses in the United States were exporters. Among the role of comparative advantage as a deter- manufacturing and agricultural fi rms, only minant of export entrepreneurship. about 15 percent were exporters. Eaton, Kor- The chapter presents results from comple- tum, and Kramarz (2011) show that only mentary benchmarking exercises. The fi rst about 15 percent of French manufacturing set is descriptive; it presents the basic statis- firms with more than 20 employees were tics for groups of countries in LAC and other exporters in 1986. regions classified by their structural charac- Although census data on enterprises are teristics. The analysis does not identify the scarce in developing countries, it is safe to portion of each export entrepreneurship indi- speculate that exporting fi rms are also rare cator that is strictly associated with country in LAC and elsewhere. Lederman (2010) features or industry characteristics. reports that about 36 percent of a developing The second set of exercises highlights country sample of more than 25,000 manu- “conditional” international comparisons. facturing fi rms surveyed between 2000 and This approach relies on econometric analy- 2006 reported exporting (these data are from sis to decompose the sources of the observed the World Bank’s Enterprise Surveys, which international differences in export entrepre- are not censuses and are probably upwardly neurship indicators into country, industry, biased in terms of the number of export- and time-period effects. ers).3 Lederman (2013) reports that the aver- The evidence presented suggests that age export intensity (the ratio of exports to export entrepreneurship in most LAC coun- total sales) ranges from less than 1 percent tries has significantly contributed to national (in the 2006 enterprise survey of Burundi) export growth, even over short time periods. to 29 percent (in the 2007 enterprise survey The picture that emerges is one of a seemingly of Bangladesh). The samples of fi rms from dynamic export sector, characterized by vig- LAC countries had average export intensi- orous entry and exit, with relatively healthy ties ranging from less than 1 percent (in the survival rates. Unfortunately, the region’s República Bolivariana de Venezuela survey export entry and survival rates appear to be of 2010) to slightly less than 15 percent (in lower than those of poorer countries after the Peru survey of 2010). Moreover, in Leder- controlling for industry characteristics and man’s global sample of more than 55,000 GDP per capita. fi rms, the correlation between export inten- There is, however, good news on the policy sity and size (measured by the number of front. Export promotion policies focused on employees) is high: average export intensity solving informational market failures seem among developing country firms with less to stimulate the entry of new LAC firms into than 15 employees was about 13 percent, exporting activities as well as enhance their whereas firms with more than 1,000 employ- likelihood of survival in export markets. ees reported average exports over total sales of about 40 percent. In sum, the vast major- ity of private sector enterprises do not export, Exporting as a transformative partly because export intensity tends to rise entrepreneurial act with the size of firms and developing econo- Exporting is difficult and thus rare among mies tend to have small firms. private enterprises. Only the best firms— The literature analyzes the export decision the largest, most productive, “superstar” as a function of both variable and fixed costs. EXPORT ENTREPRENEURSHIP 97 Variable costs are associated with transport A large body of literature suggests that and trade barriers (such as import or export more globalized economies (usually mea- taxes). The magnitude of such costs presum- sured by the share of international trade ably varies with the quantity exported. Fixed over gross domestic product [GDP] or by costs (which do not vary with the quantity indicators related to trade policies) tend to of exports) include investments necessary to grow more rapidly than countries with less establish foreign business partners and clients, globalized economies (see Sachs and War- learn about product standards (both policy- ner 1995; Frankel and Romer 1999; Alcala induced via regulations and consumer prefer- and Ciccone 2004; Wacziarg and Welch ences), market products, and conduct market 2008; Feyrer 2009; Arkolakis, Costinot, and feasibility studies and other business due dili- Rodriguez- Clare 2012; and Brückner and gence particular to each foreign market. Lederman 2012, among others). This litera- Fixed costs can be large. Using Colom- ture has its skeptics (such as Rodriguez and bian plant- level data, Das, Roberts, and Rodrik 2001), and openness to international Tybout (2007) estimate that the fi xed costs trade need not cause economic growth. Bald- of exporting were more than $400,000 in win and Robert-Nicoud (2008), for instance, 1986 (large fi rms tended to have fi xed costs argue that the impact of international inte- 10 percent lower than small firms). When gration depends on the relative magnitude entrepreneurs decide to export their prod- of two opposing effects. On the one hand, ucts or services abroad, they are taking a bet international competition may wipe out low- that export revenues will be large enough to productivity firms that must compete with cover the fi xed costs of exporting as well as imports in the domestic market. Although this the variable costs. The expected profits from one-time effect raises an economy’s aggregate exporting have to be higher than the profits productivity (by eliminating low-productivity from selling domestically. firms) and increases domestic consumer wel- Private sector enterprises that decide to fare (by making goods in the domestic market export are thus taking an entrepreneurial less expensive) the long-term growth rate can leap. They are exceptional entrepreneurs decline because of domestic firms’ perceptions who incur upfront costs in the expectation that they are less likely to “win” new varieties that their foreign sales will be more profit- of goods. Such firms may reduce their invest- able than their domestic sales. It is widely ments in both physical and knowledge capi- believed that only the most productive firms tal (for innovation). This analysis is similar are able to cover the costs of exporting and to the discussion in chapter 3 on the impacts grow in foreign markets. As discussed below, of competition on innovation. If investment most governments rely on export promotion by the typical domestic firm declines, then agencies and policies that focus on providing an economy’s aggregate growth is expected information on foreign markets to domestic to decline as well. But trade liberalization (or entrepreneurs with the objective of increas- any reduction in fixed or variable trade costs) ing national exports or diversifying the set of can reduce the marginal cost of investments in exported products. research and development (R&D) or knowl- Rephrasing the question about the fic- edge capital by reducing the price of “knowl- tional entrepreneur introduced in chapter edge capital.” As capital becomes cheaper, 1: Why should policy makers care whether private sector investment can rise, boosting Javier, a well- educated wine entrepreneur the economy’s aggregate income growth rate. from South America, succeeds in exporting Similarly, international integration can high-quality wine to the United States? The affect the incentives of the private sector concern for policy makers is not who exports and households to invest in human capi- but rather the size of the population of firms tal. Another strand of the trade literature, that are globalized. for example, links exports to the returns to 98 LATIN AMERICAN ENTREPRENEURS skills or education. It shows that under most from a global survey of EPAs conducted plausible scenarios for developing countries, by the World Bank (in 2010) suggest that increases in exports are associated with in all regions of the world the promotion increases in the relative demand for skilled of exports dominates other policy objec- or highly educated workers, thus raising the tives (figure 4.1). In fact, 60 percent of the returns to schooling and potentially stimu- 94 agencies that responded reported that lating investments in human capital (see, for stimulating export growth was the top pri- example, Brambilla and others 2012; Leder- ority, 20 percent reported that the top pri- man and Maloney 2012; and Brambilla, ority was the promotion of nontraditional Lederman, and Porto 2012 and the literature exports, and another 20 percent stated that cited therein). their top priority was the promotion of spe- In a nutshell, international integration can cific sectors. raise the prospects for quickening the pace of LAC agencies differ from agencies in other growth in the long term by raising the rate regions: their most important objective is the of accumulation of various forms of capi- promotion of nontraditional exports; overall tal (including human capital) and enhanc- export growth is their second most important ing technological upgrading. Such effects objective. Both are part and parcel of export are not preordained, however; they depend entrepreneurship, however, as will become on domestic firms’ capabilities to innovate clear through this chapter. and introduce new varieties. Indeed, it can Policy makers around the world tend to be argued that for international integration focus on export growth as an important pol- to be a source of inclusive growth, domes- icy objective. Hence, assessing the contribu- tic fi rms have to have the capacity to adapt tion of entrepreneurship to national export to competition by shifting their product and growth seems important on both analytical service varieties to sectors with higher rela- and policy grounds. tive domestic prices (Lederman 2013). As dis- The key issue is the contribution of export cussed in chapter 3, such innovation requires entrepreneurs to export growth in the long investing before (potentially) reaping the ben- run rather than their contribution in a single efits of uncertain future profits. year. On a yearly basis, the contribution of Smart public interventions in the con- new exporters is expected to be low, for two text of open, outward-oriented trade policy reasons. First, the number of new exporters regimes can help reduce the fi xed and vari- is expected to be small relative to the num- able costs of exporting for the benefit of all ber of incumbent exporting fi rms; entrepre- domestic entrepreneurs by diffusing informa- neurs are a rare breed. Second, the average tion about the idiosyncrasies of foreign mar- value of exports of new entrants is expected kets. Policy makers should thus care about to be lower than the average for established Javier’s success in foreign markets because exporters. The combination of these two his enterprise can become a conduit for other factors dictates that the short-run contribu- fi rms to obtain knowledge about the nature tion of new exporters will be small from an and magnitude of the fixed and variable costs accounting perspective. of exporting to the United States at a lower To the extent that export entrepreneurs cost than if Javier had not shown the way. can survive and attain relatively high export growth rates, their contribution to national exports tends to grow over time. Perhaps the Contribution of export easiest way to visualize the long-run contri- entrepreneurship bution of new exporters is to recognize that in the medium term today’s exporters were probably not in busi- From a policy viewpoint, increasing exports ness a century ago. In this extreme example, is usually the mandate of publicly funded the contribution of export entrepreneurs to export promotion agencies (EPAs). Data national exports approaches 100 percent EXPORT ENTREPRENEURSHIP 99 FIGURE 4.1 Strategic objectives of export promotion agencies, by region a. Europe and Central Asia b. Latin America and the Caribbean c. Middle East and North Africa 100 100 100 Country responses (%) 50 50 50 0 0 0 1 2 3 4 1 2 3 4 1 2 3 4 Strategy Strategy Strategy d. OECD countries e. Sub-Saharan Africa 100 100 Country responses (%) 50 50 0 0 1 2 3 4 1 2 3 4 Strategy Strategy Source: Lederman, Olarreaga, and Zavala 2013. Note: 1 = promote overall exports; 2 = promote nontraditional exports only; 3 = promote specific sectors; 4 = promote industrial clusters and other objectives. Survey covered 96 countries, but only 94 responded to this set of questions. OECD = Organisation for Economic Co-operation and Development. over time as older firms exit. Hence, the issue not exporting in 2004 but were exporting is how quickly this process of renewal of in 2005. 5 We define a subset of the 2005 exporting firms occurs; the speed of renewal exporter entrant cohort that includes only is associated with entry and exit rates, as well fi rms that continued to export continuously as the probability of survival of new entrants through 2009. We define as incumbents and their growth rates. fi rms that were already exporting in 2004 Given our interest in long- run export and continued to export through 2009. growth, we focus first on the contribution For each of these three groups of fi rms, we of entrepreneurs to exports over the period examine the changes in their contribution of time that is the longest possible period for to total exports and in the average size of which data are available for a large sample firms (measured by their exports). The 2005 of countries: 2004– 09. These data come cohort of new exporters includes both fi rms from the Exporter Dynamics Database, a that began exporting in 2005 but did not new database assembled by the World Bank’s continue to export through 2009 and fi rms Development Research Group (Cebeci and that entered in 2005 and survived through others 2012).4 2009. The sample of incumbents covers In each LAC country, we define as the firms that exported every year between 2004 2005 exporter entrant cohort firms that were and 2009. 100 LATIN AMERICAN ENTREPRENEURS TABLE 4.1 Number of new and incumbent exporters in seven countries in Latin America and the Caribbean, 2005–09 Dominican Year Brazil Chile Costa Rica Republic El Salvador Guatemala Peru Number of firms in 2005 exporter entrant cohort 2005 4,209 2,269 884 976 746 1,176 2,375 2006 2,272 856 488 309 303 473 1,115 2007 1,919 693 392 310 261 420 844 2008 1,662 619 340 271 237 383 689 2009 1,364 530 291 248 216 318 575 Number of firms in 2005 exporter entrant cohort surviving until 2009 984 314 210 104 128 221 392 Number of continuous exporters in 2004–09 8,472 2,345 985 586 942 1,566 1,713 Total number of exporters 2005 19,868 6,420 2,356 2,381 2,375 3,980 5,701 2006 19,102 6,535 2,824 2,043 2,360 4,022 6,147 2007 19,624 7,402 2,862 2,951 2,499 4,174 6,351 2008 19,087 7,677 2,778 2,515 2,558 4,424 6,833 2009 18,177 6,934 2,697 2,754 2,501 4,309 7,026 Source: Calculations based on data from the World Bank Exporter Dynamics Database. FIGURE 4.2 Share of total exports accounted for by new export Table 4.1 shows that the number of entrants in seven countries in Latin America and the Caribbean, exporter entrants in 2005 is large for most 2005– 09 LAC countries relative to the total number of exporters (this fi nding is consistent with 15 the high rates of exporter entry documented in the following section). Across countries, the number of new exporters in 2005 that Share of total exports (%) remain in export markets continuously 10 through 2009 is substantially smaller, rang- ing from 10 percent of the 2005 exporter entrants cohort in the Dominican Republic to 5 almost a quarter in Brazil. The share of national exports contributed by new entrants rose over time in six of the seven countries (figure 4.2). Only in Peru 0 2005 2006 2007 2008 2009 was the share of total exports from the new 2005 cohort lower in 2009 than in 2005. El Salvador Peru Perhaps more important, in all seven coun- Guatemala Costa Rica tries the share of total exports contributed by Brazil Chile the incumbent exporters that continuously Dominican Republic exported during 2000 – 09 declined over time (figure 4.3).6 The decline in all coun- Source: Calculations based on data from the World Bank Exporter Dynamics Database. tries except Costa Rica was 4– 5 percent- age points in just four years. The decline in Costa Rica was about 2 percentage points, but it occurred in 2008. Continued declines of these magnitudes would imply that a EXPORT ENTREPRENEURSHIP 101 country’s export base of firms would be com- FIGURE 4.3 Share of total exports accounted for by continuous pletely renewed in about 80–100 years. exporters in seven countries in Latin America and the Caribbean, Complementary evidence from other 2004–09 research indicates that the renewal of the pool of fi rms that sustains national exports 100 can occur more quickly than implied by the exporter dynamics portrayed in figures 4.2 Share of total exports (%) 90 and 4.3. Eaton and others (2007) show that in a sample of Colombian fi rms from 1996 to 2005, new exporters accounted for almost 80 half of exporting firms in any given year but contributed little to annual export growth. 70 The few fi rms that survived as exporters for more than a year, however, grew rapidly and accounted for about half of the country’s total 60 merchandise export growth after a decade. 2005 2006 2007 2008 2009 Lederman, Rodriguez- Clare, and Xu (2011) report that the 1999 cohort of entrants into Chile Dominican Republic Guatemala exporting accounted for almost 40 percent Peru Costa Rica of Costa Rica’s total merchandise exports by El Salvador Brazil 2005, albeit with significant turnover. This Source: Calculations based on data from the World Bank Exporter Dynamics Database. evidence suggests that firm entry into export activities and survival are as important as exports from incumbent fi rms for aggregate chapter 2, table 4.2 shows annual growth export growth in the medium term.7 rates of the dollar value of exports of the Mimicking the analyses of firm dynam- 2005 cohort of new exporters and the growth ics and employment generation discussed in rates of continuous exporters. TABLE 4.2 Export growth by new entrants and incumbents in seven countries in Latin America and the Caribbean, 2004–09 (percent) Dominican Year Brazil Chile Costa Rica Republic El Salvador Guatemala Peru Growth in total exports of 2005 exporter entrant cohort 2005/06 84.9 58.7 55.4 175.8 177.5 144.6 18.1 2006/07 5.7 17.5 18.2 47.2 –0.3 49.5 32.0 2007/08 15.5 20.6 13.8 19.8 43.8 31.5 1.3 2008/09 –14.9 –6.2 –6.5 –1.4 –30.2 11.7 –33.2 Growth in total exports of 2005 exporter entrant cohort surviving until 2009 2005/06 165.5 133.9 127.1 399.3 262.0 320.4 135.7 2006/07 15.8 53.2 34.2 60.8 1.6 81.3 63.7 2007/08 12.8 23.4 23.3 21.4 44.6 36.5 27.6 2008/09 –20.0 –6.5 –3.9 2.0 –30.1 13.3 –30.4 Growth in total exports of continuous exporters in 2004–09 2005/06 17.0 46.4 18.2 8.1 0.1 8.4 42.7 2006/07 14.6 16.6 9.2 –0.1 14.0 14.5 16.7 2007/08 20.3 1.7 2.1 12.3 23.5 9.4 6.7 2008/09 –28.7 –28.8 –9.4 –20.2 –22.0 –10.0 –14.3 Source: Calculations based on data from the World Bank Exporter Dynamics Database. 102 LATIN AMERICAN ENTREPRENEURS The first-year export growth rates for the has the potential to sustain high growth rates cohort of 2005 entrants are striking. They of national exports. The data in table 4.1 on range from 18 percent to more than 177 per- the number of new exporters in 2005 implies cent between 2005 and 2006. Growth rates that the rate of export entrepreneurship for the group of entrants that continuously (defi ned as the ratio of new exporters to the exported after 2005 range from 135 per- total number of exporters) could be high in cent to almost 400 percent. These growth LAC. What is a “normal” export entry rate? rates are upwardly biased, because they Where do LAC countries stand relative to underestimate first-year exports (in 2005), other countries? To answer these questions, because fi rms enter into exporting activities the analyses in this section rely on the World throughout the calendar year— that is, the Bank’s Exporter Dynamics Database.9 value of exports reported by new entrants in To facilitate the descriptive international 2005 was accumulated over several months, comparisons with LAC countries, Fer- whereas their annual exports reported in nandes, Lederman, and Gutierrez- Rocha 2006 cover all 12 months of the year. Ber- (2013) selected comparators for three types nard and others (2007) label this bias the of economies: natural resource economies “partial-year effect.” To account for this bias, (including Chile, Colombia, Peru, and Costa it is prudent to adjust the observed growth Rica); simple processing or assembly econo- rates of the 2005 entrant cohorts by subtract- mies (including Guatemala, the Dominican ing 30 percent from the reported growth rates Republic, El Salvador, and Nicaragua); and for 2005– 06.8 Even after this adjustment, in economies with a broad export manufac- every country, the 2005– 06 growth rates of turing base (including Brazil and Mexico).10 the entrants cohort that survived until 2009 The natural resource countries were chosen are much higher than the growth rates of the based on net exports of natural resources incumbent exporters. during 1980–2005 (see Lederman and Malo- Growth rates of the surviving 2005 cohort ney 2012).11 The simple processing countries tend to be higher than those of incumbents in have large shares of exports of apparel and subsequent years as well, although not for all textiles in their total exports, according to countries. In Brazil, incumbents’ exports out- the database of the World Integrated Trade paced exports by the 2005 cohort in 2006/07 Solution (WITS)/United Nations Commod- and 2007/08 but not during the global ity Trade Statistics (COMTRADE).12 Coun- financial crisis of 2008/09. In El Salvador tries with a broad manufacturing export base and Peru, incumbents’ exports grew more had large shares of manufacturing exports in rapidly than exports by the 2005 cohort merchandise exports in 1990–2010, accord- in some years. Only in these two countries ing to the World Development Indicators. In did exports of incumbents decline less than addition to comparator countries in each of exports of the 2005 cohort during 2008/09. these groups, the figures in the next section We return to the role played by export entre- include data for “LAC countries” (the aver- preneurship of incumbent exporters in the age across the region), “World higher” (the following section, which provides evidence average across higher-income countries in the that incumbents tried to cushion the blow of Exporter Dynamics Database), and “World the decline in foreign demand in 2008/09 by lower” (the average across lower- income introducing new products and attempting to countries in the database). export to new foreign market destinations. Entrepreneurship and export growth Descriptive benchmarking during good times and bad of export entrepreneurship Figure 4.4 shows the average annual growth The evidence from the literature and from rate of total exports for each country in the new data compiled for this report sug- 2005– 07 (the steady growth period) and gest that export entrepreneurship is rare but 2008– 09 (the global crisis). It presents the EXPORT ENTREPRENEURSHIP 103 FIGURE 4.4 Export growth and its components in selected countries, 2005– 07 and 2008– 09 a. Export growth, 2005–07 b. Export growth, 2008–09 Chile Ecuador resources Peru Costa Rica resources Natural Natural Ecuador Peru South Africa Colombia Costa Rica Chile South Africa Average Manufacturing Simple processing Nicaragua Average Manufacturing Simple processing Bangladesh Cambodia Cambodia Bangladesh Guatemala Nicaragua Dominican Republic Guatemala El Salvador Dominican Republic El Salvador Egypt, Arab Rep. Brazil Egypt, Arab Rep. Mexico Brazil Mexico countries of LAC countries LAC countries of LAC LAC countries 0 0.1 0.2 0.3 0.4 –0.2 –0.1 0 0.1 0.2 0.3 Export growth rate Export growth rate LAC countries Other countries c. Decomposition, 2005–07 d. Decomposition, 2008–09 Chile Ecuador resources Costa Rica Natural Peru resources Natural Ecuador Peru South Africa Colombia Costa Rica Chile South Africa Average Manufacturing Simple processing Nicaragua Average Manufacturing Simple processing Bangladesh Cambodia Cambodia Bangladesh Guatemala Nicaragua Dominican Republic Guatemala El Salvador Dominican Republic El Salvador Egypt, Arab Rep. Brazil Egypt, Arab Rep. Mexico Brazil Mexico countries of LAC countries LAC countries of LAC LAC countries –0.1 0 0.1 0.2 0.3 0.4 –0.2 –0.1 0 0.1 0.2 0.3 Export growth rate Export growth rate Incumbents Exiters Entrants Source: Fernandes, Lederman, and Gutierrez-Rocha 2013, based on data from the World Bank Exporter Dynamics Database. Note: Data for Ecuador in panel a are for 2006– 07. Colombia is not included in panels a and c because data were not available. LAC = Latin America and the Caribbean. average contribution of the three terms in the explaining export growth in all LAC coun- export growth decomposition (new, incum- tries as well as comparator countries in bent, and exiting exporters). natural resources, processing, and broad man- During the steady growth period, incum- ufacturing export base countries. This fi nd- bent exporters played the dominant role in ing confi rms the evidence from high-income 104 LATIN AMERICAN ENTREPRENEURS countries (such as the United States) reported in practice relative to the uncertainty about by Bernard and others (2007). New export- the sales gains from export success. ers contributed very little to export growth in Chile, Peru, and Costa Rica; they played a Exports of new products nonnegligible role in the Dominican Repub- lic and, to a lesser extent, in Brazil, Ecuador, Export entrepreneurship is present when Guatemala, and Nicaragua in LAC as well as incumbent exporters export new products; in Bangladesh and Cambodia (the compara- it is similar to the innovation of introduc- tors for processing countries). Exiting export- ing new products discussed in chapter 3. To ers reduced total export growth in Costa Rica examine this dimension of innovative entre- and Ecuador, as well as in the Arab Republic preneurship, we focus on incumbent export- of Egypt (the comparator for broad manufac- ers in each country and consider products turing export base countries). Across country defi ned at the six- digit level of the Harmo- types, the contribution of new exporters was nized System (HS) of trade classification.13 more important in LAC countries experienc- Figure 4.5 presents product entry rates for ing moderate export growth (Costa Rica, the incumbent exporters in each country during Dominican Republic) than in LAC countries 2005– 07 and 2008– 09.14 Incumbent export- experiencing fast export growth (Chile, Peru) ers displayed a tremendous degree of experi- between 2005 and 2007. mentation along the product dimension Average export growth rates were lower during the steady growth period: on average during the global recession of 2008– 09. more than a third of the products exported by They were negative for the most developed incumbents in a given year were not exported LAC countries— Brazil, Chile, Colombia, the previous year. Within LAC, the rates Costa Rica, Mexico, and Peru. Incum- of new product introduction by incumbent bent exporters played a dominant role in exporters were somewhat higher in process- explaining the export decline in those coun- ing countries, with the Dominican Repub- tries (except Costa Rica) and in increasing lic exhibiting the highest rate (42 percent). exports in Ecuador and Nicaragua. Exiting Within the group of natural resource coun- exporters contributed significantly to the tries, incumbent exporters in LAC exhibited export decline in Brazil and Costa Rica and product entry rates that were more than 20 to reduced export growth in the Dominican percentage points lower than the rates for Republic, Guatemala, and Nicaragua during incumbent exporters in South Africa. The the crisis. global recession did not reduce export entre- This evidence on LAC countries can be preneurship by incumbent exporters along interpreted in the light of the literature’s the product dimension in the LAC region as a focus on the fi xed costs of exporting. High whole, actually increasing in a few countries entry rates are expected when either fixed (Costa Rica, the Dominican Republic, and costs are low or uncertainty is high. Because Guatemala). we know from other sources that entry costs To contrast with the patterns based on can be high, the high exit rates offer another product entry rates, we present product exit piece of the puzzle. Exit rates are likely to be rates for incumbent exporters in each coun- high because weaker (possibly less produc- try (panels c and d of figure 4.5). During the tive) fi rms enter when entry costs are low or steady growth period, product exit rates were when the probability of a large payoff is high. high in all LAC countries: on average 29 per- What the evidence for the LAC region sug- cent of the products exported by incumbents gests is that the sunk costs of entering export in a given year were dropped by the following markets—which play such a crucial role in year. Within LAC, product exit rates were the models of heterogeneous firms and trade similar across groups of countries. Among pioneered by Eaton and Kortum (2002) and natural resource countries, product exit Melitz (2003)— do not seem to be very large rates of incumbent exporters were more than EXPORT ENTREPRENEURSHIP 105 FIGURE 4.5 Product entry, exit, and first-year survival rates of incumbent exporters in selected countries, 2005– 07 and 2008– 09 a. Entry averages, 2005–07 c. Exit averages, 2005–07 e. Survival averages, 2005–07 South Africa South Africa Chile resources Peru Peru Peru Natural Costa Rica Costa Rica Ecuador Chile Chile Costa Rica Ecuador Ecuador South Africa Dominican Republic Cambodia Guatemala Average Manufacturing Simple processing El Salvador El Salvador Bangladesh Cambodia Bangladesh Nicaragua Nicaragua Dominican Republic El Salvador Bangladesh Nicaragua Dominican Republic Guatemala Guatemala Cambodia Mexico Mexico Mexico countries of LAC LAC countries LAC countries LAC countries 0 0.2 0.4 0.6 0 0.1 0.2 0.3 0.4 0.5 0 0.1 0.2 0.3 Product entry rate Product exit rate Survival rate of new products b. Entry averages, 2008–09 d. Exit averages, 2008–09 f. Survival averages, 2008–09 South Africa South Africa Colombia resources Costa Rica Costa Rica Peru Natural Peru Peru Ecuador Chile Ecuador Chile Ecuador Chile Costa Rica Colombia Colombia South Africa Dominican Republic Cambodia Cambodia Average Manufacturing Simple processing Bangladesh Dominican Republic Guatemala Cambodia Bangladesh El Salvador El Salvador Nicaragua Bangladesh Nicaragua El Salvador Nicaragua Guatemala Guatemala Dominican Republic Mexico Mexico Mexico countries of LAC LAC countries LAC countries LAC countries 0 0.2 0.4 0.6 0 0.1 0.2 0.3 0.4 0.5 0 0.1 0.2 0.3 Product entry rate Product exit rate Survival rate of new products LAC countries Other countries Source: Fernandes, Lederman, and Gutierrez-Rocha 2013, based on data from the World Bank Exporter Dynamics Database. Note: Data for Ecuador in panels a, c, and e are for 2006– 07. Colombia is not included in panels a, c, and e because data were not available. Brazil is not included because of lack of exporter-level customs data. LAC = Latin America and the Caribbean. 106 LATIN AMERICAN ENTREPRENEURS 20 percentage points lower in LAC than in Figure 4.6 shows the average export value South Africa. The crisis increased the prod- of new products relative to incumbent prod- uct exit rates of incumbent exporters sub- ucts for incumbent exporters. Exports of new stantially only in the Dominican Republic; products were very small, ranging from less it caused a moderate increase in Costa Rica, than 2 percent of incumbent product exports Mexico, and Nicaragua. in Chile and Peru to 7.3 percent in Guatemala The average survival rate during the during the steady growth period. These differ- steady growth period indicates tremendous ences may be linked to the level of maturity attrition: more than 70 percent of the new and experience as an exporting country, which products incumbent exporters started to is much higher in Chile than in Guatemala. export in a given year were not exported the The value of new products relative to incum- following year. Incumbent exporters in LAC bent products increased in the crisis period in natural resource countries exhibited substan- Chile, Costa Rica, El Salvador, Mexico, and tially higher new product survival rates than Peru. In these countries, incumbent exporters incumbent exporters in comparator South started to export new products at a relatively Africa, however. Among processing coun- larger scale during the global recession. tries, Guatemala exhibited the highest new Figure 4.7 shows the average annual product survival rate. The global recession growth rate in total exports of incumbent reduced the survival rates of new products exporters. It shows the average contribution of incumbent exporters substantially in the of the three terms in the incumbent export- Dominican Republic and moderately in other ers’ export growth decomposition (new, LAC countries. incumbent, and exiting products). FIGURE 4.6 Size of new product exports relative to incumbent products in selected countries, 2005– 07 and 2008– 09 a. Averages, 2005–07 b. Averages, 2008–09 Ecuador Colombia resources South Africa Ecuador Natural resources Natural Costa Rica South Africa Peru Costa Rica Chile Peru Bangladesh Chile Average Manufacturing Simple processing Cambodia Bangladesh Average Manufacturing Simple processing Guatemala El Salvador Nicaragua Nicaragua El Salvador Guatemala Dominican Republic Cambodia Dominican Republic Mexico Mexico countries of LAC countries of LAC LAC countries LAC countries 0 0.05 0.1 0.15 0.2 0 0.02 0.04 0.06 0.08 0.1 Size of new product exports Size of new product exports relative to incumbent products relative to incumbent products LAC countries Other countries Source: Fernandes, Lederman, and Gutierrez-Rocha 2013, based on data from the World Bank Exporter Dynamics Database. Note: Data for Ecuador in panel a are for 2006– 07. Colombia is not included in panel a because data were not available. Brazil is not included because exporter-level customs data were not available. LAC = Latin America and the Caribbean. EXPORT ENTREPRENEURSHIP 107 FIGURE 4.7 Export growth of incumbent exporters in selected countries and its decomposition along the product dimension, 2005– 07 and 2008– 09 a. Export growth, 2005–07 b. Export growth, 2008–09 Peru Ecuador resources Chile Costa Rica Natural resources Natural Ecuador Peru South Africa Colombia Costa Rica Chile Nicaragua South Africa Average Manufacturing Simple processing Bangladesh Bangladesh Average Manufacturing Simple processing Guatemala Cambodia Cambodia Nicaragua Dominican Republic Dominican Republic El Salvador Guatemala El Salvador Mexico Mexico countries of LAC countries of LAC LAC countries LAC countries 0 0.1 0.2 0.3 –0.2 –0.1 0 0.1 0.2 Export growth rate Export growth rate LAC countries Other countries c. Decomposition, 2005–07 d. Decomposition, 2008–09 Peru Ecuador resources Chile Costa Rica Natural resources Natural Ecuador Peru South Africa Colombia Costa Rica Chile Nicaragua South Africa Manufacturing Simple processing Bangladesh Bangladesh Average Manufacturing Simple processing Guatemala Cambodia Cambodia Nicaragua Dominican Republic Dominican Republic El Salvador Guatemala El Salvador Mexico Mexico countries Average of LAC countries of LAC LAC countries LAC countries –0.1 0 0.1 0.2 0.3 –0.2 –0.1 0 0.1 0.2 0.3 Export growth rate Export growth rate Incumbent destinations New destinations Exiting destinations Source: Fernandes, Lederman, and Gutierrez-Rocha 2013, based on data from the World Bank Exporter Dynamics Database. Note: Data for Ecuador in panel a are for 2006– 07. Colombia is not included in panels a and c because data were not available. LAC = Latin America and the Caribbean. 108 LATIN AMERICAN ENTREPRENEURS During the steady growth period, incum- destination dimension during the steady bent products contributed the largest share to growth period: on average a quarter of the the growth of incumbent exporters in every destinations served by incumbents in a given LAC country. New products represented a year were not served the previous year (fig- significant share of export growth of incum- ure 4.8).15 Mexico exhibited the lowest des- bent exporters only in the Dominican Repub- tination entry rate by incumbent exporters; lic, El Salvador, Guatemala, and Nicaragua. entry rates did not differ much across natural One possible explanation for the importance resource and processing countries in LAC. of new products in these countries is the However, among both natural resource and entry into force of the Central America Free processing countries, incumbent exporters Trade Agreement (CAFTA) with the United in LAC exhibited much lower destination States, which granted incumbent exporters entry rates than incumbent exporters in the access to a very large market. In El Salvador, comparator countries. The global recession the reduction in exports because of products did not reduce export entrepreneurship by dropped by incumbent exporters more than incumbent exporters along the destination compensated for the increase in exports of dimension in LAC; in the case of the Domini- new products. The importance of new prod- can Republic, it increased it substantially. ucts in explaining export growth of incum- During the steady growth period, 17 per- bent exporters was smaller in LAC than in its cent of the destinations served by incumbents comparators in natural resource and process- in LAC in a given year were dropped by the ing countries. following year. Because product exit rates During the global recession, exports from were lower than entry rates, there was posi- incumbent fi rms declined in the most devel- tive net entry into new destinations between oped LAC countries— Chile, Colombia, 2005 and 2007. As was the case for entry Costa Rica, Mexico, and Peru— as well as in rates, Mexico exhibited the lowest destina- South Africa; exports of incumbents grew in tion exit rate. Within natural resource and the LAC processing countries (as well as their processing countries, exit rates were simi- comparators) and in Ecuador. Incumbent lar across LAC countries. However, among products explained most of the export decline both natural resource and processing coun- for incumbent exporters in Chile, Colombia, tries, incumbent exporters in LAC countries Mexico, and Peru and most of the export exhibited much lower destination exit rates growth for incumbent exporters in Ecuador than did incumbent exporters in comparator and Nicaragua. During the crisis, new prod- countries. The crisis did not alter destination ucts became the major contributor to export exit rates by incumbent exporters in LAC, growth of incumbent exporters in the Domin- except in the Dominican Republic, where ican Republic, El Salvador, and Guatemala, exit rates increased substantially. and they accounted for a large share in Nica- The average survival rate during the ragua and Ecuador. Participation in CAFTA steady growth period indicated a high degree may have partially insulated the LAC process- of attrition. More than 60 percent of the new ing countries’ incumbent exporters from the destinations served by incumbent export- crisis (despite the decline in U.S. demand) by ers in a given year were no longer served the fostering entrepreneurship through the intro- next year. Within LAC, natural resource duction of new products. This pattern of the countries exhibited slightly higher survival crisis fostering entrepreneurship was also evi- rates than other countries. Among natural dent in Ecuador. resource countries, incumbent exporters in LAC exhibited substantially higher survival rates than incumbent exporters in South Exports to new destinations Africa. Among processing countries, all LAC Incumbent exporters in LAC engaged in a countries exhibited lower survival rates than high degree of experimentation along the Bangladesh but higher survival rates than EXPORT ENTREPRENEURSHIP 109 FIGURE 4.8 Destination entry, exit, and first-year survival rates of incumbent exporters in selected countries, 2005– 07 and 2008– 09 a. Entry averages, 2005–07 c. Exit averages, 2005–07 e. Survival averages, 2005–07 South Africa South Africa Chile resources Chile Chile Costa Rica Natural Peru Peru Peru Ecuador Costa Rica Ecuador Costa Rica Ecuador South Africa Cambodia Cambodia Bangladesh Average Manufacturing Simple processing Bangladesh Bangladesh Guatemala Dominican Republic Dominican Republic El Salvador Nicaragua El Salvador Nicaragua El Salvador Nicaragua Dominican Republic Guatemala Guatemala Cambodia Egypt, Arab Rep. Egypt, Arab Rep. Mexico Mexico Mexico Egypt, Arab Rep. countries of LAC LAC countries LAC countries LAC countries 0 0.1 0.2 0.3 0.4 0 0.1 0.2 0.3 0 0.1 0.2 0.3 0.4 Destination entry rate Destination exit rate Survival rate of new destinations b. Entry averages, 2008–09 d. Exit averages, 2008–09 f. Survival averages, 2008–09 South Africa South Africa Colombia resources Chile Costa Rica Costa Rica Natural Ecuador Chile Chile Costa Rica Peru Ecuador Peru Ecuador Peru Colombia Colombia South Africa Dominican Republic Cambodia Bangladesh Average Manufacturing Simple processing Cambodia Bangladesh Cambodia Bangladesh Dominican Republic Guatemala El Salvador Guatemala El Salvador Nicaragua El Salvador Nicaragua Guatemala Nicaragua Dominican Republic Egypt, Arab Rep. Egypt, Arab Rep. Mexico Mexico Mexico Egypt, Arab Rep. countries of LAC LAC countries LAC countries LAC countries 0 0.1 0.2 0.3 0.4 0 0.1 0.2 0.3 0 0.1 0.2 0.3 0.4 Destination entry rate Destination exit rate Survival rate of new destinations LAC countries Other countries Source: Fernandes, Lederman, and Gutierrez-Rocha 2013, based on data from the World Bank Exporter Dynamics Database. Note: Data for Ecuador in panel a are for 2006– 07. Colombia is not included in panels a, c, and e because data were not available. Brazil is not included because exporter-level customs data were not available. LAC = Latin America and the Caribbean. 110 LATIN AMERICAN ENTREPRENEURS Cambodia. The global recession was associ- export growth decomposition (new, incum- ated with a slight reduction in first-year sur- bent, and exiting destinations). During the vival rates of new destinations of incumbent steady growth period, incumbent destina- exporters across LAC. tions accounted for the largest share of the Exports to new destinations were gener- growth of incumbent exporters in every LAC ally small, ranging from less than 5 percent country. New destinations contributed a sig- of exports to incumbent destinations in Chile nificant share to export growth of incumbent to 12 percent in Ecuador (figure 4.9). As in exporters in the Dominican Republic, Ecua- the case of exports of new products, we can dor, El Salvador, Guatemala, and Nicaragua. only speculate that the differences between New destinations contributed minimally to Chile and Ecuador reflect Chile’s longer time annual export growth in the LAC countries under an open trade regime, which may have whose incumbent exporters experienced the fostered outward- oriented firms with lon- fastest growth (Chile and Peru). In Costa ger exporting experience. During the crisis, Rica, El Salvador, and Peru the reduction the value of exports to new destinations by in exports due to destinations dropped by incumbent exporters to exports to incumbent incumbent exporters almost compensated destinations increased in LAC as a whole and for the increase in exports due to their new in most individual countries. destinations. Among incumbent exporters in Figure 4.10 shows the average contribution LAC, new destinations were less important of the three terms in the incumbent exporters determinants of export growth in processing FIGURE 4.9 Exports to new destinations as a share of total exports by incumbent exporters in selected countries, 2005– 07 and 2008– 09 a. Averages, 2005–07 b. Averages, 2008–09 Ecuador Ecuador resources Peru Peru Natural resources Natural South Africa South Africa Costa Rica Colombia Chile Costa Rica Bangladesh Chile Average Manufacturing Simple processing Nicaragua Guatemala Average Manufacturing Simple processing Guatemala Bangladesh El Salvador Nicaragua Dominican Republic El Salvador Cambodia Dominican Republic Cambodia Egypt, Arab Rep. Mexico Egypt, Arab Rep. Mexico countries of LAC countries of LAC LAC countries LAC countries 0 0.1 0.2 0.3 0 0.05 0.10 0.15 Size of exports to new destinations Size of exports to new destinations relative to incumbent destinations relative to incumbent destinations LAC countries Other countries Source: Fernandes, Lederman, and Gutierrez-Rocha 2013, based on data from the World Bank Exporter Dynamics Database. Note: Data for Ecuador in panel a are for 2006– 07. Colombia is not included in panel a because data were not available. Brazil is not included because exporter-level customs data were not available. LAC = Latin America and the Caribbean. EXPORT ENTREPRENEURSHIP 111 FIGURE 4.10 Export growth of incumbent exporters in selected countries and its decomposition along the destination dimension, 2005– 07 and 2008– 09 a. Export growth, 2005–07 b. Export growth, 2008–09 Peru Ecuador resources Chile Costa Rica Natural resources Natural Ecuador Peru South Africa Colombia Costa Rica Chile Nicaragua South Africa Average Manufacturing Simple processing Bangladesh Bangladesh Average Manufacturing Simple processing Guatemala Cambodia Cambodia Nicaragua Dominican Republic Dominican Republic El Salvador Guatemala El Salvador Egypt, Arab Rep. Mexico Egypt, Arab Rep. Mexico countries of LAC countries of LAC LAC countries LAC countries 0 0.1 0.2 0.3 –0.2 –0.1 0 0.1 0.2 Export growth rate Export growth rate LAC countries Other countries c. Decomposition, 2005–07 d. Decomposition, 2008–09 Peru Ecuador resources Chile Costa Rica Natural resources Natural Ecuador Peru South Africa Colombia Costa Rica Chile Nicaragua South Africa Average Manufacturing Simple processing Bangladesh Bangladesh Average Manufacturing Simple processing Guatemala Cambodia Cambodia Nicaragua Dominican Republic Dominican Republic El Salvador Guatemala El Salvador Egypt, Arab Rep. Mexico Egypt, Arab Rep. Mexico countries of LAC countries of LAC LAC countries LAC countries –0.1 0 0.1 0.2 0.3 0.4 –0.2 –0.1 0 0.1 0.2 03 Export growth rate Export growth rate Incumbent destinations New destinations Exiting destinations Source: Fernandes, Lederman, and Gutierrez-Rocha 2013, based on data from the World Bank Exporter Dynamics Database. Note: Data for Ecuador in panel a are for 2006– 07. Colombia is not included in panels a, c, and d because data were not available. LAC = Latin America and the Caribbean. 112 LATIN AMERICAN ENTREPRENEURS countries than in comparator Bangladesh, Export entrepreneurship and they were less important to countries across industries with broad manufacturing export bases than Figure 4.11 presents annual export entry rates in comparator Egypt. for 15 industries. Minerals and base metals During the global recession, exports of appear in the middle of the pack. However, incumbent exporters declined in the most they are outperformed by some eye-catching developed LAC countries but increased in the manufacturing industries (such as transport processing countries and Ecuador. Incum- vehicles) that have been central for the indus- bent destinations accounted for most of the trial resurgence of certain countries, such as export decline of incumbent exporters in Mexico. Industries related to agriculture— Chile, Colombia, and Mexico. New desti- vegetable products and oils; food, beverages, nations played a dominant role in boosting and tobacco; and live animals—tend to have the exports of incumbent exporters in the Dominican Republic, El Salvador, Guate- mala, and Nicaragua, and they played as FIGURE 4.11 Conditional benchmarking of important a role as incumbent destinations export entry rates by sector, 2005– 09 in increasing exports of incumbent exporters in Ecuador. For the LAC processing countries Food, beverages, and Ecuador, the crisis was associated with and tobacco an increase in export entrepreneurship by Live animals incumbent exporters through the exploration Vegetable products of new export destinations. and oils Chemicals Econometric benchmarking Paper of export entrepreneurship Minerals The descriptive benchmarking suggests that Plastics both new and incumbent exporters in LAC engaged in export entrepreneurship during Textiles (including leather) the crisis years. It is difficult to derive fi rm Base metals conclusions about LAC’s relative standing from descriptive statistics, however, because Others year, country, and industry characteristics Wood and articles may jointly affect entrepreneurial outcomes. thereof Cement, ceramics, For example, the finding that Chile and Peru and glass had both the highest export growth rates and Transportation the smallest number of new exporters dur- vehicles ing 2005– 09 does not necessarily mean that Electrical machinery both solely reflect these economies’ charac- Mechanical teristics (such as relatively open and mining- machinery dependent economies). –0.15 –0.10 –0.05 0 0.05 0.10 Entry rates by sector This section presents the results of a sec- relative to benchmark ond set of benchmarking exercises that rely Natural resource Industry on an econometric decomposition of the sources of international differences in export Source: Estimations based on data from the World Bank Exporter Dynam- entrepreneurship indicators. This method- ics Database. Note: Figure shows estimates of each sector’s dummy variable coefficient ology entails the estimation of industry, from an econometric model that also includes country and year dummies. country, and year effects on the export entre- Industries are defined at the two- digit level of the Harmonized System. The excluded benchmark industry is apparel and footwear. The vertical preneurship indicators observed at the fi rm axis measures the probability of observing export-firm entry in each sec- level (full results are available upon request). tor relative to apparel and footwear in percentage points. EXPORT ENTREPRENEURSHIP 113 relatively lower entry rates. The data thus do Export entrepreneurship not support the notion that entrepreneurship across countries is relatively weak in mining; to some extent, We now turn to the conditional benchmark- they support the notion in agriculture. ing of LAC countries. Figures 4.13 and 4.14 Figure 4.12 shows the industries’ relative present the results for two export entrepre- standing in terms of average survival rates neurship indicators, entry and survival rates. one year after entry into exporting. The cor- (Results for exit rates and size of exports at relation with the rankings for entry rates is entry are available upon request.) – 0.977: industries with lower entry rates tend to have higher export survival rates. Mining does not stand out, but agriculture exhibits higher survival rates. FIGURE 4.13 Conditional benchmarking of export entry rates in selected countries, 2005– 09 FIGURE 4.12 Conditional benchmarking of one- Chile year export survival rates by sector, 2005– 09 El Salvador Costa Rica Colombia Food, beverages, and tobacco Bulgaria Mexico Live animals Guatemala Vegetable products Macedonia, FYR and oils Peru Chemicals Ecuador Jordan Paper Morocco Minerals Mauritius South Africa Plastics Senegal Textiles Mali (including leather) Dominican Republic Base metals Nicaragua Bangladesh Others Kenya Wood and articles Burkina Faso thereof Tanzania Cement, ceramics, Pakistan and glass Cameroon Transportation vehicles Niger Electrical Cambodia machinery Malawi Mechanical Iran, Islamic Rep. machinery Uganda –0.06 –0.04 –0.02 0 0.02 0.04 0.06 0.08 –20 –15 –10 –5 0 5 10 15 20 Average 1-year survival probabilities by sector relative to benchmark Marginal effect Natural resource Industry LAC countries Other countries Source: Estimations based on data from the World Bank Exporter Dynam- Source: Estimations based on data from the World Bank Exporter Dynam- ics Database. ics Database. Note: Figure shows estimates of each sector’s dummy variable coefficient Note: Figure shows estimates of each country’s dummy variable coef- from an econometric model that also includes country and year dummies. ficient from an econometric model that also includes industry and year Industries are defined at the two- digit level of the Harmonized System. dummies. Industries are defined at the two- digit level of the Harmonized The excluded benchmark industry is apparel and footwear. The vertical System. The excluded benchmark country is Albania. The vertical axis axis measures the probability of export-firm survival in each sector relative measures the probability of export-firm survival in each country relative to apparel and footwear in percentage points. to Albania in percentage points. LAC = Latin America and the Caribbean. 114 LATIN AMERICAN ENTREPRENEURS FIGURE 4.14 Conditional benchmarking of one- Islamic Republic of Iran and Uganda) are on year export survival rate in selected countries, the right. As the descriptive benchmarking 2005– 09 suggested, entry rates may be higher in more difficult business environments. Like the sub- Chile sistence entrepreneurs discussed in chapter 1, El Salvador entrepreneurs operating in difficult environ- Costa Rica ments may be more likely to take risks and Colombia to enter export markets precisely because Bulgaria domestic profits may be relatively low. Mexico The negative correlation between entry Guatemala Macedonia, FYR rates and survival probabilities (– 0.847) is Peru less strong than the correlation for the indus- Ecuador try benchmarking (compare figures 4.13 Jordan and 4.14). In addition, relatively successful Morocco LAC exporters during 2005– 09 tend to have Mauritius relatively low (conditional) entry rates and South Africa relatively high (conditional) survival rates. Senegal Nicaragua appears to be an outlier in this Mali context: it had the highest conditional entry Dominican Republic Nicaragua rate in the sample of LAC countries and a Bangladesh high survival rate for new exporters. Kenya The higher survival probabilities of the Burkina Faso LAC countries in this sample are explained Tanzania by their higher levels of development. Fig- Pakistan ure 4.15 presents the country-specific export Cameroon entry and survival rates after controlling for Niger the level of development of each country and Cambodia comparative advantage. The evidence is clear: Malawi Iran, Islamic Rep. after taking into account the level of develop- Uganda ment, LAC countries underperform in terms –8 –6 –4 –2 0 2 4 6 8 10 12 of export entry rates and are not overachiev- Marginal effect ers in terms of survival rates. (Comparative LAC countries Other countries advantage did not affect the ranking of coun- tries, however.) In fact, after controlling for Source: Estimations based on data from the World Bank Exporter Dynam- GDP per capita, only the Islamic Republic of ics Database. Iran has lower conditional survival rates than Note: Figure shows estimates of each country’s dummy variable coef- ficient from an econometric model that also includes industry and year the LAC countries in the sample. dummies. Industries are defined at the two- digit level of the Harmonized The benchmarking exercises reveal impor- System. The excluded benchmark country is Albania. The vertical axis measures the probability of export-firm survival in each country relative to tant findings about the nature of export entre- Albania in percentage points. LAC = Latin America and the Caribbean. preneurship. Economies that enjoyed relative high export growth rates also tended to have relatively low export entry rates, with entrants Most LAC economies appear on the left coming in at smaller sizes (compared with side (in red) of the rankings of entry rates in incumbent exporters), but their survival rates figure 4.13 (indicating poor performance). tended to be higher. Survival thus appears to Some of the best-performing countries in be the dominant variable underpinning export terms of export growth (including Chile, growth; it may also reflect adequate business Costa Rica, Colombia, and Guatemala) are environments. However, as emphasized ear- on the left, whereas economies with relatively lier, when the going gets tough, as it did dur- harsh business environments (such as the ing the global crisis of 2008– 09, incumbent EXPORT ENTREPRENEURSHIP 115 FIGURE 4.15 Conditional benchmarking of The role of comparative advantage export entry and one-year survival rate in selected countries after controlling for GDP per capita and To assess the role of comparative advantage, comparative advantage, 2005– 09 Ana M. Fernandes and Daniel Lederman (World Bank) estimated the models of export entrepreneurship used for the conditional Chile El Salvador benchmarking by adding an indicator of Costa Rica revealed comparative advantage (RCA) pro- Colombia posed by Vollrath (1991) as well as the level Bulgaria of development. This indicator takes into Mexico account the structure of a country’s trade on Guatemala Macedonia, FYR both the export and import sides and data Peru at the HS six-digit level. In addition to coun- Ecuador try fi xed effects, broad industry fixed effects Jordan (defi ned at the two-digit level aggregation), Morocco and year effects, the regressions include the Mauritius South Africa Vollrath index of RCA. The results for this Dominican Republic variable are shown in figure 4.16. Mali Consistent with the previous fi ndings on Senegal the quality of the business environment and Nicaragua global economic conditions, the RCA has a Bangladesh Kenya negative partial correlation with a country’s Burkina Faso Tanzania Pakistan FIGURE 4.16 Partial effects of 1 percent increase in Cameroon index of revealed comparative advantage on export Iran, Islamic Rep. entrepreneurship indicators in seven countries in Cambodia Latin America and the Caribbean, 2005–09 Malawi Niger 5 Uganda 3.4 2.8 Percentage point increase –0.3 –0.2 –0.1 0 0.1 0.2 0.3 0.4 0 Log of GDP per capita Average entry rate Average 1-yr survival rate of new entrants –5 –6.0 Source: Estimations based on data from the World Bank Exporter Dynam- ics Database. –10 Note: Figure shows estimates of each country’s dummy variable coefficient from an econometric model that also includes (the log of) gross domestic product (GDP) per capita (adjusted for purchasing power parity), the –15 –14.0 Vollrath (1991) index of revealed comparative advantage at the six- digit level of the Harmonized System classification, industry dummies, and Average Average Average Average year dummies. Industry dummies are defined at the two- digit level. The entry exit rate 1-yr survival size of excluded benchmark country is Albania. The vertical axis measures the rate rate of new new probability of export-firm survival in each country relative to Albania in entrants entrants percentage points. yr = year. Source: Estimations based on data from the World Bank Exporter Dynam- ics Database. exporters become more entrepreneurial by Note: Figure shows the estimated coefficients on Vollrath’s (1991) index of revealed comparative advantage (RCA). The other variables included in the seeking new products and, to a lesser extent, econometric estimations are (the log of) gross domestic product (GDP) per new export destinations. In addition, analyz- capita (adjusted for purchasing power parity) and sector, country, and year dummies. Industries were defined at the two-digit level of the Harmonized ing the role of comparative advantage as a System; the RCA indexes were computed at the six-digit level. Differences in determinant of export entrepreneurship out- the magnitudes of the effects across the four indicators of export entrepre- neurship reflect differences in the units of measurement: the average size of comes might shed further light on the whether new entrants is measured as the ratio of average exports of new entrants over necessity is the mother of innovation. average exports of the average incumbent exporter in the sector. yr = year. 116 LATIN AMERICAN ENTREPRENEURS average export entry rate— that is, when that export promotion does little to explain countries offer favorable conditions, such as export growth by increasing fi rm entry into endowments or other policy-driven factors, export activities. Bernard and Jensen (2004) for a given product, entry falls, exit and the find that export promotion across states probability of survival rise, and the average has no statistically significant impact on the size of new entrants relative to incumbents probability of exporting in a sample of U.S. falls (by 14 percentage points of the average manufacturing firms. Görg, Henry, and value of exports of incumbent fi rms). Over- Strobl (2008) find that export promotion all, the data speak loud and clear: the key grants offered to Irish manufacturing fi rms for success in export markets is entry com- had no impact on the probability of export- bined with survival, which tends to increase ing but did affect the level of exports. Our when conditions are favorable. LAC appears results, based on different data, contradict to underperform poorer economies in terms these findings. of entry, and it does not overachieve in terms There is also a growing body of litera- of survival (after controlling for its level of ture on export promotion and its impact on development). The question about the impact exporting firms’ intensive and extensive mar- of export promotion policies and how they gins (where the extensive margin is defi ned affect the various dimensions of export entre- either as the introduction of new products or preneurship remains. entry into new export destinations). Based on a sample of Peruvian firms, Volpe and Carballo (2008) conclude that export pro- Export promotion policies motion affects exports mainly along the Through what mechanisms do export promo- extensive margin, in terms of both markets tion services affect export growth? Do such and products; it has little impact on intensive mechanisms promote firm entry and survival margins. Using product-level data, Volpe, in exporting activities? Do they help incum- Carballo, and Gallo (2011) confirm this bents by increasing the share of exports in finding in a sample of LAC countries for the total sales? Or do they operate through both period 1995–2004. channels? The finding of Volpe and Carballo (2008) The answers to these questions can shed and Volpe, Carballo, and Gallo (2011) that light on the social desirability of export pro- export promotion works mainly through the motion programs. Indeed, the economic jus- extensive margins of products and markets tification for export promotion is often based does not necessarily contradict the findings of on the existence of asymmetric information Bernard and Jensen (2004) and Görg, Henry, and other externalities associated with the and Strobl (2008) that export promotion has collection of information on market condi- no impact on the probability of a firm export- tions and business opportunities in inter- ing. Volpe and Carballo (2008) used a sample national markets (Hausmann and Rodrik of exporting fi rms, and the results reported 2003). Private firms have no incentives to by Volpe, Carballo, and Gallo (2011) are not share this information with potential compet- based on fi rm-level data. Thus, they cannot itors after incurring the costs of discovering address the question of whether export pro- how to export profitably. This market failure motion raises the probability of firms becom- justifies government intervention. Given the ing exporters. nature of the market failure, such interven- None of these studies distinguishes the tions should affect fi rms’ extensive margins impact of export promotion on entry into (that is, the decision to enter and survive in export markets from its impact on survival in export markets), not their intensive margins exporting activities. The distinction is impor- (that is, the decision on how much to export). tant given the large number of firms that The literature on export promotion and enter and exit export activities after one year, fi rm entry is thin, but the evidence suggests as discussed earlier. EXPORT ENTREPRENEURSHIP 117 To identify the impact of export promo- propensity matching to control for the fact tion activities on fi rm entry, exit, survival, that export promotion services are not ran- and export intensity, Lederman, Olarreaga, domly allocated across firms. Overall, the and Zavala (2013) used firm surveys from results suggest that entry and survival mar- seven Latin American countries from 2006 gins are the main channels through which and 2010. They estimated a multinomial logit export promotion agencies affect export model to explain the probability of observing growth and that they tend to be unsuccessful four potential paths of the status of a fi rm: at increasing export intensity, thus highlight- from nonexporting to exporting (entry), con- ing the role of fixed costs of entry in export- tinuity in exporting (survival), from export- ing activities. ing to nonexporting (exit), and continuity of nonexport status. The variable of interest is whether the firm used the services of an EPA Concluding remarks between 2006 and 2010. The authors also There is good news on the policy front. If explored the treatment effect of EPA services the costs of entry into export markets reflect on the change in the share of exports over the need to gather information about the total sales (export intensity) within firms, characteristics of foreign markets, export which provide estimates of the effect of EPAs promotion policies that focus on providing on the intensive margin of exports. such information could both increase entry The results, some of which appear in and enhance the chances of survival of entre- table 4.3, suggest that having used export preneurs in global markets. Preliminary evi- services significantly increases the probabil- dence suggests that LAC export promotion ity of entry and survival (with respect to the policies are having exactly these effects, but probability of exiting export markets). It also the region’s superstars still appear to have decreases the probability of remaining a non- relatively low entry rates into exporting, exporter. In contrast, fi rms that use export possibly revealing an innovation gap, with services do not seem to increase their export unimpressive survival rates to boot. Chap- intensity. These results appear in the (uncon- ter 5 continues the exploration of high- end ditional) descriptive data and in the estimate entrepreneurs by examining the performance of conditional EPA treatment effects; they are of superstar firms that penetrate foreign mar- robust to the use of three different types of kets by exporting capital. TABLE 4.3 Treatment effects of export promotion agencies in seven countries in Latin America and the Caribbean Outcome Size of Size of control Average treatment Bootstrapped variable treatment group group effect on outcome standard errors t-statistic Exit 401 1,134 0.000 0.018 – 0.01 Nonexporter 401 1,134 – 0.403 0.034 –11.96*** Survival 401 1,134 0.344 0.038 9.10*** Entry 401 1,134 0.059 0.023 2.58*** ⌬exp_int 401 1,133 0.014 0.012 1.15 ⌬exp_int 265 261 0.035 0.018 1.92 Source: Lederman, Olarreaga, and Zavala 2013, based on data from World Bank Enterprise Surveys. Note: The propensity score was estimated using a logit on Size, FDI, Web, and Email. Size is the log of the firm’s full-time employment, FDI is foreign owner- ship of the firm as a share of total ownership, Web is a dummy for whether the firm has a website, and Email is a dummy for whether the firm communicates with clients via email. Exit is a dummy indicating that the firm exported in 2006 but not in 2010, Nonexporter is a dummy indicating that the firm did not export in either 2006 or 2010, Survival is a dummy indicating that the firm exported in both 2006 and 2010, Entry is a dummy indicating that the firm exported in 2010 but not in 2006, and ⌬exp_int is the change in the firm’s total exports as a fraction of total sales between 2006 and 2010. Separation of firms into treatment and control groups was done using three different matching methods: kernel, stratification (with four blocks), and nearest neighbor. This table reports on the latter. The average treatment effect is reported as the difference in means between treatment and control groups. The last row in the table corresponds to estimations with the subsample of firms that were exporters in 2006. Bootstrapped standard errors were estimated with 50 repeti- tions. *** p < 0.01. 118 LATIN AMERICAN ENTREPRENEURS Notes in developing countries. A growing body of literature uses tariff (not firm) data showing 1. This chapter draws heavily on the work of that the intensive margin (that is, exporting Daniel Lederman and his coauthors, including more of the same product) explains most Ana M. Fernandes (Development Research export growth (see Felbemayr and Kohler Group, World Bank) and Marcelo Olarreaga 2006; Helpman, Melitz, and Rubinstein (University of Geneva). 2008; and Amiti and Freund 2010). 2. The term export superstars was coined by 8. In a study commissioned for this report, Fer- Freund and Pierola (2012). nandes, Lederman, and Gutierrez-Rocha 3. The firm-level data used in Lederman (2010) (2013) show that in Peru, the probability come from numerous World Bank’s Business of entry is more or less the same across the Environment and Enterprise Performance Sur- months of the year. This lack of systematic sea- veys (BEEPS) and Investment Climate Surveys sonality in export entry implies that the first- (ICS), conducted in various countries between year exports of the new entrant cohort should 2000 and 2006. The coverage of these data in probably be multiplied by 2. The reported terms of the sampling of firms is different. The growth rates of the entry cohort for 2005–06 BEEPS use quota sampling, in which 10 per- can therefore be adjusted by subtracting 30.1 cent of selected firms are small (2–49 employ- percent (the natural logarithm of 2) from the ees), another 10 percent are large (250–999 observed rates. employees), and the rest are randomly selected 9. This section draws heavily on the study by between these two extremes. The ICS sam- Fernandes, Lederman, and Gutierrez-Rocha pling differs across countries. In some cases, (2013), which was commissioned for this quotas by sector and size are used. In others, report. existing industrial census shares by industries 10. The comparators selected are based on the and size are used as benchmark sampling quo- availability of data on export entrepreneur- tas. Thus, there may be some selectivity of the ship in the Exporter Dynamics Database. sampled firms, which may raise doubts about 11. Most LAC countries are net exporters of the randomness of the sample. energy, mining, or agriculture. Costa Rica is a 4. The Exporter Dynamics data cover all export- net exporter of various agricultural commodi- ing firms in each country—that is, the data- ties. It is also a major exporter of Intel super- base provides a census of exporters but not conductors. The data used here exclude Intel a census of all firms operating in each coun- exports, following the literature (for example, try, because it records only export trans- Lederman, Rodriguez-Clare, and Xu 2011), actions, not domestic sales. The database partly because the story of Intel is well known. is available at http://econ.worldbank.org/ Furthermore, although Brazil and Mexico are exporter-dynamics-database. also net exporters of commodities, they have 5. For this analysis, we keep in the sample only much more diversified export structures (as the LAC countries with data for all years from measured by standard indicators, such as the 2004 to 2009 so that we can define the cohort Herfindahl index of export revenue concen- of 2005 exporter entrants and follow it until tration) as well as large shares of manufac- the end of the sample period in 2009. tured exports in total merchandise exports. 6. The shares of total exports in figures 4.2 and 12. Although Mexico could be classified as a sim- 4.3 need not add up to 100 percent, because ple processing country because of the impor- the figures omit the contribution of exporters tance of the maquila sector for its economy, that began exporting between 2006 and 2009. we classify it as a country with a broad manu- 7. An important body of literature shows that facturing base. the survival of new export “relationships” is 13. Brazil, New Zealand, and Spain were not an important determinant of export growth, included in the analysis in this section because at least in developing countries. This literature we did not have the raw exporter-level cus- focuses on products at the tariff-line level. toms data for those countries necessary to Evenett and Venables (2002) and Besedes and compute the measures used here. Egypt was Prusa (2011) show that growth in the value of not used because its exporter-level customs new export products or new export markets data are provided at the four-digit (not the accounts for a large share of export growth six-digit) level. EXPORT ENTREPRENEURSHIP 119 14. The average number of HS six-digit products Research Working Paper 6007, World Bank, exported per incumbent exporter in LAC Washington, DC. countries ranged from 5.8 in Ecuador to 8.9 Cebeci, T., A. Fernandes, C. Freund, and M. in Peru. Pierola. 2012. “Exporter Dynamics Data- 15. Brazil, New Zealand, and Spain were not base.” Policy Research Working Paper 6229, included in the analysis in this section, because World Bank, Washington, DC. the raw exporter-level customs data necessary Das, S., M. J. Roberts, and J. R. Tybout. 2007. to compute the measures were not available. “Market Entry Costs, Producer Heterogeneity, The average number of destinations served and Export Dynamics.” Econometrica 75 (3): per incumbent exporter in LAC countries 837–73. ranged from 2.6 in Mexico and Nicaragua to Eaton, J., M. Eslava, M. Kugler, and J. Tybout. 4.7 in Chile. 2007. “Export Dynamics in Colombia: Firm- Level Evidence.” NBER Working Paper 13531, National Bureau of Economic Research, Cam- References bridge, MA. Alcala, F., and A. Ciccone. 2004. “Trade and Pro- Eaton, J., and S. Kortum. 2002. “Technology, ductivity.” Quarterly Journal of Economics Geography and Trade.” Econometrica 70 (5): 119 (2): 612– 45. 1741–79. Amiti, M., and C. Freund. 2010. “Anatomy of Eaton, J., S. Kortum, and F. Kramarz. 2011. “An China’s Export Growth.” In China’s Grow- Anatomy of International Trade: Evidence from ing Role in World Trade, edited by R. Feenstra French Firms.” Econometrica 79 (5): 1453–98. and S. Wei, 35–36. Cambridge, MA: National Enterprise Surveys (database). World Bank Group, Bureau of Economic Research. http://www Washington, DC. http://www.enterprise .nber.org/chapters/c10451. surveys.org/. Arkolakis, C., A. Costinot, and A. Rodriguez- Evenett, S., and A. Venables. 2002. “Export Clare. 2012. “New Trade Models, Same Old Growth in Developing Countries: Market Gains?” American Economic Review 102 (1): Entry and Bilateral Trade Flows.” University of 94–130. Bern Working Paper, Bern, Switzerland. Baldwin, R. E., and F. Robert- Nicoud. 2008. Exporter Dynamics Database. World Bank, “Trade and Growth with Heterogeneous Washington, DC. http://econ.worldbank.org Firms.” Journal of International Economics /exporter-dynamics-database. 74 (1): 21–34. Felbermayr, G., and W. Kohler. 2006. “Exploring Bernard, A. B., and J. B. Jensen. 2004. “Why the Intensive and Extensive Margins of World Some Firms Export.” Review of Economics Trade.” Review of World Economics 142: and Statistics 86 (2): 561– 69. 642–74. Bernard, A. B., J. B. Jensen, S. J. Redding, and Fernandes, A. M., D. Lederman, and M. Gutierrez- P. K. Schott. 2007. “Firms in International Rocha. 2013. “Export Entrepreneurship and Trade.” Journal of Economic Perspectives 21 Trade Structure in Latin America during Good (3): 105–30. and Bad Times.” Policy Research Working Besedes, T., and T. J. Prusa. 2011. “The Role of Paper 6413, World Bank, Washington, DC. Extensive and Intensive Margins and Export Feyrer, J. 2009. “Trade and Income: Exploiting Growth.” Journal of Development Economics Time Series in Geography.” NBER Working 96 (2): 371–79. Paper 14910, National Bureau of Economic Brambilla, I., R. Dix-Carneiro, D. Lederman, Research, Cambridge, M.A. and G. Porto. 2012. “Exports, Skills and the Frankel, J., and D. Romer. 1999. “Does Trade Wages of Seven Million Latin American Work- Cause Growth?” American Economic Review ers.” World Bank Economic Review 26 (1): 89 (3): 379–99. 34– 60. Freund, C., and M. Pierola. 2010. “Export Brambilla, I., D. Lederman, and G. Porto. Entrepreneurs: Evidence from Peru.” Policy 2012. “Exports, Export Destinations and Research Working Paper 5407, World Bank, Skills.” American Economic Review 102 (7): Washington, DC. 3406–38. ———. 2012 . “Export Superstars.” Policy Brückner, M., and D. Lederman. 2012. “Trade Research Working Paper 6222, World Bank, Causes Growth in Sub-Saharan Africa.” Policy Washington, DC. 120 LATIN AMERICAN ENTREPRENEURS Görg, H., M. Henry, and E. Strobl. 2008. “Grant Accounting of Costa Rica’s Export Growth Support and Exporting Activity.” Review of during 1997–2007.” World Bank Economic Economics and Statistics 90 (1): 168–74. Review 25 (3): 543– 61. Hausmann, R., and D. Rodrik. 2003. “Develop- Melitz, M. J. 2003. “The Impact of Trade on ment as Self-Discovery.” Journal of Develop- Intra- Industry Reallocations and Aggregate ment Economics 72: 603–33. Industry Productivity.” Econometrica 71 (6): Helpman, E., M. Melitz, and Y. Rubinstein. 1695–1725. 2008. “Estimating Trade Flows: Trading Part- Rodríguez, F., and D. Rodrik. 2001. “Trade Policy ners and Trading Volumes.” Quarterly Journal and Economic Growth: A Skeptic’s Guide to of Economics 123: 441– 87. the Cross-National Evidence.” In NBER Mac- Lederman, D. 2010. “An International Multilevel roeconomics Annual, 2000, vol. 15, edited by Analysis of Product Innovation.” Journal of B. Bernanke and K. Rogoff. Cambridge, MA: International Business Studies 41 (4): 606–19. MIT Press. ———. 2013. “International Trade and Inclu- Sachs, J., and A. Warner. 1995. “Economic sive Growth: A Primer.” Indian Growth and Reforms and the Process of Global Integra- Development Review 6 (1): 88–112. tion.” Brookings Papers on Economic Activity Lederman, D., and W. Maloney, eds. 2007. Nat- 1: 1–118. ural Resources, Neither Curse nor Destiny. Vollrath, T. 1991. “A Theoretical Evaluation Latin American Development Forum Series. of Alternative Trade Intensity Measures of Stanford, CA: Stanford University Press for the Revealed Comparative Advantage.” Review of World Bank. World Economics 2: 265– 80. Lederman, D., and W. Maloney. 2012. Does What Volpe, C., and J. Carballo. 2008. “Is Export Pro- You Export Matter? In Search of Empirical motion Effective in Developing Countries? Guidance for Industrial Policies. Latin Ameri- Firm-Level Evidence on the Intensive and the can Development Forum Series. Washington, Extensive Margins of Exports.” Journal of DC: World Bank. International Economics 76 (1): 89–106. Lederman, D., M. Olarreaga, and L. Zavala. Volpe, C., J. Carballo, and A. Gallo. 2011. “The 2013. “Export Promotion and Firm Entry Impact of Export Promotion Institutions on and Survival: Evidence from a Panel of Latin Trade.” Applied Economics Letters 18 (2): American Firms.” World Bank, Office of the 127–32. Chief Economist for Latin America and the Wacziarg, R., and K. H. Welch. 2008. “Trade Caribbean, Washington, DC, and University Liberalization and Growth: New Evidence.” of Geneva Department of Economics, Geneva. World Bank Economic Review 22 (2): 187–31. Lederman, D., A. Rodriguez- Clare, and D. Xu. World Bank. 2010. Global Survey of Export Pro- 2011. “Entrepreneurship and the Extensive motion Agencies. Washington, DC. Margin in Export Growth: A Microeconomic Foreign Direct Investment, Multinational Corporations, 5 and Innovation Multinational corporations (MNCs) employ a large share of the labor force, pay higher wages than other firms, are more productive, and have the potential to trigger positive spillovers on local firms through knowledge and technological transfers. Affiliates of foreign MNCs in Latin America and the Caribbean tend to be less innovative than multinational affiliates in other regions, but the productivity gains associated with their entry are greater than in other regions. Multinationals from the region (multilatinas) tend to make horizontal investments abroad rather than participate in global value chains and tend to be less innovative than MNCs from other middle-­ income regions. T  he past three decades were char- of the production process. All these processes acterized by a dramatic increase have played a role in reshaping the global in foreign direct investment (FDI) landscape of FDI flows.1 flows across the world. According to data A byproduct of the global rise in FDI flows reported by the United Nations Conference is the consolidation of multinational corpora- on Trade and Development (UNCTAD), the tions (MNCs) in the world economy. 2 MNCs United Nations’ trade and development unit, earned $12.4 trillion in 2010, almost 20 per- between 1980 and 2011 the dollar amount of cent of world gross domestic product (GDP), global FDI inflows increased at the stagger- a larger share than any economy except the ing average rate of 8.7 percent a year. United States. 3 Why should policy makers A number of changes in the world econ- care about MNCs in a report on entrepre- omy lie behind this pattern, changes that neurship? The previous chapters described have affected investment opportunities in various dimensions of entrepreneurial acts high-­income and developing countries alike. associated with the process of creation, They include the sharp reduction in transport growth, and consolidation of a firm. Chap- and communication costs since the 1970s, ter 2 discussed the decision to enter the local the wave of opening to foreign activity trig- market. Chapter 3 described the process of gered by economic reforms, the recovery of innovation of incumbent firms, including economic performance in developing coun- the decision of whether and how much to tries since the 1990s, and the fragmentation innovate processes or products. Chapter 4 121 122   LATIN AMERIC AN ENTREPRENEURS focused on the decision to export to foreign (LAC) are less innovative than multi- markets, an entrepreneurial act that is limited national affiliates operating in other to a very small subset of firms, typically the middle-­ i ncome regions in several most productive. dimensions. In particular, MNCs in This chapter examines MNCs, perhaps LAC are less likely to introduce new the ultimate manifestation of an entrepre- products and to have international neurial firm. These companies enter foreign quality certifications. Evidence from markets to sell their products and organize U.S. MNCs shows that the share of their production and distribution processes research and development (R&D) per- in a more efficient manner. Firms establish- formed by subsidiaries operating in ing foreign affiliates must incur a number of LAC is smaller than the share of R&D costs associated with their activities, such as performed by Asian subsidiaries. More- acquiring information about foreign mar- over, whereas the shares of R&D per- kets, paying establishment fees, and hiring formed by subsidiaries in Asia and the and training new employees. The magnitude Middle East have increased over time, of these costs implies that only firms with the share performed by subsidiaries in the highest productivity will international- LAC has contracted. ize (Helpman, Melitz, and Yeaple 2004). • Despite the relatively low levels of Going back to the typology of entrepreneurs innovation undertaken by multina- presented in chapter 2, MNCs represent the tional affiliates operating in LAC, the very high-­ end segment of transformational entry of foreign MNCs appears to have firms. increased productivity in the region—­ Transformational firms play a fundamen- more, in fact, than in any other region. tal role in spurring economic growth and This a priori surprising result is likely development. MNCs employ a large por- a result of the low productivity levels tion of the labor force, pay higher wages of firms in LAC compared with firms than other firms, and are more productive from other regions—­ L AC firms start than other firms (Lipsey 2002). More impor- from a lower base, making the rela- tant, the coexistence of MNCs and local tive impact of productivity gains larger. non-MNC firms gives rise to the possibility Most of these productivity gains are a of knowledge and technological spillovers, result of knowledge and technological which can enhance developing countries’ transfers from multinational affiliates growth prospects. Hence, understanding the to local firms, especially through local behavior of MNCs and the factors that allow suppliers. them to excel is highly relevant for policy • Multilatinas tend to focus on horizon- makers in LAC. tal investments abroad rather than par- This chapter characterizes some of the ticipation in global value chains. The defining traits of two types of MNCs in the leading reason why MNCs from LAC region: foreign MNCs and emerging Latin cross borders is to serve foreign mar- American MNCs— ­t he so- ­c alled multi­ kets. In contrast, MNCs from other latinas. It uses new data sources that reveal emerging regions internationalize to the types of markets MNCs enter, the way take advantage of lower labor costs and they interact with their foreign affiliates, and access export promotion zones. their innovation efforts.4 • Multilatinas tend to be less innovative Four main findings emerge from this than other MNCs. Their R&D expen- chapter: diture per $1,000 of revenue is low rela- tive to their counterparts in other devel- • Multinational affiliates operating in oping countries, and their management Latin America and the Caribbean practices are far from best practices. F o rei g n D irect I n v estment , M ultinati o nal C o rp o rati o ns , and   I nn o v ati o n   123 The rest of the chapter is organized as fol- in 2004, 78 of 110 countries were actively lows. The first section explores some salient offering fiscal or financial concessions to for- characteristics of MNCs operating in LAC eign companies that decided to set up produc- and quantifies the productivity spillovers tion or other facilities within their borders on local firms. The second section describes (see Harding and Javorcik 2011, 2012, for a some of the characteristics of multilatinas. description of the census). The last section provides some concluding Are these policy choices justified? Is the remarks. relatively large number of MNCs operat- ing in LAC beneficial for the region? Luring foreign firms into developing economies is Foreign multinational potentially appealing for two broad reasons. corporations in Latin America The first, and perhaps more obvious one, and the Caribbean is enhancing factor accumulation. Foreign After some decades of adjustments, LAC firms are likely to add to the capital stock sailed through the wave of globalization of the host economy by building factories in the first decade of the 21st century with and investing in machinery and equipment. unprecedented economic strength. A series of This reason alone hardly explains the vigor- institutional and policy changes undertaken ous efforts exerted by countries to attract over the past two decades, especially but not MNCs, however. Policy makers believe that exclusively in the macro­ f inancial terrain, the overall benefits of foreign presence go allowed the region to enjoy a decade of solid beyond factor accumulation. Proponents of growth and macroeconomic stability during attracting FDI suggest that foreign presence the 2000s. benefits the host country in a second and Not surprisingly, these factors made LAC more important way: by bringing advanced an appealing destination for foreign inves- technologies and know-­ how that lead to tors. Almost 70 percent of the countries aggregate productivity improvements and in the region show levels of FDI inflows in positive externalities to local firms through 2010 above those predicted by their GDP and technological spillovers.6 population (figure 5.1, panel a). 5 A similar This discussion makes the crucial assump- picture emerges when looking at the revenues tion that MNCs are technologically superior of multinational affiliates in the region (fig- to local firms, an idea that is supported by ure 5.1, panel b). at least three arguments. First, most of these Two groups of countries in LAC deserve corporations come from high-­ income econo- special attention. Brazil and Mexico, LAC’s mies, which are closer to the technological largest economies, have not only achieved frontier. The theory of MNCs goes beyond levels of FDI that exceed those predicted by mere country advantages; it argues that their country characteristics, they also place MNCs rely heavily on intangible assets, such among the world’s top 15 recipients of FDI as firm-­ established specific technologies, well-­ flows, above India and South Africa. In con- brand names, and know-­ how or manage- trast, Guatemala, Haiti, and República Boli- ment techniques that give them an “owner- variana de Venezuela appear to be lagging in ship advantage” over other organizations (see terms of attracting foreign firms. Dunning 1988). Subsidiaries operating in Policy makers in developing countries developing economies could therefore poten- place attracting FDI and MNCs high on their tially benefit from aggregate technological agendas. They use incentives such as income advantages from the MNC as a whole.7 tax holidays, tariff exemptions, and subsidies Second, recent theoretical work high- for infrastructure to attract foreign firms. lighting firm heterogeneity points out that According to a census of investment promo- only the most productive establishments can tion agencies carried out by the World Bank afford the extra cost of setting up production FIGURE 5.1  Inward foreign direct investment and multinational activity in Latin America and the Caribbean a. FDI in ows into LAC in the 2000s: observed and benchmarked values b. Revenue of MNC foreign subsidiaries: observed and benchmarked values by host country Cuba Dominica Guyana LAC countries Haiti Dominica St. Vincent and the Grenadines Haiti Other countries Grenada Guyana Azerbaijan and economies St. Kitts and Nevis Grenada Paraguay Armenia Benchmark Saint Lucia Uzbekistan Antigua and Barbuda Moldova Kuwait Nicaragua Moldova Macedonia, FYR Macedonia, FYR Georgia Nicaragua Dominican Republic Armenia Bolivia Honduras El Salvador Guatemala Uzbekistan Oman Bosnia and Herzegovina Costa Rica Albania Philippines Ecuador Paraguay Guatemala Belarus Honduras Jamaica El Salvador Azerbaijan United Arab Emirates Georgia Bolivia Slovenia Saudi Arabia Latvia Albania Lithuania Trinidad and Tobago Trinidad and Tobago Uruguay Belarus Ecuador Uruguay Oman Kuwait Costa Rica India Turkmenistan Kazakhstan Dominican Republic Venezuela, RB Philippines Bosnia and Herzegovina Venezuela, RB Latvia Greece Israel Serbia Croatia New Zealand Croatia Lithuania Slovak Republic Bulgaria South Africa China Peru Serbia Bulgaria Peru Hungary Malaysia Ukraine Indonesia Indonesia Finland Greece Malaysia Colombia Portugal New Zealand Romania Chile Argentina Portugal Colombia Thailand Korea, Rep. Turkey Czech Republic Finland Israel Norway Hungary Kazakhstan Australia United Arab Emirates Argentina Thailand Korea, Rep. Denmark Romania Japan Japan Chile Austria Austria Denmark Turkey Ireland Ireland Poland Czech Republic Saudia Arabia Poland Switzerland Sweden Sweden Brazil India Canada Italy Singapore Mexico Mexico Australia Russian Federation Norway Singapore Russian Federation Brazil Belgium Netherlands Switzerland Spain Italy Canada Spain Hong Kong SAR, China France Germany France Hong Kong SAR, China Belgium Netherlands China Germany United Kingdom United States United States United Kingdom 1 10 100 1,000 10,000 100,000 1 10 100 1,000 10,000 100,000 FDI in ows (US$ millions) Revenue produced by MNC a liates (US$) Source: World Bank, based on data from UNCTAD’s FDI database and Orbis. Note: Diamonds in panel a represent the predicted value of a regression of log foreign direct investment (FDI) inflow, after controlling for log average gross domestic product (GDP) and average population in the 2000s using all countries and economies with available FDI data. Only comparable countries are displayed in the graph. Diamonds in panel b represent the predicted value of a regression of log revenue, after controlling for log average GDP and average population in the 2000s. Panel b uses the latest available information of firms that were active in 2011. LAC = Latin America and the Caribbean, MNC = multinational corporation. 124 F o rei g n D irect I n v estment , M ultinati o nal C o rp o rati o ns , and   I nn o v ati o n   125 facilities in a foreign country. MNCs are thus FIGURE 5.2  Difference in number of patents held predicted to come from the upper tier of the by multinational parent and local firms in home productivity distribution of firms in the home country in selected country groups, countries, and country (Helpman, Melitz, and Yeaple 2004). economies, 2010–11 Third, by definition MNCs are multi- plant, multilocation organizations and thus 1.8 typically larger than non-MNCs. Their size 1.6 advantage allows them to operate more effi- Di erence in patents (log) 1.4 ciently by benefiting from economies of scale 1.2 and scope. Origin, selection, and economies 1.0 of scale all point in the direction of higher 0.8 efficiency and better technologies from 0.6 MNCs. The data support the view that on aver- 0.4 age, MNCs are more productive and innova- 0.2 tive than other firms. In 2002, for example, 0 EAP4 ECA LAC5 High- India China MNCs accounted for almost half of total income R&D expenditure and almost 70 percent of economies business R&D (Javorcik 2010). Patenting is another area where MNCs Source: World Bank, based on data from Orbis. Note: Calculations based on the latest available information of firms that have a clear advantage. Across regions, head- were active in 2011. Bars represent estimated coefficients of a dummy quarters of MNCs hold more patents than variable taking the value 1 if the firm is a multinational parent firm in a regression of ln (1 + patents in 2010) using all firms from a given country local firms in the country where the head- (excluding foreign-­owned firms). Additional controls include the firm’s quarters is located. Figure 5.2 shows the aver- revenue in 2006, industry fixed effects, and country fixed effects. Standard errors are clustered at the industry level. For countries and economies age difference between the number of patents included in each group, see note 8. held by parent firms of MNCs and local firms in the country origin of the MNC, control- ling for country, firm, and sectoral character- and productivity advantage of MNCs trans- istics. Thus, in Chile, it compares the number lates into technological advantages of their of patents of Concha y Toro, a multinational affiliates in the developing world. The empir- wine company, with the number of patents ical literature shows evidence of such an held by nonmultinational winemaking com- effect. Studies on Mexico (Blomström 1983); panies in Chile. These data reveal very large Uruguay (Kokko, Zejan, and Tansini 2001); differences in the patent gap between MNCs and República Bolivariana de Venezuela and local firms across regions. China and (Aitken and Harrison 1999) find evidence of India have the largest gaps, followed by high-­ higher labor productivity in foreign-­ owned income and LAC5 countries; Eastern Europe firms than local firms. Although part of this and Central Asia (ECA) and East Asia and productivity advantage is explained by higher Pacific (EAP4) have the smallest differences.8 capital intensity, differences in other inputs These differences do not necessarily mean may also be responsible. Work by Bloom that MNCs from ECA or EAP4 countries and others (2012), for example, shows that have fewer patents than MNCs from LAC5 foreign-­ owned firms in Argentina, Brazil, countries; they could reflect the poor perfor- Chile, and Mexico have better management mance of local firms in LAC relative to their practices than local firms and that the quality counterparts in ECA and EAP4 (see chapter of management practices by foreign-­ owned 3). Later in this chapter we return to the com- firms in LAC is much closer to best practices parison of MNCs from different regions. than to local practices, giving support to the More important to developing countries is idea that multinational affiliates “import” the extent to which the overall technological knowledge from headquarters.9 126   LATIN AMERIC AN ENTREPRENEURS Multinational affiliates in the region is not significant from a statistical point of appear to be more innovative than local view); and almost 13 percentage points more firms in almost every dimension. Figure 5.3 likely to adopt foreign technologies. The shows the difference between the proportion differences are even larger when compar- of foreign-­owned and local manufacturing ing efforts to improve quality. MNCs are firms that engage in a number of entrepre- 21 percentage points more likely to engage neurial activities after controlling for country in quality-­improving investments and almost and sectoral characteristics.10 25 percentage points more likely to have Everything else equal, the likelihood of a international quality certifications than local firm in LAC introducing or producing a good firms, perhaps indicating their higher likeli- that is new to the market is about 11 percent- hood of exporting. age points higher for foreign-­ owned firms, The evidence so far appears to suggest and the likelihood of introducing a new pro- that multinational affiliates operating in LAC cess is about 5 percentage points higher (fig- are able to overcome the obstacles that deter ure 5.3). Foreign-­owned firms are also about innovation by local firms in the region. Is 5 percentage points more likely to file for a this really the case? A comparison of multi- patent, trademark, or copyright or to collab- national affiliates across regions suggests it orate with other institutions for innovation is not: although multinational affiliates are purposes; 6 percentage points more likely more productive than local firms, foreign-­ to invest in R&D (although this difference owned firms in LAC are less innovative than their counterparts in other regions. Figure 5.4 uses data from World Bank Enterprise Surveys to compare the likeli- FIGURE 5.3  Difference in innovation between multinational hood that manufacturing multinational affiliates and local firms in the host economy in Latin America and affiliates in different country groups intro- the Caribbean, 2010 duce a new product. Foreign subsidiaries in non-­ Caribbean LAC countries are on aver- Has an international quality certi cation age almost 20 percentage points less likely to Invested to improve quality introduce new products than foreign subsid- control or obtain certi cation iaries of high-­income countries. The picture Uses foreign technology is even gloomier for multinational affiliates in the Caribbean, which are almost 40 per- New or signi cantly improved product centage points less likely to introduce new Invested in research and development products than their high-­ i ncome counter- parts. Countries in ECA and EAP4 have an New or signi cantly improved process average propensity to innovate that is close Cooperates on innovation with others income to that of affiliates operating in high-­ economies. Filed for patent, trademark, or copyright The underperformance of subsidiar- 0 5 10 15 20 25 ies operating in LAC relative to subsidiar- Additional likelihood by MNC a liates ies operating in other regions is also evident (percentage points) from their participation in the production of Signi cant at 10% Not signi cant at 10% knowledge. In 2008, the share of total R&D by foreign affiliates of U.S. MNCs coming Source: World Bank, based on data from Enterprise Surveys. from LAC countries was almost 70 percent- Note: Bars are the coefficients of a dummy variable taking the value 1 if the firm is foreign owned age points lower than that coming from sub- in a regression of innovation variables. Additional controls include country and industry fixed effects. Countries include Antigua and Barbuda, Argentina, Chile, Colombia, Costa Rica, Dominica, sidiaries operating in Europe and Canada the Dominican Republic, Ecuador, El Salvador, Grenada, Guatemala, Guyana, Honduras, Jamaica, (figure 5.5). Differences in characteristics Mexico, Nicaragua, Panama, Paraguay, Peru, St. Kitts and Nevis, St. Lucia, St. Vincent and the Grena- dines, Suriname, Trinidad and Tobago, and Uruguay. Standard errors are clustered at the industry between Canada and countries in LAC and level. MNC = multinational corporation. Europe explain part of these differences. F o rei g n D irect I n v estment , M ultinati o nal C o rp o rati o ns , and   I nn o v ati o n   127 A broader comparison shows equally dis- FIGURE 5.4  Product innovation by foreign multinational affiliates couraging results, however. LAC’s participa- in selected country groups and economies, 2010 tion was 16 percentage points lower than that of Asia and only 1 percentage point higher 100 than that of the Middle East. Moreover, the 90 share of R&D undertaken by subsidiaries of 80 U.S. MNCs operating in LAC has fallen over 70 time: in 2008, only 3 percent of total R&D 60 Percent by overseas affiliates of U.S. MNCs came 50 from LAC, down from 5 percent in 1998. In 40 contrast, between 1998 and 2008, the share 30 of overseas R&D by Asian affiliates increased 20 9 percentage points and the share by Middle 10 Eastern affiliates rose 2 percentage points.11 0 Caribbean Other LAC5 ECA EAP4 High- In sum, the presence of multinational LAC income affiliates appears to benefit countries in LAC economies by raising their productivity and innovation LAC countries Other countries and economies activities, albeit less so than in other regions. The factors that deter innovation by local Source: World Bank, based on data from Enterprise Surveys. Note: Bars represent the coefficient regional dummies in a regression of innovation variables using firms also appear to constrain the ability of only foreign-­owned firms. Additional controls include industry fixed effects. Standard errors are foreign-­owned firms to tap the pool of intan- clustered at the industry level. All coefficients are significant at the 10 percent level. For countries and economies included in each group, see note 8. gible assets held by the MNC and to innovate in the host country. (Chapter 6 explores the factors that are likely to be behind this lack of innovation in the region.) FIGURE 5.5  Research and development by foreign affiliates of U.S. multinational corporations in selected regions, 1998 and 2008 Spillovers and aggregate productivity gains from foreign-­owned firms 100 90 Although the direct impact of innovation 80 by multinational affiliates is an important 70 channel through which MNC activity fosters 60 Percent growth and innovation in the host country, 50 40 efforts by policy makers to attract foreign 30 firms are based on the belief that MNCs can 20 lead to productivity improvements in local 10 firms and, through this channel, at the aggre- 0 1998 2008 gate level. This motivation is of particular LAC Europe and Africa Middle Asia importance in LAC, where low productivity Canada East is widely recognized as the region’s Achilles heel (Pagés-­Serra 2010; de la Torre and oth- Source: National Science Board 2012. ers 2011). Note: LAC = Latin America and the Caribbean. Has multinational entry in LAC led to positive productivity spillovers to local firms and aggregate productivity improvements? If The first is competition in product and factor so, how large are these effects? markets. The arrival of MNCs clearly affects Before trying to answer these questions, the profitability of local firms. The pres- it is important to point out two economic ence of MNCs most likely increases prod- channels through which MNCs could affect uct and factor market competition, which domestic productivity and entrepreneurship. may depress goods prices and exert upward 128   LATIN AMERIC AN ENTREPRENEURS pressure on factor prices.12 Greater compe- variety of inputs produced by Intel’s suppliers tition should lead to a reduction in profits that were previously unavailable in the local among local firms, perhaps even precipitat- economy. MNCs are also a useful conduit ing the exit of less productive firms.13 This to inform local firms about new technolo- shutdown of inefficient firms is not necessar- gies, new marketing techniques, and export ily bad news for the aggregate economy: in markets (see Aitken, Hanson, and Harrison a healthy economic environment, it leads to 1997; Javorcik and Spatareanu 2005; Chen the reallocation of resources toward the most and Swenson 2008). productive firms, increasing aggregate pro- Worker turnover is yet another way in ductivity (see Alfaro and Chen 2013).14 which local firms may appropriate part of the The second channel is direct and indirect MNCs’ intangible assets. The accumulation knowledge transfers. Direct knowledge trans- of experience and the training received dur- fers could arise, for instance, through the ing workers’ tenure at an MNC can enable engagement of MNCs with local suppliers to them to take part of the firm’s stock of raise the quality of inputs. Employee train- knowledge with them if they decide to move ing, quality control, the lending or leasing of (Fosfuri, Motta, and Rønde 2001). Using machinery, and the provision of advice on the employer-­ employee matched data for Bra- firm’s business strategy are some of the com- zil, Poole (2013) finds a positive correlation monly observed support activities provided between the share of former MNC employ- by MNCs to local suppliers (Moran 2001; ees and the wages paid to incumbent work- Javorcik and Spatareanu 2005). ers with no prior MNC affiliation, suggesting An often cited case is that of Intel’s plant that the presence of employees with former in Costa Rica. Larrain, Lopez-­ C alva, and MNC experience raises the productivity of Rodriguez-­ Clare (2001) show that 35 per- other workers in the firm.15 cent of local service providers and 17 percent Local firms can also benefit from public of input providers received training from goods arising from the presence of MNCs. Intel. Doing business with Intel Costa Rica For instance, part of the commitments made also appears to have led to organizational by local governments to attract foreign firms changes: 18 percent of the goods provid- may include public investments, which then ers reported changes in their organizational become available to local firms. Similarly, the structure because of their activities with presence of MNCs may induce skill upgrades Intel, and 10 percent reported being associ- in the local economy that could benefit indig- ated with foreign firms after the arrival of the enous firms (box 5.1). semiconductor manufacturer. Although aggregate productivity increases Knowledge transfers from multinational with MNC activity irrespective of the chan- presence are not limited to transfers aris- nels at work, the effect on local firms depends ing from their dealings with local suppliers: on whether competition or knowledge trans- intangible assets of MNCs, such as manage- fers dominate. Local firms with more expo- ment practices, business models, or special sure to the competition channel, such as firms inputs and services, may also become avail- in the sectors in which the MNCs operate, able to local firms outside the MNCs’ pro- typically suffer from negative spillovers, duction chain. MNCs may require local whereas firms with more exposure to the suppliers to provide inputs or services new knowledge channel, such as local suppliers, to the host economy that in turn can be sold are more likely to benefit from positive pro- to other local firms (Rodriguez-­ Clare 1996; ductivity spillovers.16 Quantifying the relative Blalock and Gertler 2005). Larrain, Lopez-­ importance of the competition and knowl- Calva, and Rodriguez-­ C lare (2001), for edge transfer channels is crucial to assessing example, document that local firms in Costa the impact of MNC entry on firm-­ level and Rica unrelated to Intel gained access to a aggregate productivity. F o rei g n D irect I n v estment , M ultinati o nal C o rp o rati o ns , and   I nn o v ati o n   129 BOX 5.1  Can a whale in a swimming pool create a splash? Intel and the upgrading of tertiary education in Costa Rica In November 1996, Intel, the world’s largest semi- • A one-­ year certificate program and a one-­year conductor producer, announced that it would con- associate degree at ITCR focused on new tech- struct a new $300 million assembly and test plant nical fields, such as semiconductor manufac- in Costa Rica. The investment community was ini- turing and microelectronics, and, later, materi- tially stunned by Intel’s announcement. After all, als science Costa Rica, a relative small economy, was chosen • Links with UCR’s School of Physics and tech- over some of LAC’s biggest economies. Even Intel nological and vocational schools for electronics officials recognized that the decision was bold. Bob • Support for the electrical, electronics, comput- Perlman, an Intel vice-­ president, stated that bring- ing, and industrial engineering fields. ing Intel to Costa Rica was “like putting a whale The benefits of these programs have exceeded in a swimming pool.” The country’s economic and those that came directly from Intel. A survey of political stability, its proximity to the United States, 20 Costa Rican firms identified by CINDE, Costa and its pro-­business environment were all important Rica’s investment promotion agency, as potential factors in Intel’s decision (Spar 1998). competitors of Intel in the labor market revealed Some bottlenecks and limitations in the Costa that all but one saw the arrival of Intel as positive for Rican economy raised concerns for Intel. One was the accumulation of human capital. Of these firms, education. Costa Rica had a high literacy rate and a eight had hired a graduate from ITCR’s one-­ year good education system, but the low number of engi- certificate, and all reported benefiting from the cre- neers and workers with technical skills was consid- ation of this program. ered a constraint on Intel’s operation. The spillovers of Intel’s presence can also be To overcome this hurdle, the Costa Rican gov- seen in the stock of engineers in Costa Rica. The ernment, Intel, and major academic institutions and number of graduates from engineering programs in technical schools joined forces to help strengthen Costa Rica reportedly increased by almost 40 per- the country’s educational system. They developed cent between 2002 and 2011, from 1,580 to almost a series of programs and relationships designed to 2,200 (UNESCO [United Nations Educational, Sci- increase both the number of graduates in engineer- entific and Cultural Organization]). ing and technical degrees and the proficiency of In sum, Intel’s presence has played an important the graduates. These programs and relationships role in increasing the stock and the quality of work- included the following: ers with technical skills in Costa Rica. The ben- • Programs and enhanced curricula at the three efits associated with these achievements go beyond major educational institutions—­ Tecnológico the boundaries of Intel. To put it in Bob Perlman’s de Costa Rica (ITCR), Universidad de Costa words, the whale in the swimming pool created a Rica (UCR), and Instituto Nacional de Apren- big splash. dizaje (INA)—­ 2003 especially during 1999–­ • English reinforcement program at ITCR Calva, and Rodriguez-­ Sources: Spar 1998; Larrain, Lopez-­ Clare 2001; World Bank Group 2006. This is precisely what Alfaro and Chen competition in explaining these gains. They (2013) do. They use a standard economic use a sample of manufacturing firms taken model of MNC activity similar to the one from Orbis from 60 countries, 5 of which presented in Helpman, Melitz, and Yeaple (Argentina, Brazil, Chile, Colombia, and (2004) to estimate the aggregate produc- Mexico) are in LAC. tivity gains from MNC entry as well as the The results are striking: doubling MNC relative weight of knowledge transfers and entry into LAC countries would increase 130   LATIN AMERIC AN ENTREPRENEURS aggregate productivity by 3.8 percent (fig- to a large gap between foreign-­ owned and ure 5.6). This number is six times higher local firms and to low productivity by local than in ECA or high-­ income economies and firms (Pagés-­Serra 2010), suggesting a large seven times higher than in China. More potential for spillovers in the region even in important, and in contrast to other regions, the absence of vigorous innovation activity knowledge spillovers run the entire show from MNCs operating in the region. In con- in LAC, explaining 100 percent of the esti- trast, in other regions there appears to be a mated aggregate productivity gains from smaller gap between multinational affiliates MNC entry.17 and local firms. Although the large spillovers Alfaro and Chen (2013) find for LAC may seem at odds Capitalizing on the spillovers with the poor performance of the region from multinational activity: in terms of productivity and the under- Scope for policy intervention performance of MNCs operating in LAC in terms of innovation, all these pieces are The policy implications of this surprising consistent. Indeed, the marginal productiv- result are immense. In particular, it highlights ity gains for local firms from MNC entry the large premium of policies that foster the through knowledge spillovers are expected attraction of MNCs and their spillovers to to depend on both the technological gap local firms. What types of policy interven- between local firms and multinational affili- tions could yield these goals? ates and the productivity level of local firms: Policy makers have typically pursued spillovers are likely to be larger, the larger three sets of policies to achieve these objec- the technological gap between local and tives. The first are policies aimed at attracting foreign firms and the lower the productivity FDI and MNCs, such as tax holidays or cash of local firms. The evidence in LAC points incentives. One policy tool that has proven very effective in attracting FDI is the estab- lishment of investment promotion agencies FIGURE 5.6  Sources of predicted productivity (IPAs). IPAs actively look for foreign inves- gains associated with entry of multinational tors and provide them with valuable sectoral corporations, by country groups, countries, and economies, 2002–­10 and country information during their deci- sion process. Using information from a survey of actual 4.0 and potential foreign investors, Kenyon and 3.5 Margalit (2012) show that the information 3.0 provided by IPAs about local markets to Percentage points foreign investors is crucial in their decision-­ 2.5 making process. Harding and Javorcik (2011) 2.0 show that FDI into sectors targeted by IPAs is 1.5 larger than FDI to other sectors. 1.0 These agencies have flourished in LAC since the 1990s. Are they working? More 0.5 efforts are needed to improve the quality of 0 these institutions, as suggested by the results High-income China ECA LAC5 economies of the 2012 Global Investment Promotion Knowledge spillover Market reallocation Best Practices (GIPBP) report prepared by the World Bank Group. The GIPBP assesses Source: Alfaro and Chen 2013. two aspects of the information facilitation Note: Bars represent total productivity gains from doubling the probability of entry by a multinational corporation, estimated though a structural role of IPAs: their ability to handle inqui- model. For countries and economies included in each group, see note 8. ries from foreign investors in a professional F o rei g n D irect I n v estment , M ultinati o nal C o rp o rati o ns , and   I nn o v ati o n   131 and speedy manner and the clarity and con- minimum local content requirements, labor tent of their websites. Harding and Javor- requirements, and import substitution poli- cik (2013) find that countries with IPAs cies. This set of policies is more complex to that are better able to handle queries and evaluate at a regional or even country level, have clearer information on their websites because doing so requires in-­ depth analysis tend to attract more FDI. The 2012 GIPBP of their detailed characteristics and interac- highlights that IPAs from LAC still suffer tions. Their characterization goes beyond the from weaknesses in handling inquiries from scope of this chapter. potential investors and serious deficiencies Fine-­tuning the balance among these three in their websites, especially in countries in policy areas is fundamental and constitutes South America. As a result, IPAs from LAC a serious challenge for policy makers, as score 48 on a 100-­ point scale—­ far from the policies aimed at strengthening one objec- 64 scored by IPAs from high-­ income Organ- tive may end up weakening the other. Take, isation for Economic Co-operation and for, instance, minimum content require- Development (OECD) countries. ments. Such requirements strengthen the A second set of policies concentrates on link between multinational affiliates and improving the general business environ- local firms and could generate larger knowl- ment. Policies that improve human capital edge spillovers. But in countries where the or institutions fall in this category. These quality of inputs produced by local firms is policies typically seek to achieve multiple poor, they could discourage new MNCs from objectives at the same time. They are good entering the host country and limit the incen- ways to attract FDI while reducing the bar- tives of incumbent multinational affiliates to riers that hinder both the ability of multina- innovate. The balance among the three types tional affiliates to innovate and the capacity of policies will depend on the specificities of of local firms to absorb knowledge transfers each country. from MNCs. These policies are of tremendous impor- tance in achieving the goals mentioned above Multilatinas because LAC still suffers from substantial Foreign presence is an important aspect of deficiencies in areas such as human capital globalization and economic integration for and financial access, some of which are likely developing countries, but not the only one. to prevent it from exploiting the full potential An equally important feature of the increase of MNC activity.18, 19, 20 in global FDI flows is the role played by These barriers have implications for the developing countries as a source of FDI and allocation of the productivity gains from the rise of MNCs from these countries. knowledge transfers in the host economy. Aggregate FDI outflows from LAC grew Indeed, many studies find that MNC spill- at an annual rate of 16 percent from 1980 overs in LAC are concentrated among the to 2011, rising by a factor of 15 as share of largest firms. 21 As a result, barriers in the GDP over the same period (from 0.13 per- absorptive capacity of local firms could cause cent of LAC’s GDP to 1.9 percent). LAC’s MNC activity to accentuate productivity dif- weight in total FDI outflows also increased. ferences across firms in a region that suffers In the 1980s, a meager 1.2 percent of total from a very uneven productivity distribution FDI outflows came from LAC countries; in (Busso, Madrigal, and Pagés-Serra 2012), the 2000s, this number reached 5.2 percent. something that can create political economy Many multilatinas are now global players, constraints to the attraction of FDI. with 18 of them among Boston Consulting A third set of policies attempts to Group’s list of top 100 firms from emerg- strengthen the links between foreign-­ owned ing markets to watch (BCG 2006; Santiso firms and the local economy. They include 2008). 132   LATIN AMERIC AN ENTREPRENEURS Can multilatinas be another vehicle for FIGURE 5.7  Share of total revenues of innovation upgrades in LAC? The interna- multilatinas by country or country group of origin, tionalization of these firms could give them 2010–11 access to technologies and know-­ how avail- able in foreign markets, which they can 1% import back home. In addition to giving 2% 3% multilatinas access to existing top technolo- gies, the internationalization of these firms 4% can boost their own innovation potential. For instance, access to a large pool of skilled workers and more developed financial mar- 39% 16% kets may allow these firms to overcome some of the constraints to innovation they face at home. 22 Good knowledge of the workings of the home country’s economy (institutions, markets, and so forth) and the tight connec- 31% tion with local firms might suggest that the potential for knowledge spillovers from mul- tilatinas to the local economy is large.23 In fact, the evidence indicates that the Brazil Peru potential for multilatinas to bring wide- Mexico Argentina spread productivity and innovation gains Chile Trinidad and Tobago into the region is limited, for a few reasons. Colombia Uruguay First, the emergence of multilatinas is heav- Costa Rica Caribbean ily tilted toward a very small number of Venezuela, RB Other LAC countries. Indeed, multilatinas from Brazil, Chile, Colombia, and Mexico account for Source: World Bank, based on data from Orbis. Note: Calculations are based on the latest available information of firms 90 percent of the revenue earned by these that were active in 2011. For countries included in Other LAC, see note 8. firms (figure 5.7). Interestingly, MNCs from LAC = Latin America and the Caribbean. Costa Rica, a relative small country in the region, have positioned themselves as impor- by their income and population. Even LAC’s tant players, accounting for about 3 percent two top performers in absolute numbers, of total revenue and standing above their Mexico and Brazil, are far from their bench- counterparts from countries such as Argen- mark level. There are, however, some bright tina, Peru, and República Bolivariana de spots. Chile and Colombia are not only Venezuela. among the top four multilatina -­producing To be sure, country characteristics such as countries, they also overperform relative to GDP and population partly explain the het- their country characteristics. Multilatinas erogeneity in the performance of multilatinas from the Caribbean countries also appear to from different countries. Figure 5.8 takes excel once the size of their economies is taken these differences into consideration by pre- into account. senting both the observed dollar amount of Anecdotal evidence suggests that there are the revenue produced by MNCs from differ- large differences in the origins of multilatinas ent countries and the level predicted by GDP and MNCs from other regions. In the United and population by means of a multivariate States—­ and to a lesser extent East Asia and regression. the Pacific and the Republic of Korea—­ most The picture that emerges from figure 5.8 is MNCs are private firms that took the leap discouraging. Revenues of multilatinas from and started operating in foreign markets. In most LAC countries are lower than predicted contrast, many multilatinas, especially the FIGURE 5.8  Actual and benchmarked revenue of multinational corporations in selected countries and economies, relative to given characteristics, 2010–11 Nicaragua LAC countries Grenada larger ones, were public sector companies Suriname Other countries Haiti or economies that were privatized in the wave of liberal- Paraguay Georgia Benchmark ization of the 1990s and enjoyed monopoly Honduras Cuba power for long periods of time (Casanova El Salvador Albania and others 2009). Macedonia, FYR Bolivia Moldova Dominican Republic Armenia Drivers of internationalization Guyana Romania Bosnia and Herzegovina In addition to differences in origin, there also Costa Rica Dominica appear to be differences in the motivation Guatemala Oman for internationalization of multilatinas and Bulgaria Ecuador MNCs from other regions. Tapping into for- Serbia eign countries to open up new markets and Peru Uzbekistan to diversify country risk is one of the leading Philippines St. Kitts and Nevis reasons why LAC firms cross borders (Alfaro Latvia Croatia and Hammel 2006; Casanova and others St. Vincent and the Grenadines Belarus 2009). Kenyon and Margalit (2012) asked Uruguay Trinidad and Tobago firms in four emerging market countries—­ Indonesia Antigua and Barbuda Brazil, India, Korea, and South Africa—­ St. Lucia Azerbaijan about the main motivations for investing in Lithuania Venezuela, RB other emerging markets. The firms were ran- Ukraine Slovak Republic domly drawn from registries that included all Argentina Czech Republic firms in each country with annual revenues Slovenia Hungary of at least $25 million and that operated in Kazakhstan Turkey one of five sectors: finance and insurance, Thailand New Zealand manufacturing, wholesale trade, retail trade, Chile Colombia and transportation and warehousing. Kuwait The results yield two conclusions. First, Greece Poland the decision by Brazilian firms about where Israel Saudi Arabia to invest is driven primarily by market oppor- Hong Kong SAR, China Malaysia tunities. In particular, the presence of key Portugal Russian Federation customers and attractive domestic markets Mexico Singapore are key factors (figure 5.9, panel a). Second, United Arab Emirates Brazil Brazilian firms are much more focused on the Norway Australia opportunities offered by the domestic market Austria Denmark than are firms from other countries. They are Finland India significantly less concerned than their coun- Korea, Rep. Belgium terparts about the quality of the workforce, Ireland Canada labor costs, and regulatory transparency of Spain Sweden the host country (figure 5.9, panel b). China Italy The fact that multilatinas internationalize Netherlands Switzerland in order to expand markets implies that their Japan Germany foreign subsidiaries tend to operate in the France same sector as the headquarters. Figure 5.10 United Kingdom United States divides the subsidiaries of MNCs from differ- 0.001 0.01 0.1 1 10 100 1,000 100,000 ent regions into three groups: firms operat- 10,000 10,000,000 ing in the same sectors as the headquarters Revenue of multilatinas (US$ thousands) (horizontal), firms providing inputs to the Source: World Bank, based on data from Orbis. headquarter (upstream of headquarters), and Note: Calculations are based on the latest available information of firms that were active in 2011. Diamonds represent the predicted value of a regression of log revenue, after controlling for log average gross domestic product and average population in the 2000s. LAC = Latin America and the Caribbean. 133 134   LATIN AMERIC AN ENTREPRENEURS FIGURE 5.9  Factors driving Brazilian firms to cross borders, 2010–11 a. Drivers of internationalization in Brazil b. Di erences in drivers of internationalization: Additional likelihood of Brazilian rms relative to those from other emerging countries Presence of foreign investors Domestic market 10% Availability of joint 14% venture partners Export processing 7% zones Domestic market Regional market Security Presence of key client/sellers Labor costs 28% 27% Political stability Transparency –25 –20 –15 –10 –5 0 5 10 15 20 14% Percent Source: Kenyon and Margalit 2012. Note: Comparator countries are India, the Republic of Korea, and South Africa. FIGURE 5.10  Sectoral position of foreign subsidiaries relative firms obtaining inputs from the headquarters to headquarters in selected country groups, countries, and (downstream from headquarters).24 economies, 2010–11 The results show that the pattern of link- ages between headquarters and subsidiaries 60 observed for multilatinas stands in sharp contrast to that observed in other regions. By Share of foreign subsidiaries (%) 50 and large, multilatinas establish horizontal 40 links with their subsidiaries: almost half of their foreign subsidiaries operate in the same 30 sector as their headquarters. In contrast, for- eign subsidiaries of MNCs from other regions 20 tend to establish vertical linkages with their 10 headquarters. For example, about 40 percent of foreign subsidiaries of Asian MNCs oper- 0 ate in the same sector as the headquarters. LAC China and EAP4 High-income ECA India economies This number is even lower in ECA and high-­ Location of headquarters income countries, where only 35 percent of subsidiaries operate in the same sector as the Upstream (relative to parent) Horizontal parent company. Downstream (relative to parent) An implication of this pattern is the lim- ited scope for multilatinas to transfer knowl- Source: World Bank, based on data from Orbis. Note: Calculations are based on the latest available information of firms that were active in 2011. edge to the home economy through their Sectoral position was calculated using the input-­output matrix for the United States. A subsidiary involvement in global value chains. Global is defined as downstream if the parent company’s sector is a net supplier of the subsidiary’s sector. A subsidiary is defined as upstream if the subsidiary’s sector is a net supplier of the parent com- value chains are the ultimate manifestation pany’s sector. For countries and economies included in each group, see note 8. of the fragmentation of the production pro- cess of MNCs, whereby each subsidiary in the organization produces inputs based on its comparative advantage.25 This fluid move- ment of tangible and intangible inputs within F o rei g n D irect I n v estment , M ultinati o nal C o rp o rati o ns , and   I nn o v ati o n   135 MNCs leads to enhancements in the transfer FIGURE 5.11  Origin of revenues of multilatinas, 2010 of a wide array of technologies and knowl- edge across borders. In 2010, by revenue The market-­ d riven orientation of mul- 0.3% Brazil tilatinas has also led to a specific sequence 2.7% 0.2% Mexico 1.3 % 1.9% 0.0% 1.9% of geographical expansion in which firms 0.9% Chile 1.4% establish operations in neighboring countries 2.4% United States before crossing beyond regional borders. In 3.2% Colombia fact, most subsidiaries of multilatinas remain Spain constrained to LAC: nearly 85 percent of 5.5% Argentina 36.1% the revenues of their subsidiaries come from Bermuda within the region (figure 5.11). Although this 8.0% United Kingdom pattern leads to regional integration, which Canada could have important benefits, it prevents High-income multilatinas from seizing some of the poten- economies 12.7% tial innovation boosters from operating in Other LAC 22.7% ECA non-LAC, especially high-­ income, countries. EAP4 China Innovation deficit Other The scope for productivity gains from the emergence of multilatinas is hindered by Source: World Bank, based on data from Orbis. Note: For countries and economies included in each group, see note 8. their underperformance in terms of innova- tion. Maloney and Sarrias (2013) show that although they engage in better management manufacturing sector invest on average only practices than local firms, multilatinas lag $0.6 per $10,000 of revenue (figure 5.13). owned firms in all LAC countries for foreign-­ This spending stands in sharp contrast with which data were available (figure 5.12) (See R&D intensity by manufacturing MNCs chapter 3 for a detailed description of the income economies and even other from high-­ management practices data.) developing countries and regions, such as Multilatinas also fall behind their coun- China and EAP4. For example, MNCs terparts from other regions in terms of from EAP4, the region with the lowest aver- R&D investments. Multilatinas from the age R&D intensity among the three regions FIGURE 5.12  Management practices by firms in the United States and selected countries in Latin America and the Caribbean, by type of firm a. Argentina b. Brazil c. Chile d. Mexico e. United States Local rms Local rms Local rms Local rms Local rms Domestic Domestic Domestic Domestic Domestic multinationals multinationals multinationals multinationals multinationals Foreign Foreign Foreign Foreign Foreign multinationals multinationals multinationals multinationals multinationals 0 1 2 3 4 0 1 2 3 4 0 1 2 3 4 0 1 2 3 4 0 1 2 3 4 Mean management score Source: Maloney and Sarrias 2013. 136   LATIN AMERIC AN ENTREPRENEURS FIGURE 5.13  Research and development by multinational by some LAC–­ specific characteristics that corporations in selected country groups, countries, and economies, prevent them from excelling. 2010–11 Understanding the exact causes hamper- ing innovation in LAC or the exact policy interventions to relax these constraints is US$ of R&D per ten thousand US$ of revenue 30 a daunting task that goes beyond the scope 25 of this report. However, chapter 6 discusses and characterizes the key factors that may 20 be hindering LAC’s innovation potential 15 and puts on the table broad policy areas of intervention. 10 5 Notes 1. Freund and Weinhold (2002) document the 0 High- China EAP4 ECA LAC5 India positive effect of the expansion of the Inter- income net on trade in services. Hummels (2007) economies documents the effect of reductions in trans- port costs on trade. Sachs and Warner (1995) Source: World Bank, based on data from Orbis. describe the economic reforms undertaken in Note: For countries and economies included in each group, see note 8. the 1990s. 2. Many parts of this chapter treat FDI and MNCs as if they were the same thing. There mentioned earlier, invest $17 on R&D for are, however, important differences to keep in every $10,000 of revenue—­ almost 30 times mind when analyzing the effects of the two the R&D intensity of multilatinas.26 variables. FDI is a form of investment that In sum, the picture emerging from this creates an asset held by the home economy. analysis is that the scope for multilatinas This asset can come from the creation of a to generate technological and productiv- new firm or project or from the acquisition ity spillovers to LAC countries is limited. of an existing firm or project. FDI does not Multilatinas are concentrated among very necessarily imply control over the firm or few countries (Brazil, Chile, Colombia, and project. In contrast, multinational activity is Mexico), and the size of these firms is smaller associated with control of production and employment decisions in the host economy than predicted by the level of development of by a foreign-­owned firm. their country of origin. Moreover, the expan- 3. This calculation was made using Bureau van sion of multilatinas is horizontal and oriented Dijk’s Orbis dataset. For more information on toward regional markets, limiting the scope the data, see box 2.1 in chapter 2. for technological gains from participating in 4. This chapter relies on two primary data global value chains and serving high-­ income, sources: the World Bank’s Enterprise Surveys technologically advanced, economies. Multi- and Orbis. For more information on these latinas are also less likely than their counter- datasets, see box 2.1 in chapter 2. parts from other regions to innovate. 5. For each variable of interest, we ran a regres- sion using all countries for which information was available, controlling for the natural loga- Concluding remarks rithm of GDP in constant 2000 U.S. dollars and the natural logarithm of population. The results presented in this chapter lead to 6. Romer (1993), for instance, argues that one important conclusion: something in the MNC presence can lead to a narrowing of business environment in LAC deters inno- the “object gap” and “ideas gap” in develop- vation even among the high-­ end segment of ing countries. The “object gap” refers to the transformational firms. Affiliates of foreign shortage of physical goods, such as factories MNCs and multilatinas alike are constrained and roads, in developing countries. The “ideas F o rei g n D irect I n v estment , M ultinati o nal C o rp o rati o ns , and   I nn o v ati o n   137 gap” refers to the shortage of knowledge used be reflecting differences in the propensity to to create value added in the modern economy. conduct R&D instead of capturing differences 7. Following the literature in international eco- across types of firms. Industry fixed effects nomics, we label the country that receives correct for the fact that MNC affiliates may the MNC or FDI the “host country” and the have a propensity to locate in sectors that have country of origin of the capital the “home a natural bias toward, say, conducting R&D. country.” 11. Distance to headquarters may also be a fac- 8. Throughout this chapter we use the follow- tor explaining these patterns. For instance, ing groups of economies unless otherwise Keller and Yeaple (2013) explore the relation noted. LAC5 includes Argentina, Brazil, Chile, between trade costs, which are expected to be Colombia, and Mexico. Other LAC includes associated with distance, and the way in which Bolivia, Costa Rica, the Dominican Republic, knowledge is produced and transferred within Ecuador, El Salvador, Guatemala, Honduras, the boundaries of MNCs. The theoretical Nicaragua, Paraguay, Peru, Uruguay, and model they present predicts that subsidiaries in República Bolivariana de Venezuela. Carib- locations farther away from headquarters will bean includes Antigua and Barbuda, Cuba, rely less on imported knowledge, embodied in Dominica, Grenada, Guyana, Haiti, Jamaica, sophisticated goods, and more on knowledge St. Kitts and Nevis, St. Lucia, St. Vincent and produced in the host economy. This may be the Grenadines, Suriname, and Trinidad and one reason why U.S. MNCs have located their Tobago. ECA (Eastern Europe and Central R&D more in Asia or the Middle East and less Asia) includes Albania, Armenia, Azerbaijan, in LAC. Belarus, Bosnia and Herzegovina, Bulgaria, 12. For instance, industry wages in Mexico and Georgia, Kazakhstan, Latvia, Lithuania, FYR República Bolivariana de Venezuela have been Macedonia, Moldova, Romania, the Russian shown to increase with foreign production Federation, Serbia, Turkey, Turkmenistan, (Aitken, Harrison, and Lipsey 1996). Ukraine, and Uzbekistan. EAP4 includes Indo- 13. Ramondo (2009) shows that exit rates among nesia, Malaysia, the Philippines, and Thailand. the least productive firms in Chile increase High-income economies include Australia; with foreign presence. Canada; Hong Kong SAR, China; Israel; Japan; 14. In addition, competition could lead to dynamic the Republic of Korea; Kuwait; New Zealand; changes in productivity. The expectation of Oman; Saudi Arabia; Singapore; Switzerland; lower future profits causes the productiv- the United Arab Emirates; the United States; ity of new entrants to be higher. Competi- and all countries in the European Union not tion also reshapes the innovation decision of included in ECA. The set of economies from entrants. The direction of this change depends each group used in figures throughout this on whether the “escape competition” effect chapter varies according to data availability. faced by incumbents outweighs the potentially 9. MNC affiliates have a productivity and man- lower postinnovation profits and higher costs agerial advantage over local firms in other of innovation these firms face. developing and high-­ income regions as well. 15. Other empirical studies find similar results. Lipsey (2002), for instance, provides a thor- Görg and Strobl (2005) find that firms in ough review of the empirical work on pro- Ghana run by owners with past MNC affili- ductivity differences among foreign-­ owned ation have higher productivity than compa- and local firms. He finds that foreign-­ owned rable local firms. Balsvik (2011) finds that the firms are more productive than local firms productivity of local producers in Norway is almost everywhere. Bloom and others (2012) positively correlated with the share of employ- examine management differences in 16 non- ees with prior MNC experience. LAC countries. Their results are similar to the 16. The empirical literature finds almost com- results for LAC countries. plete support for the presence of productiv- 10. Country fixed effects are included to take ity spillovers of MNCs through backward account of the fact that on average, some linkages (to local suppliers); such spillovers countries may have a larger share of MNCs have been found in Brazil (Lopez-­ Cordova or a greater propensity to, say, conduct and Mesquita Moreira 2004), Colombia R&D. If this is the case, differences in R&D (Kugler 2006), and Mexico (Lopez-­ Cordova between MNC affiliates and local firms may 2003). The evidence for horizontal spillovers 138   LATIN AMERIC AN ENTREPRENEURS and spillovers through forward linkages 23. Van Pottelsberghe de la Potterie and Lich- is less conclusive. Evidence of horizon- tenberg (2001) use data for 13 industrial tal spillovers from MNC activity has been economies and show that the growth effect found in Brazil (Lopez-­ Cordova and Mes- of knowledge spillovers from outward FDI is quita Moreira 2004) and Mexico (Lopez-­ positive and greater than that of inward FDI. Cordova 2003) following periods of trade 24. Classification of sectors as upward or down- integration. Ramondo (2009) finds posi- ward relative to the headquarters was done tive horizontal spillovers on Chilean firms. using U.S. input-­ output tables. We assumed Kugler (2006) finds no significant horizon- that the sectoral linkages observed in the United tal spillovers from MNC activity on Colom- States are similar to those in other countries. bian manufacturing firms. Lopez-­ Cordova 25. A consequence of the emergence of global and Mesquita Moreira (2004) find positive value chains is the increased importance of and significant spillovers for Mexico but not intrafirm trade in the global economy. Brazil. (For a comprehensive review of the 26. The low average R&D intensity observed in literature on spillovers, see Harrison and multilatinas is driven partly by the prevalence Rodriguez-­Clare 2010.) of multilatinas with zero R&D. Although the 17. The estimated contributions of knowledge average R&D intensity of multilatinas is much transfers to aggregate productivity from higher once firms with zero R&D are excluded, Alfaro and Chen (2013) are likely to have it remains much lower than other regions. an upward bias, because small firms in LAC, which are more prone to suffer from factor market competition, are underrepresented in References Orbis. Taking into account that the negative Aedo, C., and I. Walker. 2012. Skills for the 21st impact of MNC is underestimated in the anal- Century in Latin America and the Caribbean. ysis, we could consider the above estimates as Washington, DC: World Bank. upper bounds of the true effects. Aitken, B., G. Hanson, and A. Harrison. 1997. 18. Borensztein, De Gregorio, and Lee (1998) “Spillovers, Foreign Investment, and Export find that the relation between inward FDI Behavior.” Journal of International Econom- and growth is positive and significant only for 32. ics 43 (1): 103–­ countries with a minimum level of education. Aitken, B., and A. Harrison. 1999. “Do Domestic Alfaro and others (2003) find a similar result Firms Benefit from Direct Foreign Investment? for financial development. They show that the Evidence from Venezuela.” American Eco- effect of FDI on growth is higher in countries nomic Review 89 (3): 605–­ 18. with a higher level of financial development. Aitken, B., A. Harrison, and R. Lipsey. 1996. 19. Aedo and Walker (2012) show that LAC high “Wages and Foreign Ownership: A Com- school students score below their peers from parative Study of Mexico, Venezuela and the other countries on the standardized math and United States.” Journal of International Eco- reading Programme for International Student nomics 40 (3–­4): 345–­71. Assessment (PISA) tests. Chapter 2 discusses Alfaro, L., C. Areendam, S. Kalemli-­ Ozcan, and the region’s underperformance in terms of S. Selin. 2003. “FDI and Economic Growth: financial development. The Role of Local Financial Markets.” Journal 20. Javorcik and Spatareanu (2005) use a survey of International Economics 61 (1): 512–­ 33. of firms in the Czech Republic and Lithu- Alfaro, L., and M. Chen. 2013. “Market Real- ania to show that the likelihood of a local location and Knowledge Spillovers: The Gains owned firm acting as a supplier for a foreign-­ from Multinational Production.” NBER Work- firm increases with the local firm’s access to ing Paper 18207, National Bureau of Economic financing. Research, Cambridge, MA. 21. Blyde, Kugler, and Stein (2004) find that spill- Alfaro, L., and E. Hammel. 2006. “Latin Ameri- overs from MNC activity in República Boli- can Multinationals.” In The Latin American variana de Venezuela are concentrated among Competitiveness Report, Essay 2.11. New large firms. Kokko, Zejan, and Tansini (2001) York: Oxford University Press for the World find a similar result for Uruguay. Economic Forum and the Center for Interna- 22. Alfaro and Hammel (2006) suggest that access tional Development. to finance is one of the motivations for firms Balsvik, R. 2011. “Is Labor Mobility a Channel from LAC to start operating in foreign markets. for Spillovers from Multinationals? Evidence F o rei g n D irect I n v estment , M ultinati o nal C o rp o rati o ns , and   I nn o v ati o n   139 from Norwegian Manufacturing.” Review of Dunning, J. 1988. “The Eclectic Paradigm of Economics and Statistics 93 (1): 285–­ 97. International Production: A Restatement and BCG (Boston Consulting Group). 2006. The New Some Possible Extensions.” Journal of Inter- Global Challenges: How 100 Top Compa- national Business Studies 19 (1): 1–­31. nies from Rapidly Developing Economies Are Enterprise Surveys (database). World Bank Group, Changing the World. Boston: BCG. Washington, DC. http://www.enterprise Blalock, G., and P. J. Gertler. 2005. “Foreign surveys.org/. Direct Investment and Externalities: The Case Fosfuri, A., M. Motta, and T. Rønde. 2001. “For- for Public Intervention.” In Does Foreign eign Direct Investment and Spillovers through Investment Promote Development? edited by Workers’ Mobility.” Journal of International T. Moran, E. Graham, and M. Blomström, Economics 53: 205–­22. 73–­ 106. Washington, DC: Center for Global Freund, C., and D. Weinhold. 2002. “The Inter- Development. net and International Trade in Services.” Blomström, M. 1983. “Foreign Investment, Tech- American Economic Review 92 (2): 236–­ 40. nical Efficiency and Structural Change: Evi- Görg, H., and E. Strobl. 2005. “Spillovers from dence from the Mexican Manufacturing Indus- Foreign Firms through Worker Mobility: An try.” Ekonomiska Studier, Utgivna av Nation- Empirical Investigation.” Scandinavian Jour- alekonomiska Institutionen 12, Vid Göteborgs nal of Economics 107 (4): 693–­ 709. Universitet, Gothenberg, Sweden. Harding, T., and B. Javorcik. 2011. “Roll out the Bloom, N., C. Genakos, R. Sadun, and J. Van Red Carpet and They Will Come: Investment Reenen. 2012. “Management Practices across Promotion and FDI Inflows.” Economic Jour- Firms and Countries.” Academy of Manage- nal 121 (557): 1445–­ 76. ment Perspectives 26 (1): 12–­ 33. ———. 2012. “Foreign Direct Investment and Blyde, J., M. Kugler, and E. Stein. 2004. “Export- Export Upgrading.” Review of Economics ing versus Outsourcing by MNC Subsidiaries: 80. and Statistics 94 (4): 964–­ Which Determines FDI Spillovers?” Discus- ———. 2013. “Investment Promotion and FDI sion Paper in Economics and Econometrics Inflows: Quality Matters.” CESifo Economic 0411, University of Southampton, School of Studies 59 (2): 337–­59. Social Sciences, Economics Division, South- Harrison, A., and A. Rodriguez-­ C lare. 2010. ampton, U.K. “Trade, Foreign Investment, and Indus- Borensztein, E., J. De Gregorio, and J.-W. Lee. trial Policy.” In Handbook of Development 1998. “How Does Foreign Investment Affect Economics, vol. 5, edited by D. Rodrik and Growth?” Journal of International Economics M. Rosenzweig, 4040 –­ 214. Amsterdam: 45 (1): 115–­72. Elsevier. Busso, M., L. Madrigal, and C. Pagés-­ Serra. 2012. Helpman, E., M. Melitz, and S. Yeaple. 2004. “Productivity and Resource Misallocation “Exports versus FDI with Heterogeneous in Latin America.” IDB Working Paper IDB–­ Firms.” American Economic Review 94 (1): WP–­ A merican Development Bank, 306, Inter-­ 300–­16. Washington, DC. Hummels, D. 2007. “Transportation Costs and Casanova, L., M. Fraser, H. Hoeber, A. Goldstein, International Trade in the Second Era of Glo- R. Molina, C. Arrunda, and A. Almeida. 2009. balization.” Journal of Economic Perspectives From Multilatinas to Global Latinas: The New 21 (3): 131–­54. Latin American Multinationals. Washing­- Javorcik, B. S. 2004. “Does Foreign Direct Invest- ton, DC: Inter-American Development Bank. ment Increase the Productivity of Domestic Chen, H., and D. L. Swenson. 2007. “Multina- Firms? In Search of Spillovers through Back- tional Firms and New Chinese Export Transac- ward Linkages.” American Economic Review tions.” Canadian Journal of Economics 41 (2): 94 (3): 605–­27. 596–­618. ———. 2010. “Foreign Direct Investment and de la Torre, A., T. Didier, C. Calderon, T. Cordella, International Technology Transfer.” In Ency- and S. Pienknagura. 2011. “Latin America and clopaedia of Financial Globalization, edited the Caribbean’s Long-­ Term Growth: Made by G. Caprio. Amsterdam: Elsevier. in China?” World Bank, Office of the Chief Javorcik, B., and M. Spatareanu. 2005. “Dis- Economist, Latin America and the Caribbean entangling FDI Spillover Effects: What Do Region, Washington, DC. Firm Perceptions Tell Us?” In Does Foreign 140   LATIN AMERIC AN ENTREPRENEURS Investment Promote Development? edited by National Science Board. 2012. “Science and Engi- T. Moran, E. Graham, and M. Blomström, neering Indicators 2012.” National Science 45–­ 72. Washington, DC: Center for Global Foundation, Arlington, VA. Development. Orbis (database). Bureau van Dijk, Amsterdam. Keller, W., and S. Yeaple. 2013. “The Gravity of http://w w w.bvdinfo.com /en-­u s/products Knowledge.” American Economic Review 103 /company-­information/international/orbis-­(1). (4): 1414-­1444. Pagés-­Serra, C., ed. 2010. The Age of Productiv- Kenyon, T., and Y. Margalit. 2012. “Motivations ity: Transforming Economies from the Bottom and Impediments for Brazilian Foreign Invest- Up. Washington, DC: Inter-American Devel- ment: Evidence from a New Firm-­ Level Study.” opment Bank. Unpublished manuscript. Poole, J. 2013. “Knowledge Transfers from Mul- Kokko, A., M. Zejan, and R. Tansini. 2001. tinational to Domestic Firms: Evidence from “Trade Regimes and Spillover Effects of FDI: Worker Mobility.” Review of Economics and Evidence from Uruguay.” Weltwirtschaftliches 406. Statistics 95 (2): 393–­ Archiv 137 (1): 124–­ 49. Ramondo, N. 2009. “The Effects of Foreign Kugler, M. 2006. “Spillovers from Foreign Direct Plants on Industry Productivity: Evidence from Investment: Within or between Industries?” Chile.” Scandinavian Journal of Economics Journal of Development Economics 80 (2): 111 (4): 789–­ 809. 444–­77. Rodriguez-­ Clare, A. 1996. “Multinationals, Link- Calva, and A. Rodriguez-­ Larrain, F., L. F. Lopez-­ ages, and Economic Development.” American Clare. 2001. “Intel: A Case Study of Foreign Economic Review 86 (4): 852–­ 73. Direct Investment in Central America.” In Romer, P. 1993. “Idea Gaps and Object Gaps in Economic Development in Central Amer- Economic Development.” Journal of Monetary ica, vol. I: Growth and Internationalization, Economics 32 (3): 543–­ 73. edited by Felipe Larrain B. Cambridge, MA: Sachs, J., and A. Warner. 1995. “Economic Harvard University Press. Reform and the Process of Global Integra- Lipsey, R. E. 2002. “Home and Host Country tion.” Brookings Papers on Economic Activity Effects of FDI.” NBER Working Paper 9293, 1: 1–­118. National Bureau of Economic Research, Cam- Santiso, J. 2008. “La emergencia de la ‘multilati- bridge, MA. nas.’” Revista de la CEPAL 95: 7–­ 30. Lopez-­ Cordova, J. E. 2003. “NAFTA and Manu- Spar, D. 1998. “Attracting High Technology facturing Productivity in Mexico.” Economía Investment: Intel’s Costa Rican Plant.” World 4 (1): 55–­98. Bank, Foreign Investment Advisory Service Lopez-­ Cordova, J. E., and M. Mesquita Moreira. Occasional Paper 11, Washington, DC. 2004. “Regional Integration and Productiv- UNESCO (United Nations Educational, Scien- ity: The Experiences of Brazil and Mexico.” In tific and Cultural Organization). 2013. Data Integrating the Americas: FTAA and Beyond, Center, Institute for Statistics. http://stats.uis edited by A. Estevadeordal, D. Rodrick, .unesco.org/. A. Taylor, and A. Velascods. Cambridge, van Pottelsberghe de la Potterie, B., and F. Lich- MA: David Rockefeller Center Series on Latin tenberg. 2001. “Does Foreign Direct Invest- American Studies, Harvard University. ment Transfer Technology across Borders?” Maloney, W. F., and M. Sarrias. 2013. “Conver- Review of Economics and Statistics 83 (3): gence in Management Quality across Coun- 490–­97. tries.” Unpublished manuscript. World Bank Group. 2006. The Impact of Intel in Moran, T. 2001. Parental Supervision: The New Costa Rica: Nine Years after the Decision to Paradigm for Foreign Direct Investment and Invest. Washington, DC: World Bank Group. Development. Washington, DC: Institute for ———. 2012. Global Investment Promotion Best International Economics. Practices 2012. Washington, DC: World Bank Group. Toward an Enabling Environment for Innovative Entrepreneurs 6 Creating an enabling environment for innovative entrepreneurship is difficult because it involves multiple policy areas that interact in complex ways. Of the usual suspects—­ inadequate pro- tection of intellectual property rights and contracts, access to finance, competition, openness to international trade, and human capital—­ the region appears to underperform other regions most clearly on the human capital front and the lack of competition in nontradable industries. More research is needed before definitive conclusions can be reached, but some evidence sug- gests that the small share of engineers in the population and excessive concentration of domes- tic nontradable markets in a few firms may be major factors behind the region’s innovation deficit. Intellectual property rights might also be important, but more research is definitely needed on this complex policy area. I n an op-­ e d piece in the New York firm. Eventually, Shiller and his two partners Times, Nobel Prize laureate Robert sold the company (and the rights to use his Shiller argued that innovation remains economic ideas about how to measure the the engine of growth in market economies. evolution of the real estate market in the His opening salvo was “Capitalism is cul- United States) for an undisclosed but pre- ture. To sustain it, laws and institutions are sumably attractive sum. His emphasis on the important, but the most fundamental role is spirit of capitalism—­ a Weberian idea–­ a nd played by the basic human spirit of indepen- his lukewarm tribute to laws and institutions, dence and initiative” (Shiller 2013). however, suggest that Shiller is a bit skeptical Shiller was writing as both an academic about the ability of public policies to become and an entrepreneur, narrating his attempts drivers of innovation. to secure financing for a new business ven- Javier—­ t he fictional entrepreneur intro- ture that would commercialize an idea that duced in chapter 1—­ c aught the Weberian emerged out of his academic research. He spirit early on, making choices that were expressed frustration with the lack of interest qualitatively different from those of his sib- shown by potential investors until he mort- lings. He took financial risks, which paid gaged his home to help finance the young off handsomely. As suggested in chapter 2, 141 142   LATIN AMERIC AN ENTREPRENEURS entrepreneurs like Javier are rare but impor- What are the elements of an tant for the development of Latin America enabling environment for and the Caribbean (LAC): although their innovative entrepreneurs? numbers are small, they employ more than half of all formal sector salaried workers in Chapter 1 concluded that regulatory barri- the region. If LAC had more entrepreneurs ers to entry are unlikely to be a major con- like Javier, the number of formal salaried jobs straint for LAC entrepreneurship and that it would probably be larger and the number of is difficult to find conclusive evidence that the low-­growth entrepreneurs smaller. region’s culture is less inclined toward entre- Chapter 3 argued that LAC can be char- preneurship than other cultures. Rather, we acterized by both its large number of entre- must search for answers in Shiller’s “laws preneurs and by their underperformance in and institutions.” Chapter 2 provided some terms of innovation. Shiller seems to believe additional clues. It identified important corre- that entrepreneurship and innovation go lates of innovation by incumbent firms, such hand in hand, drawing little distinction as regulation of entry, competition, access to between the two. But some dimensions of finance (especially by young enterprises), and entrepreneurship (such as firm creation and entrepreneurial skills. survival) are clearly present in LAC even as there is a deficit along the innovation dimen- Interactions and complexity sion. Although we embrace the presump- Pinpointing the enablers of innovative entre- tion that the spirit of innovation is a driver preneurship is fraught with complexity. The of modern market economies around the difficulty may stem from the intricate inter- world, not just in the United States, a chal- actions between the various dimensions of lenge for policy makers remains figuring the enabling environment that matter for out how to shape policies, including laws innovation; it may also reflect the fact that and institutions, to enhance the incidence both entrepreneurial innovation and its of transformational entrepreneurship so possible determinants may be affected by that surviving entrepreneurs become true common factors and, hence, jointly deter- innovators. mined. For instance, an economy’s laws and Chapters 4 and 5 focused on the region’s institutions—­ its contractual environment—­ top high-­ end entrepreneurs—­ namely, large might simultaneously affect firms’ access formal enterprises that compete in global to credit and innovation. Young firms con- markets through exports of goods and capi- sidering whether to invest in research and tal. They show that LAC economies tend to development (R&D) to develop new prod- underperform in terms of export entrepre- ucts or services might have access to credit neurship and that multinational corpora- in economies where intellectual property tions headquartered in LAC (multilatinas) rights are well established in law. Similarly, tend to be less innovative than similar firms an institutional and legal environment in elsewhere. which contracts are unevenly enforced or What should leaders and policy makers in economic transactions depend on informal LAC focus on? Where should they look for enforcement mechanisms might limit access insights into the fundamental drivers of both to finance for young firms and reduce risky entrepreneurship and innovation by high-­ end investments in innovation. A long history entrepreneurs in the region? Perhaps looking of macroeconomic and financial instabil- at the region’s laws and institutions is not a ity might undercut both the incentives to bad place to start. This chapter examines ele- innovate—­ by, for instance, widening the ments that might be the cornerstones of an gap between downside risks and upside enabling environment to foster innovative potential—­ a nd the availability of suitable entrepreneurs. financial services for firms. Furthermore, in T o w ar d an E na b lin g E n v ir o nment f o r I nn o v ati v e E ntrepreneurs   143 economic environments in which a few firms FIGURE 6.1  Actual and benchmarked access to credit by young enjoy economic rents because of the lack of firms in selected countries antitrust laws or competition laws that lack teeth, one might observe little innovation, Jamaica LAC countries because high-­ e nd entrepreneurs might be Honduras Other countries Uzbekistan able to make hefty profits with little innova- Benchmark Suriname tive effort and would thus see little benefit Ukraine to changing the mix of products or services Kazakhstan they offer. The numerous potential elements Indonesia Grenada of an enabling environment for innovative Philippines entrepreneurship mean that any analysis St. Lucia must remain speculative but comprehensive. Russian Federation Dominica The following sections discuss potential Poland elements of such an enabling environment. Argentina They cover a broad swath of economic and Azerbaijan institutional characteristics, including access Mexico St. Vincent and the Grenadines to finance, intellectual property rights, con- Albania tractual certainty, competition in nontradable Moldova industries, competition in tradable industries, Latvia Armenia and human capital. The discussion focuses Hungary on each topic separately before summariz- Nicaragua ing the main findings of the benchmarking Bulgaria exercises. Romania Guatemala Colombia Turkey Access to finance by young firms Belarus El Salvador Entrepreneurs decide whether to invest in Antigua and Barbuda innovation and, if so, how much to invest. St. Kitts and Nevis A key consideration is the costs of the nec- Uruguay Dominican Republic essary investments in innovation (investment Lithuania in R&D). Financial markets play a role in Ecuador determining this cost. In some environments, Trinidad and Tobago young firms may be credit constrained or Guyana Croatia financial institutions may lend to them only Slovak Republic at exorbitant interest rates, thus raising the Macedonia, FYR costs of innovation investments. Lack of Georgia Bosnia and Herzegovina access to financing was Shiller’s main obsta- Serbia cle to commercializing his ideas. Paraguay To push the debate forward without offer- Costa Rica ing country-­specific policy prescriptions, we Chile Venezuela, RB rely on simple international benchmarking. Slovenia Figure 6.1 presents evidence compiled from Czech Republic the World Bank’s Enterprise Survey data- Peru Brazil base. It shows the percentage of surveyed Bolivia firms in each country that reported being 0 20 40 60 80 100 less than five years old and having access to Firms less than 5 years old (%) credit. The bars represent the actual share of firms that meet these criteria; the dots repre- Source: World Bank, based on data from World Development Indicators and 2006–­10 World Bank sent the share of firms predicted by country Enterprise Surveys. Note: Bars show the percentage of firms that are five years old or younger and have access to credit. Dots show the predicted percentage of firms from a regression that includes (the log of) population and gross domestic product (GDP) adjusted for purchasing power parity as explanatory variables. The regression used all available countries. The figure presents only comparator countries. LAC = Latin America and the Caribbean. 144   LATIN AMERIC AN ENTREPRENEURS characteristics. The LAC economies in the the problem may affect many developing sample (the dark bars) tend to be above the countries, not only countries in LAC. median and in most cases near or above their predicted shares. Some notable exceptions Property rights and are several small Caribbean economies plus contractual certainty Jamaica and Mexico. It is difficult to conclude that for the The expected payoff from an investment in region as a whole, lack of access to finance innovation affects an entrepreneur’s deci- is a main driver of the underperformance in sion to take the necessary risk. It depends innovation. To be sure, as documented in on the probability of discovering a profitable the World Bank’s flagship report on finan- idea as well as on the ability of inventors to cial development in LAC (de la Torre, Ize, appropriate the commercial windfalls of their and Schmukler 2012), the region’s gap in investments in innovation. The most directly bank credit to the private sector (relative to relevant set of laws and institutions is argu- a carefully constructed international bench- ably related to intellectual property rights mark) is not only significant but appears to (IPRs) and contractual certainty. have been growing over the past 15 years. Laws and regulations define the number However, much of this gap appears to be and types of industries subject to IPRs, the explained by LAC’s turbulent macro and number of international agreements on IPRs financial history and by a shortage of prom- to which a country is a signatory, and the ising productive projects (that is, a shortage legal recourse available to patent holders in of innovation) rather than by credit ration- case of an alleged infringement of their IPRs. ing and credit supply-­ side constraints per se. Figure 6.2 displays the benchmarking of Moreover, a relevant constraint for bank Park’s (2008) index of IPRs. This index is the credit supply in LAC may be weaknesses in sum of five components: coverage of patents the contractual (rather than the informa- in eight industries; participation in five inter- tional) environment, and contractual weak- national IPR treaties; duration of protection nesses and property rights may be a common (relative to a global standard, such as 15–­ 20 factor that undermines both the supply of years for patents); the existence of up to three bank credit and entrepreneurial innovation, enforcement mechanisms; and the existence as discussed below. of up to three types of restrictions on patent Finance can come from various sources, rights. As of 2005, the Park index scores of not just from banks but also from venture all LAC countries in the sample except Chile capital and capital markets. Chapter 3 dis- were below the median. Some of these coun- cussed recent data on the size and destination tries appear to have de facto IPRs that exceed of venture capital deals in LAC, concluding those predicted by their size and level of devel- that such transactions are large and pursue opment, however. These countries include traditional (natural resource–­ related) sec- (in ascending order of the index score) Haiti, tors, presumably because expected profits are Jamaica, Bolivia, El Salvador, Colombia, high. Ongoing research by Didier, Levine, Ecuador, and Argentina. Although LAC does and Schmukler (2013) in a sample of six not underperform relative to the predicted lev- countries (Argentina, Brazil, Chile, Colom- els of IPR protection, it lags comparator coun- bia, Mexico, and Peru) shows that firms tries in actual terms, which may explain why that issue bonds or equity tend to be much it also lags in innovation by high-­ end entre- larger than those that do not (3,484 versus preneurs. The IPR policy area might therefore 859 employees on average). This relationship be a potentially fruitful avenue to explore. is also apparent in the authors’ global sample Intellectual property rights, however, of 51 countries, however. Thus, although it is are complex relative to other legal areas. plausible that financial markets do not meet Hence, the establishment of well-­ defined and the financing needs of small and young firms, enforceable intellectual property rights is T o w ar d an E na b lin g E n v ir o nment f o r I nn o v ati v e E ntrepreneurs   145 inextricable from the quality and functional- FIGURE 6.2  Actual and benchmarked index of intellectual ity of the broader legal and judicial system. property rights in selected countries or economies, 2005 If there are deficits in the legal definition and enforcement of more tangible property rights, Guyana LAC countries the difficulties in setting up a suitable intel- Thailand Indonesia Other countries lectual property right system are a fortiori Dominican Republic or economies Costa Rica going to be greater. Thus, fixing intellectual Paraguay Benchmark property right regimes is in many cases likely Haiti Nicaragua to involve accompanying broader reforms to Saudi Arabia the legal and the judiciary frameworks. Honduras Grenada Like the potential effects of IPRs on the Guatemala Venezuela, RB expected payoff of investments in innovation, Peru an economy’s contractual environment can Jamaica Uruguay affect economic incentives for private invest- Bolivia El Salvador ments, including in R&D. The contractual Malaysia environment can also affect other elements Brazil Ukraine of the enabling environment for innovative Russian Federation entrepreneurship, such as access to finance. Colombia Ecuador Figure 6.3 presents international com- Trinidad and Tobago Hong Kong SAR, China parisons for an indicator of contract cer- India tainty from the International Country Risk Mexico Argentina Guide (ICRG), a private firm that assesses Lithuania the sources of country risk for international Turkey New Zealand investors and other private sector clients. This China Israel type of “expert” indicator is imperfect; it is Australia used because it is difficult to find alternative Norway Romania indicators of such a complex phenomenon. Philippines Poland Figure 6.3 suggests that LAC as a whole Slovak Republic does not underperform in terms of contract Singapore Chile certainty. It suggests that there are two types Greece of LAC countries: those with high contract Switzerland Spain viability and those that underperform. The Austria Czech Republic high contract viability group includes Chile, Korea, Rep. the Dominican Republic, Mexico, Trinidad Portugal Germany and Tobago, and Uruguay, among others. Hungary Bulgaria The underperformers include Brazil, Costa Sweden Rica, and El Salvador, among others. This United Kingdom Belgium bifurcated picture contrasts in part with France the findings of the 2011 flagship report on Canada Italy financial development in LAC (de la Torre, Japan Finland Ize, and Schmukler 2012), which identi- Ireland fies contractual weaknesses as an important Denmark Netherlands driver of credit depth and access to long-­ term United States finance, which itself may affect entrepre- 2 3 4 5 neurship. These nuances suggest that more Park index (as of 2005) research needs to be done to understand the complex and multifaceted relation between Source: World Bank, based on data from World Development Indicators and Park 2008. Note: Bars show the 2005 Park index for each country. Dots show the predicted percentage of firms the contractual environment (in particular from a regression that includes (the log of) population and gross domestic product (GDP) adjusted the governance side), access to finance, and for purchasing power parity as explanatory variables. The regression used all available countries. The figure presents only comparator countries. LAC = Latin America and the Caribbean. entrepreneurship. FIGURE 6.3  Actual and benchmarked contract certainty in selected countries or economies, 2012 Venezuela, RB LAC countries Competition in nontradable industries Bolivia Argentina Other countries Portugal Potential innovators assess the potential pay- or economies Honduras Haiti Benchmark off from innovation relative to the profits from China continuing to produce the same set of prod- Belarus Greece ucts or services with the same level of quality, Ecuador Indonesia technology, and management practices. When Spain Peru competitive pressures are low, enterprises may Costa Rica Hungary choose to invest little in innovation, enjoying El Salvador the rents from market power. Although it is Italy Moldova plausible that too much competition can actu- Ukraine United Kingdom ally reduce incentives to innovate by firms, Slovenia Russian Federation especially for firms with low capabilities, it is Suriname Australia likely that most of LAC suffers from too little Kuwait Cuba competition (see chapter 3). Ireland This section benchmarks LAC economies Kazakhstan Guyana in terms of market concentration in indus- Serbia France tries that are arguably not subject to inter- Philippines Brazil national competition (the following section Bulgaria discusses the role of competition in trad- Romania India Latvia ables). The distinction between tradables and Austria nontradables is important. Domestic market Belgium Slovak Republic concentration could be high in the sense that Mexico Paraguay few domestic firms participate in an industry, Nicaragua but if domestic firms compete with imports, Armenia Turkey Colombia domestic market concentration would be a Thailand poor proxy for competition. To avoid this Lithuania Netherlands problem, we examine data from 17 sectors Czech Republic Albania that seem to be nontradable service industries Poland Guatemala (and for which there is sufficient information Jamaica Dominican Republic across countries to conduct the benchmark- Denmark Finland ing exercises).1 Saudi Arabia The results are shown in figure 6.4. 2 LAC Israel Japan countries seem to have excessively concen- Malaysia Trinidad and Tobago trated domestic markets in nontradables; Korea, Rep. Croatia most countries appear at the upper end of Norway Azerbaijan the distribution of the market concentration Uruguay Hong Kong SAR, China index. Moreover, all but two LAC countries Germany (Colombia and Brazil) exhibit average levels Oman Chile of market concentration that are higher than New Zealand Canada the levels of countries with similar popula- Switzerland United States tions and gross domestic products (GDPs). Singapore Sweden (Argentina appears to have a relatively 1 2 3 4 low level of concentration, but it did not ICRG index (as of 2012) appear in the regression analysis because of data limitations.) Consequently, lack of Source: World Bank, based on data from World Development Indicators and the International Coun- competition appears to be a strong candi- try Risk Guide (ICRG). Note: Bars show the 2012 contract viability index for each country, as reported by ICRG. Dots show date for explaining the region’s lackluster the benchmark predicted by a regression with (log of) population and gross domestic product innovation. (GDP) adjusted for purchasing power parity as the explanatory variables. The regression used all available countries. The figure presents only comparator countries. Bolivia and República Bolivari- ana de Venezuela are not covered by the ICRG data. LAC = Latin America and the Caribbean. 146 T o w ar d an E na b lin g E n v ir o nment f o r I nn o v ati v e E ntrepreneurs   147 Competition in tradables FIGURE 6.4  Actual and benchmarked index of competition in 17 nontradable industries in selected countries or economies Some industries face competition from for- eign competitors. The literature on interna- United States LAC countries tional trade and growth has tended to focus Bulgaria Romania Other countries on the ratio of international trade flows Poland or economies (the sum of imports plus exports) to GDP. Canada Hungary Benchmark This variable tends to rise with the share of Russian Federation Lithuania domestic consumption that is satisfied by Czech Republic Norway imports from abroad and with the share of Colombia Latvia domestic production sold to consumers in Korea, Rep. foreign countries. We use this ratio—­ usually Japan Portugal called the “openness” ratio—­ as the proxy for United Kingdom Switzerland the extent of competition affecting tradable Macedonia, FYR China industries. Italy Ireland Figure 6.5 benchmarks the region’s level Germany of openness as of 2012. By and large, LAC Croatia Serbia countries are either above the median coun- Belarus Thailand try in the sample or have trade shares above Spain Sweden what is expected given their geographic char- Austria Finland acteristics and size. There are two important Netherlands exceptions, Brazil and Colombia. Brazil has Argentina Greece the lowest level of openness in the sample; it Denmark Brazil is underperforming by about 3–­ 5 percentage Bosnia and Herzegovina Singapore points of GDP. Two of its Mercosur (South- Australia Belgium ern Cone Common Market) partners also Turkey appear to be unexceptional. Argentina seems Mexico France to have the level of trade that is expected Philippines Malaysia given its characteristics, and Uruguay should Moldova Israel be trading more than it was in 2012. Hence, New Zealand Kuwait it is potentially relevant for future research to Hong Kong SAR, China assess whether the South American trading Ecuador Kazakhstan bloc could be opened further. Indonesia Chile Colombia is an interesting case, as it Peru Albania belongs to the much touted recent trade ini- Saudi Arabia India tiative called the Pacific Alliance (Alianza Uruguay del Pacifico, in Spanish). This initiative, Dominican Republic Oman launched in 2012, includes Chile, Mexico, Guatemala Bolivia and Peru. Chile and Mexico are extremely Jamaica United Arab Emirates open to trade, given their geographic charac- Paraguay teristics and size. Peru is performing accord- Venezuela, RB Trinidad and Tobago ing to expectations. The Alliance, however, is Nicaragua El Salvador supposed to be a group of countries oriented Costa Rica Honduras toward free trade and deep economic inte- 0 0.2 0.4 0.6 0.8 gration. Colombia has become such a coun- Her ndahl index try, but only in recent years. Its free trade agreement (FTA) with the United States was Source: World Bank, based on data from World Development Indicators and firm-­level data from Orbis. Note: Bars show the average Herfindahl index of concentration of revenues across a selection of approved by the U.S. Congress only in 2011. two-­digit nonfinancial services sectors for which data were available for more than 80 countries. Colombia also has an FTA with Canada, A value of 1 represents a market captured entirely by a single firm (the highest level of concentra- tion); lower values indicate less concentration. Revenues were averaged across 2007–­10. Dots rep- resent a benchmark predicted value from a regression for each sector with (log of) population and gross domestic product (GDP) adjusted for purchasing power parity as explanatory variables. The regression model was estimated for each of 17 sectors separately; the dots are the averages of all sectors. The regression used all available countries. The figure presents only comparator countries. LAC = Latin America and the Caribbean. FIGURE 6.5  Actual and benchmarked index of openness to trade in selected countries, 2012 Brazil LAC countries Japan United States Other countries India Zero-in ated Poisson Argentina which was implemented in August 2011. An Australia Poisson Colombia FTA with the European Union was signed Uzbekistan Negative binomial Russian Federation in August 2013 but has not yet been imple- Philippines Greece mented; an FTA with the Republic of Korea Indonesia China is expected by 2014. Hence, Colombia’s rapid Venezuela, RB Peru move toward free international trade is a United Kingdom Dominican Republic recent phenomenon that may take time to be Turkey reflected in the data. Broadly speaking, how- Spain Italy ever, LAC does not appear to systematically France Norway underperform in terms of openness. Thus, Jamaica Uruguay tradable industries are presumably facing Canada Guatemala tougher competitive pressures than nontrad- Albania Kazakhstan able industries. Azerbaijan Armenia Turkmenistan Suriname Human capital for innovation Portugal Chile Croatia The broad agenda of human capital forma- Ecuador Finland tion in LAC and elsewhere is well known Denmark (see, for example, Aedo and Walker 2012). El Salvador Sweden Bolivia A country’s stock of human capital is often Switzerland Mexico measured by the average years of schooling of Paraguay the labor force (that is, the adult population). Haiti Bosnia and Herzegovina Because the quality of education also matters Serbia Germany for economic performance, researchers often Korea, Rep. Poland look at evidence of quality of education, as Georgia reflected in student scores on internationally Ukraine Honduras Ireland standardized scholastic tests. The few coun- Austria tries in the region that participate in such Moldova Trinidad and Tobago tests tend to perform below high-­ i ncome Nicaragua Malaysia countries and a handful of fast-­ g rowing Thailand Macedonia, FYR emerging market countries. Latvia Costa Rica Human capital for entrepreneurship and Guyana innovation needs to be assessed with some Bulgaria Slovenia Belarus nuance, as it only partially overlaps with gen- Lithuania eral curricula, as discussed in chapter 3. Since Czech Republic Netherlands at least the early 20th century, LAC has had a Hungary Slovak Republic relatively low number of engineers per capita. Belgium More recent data from UNESCO (United 0 50 100 150 200 Nations Educational, Scientific, and Cultural % of GDP Organization) show the average number of Source: World Bank, based on data from International Monetary Fund; Rose 2004; World Develop- engineers per capita from 2008 to 2010 (fig- ment Indicators; and Penn World Tables 7.1. ure 6.6). The few LAC countries included Note: Bars show the openness ratio, calculated as the sum of exports and imports of merchandise over GDP. Dots are benchmarks predicted by the gravity model of openness proposed by Frankel all have fewer engineers than the median and Romer (1999). The regression included the following explanatory variables: log of the area of country and fewer than expected given their the reporting country, log of the area of the partner country, landlocked, common border, and inter- action with border. Fitted values of openness are the sum across partner countries. Models exclude level of development. This finding holds even the following economies: (a) Liberia; Hong Kong SAR, China; and Singapore (outliers); (b) major oil for the region’s relatively large middle-­and producers with production exceeding 200,000 barrels of oil per day in 1985 (following Alcala and Ciccone 2004): Angola, Gabon, Congo, Iraq, Oman, Kuwait, Qatar, Saudi Arabia, and the United high-­income countries, such as Brazil, Chile, Arab Emirates; and (c) countries with populations of less than 500,000. Following Santos Silva and Colombia, and Mexico. 3 Indeed, if these Tenreyro (2006) on the estimation of the gravity model of trade, the figure includes benchmarks from three estimators: zero-­inflated Poisson, Poisson, and negative binomial. LAC = Latin America limited data are reflective of the region as a and the Caribbean. whole, this constraint may be the single most 148 T o w ar d an E na b lin g E n v ir o nment f o r I nn o v ati v e E ntrepreneurs   149 FIGURE 6.6  Actual and benchmarked share of important barrier to innovative entrepreneur- engineers in selected countries, 2008–­10 ship found in this brief empirical tour. What factors underlie the inclination of Honduras LAC countries LAC students toward nonscientific stud- Guyana Other ies? Two stand out: path dependency and Uruguay countries the broader socioeconomic context. For his- El Salvador Benchmark torical reasons, LAC universities have been Brazil locked into an emphasis on the humanities Argentina and law, as well as social, economic, and Indonesia political fields. This tendency may constrain Colombia the ability to switch rapidly to educating Mexico more engineers and scientists. Such a switch Serbia would require very aggressive public policy, Saudi Arabia such as the United States adopted when it Chile Turkey developed mining and engineering stud- Hungary ies in the early 20th century. Young people Netherlands may be attracted to fields of studies that are Armenia relevant to pressing problems faced by their Norway societies, which may explain why LAC may United States have formed more macro than micro econo- Latvia mists and sociologists. Given the progress the Croatia region has made in taming macroinstability, Sweden there may be more incentives for students to Bulgaria embark on scientific careers. Belgium Greece Germany What explains the Denmark region’s innovation gap? Japan The leading suspects Malaysia Portugal At least 13 LAC economies underperform Austria in terms of patenting activity (see chapter 3). Lithuania Table 6.1 indicates the areas in which each Poland country underperforms relative to both the Slovenia median in the global sample and the expected Switzerland level of performance given its size and level Spain of development, among other issue-­ specific Czech Republic relevant factors. New Zealand The first place to look is human capital. Slovak Republic The evidence is not ironclad; because of the Ireland 13 LAC economies that have a deficit in inno- Finland Ukraine vation (measured by their patenting activ- Thailand ity), data are available for only 6. But all six 0 10 20 30 countries exhibit a deficit in the number of Engineering graduates per thousand engineers per capita. (Honduras, which does inhabitants, ages 15–24 not have a deficit in patenting, because of its relatively low GDP per capita, also has a defi- Source: World Bank, based on data from World Development Indicators cit in engineers.) and UNESCO. Note: Bars show average number of engineering graduates per thousand Another place to look is competition. inhabitants, ages 15–­24. Dots show the benchmark predicted by a regres- Of the 13 countries with a deficit in inno- sion with (the log of) population and gross domestic product (GDP) adjusted for purchasing power parity as the explanatory variables. The vation, 10 have a deficit in competition in regression used all available countries. The figure presents only compara- tor countries. LAC = Latin America and the Caribbean. 150   LATIN AMERIC AN ENTREPRENEURS TABLE 6.1  Factors that may account for innovation deficits in 13 countries in Latin American and the Caribbean, 2005 Human capital for innovation Competition (number of Access to Intellectual Contractual in tradables Competition in engineers per Country Patenting finance property rights certainty (openness)a nontradables capita) Bolivia 1 0 0 1 0 1 —­ Brazil 1 0 1 1 1 0 1 Chile 1 0 0 0 0 1 1 Colombia 1 1 0 0 1 0 1 Dominican Republic 1 0 1 0 1 1 —­ Ecuador 1 0 0 1 0 0 —­ El Salvador 1 0 0 1 0 1 1 Guatemala 1 0 1 0 1 1 —­ Mexico 1 1 1 0 0 1 1 Paraguay 1 0 1 0 0 1 —­ Peru 1 0 1 1 0 1 —­ Uruguay 1 0 1 0 1 1 1 Venezuela, RB 1 0 1 1 1 1 —­ Note: A 1 indicates a variable in which a country is below the median country and below the level predicted by its level of development and size or other benchmarking explanatory variables (see notes to figures 6.1–­6.7). The median country is calculated within the sample of Latin American and Caribbean and comparator countries; samples vary with data availability. —­= Not available. a. A value of 1 on this measure indicates only that the country is below the predicted level. nontradable industries. This finding is in about why it has so many entrepreneurs stark contrast with openness, in which only 6 and so little innovation. Issues that warrant have a deficit. Simply put, the challenge lies in examination include human capital, competi- enhancing the level of competition in sectors tion, and intellectual property rights. that are not exposed to international compe- tition. This implication is consistent with the Notes call in chapter 3 to take a second look at the unfinished agenda of competition policy in 1. The 17 sectors are electricity, gas, steam, and air conditioning supply; construction of build- the region. ings; civil engineering; specialized construc- A third element of the enabling environ- tion activities; wholesale and retail trade and ment that might pose obstacles for innova- repair of motor vehicles and motorcycles; tion is IPRs. Of the 13 countries with a deficit wholesale trade, except of motor vehicles and in innovation, 8 lagged in IPR protection. motorcycles; retail trade, except of motor More research needs to be undertaken to vehicles and motorcycles; land transport and understand how contractual enforcement transport via pipelines; air transport; ware- and viability may affect access to credit and housing and support activities for transpor- entrepreneurship. Although the ICRG indica- tation; accommodation; telecommunications; tors indicate that the region is not necessarily insurance, reinsurance, and pension funding, underperforming in contract viability, other except compulsory social security; real estate activities; architectural and engineering activi- research finds that contractual weaknesses ties; technical testing and analysis; other pro- are an obstacle to access to finance. The spe- fessional, scientific, and technical activities; cific aspects that indicators measure therefore and travel agency, tour operator, reservation matter; it may be too generic to speak about service, and related activities. “contract viability.” 2. These findings should be interpreted with Our hope is that the evidence presented some caution, because the concentration in this report will feed debate in the region indicators were constructed from the Orbis T o w ar d an E na b lin g E n v ir o nment f o r I nn o v ati v e E ntrepreneurs   151 firm-­level database, which contains informa- de la Torre, A., A. Ize, and S. L. Schmukler. 2012. tion for all firms on which it collects data. Financial Development in Latin America and Sample selection biases could play a role in the Caribbean: The Road Ahead. Washing- the ranking, as better and more numerous ton, DC: World Bank. data may have been collected in some coun- Didier, T., R. Levine, and S. Schmukler. 2013. tries than in others. To cope with this poten- “Finance and Growth for Whom?” World tial risk, we averaged revenues for each firm Bank, Washington, DC. between 2007 and 2010 (which increases the Enterprise Surveys (database). World Bank Group, likelihood that a firm was sampled in these Washington, DC. http://www.enterprisesurveys four years); we included only firms with rev- .org/. enues of more than $1 million (which are Frankel, J., and D. Romer. 1999. “Does Trade more likely to be surveyed); and we included Cause Growth?” American Economic Review only countries for which information was 89 (3): 379–­ 99. available for at least 30 firms (thus dropping Park, W. 2008. “International Patent Protection, several Caribbean countries). However, some 1960–­2005.” Research Policy 37: 761–­ 66. measurement errors and selection biases may Penn World Tables Version 7.1 (database). Center persist. The selection of firms with revenues of for International Comparisons of Production, more than $1 million does not seem to affect Income and Prices, University of Pennsylvania, concentration much: in LAC, for instance, Philadelphia, PA. https://pwt.sas.upenn.edu the correlation between the Herfindahl index /php_site/pwt_index.php. with all firms in Orbis and only firms with Rose, A. K. 2004. “Do We Really Know That the more than $1 million in revenues is 0.95. WTO Increases Trade?” American Economic 3. Chile was reclassified as a high-­income coun- Review 94 (1): 98–­ 114. try in 2013. Santos Silva, J. M. C., and S. Tenreyro. 2006. “The Log of Gravity.” Review of Economics and Statistics 88 (4): 641–­ 58. Shiller, R. S. 2013. “Why Innovation Is Still Capi- References talism’s Star.” New York Times, August 17. Aedo, C., and I. Walker. 2012. Skills for the 21st World Development Indicators (database). World Century in Latin America and the Carib- Bank, Washington, DC. http://data.worldbank bean. Washington, DC: World Bank. . o r g /d a t a - ­c a t a l o g / w o r l d -­d e v e l o p m e n t Alcala, F., and A. Ciccone. 2004. “Trade and Pro- -­indicators. ductivity.” Quarterly Journal of Economics 119 (2): 612–­45. ECO-AUDIT Environmental Benefits Statement The World Bank is committed to pre- Saved: serving endangered forests and natural • 6 trees resources. Latin American Entrepre- •2 million BTUs of neurs: Many Firms but Little Innova- total energy tion is printed on recycled paper with •543 pounds of net 30 percent postconsumer fiber in accord- greenhouse gases ance with the recommended standards •2,945 gallons of for paper usage set by the Green Press waste water Initiative, a nonprofit program sup- •198 pounds of porting publishers in using fiber that solid waste is not sourced from endangered for- ests. For more information, visit www .greenpressinitiative.org.