WORLD BANK LATIN AMERICAN AND CARIBBEAN STUDIES 40008 Informality Exit and Exclusion Guillermo E. Perry William F. Maloney · Omar S. Arias Pablo Fajnzylber · Andrew D. Mason Jaime Saavedra-Chanduvi · INFORMALITY: EXIT AND EXCLUSION THE PRINCIPAL AUTHORS OF THIS BOOK ARE AS FOLLOWS: Overview: Guillermo E. Perry and William F. Maloney Chapter 1: William F. Maloney and Jaime Saavedra-Chanduvi Chapter 2: William F. Maloney and Omar S. Arias Chapter 3: Omar S. Arias Chapter 4: William F. Maloney and Mariano Bosch Chapter 5: Pablo Fajnzylber and William F. Maloney Chapter 6: Pablo Fajnzylber Chapter 7: Andrew D. Mason Chapter 8: Jaime Saavedra-Chanduvi INFORMALITY: EXIT AND EXCLUSION Guillermo E. Perry William F. Maloney Omar S. Arias Pablo Fajnzylber Andrew D. Mason Jaime Saavedra-Chanduvi Washington, D.C. ©2007 The International Bank for Reconstruction and Development / The World Bank 1818 H Street, NW Washington, DC 20433 Telephone: 202-473-1000 Internet: www.worldbank.org E-mail: feedback@worldbank.org All rights reserved 1 2 3 4 10 09 08 07 This volume is a product of the staff of the International Bank for Reconstruction and Development / The World Bank. The findings, interpretations, and conclusions expressed in this volume do not necessarily reflect the views of the Executive Directors of The World Bank 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 judgement on the part of The World Bank concerning the legal status of any territory or the endorsement or acceptance of such boundaries. Rights and Permissions The material in this publication is copyrighted. Copying and/or transmitting portions or all of this work without permission may be a violation of applicable law. The International Bank for Reconstruction and Development / The World Bank encourages dissemination of its work and will normally grant permission to reproduce portions of the work promptly. For permission to photocopy or reprint any part of this work, please send a request with complete information to the Copyright Clearance Center Inc., 222 Rosewood Drive, Danvers, MA 01923, USA; telephone: 978-750-8400; fax: 978-750-4470; Internet: www.copyright.com. All other queries on rights and licenses, including subsidiary rights, should be addressed to the Office of the Publisher, The World Bank, 1818 H Street, NW, Washington, DC 20433, USA; fax: 202-522-2422; e-mail: pubrights@worldbank.org. ISBN-10: 0-8213-7092-8 ISBN-13: 978-0-8213-7092-6 eISBN-10: 0-8213-7093-6 eISBN-13: 978-0-8213-7093-3 DOI: 10.1596/978-0-8213-7092-6 Library of Congress Cataloging-in-Publication Data Perry, Guillermo. Informality : exit and exclusion / Guillermo E. Perry and William F. Maloney ; Omar S. Arias ... [et al.]. p. cm. Includes bibliographical references. ISBN-13: 978-0-8213-7092-6 ISBN-10: 0-8213-7092-8 ISBN-13: 978-0-8213-7093-3 (electronic (13)) ISBN-10: 0-8213-7093-6 (electronic (10)) 1. Informal sector (Economics) I. Maloney, William F. (William Francis), 1959- II. Title. HD2341.P43 2007 330--dc22 2007013736 Cover photo: "Fenómeno" (1962) by Remedios Varo © 2007 Artists Rights Society (ARS), New York / VEGAP, Madrid. Cover design: Naylor Design. Contents Foreword xi Acknowledgments xiii Abbreviations xv Overview: Informality: Exit and Exclusion 1 The razón de ser of the informal sector: Adding exit to exclusion 1 Workers: A mix of opting out and exclusion 4 Firms: Little gain, high costs, or weak enforcement? 9 Need for more effective and legitimate institutions 12 Summing up: Policy implications of the report 13 Informality and the development agenda 19 Note 19 References 19 Chapter 1: The Informal Sector: What Is It, Why Do We Care, and How Do We Measure It? 21 Introduction: What is informality? 21 Informality and the relationship between the individual and the state 22 Three margins of informality 25 Measuring the informal sector 28 Correlations among measures and trends over time 35 Conclusions 37 Annex 39 Notes 39 References 40 Chapter 2: The Razón de Ser of the Informal Worker 43 Informal work: Adding exit to exclusion 43 The sectors of informal labor: Characteristics and dynamics 48 Motivations for participation in informal work 62 Conclusions 74 Notes 74 References 75 Chapter 3: Informality, Earnings, and Welfare 79 Compensating differentials, comparative advantage, and informal work 80 The question of equal pay for equal work in the informal and formal sectors 85 Informality and self-rated welfare 92 Conclusions and policy implications 98 Notes 99 References 99 v C O N T E N T S Chapter 4: The Informal Labor Market in Motion: Dynamics, Cycles, and Trends 101 Informality through the lens of gross labor force dynamics 102 Drivers of the increase in informality 112 Conclusion 125 Notes 127 References 128 Chapter 5: Microfirm Dynamics and Informality 133 Conceptual framework: Firm dynamics and institutional development 134 Microfirm dynamics in Latin America 140 Informality among microfirms 148 Conclusions 153 Notes 154 References 154 Chapter 6: Informality, Productivity, and the Firm 157 Informality among registered firms 158 Firm-level determinants of informality 160 Impact of informality on firm productivity and economic growth 171 Conclusions 175 Notes 177 References 177 Chapter 7: Informality, Social Protection, and Antipoverty Policies 179 Informality and social protection: Why policy makers should care 179 The state of social protection in Latin America and the Caribbean 184 Private risk management and rationale for public social protection 190 Challenges for social protection in the face of informality 193 Reengineering social protection to protect all citizens 199 Potential costs of social protection reform: Financing essential cover 204 Managing the transition from here to there 206 Conclusion 208 Notes 210 References 211 Chapter 8: The Informal Sector and the State: Institutions, Inequality, and Social Norms 215 Social norms, the state, and informality 217 The tax side of the social contract in Latin America 223 Inequality, taxes, and transfers 235 Informality: A reflection of a broken social contract? 239 Conclusions 244 Notes 245 References 246 Boxes Overview 1 Informal workers in Latin America and the Caribbean: Their profile, motivations, earnings, and welfare 4 2 Government failures in the design of social protection systems in Latin America and the Caribbean 8 Chapter 1 1.1 The ILO definition of informality 27 1.2 Indirect methods of estimating informality 34 1.3 Schneider and Enste in the new world: Checking MIMIC estimates against Mexican data 35 Chapter 2 2.1 Home-based work: Exploitation or flexible work arrangement? 52 2.2 Data from rotating panels in Argentina, Brazil, and Mexico 56 2.3 Informal self-employment: Risky and informal, or risky because informal? 60 2.4 Special informal employment surveys: What can we learn from them? 64 vi C O N T E N T S Chapter 4 4.1 Conceptual issues in gross worker flows 103 4.2 Simulated effects of labor market legislation on the size of the informal sector 106 Chapter 5 5.1 Patterns of entry and exit in industrialized countries 137 Chapter 7 7.1 Social protection--strengthening people's abilities to manage risk and promoting long-term productivity, growth, and development 180 7.2 The comprehensive insurance framework--providing guidance on social protection policy making 182 7.3 Extending health insurance coverage by correctly aligning risk-pooling instruments 201 7.4 Old-age protection in the new millennium: Chile's proposed pension reform 203 7.5 De-linking health coverage from employment status: Spain's shift to general-revenue financing of essential social insurance 207 Chapter 8 8.1 The extreme of informality and exclusion: Being undocumented 219 8.2 Local taxation and social norms 229 8.3 Tax compliance, social norms, and trust in the state: The contrasting cases of Chile and Argentina in the late 1990s 231 8.4 Earned income tax credits: Transfers that encourage formal employment 234 8.5 Expansion of private security services in Managua 242 8.6 Negotiating tax reform and the start of the social contract, Chile, 1990 243 Figures Overview 1 Labor market informality and income per capita 2 2 Trends in informality, by various definitions 3 3 Probability of transition between formal salaried and self-employment in Mexico 6 4 Rate of urban employment across sectors, by age in Brazil, 2002 7 5 Informal workers across firm size 9 6 Male entry into and exit from self-employment 10 7 Advantages of formalization reported by IFC-surveyed firms 11 8 Underreporting of tax and social security contributions, by firm size 11 9 Economically active population contributing to the pension system 12 10 Informality versus inequality 13 11 Tax morale and state capture 14 12 Self-employment and quality of institutions (governance) 14 Chapter 1 1.1 Margins of informality 25 1.2 Methods for measuring the informal sector 28 1.3 Selected measures of informality 29 1.4 Informality and gender: women versus men (ages 25­64) 31 1.5 Informal salaried workers (legal definition) across firm size and time 32 1.6 Distribution of manufacturing sector firms according to worker registration in Mexico 33 1.7 Distribution of manufacturing sector firms according to tax compliance in Mexico 33 1.8 Global correlation of measures of informality with GDP 37 1.9 Trends in informality by various definitions 38 Chapter 2 2.1 Relative sector sizes and wages under a nominal wage rigidity 44 2.2 Relative sector sizes and wages in the presence of a labor tax 48 2.3 Rate of urban employment across sectors, by age 53 2.4 Probabilities of transition among sectors of employment 54 2.5 Absolute mean duration of labor force status 57 2.6 Propensity to move to self-employment from different sectors 58 2.7 Transition between self-employment and out of labor force, by gender 61 2.8 Employment sector allocation by gender, marital status, and parental status in Mexico 61 vii C O N T E N T S 2.9 Propensity to move to informal salaried status from different sectors 62 2.10 International comparison of desired and actual self-employment rates 65 2.11 Earnings gain from voluntary movement to informality in Mexico 68 Chapter 3 3.1 Propensity to informal employment by firm size and economic sector, 2005­06 82 3.2 Propensity to informal employment by education and tenure, 2005­06 83 3.3 Propensity to informal employment by age, gender roles, and work preferences, 2005­06 84 3.4 Distribution of hourly earnings for workers in urban areas of Argentina and Bolivia, 2005­06 86 3.5 Earnings cost of informality in urban areas of Argentina, Bolivia, and the Dominican Republic 89 3.6 Fraction of workers who are income-poor and self-rated poor, by education and occupational group 93 3.7 Impact of informality on self-rated poverty 94 3.8 Direct impact of having access to social protection through a family member on self-rated poverty 95 3.9 Differences between actual labor incomes and the level of income needed to avoid self-rating poverty in Argentina 97 Chapter 4 4.1 Formal share of the labor force and unemployment, Brazil and Mexico 102 4.2a Probability of transition between formal salaried and self-employment, Brazil and Mexico 104 4.2b Probability of transition between formal salaried and informal salaried, Brazil and Mexico 104 4.3 Involuntary transition to self-employment in Mexico 105 4.4 Searching while employed, Mexico 105 4.5 Decreased availability of formal sector jobs without segmentation 106 4.6 Probability of transition to unemployment (separation rate), Brazil and Mexico 108 4.7 Probability of transition from unemployment (job-finding rate), Brazil and Mexico 109 4.8 Informal employment reaction to the business cycle 111 4.9 Relative sector shares and earnings, real exchange rate 112 4.10a Variation of informality rate (social protection definition) 114 4.10b Variation of informality rate (productive definition) 115 4.11 Decomposition of changes in informality (social protection definition) for urban salaried workers 115 4.12a Labor force participation, rate of formality, and unemployment, Brazil 116 4.12b Labor force participation, rate of formality, and unemployment, Colombia 116 4.13 Actual and predicted size of the industrial formal sector, Brazil 117 4.14 Informality rate for salaried workers in Greater Buenos Aires 118 4.15 Informal salaried versus formal salaried, Mexico, relative earnings and sector size 118 4.16 Evolution of informality across age groups 120 4.17 Labor legislation and related variables, Peru 123 Chapter 5 5.1 Entry and exit into self-employment among men 141 5.2 Entry and exit into self-employment among women 142 5.3 Job creation and destruction in Colombia, 1977­82 144 5.4 Job creation and destruction by entering and exiting versus continuing firms in Colombia, 1977­82 145 5.5 Employment growth of microenterprises in Nicaragua 146 5.6 Size distribution of microenterprises in Mexico and Nicaragua 147 5.7 Informality among Mexican microenterprises by number of paid workers and time in business 150 5.8 Effect of doubling labor productivity on probability of being informal 153 Chapter 6 6.1 Unreported sales and workers, by firm size 159 6.2 Effect of doubling employment size on underreporting rates 159 6.3 Effect of having started informally on underreporting rates 160 6.4 Effect of doubling labor productivity on underreporting rates 160 6.5 Advantages of formalization reported by IFC-surveyed firms 162 6.6 Disadvantages of formalization reported by IFC-surveyed firms 162 6.7 Impact of changes in costs and benefits of informality 163 6.8 Labor regulations described by firms as biggest obstacle to hiring workers 169 6.9 Estimated impact of informality on growth 172 6.10 Estimated impact of informality on labor productivity 173 viii C O N T E N T S 6.11 Effects of a 10 percent increase in tax and social security evasion at the industry/region level on individual firm productivity (pooled sample) 174 6.12 Productivity difference between tax-paying Mexican microfirms and informal firms with similar characteristics 175 6.13 Productivity and paid employment effects of exogenous increase in formality driven by Brazil's SIMPLES program 175 Chapter 7 7.1 Pension coverage rates in Latin America and the Caribbean 184 7.2 Social security coverage and GDP per capita 185 7.3 Coverage rates in Latin America and the Caribbean, mid-1990s to early/mid-2000s, by quintile of per capita income 186 7.4 Pension coverage rates in Argentina, by quintile of per capita income 187 7.5 Pension coverage among the elderly in Latin America and the Caribbean, contributory and noncontributory programs 188 7.6 Coverage of two poverty-targeted assistance programs in Mexico: Oportunidades and Seguro Popular (by decile, 2004) 188 7.7 Public social protection spending in Latin America and the Caribbean 189 7.8 Public spending on social insurance and social assistance in Latin America and the Caribbean, early 2000s 190 7.9 Nonpoor population who fell below the national poverty line due to out-of-pocket health expenditures 191 7.10 Distribution of Mexican workers' years insured by IMSS, 1997­2005, for low- and high-wage workers in IMSS in 1997 195 7.11 Density function for years of formal sector work among 60-year-olds in Uruguay (accumulated ages 18­60) 196 7.12 Cumulative distribution of years of formal sector service among 60-year-olds in Uruguay (accumulated ages 18­60) 196 Chapter 8 8.1 Informality and government effectiveness 220 8.2 Informality and corruption 220 8.3 Share of firms confident that the judiciary will enforce contractual and property rights 220 8.4 Share of firms that consider the court system fair, impartial, and uncorrupt 221 8.5 Informality and state competence indicators 221 8.6 Central government tax revenue and GDP per capita 224 8.7 Tax revenue of Latin America and Caribbean countries 226 8.8 Average tax rate of Latin America and Caribbean countries 226 8.9 Average VAT and income tax productivity in Latin American countries 227 8.10 VAT productivity and informality indicators 227 8.11 Tax morale, state capture, and perception that the government spends taxpayers' money wisely 228 8.12 Tax morale and informality indicators 232 8.13 Social spending and taxation by income quintiles 238 8.14 Informality and inequality 239 Tables Chapter 1 1.1 Correspondence of the "productive" and "legalistic" definitions of informality 31 1.2 Correlations across measures of informality 36 1A.1 Comparisons of ILO and Gasparini-Tornarolli measures of self-employment 39 Chapter 2 2.1 Distribution of formal salaried, informal salaried (social protection/legal definition), and self-employed in urban areas of Latin America (percent) 49 2.2 Distribution of informal salaried, self-employed, and unpaid workers, by firm size (percent) 50 2.3 Characteristics of informal salaried employees in Mexican microfirms 51 2.4 Informal employment and work-life occupational history, Dominican Republic and Argentina (percent of workers) 63 2.5 Preference for independent employment (percent of workers) 64 2.6 Distribution of the motivations/reasons for being in the current job as an independent worker (percent) 66 2.7 Reported reasons to be informal self-employed in Mexico (percent of workers) 67 2.8 Reported reasons to be informal self-employed in Brazil (percent of workers) 67 2.9 Distribution of the motivations/reasons for being in the current salaried job (percent) 69 ix C O N T E N T S 2.10 Reported reasons to be informal salaried in Brazil (percent of workers) 69 2.11 Principal reason for working without monetary compensation in the Dominican Republic (percent of workers) 70 2.12 Main reasons why the informal do not contribute to social security (percent of workers) 71 2.13 Main reasons why workers do not contribute to health insurance (percent of workers) 73 Chapter 3 3.1 Unconditional hourly earnings gaps (percent difference in earnings) for formal employees, informal employees, and independent workers in urban areas in Argentina, Bolivia, and the Dominican Republic, 2005 87 3.2 Impact of informality on earnings in Argentina, estimated log-earnings differences and treatment parameters 91 3.3 Correlation between income and self-rated poverty 92 3.4 Impact on the propensity to self-rate poor and be income poor (normalized logit regression coefficients) 95 3.5 Job satisfaction and informal employment in the Dominican Republic 96 3.6 Job satisfaction and informal employment in Colombia 97 Chapter 4 4.1 Evolution of informality across firm size and time: Distribution of informal and self-employed workforce 119 4.2 Subcontracting in Mexico, 1992­2002 (percent) 119 Chapter 5 5.1 Entry probabilities into Mexico's self-employment sector (percent) 140 5.2 Exit probabilities from Mexico's self-employment sector (percent) 143 5.3 Growth and contraction probabilities of Mexican microfirms (percent) 146 5.4 Employment size distribution of firms, selected countries (percent) 147 5.5 Employment size distribution of firms with employees, selected countries (percent) 148 5.6 Reasons for starting up, firm prospects, and firm licensing in Brazil 149 5.7 Informality among Mexican and Brazilian microenterprises by age, education, and previous activity of the owner 149 5.8 Informality among Mexican, Brazilian, and Colombian microenterprises by paid employment and time in business 151 5.9 Informality among Mexican, Brazilian, and Colombian microenterprises by age, education, and previous activity of the owner 152 Chapter 6 6.1 Firm-level correlates of sales and employment underreporting 167 Chapter 8 8.1 Long-run informality relationships 222 8.2 Indicators of informality and institutional indicators 223 8.3 Comparative perspective of tax burdens and structures (percent of GDP) 224 8.4 Tax structure by region, selected years, 1975­2002 (percentage of total tax revenue) 225 8.5 Tax exemptions in Latin America (percent of GDP) 227 8.6 Comparative efficiency and corruption in tax administration: Survey evidence for 2005­06 233 x Foreword I NFORMALITY IS A WAY OF LIFE IN LATIN countries more generally is partly due to the low opportunity America and the Caribbean, no matter how we cost of opening a microbusiness, the gamut of measures to measure it. In some countries, the majority of the increase formal sector productivity (improving the business workforce is not covered by labor protections; the climate, fomenting innovation, and so forth) and to ubiquitous microenterprises found on every street increase workers' skills is important. Labor market reform corner are often not registered with authorities and only is also essential because part of the rise in the share of work- partially comply with other regulations; and tax evasion by ers uncovered by labor protections in the 1990s appears to rich and poor alike is the norm. Although these features be due to the increased burden of labor costs and restric- of the Latin American and Caribbean landscape are not tions in several countries. But reform also needs to extend new, their striking increase in some countries has given to remedying the poor design of many social protections new life to the debate concerning what informality tells us that workers find of poor quality, inefficient, and of little about how our economies are functioning. value. Hence, shifting workers' cost-benefit analysis toward Informality: Exit and Exclusion, the ninth in our series of engagement with the institutions of the state--both regional flagship reports, explores this very heterogeneous through improving the benefits of being formal and sector from a variety of perspectives ranging from a concern through better monitoring--is critical to reducing their with the protection of workers, to the productivity of firms, often voluntary entry into informality. Relatedly, although to the determinants of tax evasion. The report accepts and experiments reducing the costs of registering businesses sheds more light on the exclusionary character of much have led to relatively little formalization of existing busi- informality, which leaves citizens outside formal institu- nesses, the impact on those few businesses appears substan- tions. However, it especially highlights that there is an tial and progress in this area appears important. All of these important "exit" dimension that has been understressed in reforms need to take place in a context of improving the the literature: workers, firms, and families, dissatisfied with perception of fairness and efficiency of the state, which is the performance of the state or simply not finding any ben- often seen as serving the needs of particular elites rather efit to interacting with it, opt into informality. Hence, the than the needs of the population. In the long run, this is discussion ranges from the purely economic analysis of the only way to change social norms of compliance and microeconomic incentives to reflection on the soundness of reduce the "culture of informality." the "social contract"--shorthand for how citizens of the This is a topic that the Latin America and the Caribbean region relate to the state and to each other. Region of the World Bank has been working on in a sus- It is not surprising, then, that the analysis gives rise to tained way through various regional studies and reports recommendations that span many dimensions of the policy for more than 10 years. The present volume builds on this agenda. Because the high level of informality in developing work, introduces original data sources collected for the xi F O R E W O R D project and new statistical techniques to analyze them, Chief Economist for the Latin America and the Caribbean and brings together research done in our partner coun- Region of the World Bank. tries. Both its findings and its policy recommendations are thus grounded in detailed analysis of the reality of the region. As a final note, this Flagship is the last undertaken under the direction of Guillermo E. Perry, and I would like Pamela Cox to thank him for his leadership, insights, and enthusiasm as Vice President for Latin America and the Caribbean The World Bank xii Acknowledgments I NFORMALITY: EXIT AND EXCLUSION IS THE PRODUCT OF A COLLABORATIVE EFFORT BY A NUMBER of professionals from within and outside the Bank. The report was prepared under the direction of Guillermo E. Perry by a core team comprising William F. Maloney, Omar S. Arias, Pablo Fajnzylber, Andrew D. Mason, and Jaime Saavedra-Chanduvi. Background papers were prepared by Rita Almeida (World Bank), James Alm (Georgia State University), Guillermo Beylis (World Bank), Mariano Bosch (London School of Economics and Political Science [LSE] and University of Alicante), Karla Breceda (World Bank), Monserrat Bustelo (World Bank), Dante Contreras (Universidad de Chile), Barbara Cunha (University of Chicago), Wendy Cunningham (World Bank), Norbert Fiess (University of Edinburgh), Marco Fugazza (United Nations Conference on Trade and Development [UNCTAD]), Sebastian Galiani (Washington University in St. Louis), Rodrigo García-Verdú (World Bank), Leonardo Gasparini (Centro de Estudios Distributivos Laborales y Sociales [CEDLAS] and Universidad Nacional de La Plata), Edwin Goñi (World Bank), Melanie Khamis (LSE), Fernando Landa (Unidad de Análisis de Políticas Sociales y Económicas [UDAPE], Bolivia), Santiago Levy (Brookings Institution), Norman Loayza (World Bank), Leonardo Lucchetti (University of Illinois at Urbana-Champaign), Jorge Martinez (Georgia State University), David McKenzie (World Bank), Truman Packard (World Bank), Esteban Puentes (Univer- sity of Chicago), Jamele Rigolini (World Bank), Gabriel Montes Rojas (University of Illinois at Urbana- Champaign), Lucas Ronconi (University of California at Berkeley), Claudia Sanhueza (Universidad de Chile), Mariano Tommasi (Universidad de San Andrés), Leopoldo Tornarolli (CEDLAS and Universidad Nacional de La Plata), Federico Weinschelbaum (Universidad de San Andrés), Chris Woodruff (University of California, San Diego), and Patricia Yañez (UDAPE, Bolivia). The report has also benefited from specific contributions or comments by Juliano Assunção (Pontifícia Uni- versidade Católica [PUC], Brazil), Tito Cordella (World Bank), Jean Jacques Dethier (World Bank), Maria Victoria Fazio (World Bank), Gary Fields (Cornell University), Alvaro Forteza (World Bank), Vicente Fretes Cibils (World Bank), Marcelo Giugale (World Bank), Gustavo Gonzaga (PUC, Brazil), Ricardo Furman (Inter- national Finance Corporation [IFC]), Luke Haggarty (IFC), Mary Hallward-Driemeier (World Bank), Marco Hernandez (Oxford University), Tom Kenyon (IFC), Peter Lindert (University of California at Davis), David McKenzie (World Bank), Carolina Mejia (Fedesarrollo), Ezequiel Molina (CEDLAS and Universidad Nacional de La Plata), Carmen Pages (Inter-American Development Bank), Georgina Pizzolito (World Bank), Kristtian Rada (IFC), Lauro Ramos (Instituto de Pesquisa Económica Aplicada [IPEA], Brazil), Jose Guilherme Reis (World Bank), Helena Ribe (World Bank), Rafael Rofman (World Bank), Jose Scheinkman (Princeton University), Pablo Suarez (Brown University), Judith Tendler (Massachusetts Institute of Technology), Gabriel Ulyssea (IPEA, Brazil), and Evelyn Vezza (World Bank); and by our peer reviewers Jesko Henschel xiii A C K N O W L E D G M E N T S (World Bank), Pierella Paci (World Bank), Stefano Scarpetta (Organisation for Economic Co-operation and Development [OECD]), and Simeon Djankov (World Bank). The project benefited from a data collection grant from the World Bank Development Economics Group Research Support Board to undertake surveys in Bolivia, Colombia, and the Dominican Republic, building on the 2005 survey conducted in Greater Buenos Aires, Argentina, by the Ministry of Labor, Employment, and Social Security and the National Institute of Statistics and Census of Argentina, with support from the World Bank. We very warmly thank Clara Else for her guidance and help in shepherding through the pro- posal and François Bourguignon for his support. Invaluable collaboration to develop these surveys is acknowledged to Carlos Ignacio Becerra (Departamento Administrativo Nacional de Estadística [DANE], Colombia), Dilia Cruz (Central Bank of the Dominican Republic), Olga Diaz (Central Bank of the Domini- can Republic), Pedro José Fernández (DANE, Colombia), Eduardo Freire (DANE, Colombia), Rolando Guzman (PARETO Consulting Group), Oscar Lora (Instituto Nacional de Estadística [INE], Bolivia), Gabriel Loza (UDAPE, Bolivia), Juliana Martinez Cuellar (Departamento Nacional de Planeación [DNP], Colombia), José Ramirez (UDAPE, Bolivia), Carmen Julia Reyes (Central Bank of the Dominican Repub- lic), Guillermo Rivas (DNP, Colombia), Norberto Rojas (DNP, Colombia), Luz Magdalena Salas (DNP, Colombia), Mauricio Santamaria (DNP, Colombia), Romilda Santana (Central Bank of the Dominican Republic), Johnny Suxo (INE, Bolivia), María Fernanda Tellez (DNP, Colombia), and María Eugenia Villamizar (DANE, Colombia). We are also grateful to Mauricio Cardenas (Fedesarrollo) and Sebastian Edwards (University of California, Los Angeles); to participants at the National Bureau of Economic Research workshop on Informality in Bogota; to Hartmut Lehmann (University of Bologna and Institute for the Study of Labor [IZA]) and Stefano Scarpetta (OECD); to participants at the World Bank/IZA Conference on Informality in Bertinoro; and to participants at the annual conference of the Latin American and Caribbean Economics Association for helpful feedback on background work for this report. Finally, Patricia Da Camara, Catherine Russell, and Elena Serrano coordinated the report's publication and dissemination activities, working closely with Dina Towbin, Andrés Meneses, and Shana Wagger in the World Bank's Office of the Publisher. xiv Abbreviations ATE average treatment effect LDC least developed country AUGE Acceso Universal para prestaciones integrales LSMS Living Standards Measurement Survey y Garantías Explícitas (Plan of Universal MAROP Mecanismo de Ahorro para el Retiro Access with Explicit Guarantees) Oportunidades BONOSOL Bono solidario MIMIC Multiple Indicator­Multiple Cause BPS Banco de Previsión Social Model DANE Departamento Administrativo Nacional MPG Minimum Pension Guarantee de Estadística OECD Organisation for Economic ECINF Pesquisa Economica Informal Urbana Co-operation and Development EITC earned income tax credit OLF out of the labor force ENAHO Encuesta Nacional de Hogares PASIS Pensión Asistencial ENAMIN Encuesta Nacional de Micronegocios PNAD Pesquisa Nacional por Amostra Domicilios ENEU Encuesta Nacional de Empleo Urbano PPP purchasing power parity F-I formal to informal RN Renovación Nacional FONAHPO Fondo Nacional de Habitaciones RTS simplified tax regime Populares SARE Sistema de Apertura Rápida de FONASA Fondo Nacional de Salud Empresas F-SE formal to self-employed SE-F self-employed to formal GDP gross domestic product SENA Colombian Training Institute ICBF Colombian Institute for Family Welfare SIMPLES Sistema Integrado de Pagamento de Impostos ID identity e Contribucoes as Microempresas e Empresas I-F informal to formal de Pequeno Porte ILO International Labour Organization SME small and medium enterprise IMSS Mexican Social Security Institute TT treatment on the treated INEGI Instituto Nacional de Estadística, TU treatment on the untreated Geografía e Informática (National VAT value-added tax Statistical Institute [Mexico]) WDI World Development Indicators ISSSTE Social Security Institute for Mexican WFTC working families tax credit civil servants at the federal level WGI Worldwide Governance Indicators xv OVERVIEW Informality: Exit and Exclusion There are two main types of activist reactions to discontent with organizations to which one belongs or with which one does business: either to voice one's complaints, while continuing as a member or customer, in the hope of improving matters; or to exit from the organization, to take one's business elsewhere. --Albert O. Hirschman (1981, p. 246) I NFORMALITY HAS GAINED INCREASING analysis, we hope, will serve as a foundation for more effective attention as a possible drag on growth and rising public policies. social well being, and as a force corrosive to the integrity of our societies. In fact, informality is not The razón de ser of the informal sector: Adding especially higher in Latin America and the exit to exclusion Caribbean than in other developing countries of similar per The report sees informality as a manifestation of the rela- capita income, according to two popular measures (figure 1). tionships between economic agents and a state that the However, given the long-standing negative connotations of economic literature should play an important role in reme- informality--inferior working conditions, low-productivity dying market failures, coordinating the provision of public firms, disrespect for the rule of law, to name a few--it is not goods, and maintaining a level and equitable playing field. surprising that the rise in informality over the 1990s across Of the many lenses through which informal workers have several measures (figure 2) is viewed with concern and as been viewed, the most influential lens has focused on their meriting closer investigation. exclusion from critical state benefits or the circuits of the This said, the mere fact that we need to employ multiple modern economy. This exclusion can be seen as occurring measures of informality capturing distinct approaches to along three margins, or borders, between formality and the sector suggests that we are not clear on exactly what it is informality. First, a long tradition in the labor literature sees and what we should be studying. In all likelihood, we are segmentation in the labor market as preventing workers dealing with several distinct phenomena under this conve- from leaving their holding pattern in informality and nient, but arguably unhelpful, umbrella term. For a report taking jobs in the formal sector that offer state-mandated on informality in Latin America and the Caribbean this benefits. Second, de Soto's (1989) seminal work argues that clearly poses special challenges. Fortunately, recently avail- burdensome entry regulations prohibit small firms from able data sets and the development of new techniques to crossing the line into formality and thriving. And, third, analyze them have made possible more solidly grounded some larger firms faced with excessive tax and regulatory analysis of the underlying heterogeneity and reasons for being burdens may remain partially informal as a defensive of the sector, the factors driving its evolution across different measure and, as a result, they forgo potential growth and effi- countries, and its social and economic ramifications. Such an ciency gains. This report finds that these exclusion factors 1 I N F O R M A L I T Y FIGURE 1 Labor market informality and income per capita Share of labor force not covered by a pension scheme Share of labor force self-employed Percent Percent 100 60 90 50 80 70 40 60 50 30 40 20 30 20 10 10 0 0 0 0 0 0 0 0 0 0 0 0 0 5,000 10,000 15,00 20,00 25,000 30,000 35,00 40,000 45,00 5,000 10,000 15,00 20,00 25,000 30,000 35,00 40,00 45,00 2005 GDP per capita, purchasing power parity (PPP) adjusted 2005 GDP per capita, purchasing power parity (PPP) adjusted Latin America Advanced countries Rest of the world Sources: Loayza and Rigolini 2006; World Development Indicators 2006. are indeed important, documents their negative impact on wages, may find that what they expect to get in the formal productivity and welfare, and discusses some of the reforms jobs for which they are qualified does not outweigh the for- necessary to mitigate them. gone current consumption or the greater flexibility and However, the report also highlights a second lens earnings they may get as informal workers. This is espe- through which to view informality. This lens is more akin cially true when they have social protection alternatives to Hirschman's (1970) exit: many workers, firms, and fami- through access to universal or noncontributory programs, lies choose their optimal level of engagement with the or through private means. Larger firms or skilled profes- mandates and institutions of the state, depending on their sionals may decide to underreport their operations and valuation of the net benefits associated with formality and incomes, balancing private gains from tax evasion with low the state's enforcement effort and capability. That is, they detection risks resulting from poor enforcement. make implicit cost­benefit analyses about whether to cross These two lenses, focusing, respectively, on informality the relevant margin into formality, and frequently decide driven by exclusion from state benefits and on voluntary exit against it. This view suggests that high informality results decisions resulting from private cost­benefit calculations, are from a massive opting out of formal institutions by firms complementary rather than competing analytical frame- and individuals, and implies a blunt societal indictment of works. First, individual countries differ greatly in history, the quality of the state's service provision and its enforce- institutions, and legal frameworks, so exclusionary mecha- ment capability. nisms may be more important in some and exit may be more This view leads to important divergences with many important in others. Second, the informal sector is tremen- popular stylized facts about the informal sector. Microfirm dously heterogeneous, and arguably, then, there is a contin- owners with little intention or potential to grow may see uum in the relative importance of exclusion and exit among insignificant benefits from engaging with unhelpful regula- individual workers and firms within countries. Third, in tory and tax institutions. Interpreting Hirschman literally, some cases the two can be virtually indistinguishable. A they take their business elsewhere, knowing that they will microentrepreneur concluding, through a cost­benefit fall below the radar screen of enforcing authorities. analysis, that formality is not worth the high registration Unskilled workers, who partially pay for social protection costs may be explicitly excluded or self-excluded--either benefits directly and implicitly through lower formal way, the effect is much the same. A poor worker, excluded 2 I N F O R M A L I T Y: E X I T A N D E X C L U S I O N FIGURE 2 Trends in informality, by various definitions Self-employed workers: ILO (% of employed workers) Lack of pensions (% of labor force) Colombia 1992­2005 17.3 Ecuador 1990­2004 12.8 Bolivia 1990­2000 6.9 Argentina 1992­2004 10.3 Dominican Republic 1991­2005 4.7 Nicaragua 1993­2001 8.1 Chile 1996­2005 0.1 Costa Rica 1990­2004 3.8 Costa Rica 1990­2005 1.4 Jamaica 1997­2005 2.3 Venezuela, R.B. de 1995­2004 3.8 Panama 1991­2005 2.4 Chile 1990­2003 2.7 El Salvador 1995­2004 2.7 Uruguay 1995­2004 2.2 Honduras 1996­2005 3.4 Colombia 1996­1999 0.6 Argentina 1996­2005 3.7 Brazil 1992­2002 0.5 Peru 1996­2005 6.9 Mexico 1991­2005 8.6 El Salvador 1995­2003 4.3 10 5 0 5 10 15 20 15 10 5 0 5 10 15 % of employed workers Percent Informal workers: Productive definition Informal salaried workers: Legalistic definition (% of employed workers) (% of salaried workers) Venezuela, R.B. de 1989­2003 18.5 Argentina 1992­2005 11.9 Uruguay 1992­2004 7.0 Honduras 1992­2002 6.8 Venezuela, R.B. de 1995­2003 7.7 Panama 1995­2003 5.4 Brazil (metropolitan) 1992­2002a 6.2 Colombia 1996­2004 4.8 Peru 1997­2003 4.6 Nicaragua 1993­2001 5.9 Jamaica 1996­2002 3.4 Peru (metropolitan Lima)1990­2000 4.5 Dominican Republic 1996­2004 3.1 Paraguay 1997­2003 2.6 Mexico 1990­2004b 4.0 Mexico 1996­2002 2.4 Chile 1990­2003 1.1 El Salvador 1991­2003 2.2 Argentina 1995­2005 1.9 Ecuador 1994­1998 0.7 Bolivia 1997­2002 1.5 Brazil 1990­2003 0.8 Ecuador 1994­2003 1.1 Costa Rica 1992­2003 0.4 Paraguay 1997­2003 0.9 Nicaragua 1993­2001 0.9 Colombia 1996­1999 2.4 Brazil 1992­2003 3.3 Chile 1990­2003 5.2 El Salvador 1991­2003 12.0 10 5 0 5 10 15 20 15 10 5 0 5 10 15 % of employed workers % of salaried workers Sources: Gasparini and Tornarolli 2006; International Labour Organization (ILO) Labor Statistics Database 2006; World Development Indicators 2006. Note: Although global data is available only for the definitions of informality as self-employed workers and workers not covered by a pension scheme, two other measures are calculated here based on regional data sources. In the "productive" definition, a worker is considered informal if he or she is unskilled self-employed, a salaried worker in a small firm, or a zero-income worker. In the "legalistic" definition, a salaried worker is informal if he or she does no have the right to a pension linked to employment upon retirement. a. Percent of workers without carteira (work card). b. Based on the balanced panel sample (common municipalities) for the period 1990­2004. from health care services because he or she lives in a remote can play different roles across different dimensions: a rural area or a poor urban neighborhood, may see little point microfirm owner unlicensed because of high registration in being formal and paying labor taxes for services to which costs may be de facto excluded from desired formal credit he or she has no access. circuits while opting out of contributing to poorly designed Finally, informality is a multidimensional phenomenon: state pension funds on behalf of his or her workers. The agents interact with the state along some dimensions and findings of the report indeed support the use of both lenses not others, creating a large gray area between the extremes to fully understand and address the causes and consequences of full compliance and noncompliance. Exit and exclusion of informality in the Latin American and Caribbean region. 3 I N F O R M A L I T Y Workers: A mix of opting out and exclusion significantly in both their motivations and their relative This report examines informal labor using the vast set of levels of job satisfaction. regular household surveys conducted in most countries First, on average, independent workers--the self- of the region, together with recent special modules on employed or those owning microfirms--in the countries informality collected by statistical agencies in Argentina, surveyed (with the exception of Colombia) report being as Bolivia, Colombia, and the Dominican Republic, in well-off as they would be in jobs they are qualified for in collaboration with the World Bank and line ministries. the formal sector. As a consequence, the majority are not The analyses shed light on the characteristics of informal looking for formal jobs. Most of these informal workers workers, their motivations, and their preference for the appear to choose their occupations according to their indi- benefits and nonpecuniary characteristics of jobs (for vidual needs (especially their desire for flexibility and example, flexibility, autonomy, stability, mobility) by autonomy) and abilities (their comparative advantage in examining their remuneration, self-rated welfare, and job terms of entrepreneurship). In fact, when asked, a majority satisfaction. The findings (summarized in box 1) show of workers state that they prefer the independence of being great heterogeneity within and across countries. They also self-employed. In Mexico, they were overachievers in indicate, however, that the informal sector can generally salaried work, earning relatively high wages, given their be thought of as comprising two large groups who differ human capital. Many women, in particular, report forgoing BOX 1 Informal workers in Latin America and the Caribbean: Their profile, motivations, earnings, and welfare Who are the informal? Informal employment encompasses a diverse range of FIGURE 1. B.1 people. While the report considers numerous common Distribution of informal workers in urban areas contributing to criteria to define informal employment, it focuses on the a social security system in Latin America social protection definition. Based on whether the job is a Percent salaried relationship and unregistered with social secu- 80 rity, informal employment accounts for 54 percent of 70 total regional urban employment and it comprises two 60 groups: (1) informal independent workers, making up 50 24 percent of total urban employment (ranging from 40 20 percent or less in Argentina, Brazil, Chile, and 30 Uruguay to more than 35 percent in Bolivia, Colombia, 20 the Dominican Republic, Peru, and the República 10 Bolivariana de Venezuela), and (2) informal salaried 0 workers who account for roughly 30 percent of total 05 1 regional urban employment and more than half of all 2003 2004 2003 20 2003 2003 2002 2006 2002 2006 200 1998 2005 2002 de ia Peru informal work (ranging again from 17 percent in Chile to Chile Brazil lvadorR.B. Mexico Uruguay Sa Republic EcuadorBoliv Argentina El Guatemala Colombia Nicaragua more than 40 percent in Bolivia, Ecuador, Guatemala, Mexico, Nicaragua, Paraguay, and Peru) (figure 1.B.1) Venezuela, Dominican The informal independent work sector includes Informal salaried Informal independent microfirm owners and self-employed professionals, as well as artisans, handymen, construction laborers, taxi drivers, Source: Author's estimates, based on household surveys. and street vendors. The informal salaried work sector Note: In Ecuador, Mexico, and República Bolivarina de Venezuela, coverage of independent workers is proxied by having a tertiary largely comprises domestic employees, unpaid family education. Argentina denotes Greater Buenos Aires. 4 I N F O R M A L I T Y: E X I T A N D E X C L U S I O N workers, microfirm workers, and those who work in larger still significant for many of the people in this subsector firms under informal labor arrangements. The characteris- (for example, youth and unpaid family workers). Infor- tics more strongly correlated with informal employment mal workers state they do not contribute to social are firm size (10 employees or fewer), education (comple- security and health insurance plans mainly because of tion of schooling below the secondary level), industry low incomes and also because of their employer's deci- sector (construction, agriculture, retail, and transport), sion not to offer benefits (in the case of most informal tenure (less than one year), age (youth predominantly infor- salaried workers); because they lack information on the mal salaried, self-employed mostly older workers), and benefits and functioning of social protection programs; women's household status (married women with children). and, in the case of health care, because they resort to other means of coverage, including coverage through What are their motivations, earnings, and welfare? other family members and universal services. Evidence from workers' patterns of mobility, reported For both groups of informal workers, there is sub- motivations, and self-rated welfare and job satisfaction stantial heterogeneity of motives and demographic indicate that (1) the majority of independent workers characteristics. For instance, in Argentina, the informal (approximately two-thirds) entered their jobs voluntar- self-employed report being as well-off as formal salaried ily, attach significant value to the nonpecuniary benefits workers, but the informal salaried appear worse-off (fig- of autonomous work, and choose to "exit" formal social ure 1.B.2); in the Dominican Republic, both informal self- protection systems; (2) on the contrary, the majority of employed and in-formal salaried workers are as well-off as informal salaried workers appear to be excluded from formal employees (figure 1.B.2); and, in Colombia, both more desirable jobs, both as formal salaried and as informal groups report much lower levels of satisfaction independent workers, although voluntary motives are with current jobs. FIGURE 1. B.2 Impact of informality on self-rated poverty Argentina Dominican Republic Propensity to self-rated poverty Propensity to self-rated poverty Independent Independent large firm large firm Independent Independent formal salaried formal salaried Independent Independent permanent contract permanent contract Microfirm worker Microfirm worker large firm large firm Informal salaried Informal salaried formal salaried formal salaried Temporal salaried Temporal salaried permanent contract permanent contract 0 0.5 1.0 1.5 2.0 0 0.5 1.0 1.5 2.0 Odds ratio Odds ratio Source: Arías and Luchetti (2007), based on household surveys, 1997 and 2004. Note: Self-rated poverty is based on household survey responses. 5 I N F O R M A L I T Y formal salaried work to better balance home and work counterparts, about one-third of the self-employed (over responsibilities. half in Colombia) appear to be so largely involuntarily; Moreover, even though the desired flexibility associ- they would prefer formal jobs. ated with self-employment and microfirm ownership is There is also considerable variation across countries in often associated with low income and little security, those the plausible causes of job segmentation for those groups jobs are still considered better options than the corre- of informal workers who report being involuntary. For sponding jobs that the workers could get in the formal example, in Brazil, Colombia, and Peru, the evidence economy, given their minimal human capital, access to points to an impact of increasing labor costs and rigidities other assets, and the low aggregate productivity in the on the expansion of informality in the 1990s. In other economy. In other words, many self-employed workers countries, such as Argentina, the introduction of tempo- opt for informality because their options in the formal rary contracts and subsequent weakened enforcement of sector are at least equally poor. Furthermore, they have tax and labor regulations seem to have played a role in the the possibility of using informal mechanisms or freely observed increases in the share of the labor force without available (noncontributory) social protection programs to social security contributions. In most countries, trade substitute, in part, for formal social security benefits that liberalization appears to have had modest or no effects on they would otherwise have to pay for, explicitly or implic- informality trends. itly, via (sometimes steep) payroll taxes. In summary, The presence of both exit and exclusion factors in most of the self-employed do not appear to be "excluded" informal labor markets--apparent in informal workers' from the formal sector; rather, after implicitly making a subjective motivational responses and self-rated welfare cost­benefit analysis, they opt out of formality. measures reported above--is corroborated by the A different picture emerges, however, for the majority observed patterns of labor market dynamics, an important of informal salaried workers in the countries studied. analytical tool introduced in this report. Indeed, evidence Indeed, most of the informal salaried appear to be queuing from Mexico (figure 3), suggests that, during good times, for more desirable jobs in either the formal salaried sector the number of workers who leave the formal sector to or as self-employed workers (with Mexico and the Domini- become self-employed or take an informal salaried job is can Republic being notable exceptions). For many of these nearly equal to or sometimes even greater than the num- workers, informality largely reflects the decision of the ber of those who transit from the informal to the formal firms for which they work, especially microenterprises, to sector. This is a major piece of evidence supporting the operate outside the realm of government regulations. view of integrated labor markets in which workers are On average, these workers are not choosing to opt out of freely choosing between formal and informal jobs. formal contracts and social security institutions, and they would much prefer an equivalent job in the formal sector. In these cases, exclusion rather than exit from formality FIGURE 3 appears to be the driving force behind their present Probability of transition between formal salaried and informal status. self-employment in Mexico However, considerable heterogeneity exists even within .08 these two subsectors of the informal sector, with both .18 voluntary and involuntary entrants and a great variety of .07 Formal to .16 workers' life trajectories found in each. For example, self-employment .06 (left scale) while, on average, salaried workers in microenterprises .14 appear to want a formal sector job or self-employment, .05 .12 approximately half of the employees of Mexican microen- Self-employment .04 to formal .10 terprises turn out to be unpaid family members whose (right scale) .03 .08 presence in the sector probably reflects a mix of profes- 1987q1 1991q1 1995q1 1999q1 2003q1 sional and personal considerations. Similarly, while, on Source: Bosch and Maloney 2006. average, independent workers are as well-off as their formal 6 I N F O R M A L I T Y: E X I T A N D E X C L U S I O N However, although initially following a pattern similar to informal salaried work is a point of entry to the labor mar- that of Mexico, after the constitutional reform in Brazil ket for many of the young, and, as they accumulate experi- the flows from informal to formal salaried employment ence or simply queue, they are more able to find a job in the became significantly lower than the flows in the reverse formal sector or fulfill a surprisingly common desire to be direction even in boom times--suggesting a significant self-employed. Informal salaried employment is also an degree of market segmentation. Consistent with the liter- option for many older workers who lack the skills or capital ature focused on the U.S. economy, it is also true that for- to become self-employed or get a formal salaried job, or mal sector hiring comes almost to a halt during bad who opt out of informality because, for example, they will times, perhaps reflecting more binding downward wage never accumulate enough years to secure a meaningful pen- rigidities. Hence, many entrants have no choice but to sion. The fact that participation in self-employment rises join less desirable occupations in the informal sector. As a with age is, again, similar to that of the United States, and consequence, the performance of labor markets is asym- similarly may be partly explained by voluntary entry metric throughout the business cycles in most countries: delayed by credit or human capital constraints. labor markets tend to behave more as segmented markets The findings discussed above have critical implications during downturns and recessions, and more as integrated for policy design. Whether because of outright exclusion markets during booms. and market segmentation or massive voluntary opting out Finally, both the more exclusionary and exit dimensions of formality, informality may lead to a suboptimal social of informal salaried work are suggested by workers' life- equilibrium in which many workers go unprotected from cycle employment trajectories. In most countries, young health and employment shocks and from poverty in old workers, especially the less educated, are more likely to be age. In either case, the imperative for reform is strong. informal employees than formal salaried workers, and very However, our findings caution against one-size-fits-all few are informal self-employed. Meanwhile, middle-aged solutions. Again, whereas labor markets in Mexico and the and older workers are more likely to be found in the formal Dominican Republic show a high degree of integration-- sector or to be self-employed, although many still end up for example, most informal workers choose to be as informal employees (see figure 4). This suggests that informal--except, perhaps during crises--labor markets in Argentina, Colombia, and, to a lesser extent, Brazil show many signs of segmentation. Thus, it is very useful to clar- ify the relative importance of alternative drivers of infor- FIGURE 4 Rate of urban employment across sectors, by age in Brazil, 2002 mality so that strategists may devise the most suitable policy changes. Percent 50 For example, if labor informality were essentially 45 driven by inadequate labor legislation leading to labor mar- Formal 40 ket segmentation, then the key to reducing informality 35 would be to engage in (politically demanding) labor market reform. If, however, expected low benefits and high costs of 30 Self-employed social security contributions relative to other forms of 25 protection against shocks are prominent in driving many 20 workers to opt for informality, reform of the design and pro- 15 vision of social protection benefits must be an additional 10 Informal component of the policy package to reduce informality. 5 Indeed, this report finds that there is significant room for 0 15 19 23 27 31 35 39 43 47 51 55 59 63 67 improvements in the area of social protection program Age design (box 2). Further, when considering the case of infor- Sources: Cunningham et al. 2007; author's estimates, based on mal salaried workers, it is also key to understand and Pesquisa Mensal de Emprego, September 2002. address the factors that lead firms to be partially or entirely Note: Considers the availability of a carteria de trabajo. informal, as we discuss in the next section. 7 I N F O R M A L I T Y BOX 2 Government failures in the design of social protection systems in Latin America and the Caribbean There are high human costs associated with the lack of rise to benefits for many formal sector (or potential access to appropriate risk management instruments, such formal sector) workers. as health insurance and old-age security. In addition, · Rigid, one-size-fits-all approaches to mandated social missing insurance markets and other market failures-- security programs: For example, spouses and sec- such as negative externalities associated with the absence ondary household workers who opt to enter formal of health treatment or unsupported poor elders--create a employment generally would have to pay for clear rationale for public intervention to ensure basic health coverage to which they already have a right access to social protection. Therefore, a key challenge fac- through the household head's affiliation. This dou- ing policy makers in Latin American and Caribbean coun- ble payment for a single benefit creates a powerful tries is to ensure that their citizens have suitable access incentive to choose informal jobs. Furthermore, to social risk management instruments, even in the face social security systems rarely account for different of significant informality in the region. However, ill- needs and preferences among workers at different designed interventions actually may serve to make things stages of the life cycle, and, as a consequence, high worse. They not only hinder people's ability to access basic initial payroll contributions create a significant dis- services or to manage risk, but also distort economic incentive for young workers who have other invest- incentives that may have adverse effects on productivity ment priorities (such as education and housing) to and long-term economic growth, for example, by creating join the formal sector strong incentives for many workers to remain informal. · Weak accounting for workers' mobility into and out of the Indeed, in Latin America and the Caribbean there is formal sector during their careers and across their life ample evidence of "government failure" in the design and cycles: In particular, lengthy pension vesting periods implementation of social protection, and that failure needs make many workers ineligible because of the high to be addressed as part of any actions to strengthen risk mobility between formal and informal sectors, and management among the region's citizenry. This report thus creates, ex ante, a major disincentive to choos- documents these issues in detail. At the level of specific ing formal jobs. Furthermore, lack of portability programs, design problems raise the costs of participating often leads to intermittent health insurance cover- in social security, relative to the benefits of participation, age gaps for workers who shift from formality to causing some workers to opt out of the system. Design informality and vice versa. deficiencies also impede some workers' access to benefits. Moreover, recent efforts of a number of governments in At the level of social protection systems--the constellation the region to use noncontributory or subsidized assistance of contributory social security and noncontributory social programs to compensate for low social security coverage assistance schemes--programs often compete, creating inadvertently may have complicated the quest to adverse labor market incentives and outcomes. provide effective social protection to a broader swath of Among the key weaknesses in the design of the the region's population. Indeed, the evidence suggests region's social security programs we find the following: that programs offering informal workers services on a · High payroll contributions relative to the expected bene- noncontributory or highly subsidized basis actually may fits and quality of service: Evidence shows that high "compete" with formal social security programs that contributions can act as a disincentive to formal workers have to pay for through payroll tax contributions, employment. thereby creating additional incentives for informality. · Excessive "bundling" of multiple benefit packages: For Addressing these design weaknesses will be important example, packages in some countries include not not only to ensure broader access to effective risk man- only health care, worker risk, life insurance, and agement instruments by the citizens of Latin America pensions, but also housing, child care, and sports and the Caribbean, but also to make social protection and recreation. Some of these components represent programs consistent with increased productivity and sus- "pure taxes" rather than contributions that give tained economic growth in the future. 8 I N F O R M A L I T Y: E X I T A N D E X C L U S I O N Firms: Little gain, high costs, formality can be seen as an input in the production process or weak enforcement? for which small firms have little need. Formality increases Labor informality is primarily a small firm phenomenon rapidly with firm size and productivity. As an example, with the vast majority of workers who are unregistered whereas 86 percent of Mexican microfirms with only one with social security administrations found in firms of paid employee do not pay social security contributions, fewer than five workers (figure 5). Hence, understanding 71 percent of those with five paid workers report paying the rationale behind small firms' decisions to register their social security for at least some of their employees. firms and their workers, to pay taxes, and so forth is funda- In Brazil, 76 percent of microfirms do not have an operat- mental to understanding the phenomenon of informality. ing license and 94 percent do not pay taxes. Those rates The report shows that, in nature and dynamics, these fall to 33 and 23 percent, respectively, among firms with microfirms are closely related to their advanced-country five paid workers. counterparts and should not be treated as a pathology pecu- However, the large majority of microfirms remain too liar to developing countries. As figure 6 shows, the patterns small to benefit sufficiently from formality to overcome its of entry, exit, and participation in self-employment are various costs. In Brazil, for instance, 87 percent of all very similar, suggesting that the high levels of self-employ- microfirms have no paid employees. In Nicaragua, less than ment seen in figure 1 may not reflect a fundamentally dif- 7 percent of microfirms have more than two employees ferent phenomenon but simply much more of it. This may after three years of operation. Most of these firms have no arise precisely because the lower labor productivity of the potential to grow (turnover is extremely high in this sub- formal sector in developing countries implies a lower sector), and thus their credit needs may be marginal. They opportunity cost of participating in independent work, have a limited number of clients and with most of those which worker surveys suggest is as highly regarded in Latin clients they have personal relationships, so they would not America and the Caribbean as it is in countries of the benefit much from greater access to the organized imper- Organisation for Economic Co-operation and Development sonal markets and courts usually associated with formality. (OECD). It is not surprising that a survey of informal Mexican In this light, although a sizable body of literature microfirms reveals that nearly 75 percent of the microfirms sees these firms' lack of registration with the authorities report the main reason for not registering with the author- through an exclusionary lens, this report argues that ities is that they are just too small to make it worth their FIGURE 5 Informal workers across firm size Argentina (Greater Buenos Aires) Mexico (urban areas) % informal and self-employed workers % informal and self-employed workers 50 50 1980 45 45 2004 40 40 35 35 2003 30 30 25 25 1994 20 20 15 15 10 10 5 5 0 0 1 2­5 6­25 26­100 101­500 501 1 2­5 6­10 11­15 16­50 51­100 101­250 250 Firm size (number of workers) Firm size (number of workers) Source: Author's calculations, based on Encuesta Nacional de Empleo Urbano (Mexico) and Encuesta Permanente de Hogares (Argentina). 9 I N F O R M A L I T Y FIGURE 6 Male entry into and exit from self-employment Mexico United States Percent Percent 60 60 50 Self-employment rate 50 40 40 Exit rate Exit rate 30 30 20 20 Entry rate Self-employment rate 10 10 Entry rate 0 0 20 35 55 ­65 ­65 18­ 21­25 26­30 31­ 36­40 41­45 46­50 51­ 56­60 61 18­20 21­25 26­30 31­35 36­40 41­45 46­50 51­55 56­60 61 Age Age Sources: Evans and Leighton 1989; Fajnyzlber, Maloney, and Rojas 2006; and author's calculations, using Encuesta Nacional de Empleo Urbano. while. In contrast, only 2, 8, and 4 percent of surveyed The firm-level evidence reported here presents a mixed firms, respectively, answer that the main reason is the high record on the above hypotheses. The microfirm evidence costs and time required to register or the high costs of oper- suggests that firms choosing to register do have better per- ating as registered businesses. The same appears true in the formance or, alternatively, the firms that started operations Dominican Republic, although prohibitive registration red being registered exhibit, on average, higher levels of labor tape seems to be of greater import in Argentina. In most productivity than their equivalent unregistered peers. In cases, the degree of formality increases with firm size, sug- particular, programs in Brazil and Mexico to reduce the gesting that as firms grow, their demand for the services costs of becoming formal have had a positive impact on the associated with formal institutions increases, as does the performance of those firms that decided to cross the frontier probability of detection by authorities. from informality to formality. The evidence on the effect of Having so much of the workforce in microfirms and these programs on the aggregate levels of informality and having so many of these microfirms unplugged from the productivity through time remains inconclusive (as will be formal economy in all likelihood extract some productivity discussed in chapter 6). Further, evidence from the invest- costs. First, there may be scale economies or externalities ment climate reports suggests that firms surrounded by arising from employment in large firms. To the extent that complying firms have higher rates of productivity. How- informality is associated with a preponderance of small ever, there is evidence that, in some cases, informality firms and that increasing returns to scale are relevant for reflects defensive evasion of possibly excessive regulation. their size range, informality could lead to considerable effi- All evidence considered, the report concludes that ciency losses. Second, unfair competition from informal tilting the cost­benefit analysis toward formality for a sub- firms could slow down the process of creative destruction stantial percentage of informal firms requires a fair combi- by which inefficient firms are replaced by their more nation of "carrots" and "sticks." As an example, in Brazil efficient competitors, and could negatively affect the and Mexico, microfirms constrained at the frontier coexist incentives of formal firms to innovate and adopt new tech- with entrepreneurs who show no demand for the presumed nologies. Third, if reducing the costs and increasing the benefits of formality and for whom reducing registration benefits of formality allow previously informal firms to costs would not lead to formalization under present condi- gain increased access to markets and services, the result tions. Similarly, evidence for the Dominican Republic sug- could be increased aggregate productivity growth. gests that many informal firms perceived more gains from 10 I N F O R M A L I T Y: E X I T A N D E X C L U S I O N staying under the state radar, including not only the tax have a significant number of employees without social secu- savings but also the avoidance of interaction with the state rity contributions, and the fraction of underreported sales for bureaucracy altogether. Further, a large fraction of firms tax purposes is quite high--even as high as in small firms, in that become formal allegedly do so to avoid fines and some cases (see figure 8). Firms again appear to undertake a bribes, according to surveys of many countries in the region careful cost­benefit analysis, weighing the "private" benefits (see figure 7). Thus, interventions to address regulatory constraints faced by small firms or to reduce tax rates might not be enough to achieve a major change in their FIGURE 8 Underreporting of tax and social security contributions, by firm size levels of formality. Such a goal would also require improv- ing the positive incentives for joining the formal sector, Number of employees including improvements in private and public services 5 6­10 available to formal firms (for example, credit, contract 11­20 Panama 21­50 enforcement, technical assistance, and the like), and 51­99 enhancing the level of enforcement to increase the opportu- 100 5 nity cost of remaining informal. 6­10 11­20 Nonetheless, the low aggregate productivity in the for- Brazil 21­50 mal sector would put some limits on the overall effect of 51­99 100 even well-designed, integrated programs to induce infor- 5 6­10 mal firms to become formal. Hence, a large chunk of the 11­20 Mexico efforts to reduce informality must focus on policies to 21­50 51­99 enhance productivity and growth in the formal sector, 100 5 which would increase the perceived benefits of formality, 6­10 along with the opportunity cost of remaining informal. 11­20 Bolivia 21­50 Finally, in many countries in the region, even large firms 51­99 100 show considerable tax evasion and labor informality. They 5 6­10 11­20 Argentina 21­50 51­99 FIGURE 7 100 Advantages of formalization reported by IFC-surveyed firms 5 6­10 11­20 Colombia 21­50 Avoid paying fines 51­99 100 Compliance 5 with the law 6­10 11­20 Uruguay 21­50 Avoid paying bribes 51­99 100 Gain new clients 5 6­10 11­20 Improved Peru 21­50 access to credit 51­99 100 Operation on a greater scale 5 6­10 Legal power to 11­20 demand contracts Chile 21­50 is upheld 51­99 0 1 2 3 4 100 0 10 20 30 40 50 Mean of firms with no employees Mean of firms with at least one employee Unreported sales Unreported workers Source: Investment climate surveys 2006. Source: Investment climate surveys 2006. 11 I N F O R M A L I T Y of informality (evading taxes, avoiding burdensome regula- As an example, most countries in the region are charac- tions) against the "private" costs of informality (risk of terized by "truncated welfare systems," in which those in penalties and bribes, imperfect access to markets and govern- the formal sector have access to an often generous multidi- ment services) in choosing their "degree" of formality. mensional package of social security, while those in the Administrative and tax simplification programs to reduce informal sector have much more limited access to govern- red tape and compliance costs, regulatory reviews aimed at ment benefits or formal risk management instruments. Not eliminating laws and regulations that are either anachronis- only has overall progress in increasing social security cover- tic or privately motivated, and enhanced enforcement are the age been slow in the region, but coverage has actually key policy responses with respect to the phenomenon of declined in a number of countries throughout the 1990s. partial informality within large firms. And, in nearly all countries in the region, coverage rates are significantly lower among low-income than among high- Need for more effective income workers (see figure 9). Indeed, in many countries and legitimate institutions the poorest workers and families are practically excluded Whether informality is an outcome of exclusionary policies from the system. At the same time, although independent or mechanisms, or a result of cost­benefit decisions by workers can choose to participate in formal social security firms and individuals that lead them to opt out of formal systems in many countries, very few do it voluntarily. institutions, it represents a fundamental critique of the Actually, there is evidence that many workers opt out from Latin American state at several levels. In the burdensome the system once they obtain the right to a minimum pen- business and labor market regulations, as with the poorly sion (see Gill, Packard, and Yermo 2004). designed social protection systems, the state is complying The underperformance of Latin American and Caribbean inadequately with its designated roles. This failure, in con- states may reflect deeper, poorly resolved social tensions that junction with an abiding lack of confidence in who the may be thought of as constituting a dysfunctional underly- state represents and serves and in its capacity to enforce the ing "social contract"; in Hirschman's terms, the lack of voice law, may intensify the tendency of many Latin Americans-- and loyalty. Beyond high informality, this can also be seen perhaps no more or less than citizens of other emerging in the inability of the state to redress the long-standing regions--to opt out, rendering the fulfillment of the funda- high inequality, in the weak rule of law, in the sometimes mental roles of the state all the more difficult. large share of undocumented citizenry, or in the recurrent FIGURE 9 Economically active population contributing to the pension system Percent 100 Poorest quintile Second quintile Middle quintile Fourth quintile Wealthiest quintile 90 80 70 60 50 40 30 20 10 0 Peru Colombia Mexico Argentina Chile Sources: Adapted from Rofman and Lucchetti 2006. Note: Estimates based on national household surveys from the following years: Peru (2003), Colombia (1999), Mexico (2002), Argentina (2004), and Chile (2003). 12 I N F O R M A L I T Y: E X I T A N D E X C L U S I O N (tax privileges and loopholes, weak competition laws, and FIGURE 10 cumbersome regulations). Those inefficiencies lead to low Informality versus inequality tax collections and the abuse of monopolistic power, and they % of population without a pension further erode state legitimacy and its capacity to provide 40 public goods and services and to enforce the law. Some recent PER 20 ECU ARG literature has focused on the possibility that Latin America is MEX PARBOL NIC DOM GUA locked, in fact condemned, to a bad social equilibrium where VEN HON BRA 0 URU SAL entrenched elites have no interest in responding to the voiced COL petitions of those seeking full participation. JAM CHI 20 COS PAN However, this report is more optimistic about the possi- 40 bilities for incremental reform, despite acknowledging the vital importance of enhancing the legitimacy of the state. 60 Ironically, this conclusion partly emerges from the study of coef .72, t 3.5 the distressing increases in informality in many countries 80 20 10 0 10 20 of the region over the 1990s. These increases can be traced Gini coefficient partly to sharp increases in real minimum wages in some Source: Author's estimates, based on World Development countries, inadequate macroeconomic policies (that led to Indicators 2006. Note: Figure shows partial correlations controlling for GDP artificial booms in nontradable, highly informal sectors) in per capita at PPP. some countries in the early 1990s, changes in labor market and social security legislation or weakened enforcement bouts of macroeconomic instability. For example, high capabilities, and to the increased availability of noncontrib- inequality of incomes and power is correlated with infor- utory social protection schemes for informal workers. In the mality (see figure 10) and is often associated with weak meantime, poorly designed social security systems continued institutions and state capture by both elites and organized to tax heavily workers in the formal sector. Indeed, if infor- segments of the middle class (see de Ferranti et al. 2004; mality increased in many countries during a decade that Perry et al. 2006). State capture leads to the generalized per- exhibited modest growth due at least partially to a combi- ception that the state is run for the benefit of the few, and nation of policy missteps, then a shift toward better policies thus it reinforces a social norm of noncompliance with taxes at least can reverse these upsetting trends. Furthermore, and regulations, what might be dubbed a "culture of infor- although it appears that inequality and informality share mality." Noncompliance is then further compounded by the important causal determinants, Chile shows that strong suspicion that others are not complying either--an absence and more fair institutions and a fall in informality can be of what is termed "strong reciprocity--which, in turn, built even in the midst of high income inequality. Quoting makes enforcement even more difficult." As an example, Hirschman (1971) again, we find a "bias for hope"--good in the Dominican Republic, 19 percent of small firm owners policies can have important effects. report that they do not register or contribute to pension systems because "businesses like them don't." Summing up: Policy implications of the report At a more aggregate level, after controlling for per capita Achieving significant reductions in present informality income, informality is negatively correlated with "tax levels will require, first and foremost, actions to increase the morale"--society's disposition toward tax compliance-- aggregate productivity in the economy. A more enabling which, in turn, depends inversely on perceptions of investment climate will permit formal firms to expand and "state capture" (see figure 11) and positively on perceptions pay higher wages. Raising human capital levels, especially of the quality of public spending. Furthermore, informality for the poor, will permit more workers to find remunerative measured as the share of the workforce in self-employment is jobs in a more dynamic formal sector. Without such negatively related to the quality of institutions, as indicated improvements in aggregate productivity, we will continue in figure 12. This suggests that greater exit to independent to find a very large number of microfirms, characterized by work accompanies higher distrust of the state. Inequality high turnover, weak growth prospects, and low productiv- and its correlates then lead to inefficient laws and regulations ity, that would see little benefit in engaging with formal 13 I N F O R M A L I T Y FIGURE 11 FIGURE 12 Tax morale and state capture Self-employment and quality of institutions (governance) Tax morale Share of self-employed 0.2 0.3 COS SAL GUA 0.2 0.1 PAR BRA NIC HON COL ARG 0.1 VEN BOL PER COL HON 0 CHI GUA JAM CHI 0 BRA PAN ECU MEX COS MEX ARG SAL URU PAN 0.1 ECU BOL PER 0.1 coef .07, t 4.7 coef .88, t 2.8 0.2 0.2 1.5 1.0 0.5 0 0.5 1.0 1.5 0.15 0.10 0.05 0 0.05 0.10 0.15 Government effectiveness State capture Sources: Author's estimations, based on World Development Source: Author's estimates, based on data from World Indicators; Worldwide Governance Indicators, World Bank 2005; Development Indicators 2006 and Latinobarómetro 2004. and investment climate surveys, 2005. Note: Figure shows partial correlations controlling for GDP Note: Partial correlations controlling for GDP per capita at PPP. per capita at PPP. State capture is proxy by an indicator of the Government effectiveness index measures the quality of perception about the economy being run according to the interests public service provision, the bureaucracy, the competence of civil of a few. To construct the indicator we ask: In general terms, would servants, the independence of the civil service from political you consider that the country is governed according to the interests pressures, and the credibility of the government's commitment to of a few, or is it governed for the benefit of the country? policies. Higher values correspond to a more effective government. institutions. Without generalized improvements in human labor market segmentation, as evidenced in most countries capital, many unskilled workers would continue to prefer at least during some periods, need to be reduced. We dis- self-employment, even if their earnings are low, because the cuss below some policy changes that may be especially jobs they could find in the formal sector would also deliver promising in particular circumstances. very low earnings. Such improvements in human capital Achieving a deeper change in incentives also requires have to be synchronized with those in the investment actions to change the pervasive culture of noncompliance climate, as stressed in previous studies (particularly, de that we observe in most countries in the region. Because Ferranti et al. 2003), because otherwise the demand for such social norms are, in part, the result of a lack of trust in more skilled workers will not grow at the same speed as the effectiveness of the state and the equity of its actions, the increase in supply--thereby depressing skill premiums overcoming the culture of informality probably requires (for instance, to secondary education) and further eroding major improvements in the quality and fairness of state incentives to invest in education. institutions and policies. In short, it requires building an However, there are many things that can be done to effective and inclusive social contract in which the great change the balance of incentives for those workers and majority of individuals feel compelled to participate and small firms whose implicit cost­benefit estimates place comply with state mandates. them close to the margin between formality or informality. The same can be said of larger firms that remain partially informal. To have a significant effect, such actions normally Reforming labor market policies to increase require a good balance of carrots (reforms and actions to productivity of the formal and informal sectors facilitate, reduce the costs, and increase the benefits of for- The findings of this report suggest that labor market poli- malization) and sticks (enhanced, evenhanded enforcement cies are important determinants of informality, working of such improved laws and regulations). Furthermore, those through three channels. First, excessive labor costs, excessive regulations and taxes that create some degree of whether arising through labor legislation or unrealistic 14 I N F O R M A L I T Y: E X I T A N D E X C L U S I O N union demands (such as exaggerated minimum wages, related government- or union-induced inflexibilities, may severance costs, or labor taxes and contributions), depress have a disincentive effect on technology adoption, which the number of jobs in the formal sector, often creating the accounts for up to half of the differences in levels of eco- classic segmented market. Other recent reports from the nomic development. And, as shown in this report, the level World Bank--"Minimum Wages and Social Policies: of economic development is the most important determi- Lessons from Developing Countries" (Cunningham nant of observed informality levels. 2007b) and "Job Creation in Latin America and the Overall, the present constellation of often well-intended Caribbean" (World Bank 2007)--as well as the Inter- but heavy-handed labor regulations poorly serves work- American Development Bank (IDB) (2006) publication ers and firms, and both could benefit from substantial Good Jobs Wanted, have investigated in detail the trade-offs reform. In particular, stronger enforcement of a redesigned encountered in offering strong protections to some work- labor code that combines strengthened safety nets, ers at the possible expense of excluding others (see also de well-designed worker protections, and worker representa- Ferranti et al. 2001). Minimum wages in most countries tion with the flexibility that firms need to adapt in a global are not extremely binding, but in some--such as Colom- economy has the potential to expand formal employment bia--they seem to be a deeply segmenting force that begs and reduce opting out. Simply tightening the enforcement for moderation. Higher non-wage burdens on formal jobs of existing laws, particularly in the largely informal in Brazil, Colombia, and Peru appear responsible for sub- microfirm sector, may just eliminate jobs--many of which stantial declines in formal employment. Further, the expe- the chapters in this volume will show to be of good quality rience in the OECD countries suggests that such when measured by the worker's overall welfare. At the regulations have a heavy exclusionary impact on young other extreme, attempting to reduce the weight of labor people who are trying to find jobs and who, in Latin legislation by creating special classes of less protective con- America and the Caribbean, are especially overrepresented tracts can be problematic. When well designed, such con- among the involuntary informal salaried. Nor does the tracts may offer flexibility that helps young people enter evidence of a very high degree of integration of the formal the market. However, they can often effectively create a and informal sectors in some other countries in the region parallel, unregulated "formal" sector that displaces formal necessarily imply satisfactory labor codes. Latin America's contracting. That can lead to higher job turnover and to globally very high levels of severance costs, for example, diminished incentives for training and may contribute to may substantially reduce job creation arising from growth the overall culture of informality. Provisions to accommo- without necessarily segmenting the market. Furthermore, date different non-wage costs for smaller firms and flexibil- as in the case of minimum wages, regulations sometimes ity in benefits plans (such as simplified health/pension can be binding in practice for the informal salaried sector plans) may offer an extension of the overall rubric of labor and may inhibit job creation there. protections without prejudicing the viability of these firms. Second, legislation can create incentives for voluntary Perhaps more flexible work schedules, or enhanced avail- informality. The design of social safety nets and labor legis- ability of day care, would keep more women in the formal lation needs to consider a more integrated view of the labor sector, although care must be taken to avoid measures that market, taking into account the cost­benefit analysis that might lead firms to discriminate against women in hiring workers and firms make in deciding whether to interact as a result. Finally, institutional strengthening (staffing, with formal institutions. High labor taxes or contributions training, technical assistance) of the labor ministries and that do not correspond to benefits that workers value cause coordinating of relevant public agencies (social security workers to opt out of the formal labor market. The difficul- administration, enterprise development agencies, and com- ties of managing work and children under rigid formal petitiveness councils) are needed so that those ministries work contracts also lead young mothers to opt for informal and agencies can assume their increasingly more complex independent work offering more flexibility. role of facilitating labor productivity growth. Third, the effect of labor market institutions on produc- Informality sharply decreases with education, partly tivity growth has probably been underemphasized. Theory because the opportunity cost of being independent rises. and anecdotal evidence suggest that excessive restrictions However, the substandard education and training systems on job reallocation or destruction for just cause, or other in the Latin American and Caribbean region both impede 15 I N F O R M A L I T Y the growth necessary to generate jobs in the more modern costs on society, there is a case for providing a package of sector of the economy and reduce workers' attachment to minimum essential direct cover, de-linked from the labor it. Furthermore, the poor signaling of education quality contract and financed through general taxation. In the case that results from a lack of uniform certification or accredi- of old-age security, there is also a case to provide essential tation impedes the entry of young workers into formal cover in the form of a poverty prevention pension targeted jobs. Remedying these failures, perhaps along with an toward the poor as part of a broader multipillar pension expansion of intermediation services, may reduce the system that includes provisions for individual retirement information asymmetries that young workers face. Further savings. The social costs associated with people falling into discussion of necessary reforms in this area is offered in the poverty in old age create a clear risk management rationale regional report "Raising Student Learning in Latin for providing minimum old-age support de-linked from America and the Caribbean: The Challenge for the 21st the labor contract; nonetheless, the high probability of Century" (Vegas and Petrow 2007). Ongoing upgrading of income loss in old age suggests that saving should play the the workforce through training, particularly in rapidly main role in earnings replacement during old age. evolving industries, is a central element of the national For a number of reasons, including those related to fiscal innovation system, and is critical to developing skills and institutional capacity, movement to provide essential used in the modern sector of the economy and to promot- cover in health care and old-age security, de-linked from ing productivity growth. Training systems in the region, the labor market and financed by general taxes, represents a however, need to be more competitive and responsive to long-term agenda for many countries in the region. It is market demands. important, therefore, for countries to orient their short-to- In sum, the high rate of informal employment in some medium-term policy agendas in ways that are consistent-- cases reflects classic segmentation, but in others it simply or at least not inconsistent--with their long-term vision. reflects the high costs and small benefits of employment in This will be critical if the region's governments are to the formal sector. Labor regulations need to allow for pro- ensure that the objectives of social policy--and particularly ductivity growth in both sectors, while revisiting the social risk management--are well aligned with those of design of regulations and social protection systems that higher productivity and growth. provide incentives for firms and individuals to become To this end, there will be high returns to governments informal. in the region undertaking step-by-step reforms to improve the efficiency of existing programs as well as to establish Reengineering social protection to cover greater consistency and incentive compatibility across pro- all citizens grams. Several sets of actions will contribute to short-term Central to this agenda is the need to rethink and, in fact, improvements in social protection while moving countries reengineer social protection policy and programs in much in the direction of achieving essential cover over the long of Latin America and the Caribbean. Poor access to basic term. These actions would include efforts to improve the risk management instruments, coupled with poor design of cost­benefit ratios of programs in several ways. First, com- social security schemes, serves workers poorly and provides plex, multidimensional social security benefit packages an incentive to be informal. Fundamentally, this may should be unbundled to focus on increasing access and require rethinking the traditional model of social protec- quality of those programs with high public goods content tion in which protection depends on the specific form of the (for example, health care, old-age security). Second, efforts labor contract. A broader notion of who has access to basic should be made to improve program quality and benefits risk management instruments is needed--one based on via microefficiency reforms in health care and country ensuring the basic protection and welfare of countries' citi- pension systems. Third, program design should be zens rather than of "workers" as traditionally and narrowly strengthened--for example, by revising overly burdensome defined. pension-vesting periods, fostering benefits portability, Drawing insights from the economics of insurance, this eliminating double payments for health insurance, and so report outlines a long-term vision for social protection on--to account for and enable greater worker mobility. reform in the region.1 In the case of health care, because Finally, rules, eligibility requirements, and benefits levels shocks that go "uncovered" can impose significant external across programs and institutions of social security and 16 I N F O R M A L I T Y: E X I T A N D E X C L U S I O N social assistance should be made consistent and compatible registration costs. Hence, although high entry costs may with incentives. Such types of measures will provide the keep some entrepreneurs from starting new formal busi- foundation for more effective social protection for all citi- nesses, the available evidence suggests that just lowering zens and help strengthen the alignment of social and administrative barriers to business registration may not economic policy objectives. have a major impact on aggregate levels of informality. This finding does not imply that those interventions are Improving opportunities for workers in the not important. First, lower entry costs do induce at least formal sector while reducing barriers to some entrepreneurs to open new formal businesses, and business formalization their improved performance probably would be sufficient As suggested above, the fact that few micro firms grow justification for the corresponding reforms. Second, the from small informal entities to large formal firms can be impact of administrative simplification programs may explained in two complementary ways. On one hand, the be larger when accompanied by other complementary presence of low opportunity costs for entry into the sector measures aimed either at increasing the potential benefits could be to blame because it would lead to a predominance of joining the formal sector or at reducing the costs of of low-productivity businesses with low growth prospects regulatory compliance--beyond the facilitation of firm and high failure rates. In this context, policy makers wish- registration. ing to reduce informality should focus not only on altering With regard to the first alternative, recent evidence the direct costs and benefits of formality but also on alter- from randomized experiments suggests relatively high ing the drivers of formal sector productivity, including returns to capital among very small Mexican microenter- measures to improve the investment climate and policies prises, and this suggests that considerable increases in aimed at increasing human capital accumulation. On income could be obtained through measures to increase the other hand, however, policy-induced barriers to small businesses' access to bank credit and other forms of formalization could impede microfirms' access to technolo- external finance. Similarly, in principle, formality could be gies, markets, and government services, and that would made more attractive by improving the provision of busi- explain at least some of their low growth and job creation ness development and training services available to formal rates. In this second interpretation, reductions in informal- firms, and by facilitating access to product markets ity and improvements in microfirm performance could be through public procurement opportunities and supplier achieved by lowering statutory burdens to formalization development programs aimed at increasing links with and implementing administrative simplification programs larger private firms. Other ways of making formality more aimed at reducing the transaction costs associated with attractive include improvements in the quality of legal ser- operating legitimate businesses. In particular, programs vices available to small businesses so that they may find it that facilitate business registration--for instance, through less risky to expand beyond local markets, and creating the use of Internet-based technologies and one-stop- mechanisms to provide information to entrepreneurs wish- shops--should lead to an increase in formality rates and ing to formalize their businesses (from advisory services on improved microfirm performance. taxes and regulations to information on financial and nonfi- In practice, the existing evidence from Brazil and nancial services available to them). Mexico suggests that administrative simplification pro- As for measures to reduce other costs of operating legiti- grams do lead to increases in formal firm registration rates, mate businesses, governments should consider performing with large consequent improvements in the revenue- and comprehensive regulatory reviews aimed at eliminating employment-generating capacity of the corresponding unnecessary and costly bureaucratic requirements. In this businesses. However, the number of positively affected respect, the challenge is that of distinguishing relevant from firms is relatively small in comparison with the overall size anachronistic regulations, as well as identifying those regu- of the informal microfirm sector. Moreover, at least in the lations that significantly increase the costs of operating pri- case of Mexico, it appears that most of the newly registered vate businesses and are not justified by legitimate public firms are new entrants--that is former wage workers who interests, such as the protection of public safety or the envi- decide to open new formal businesses--rather than exist- ronment. If successful, such regulatory reforms may help ing informal firms that become formal as a result of lower reduce informality both by increasing job creation in the 17 I N F O R M A L I T Y formal sector and by reducing regulatory noncompliance tration can be enhanced through better use of third-party among formally registered, medium-size and large firms-- information (for example, cross-references between tax for example, tax and social security evasion. Indeed, reporting, social security records, and data from the financial although cross-country differences in the size of the infor- system). Audits can be made more effective by the adoption mal sector are largely explained by countries' general levels of modern audit technology, such as has been the case in of development--which, in turn, are driven partially by the Chile and Spain. In most countries there is scope for reduc- quality of governance and institutions--there is also evi- ing administrative and compliance costs and increasing dence that, for given levels of per capita income, informality collections through changes in tax structure, combining tends to be higher where regulatory burdens are heavier. reductions in marginal tax rates with the elimination of Moreover, not only the quantity but also the quality of reg- exemptions and privileged regimes that create loopholes in ulations appears to matter, as is illustrated by the finding value-added, income, and property taxes, and simplification that firms facing higher levels of corruption are more likely of deductions. Collections can also be increased by facilitat- to evade taxes and social security. In that context, the objec- ing payments through the banking system, relying more tive should be to eliminate unnecessary regulations while heavily on source withholding and applying non-harsh enhancing their evenhanded enforcement and streamlining penalties for noncompliance often and consistently. Finally, administrative processes to reduce excessive red tape. criminalization of certain tax offenses in combination with Overall, a wider and integrated approach appears to be a modernized tax administration agency has been credited necessary to switch the incentives of a large fraction of with playing a key role, among other factors, in Spain's informal firms in the direction of formality. Such an success in the late 1970s and 1980s in drastically reducing approach would likely have to combine carrots (such as tax evasion and eventually doubling the ratio of tax revenue lower costs of formalization, better and more efficient gov- to gross domestic product. ernment services, and greater access to government- and Second, reforms should also emphasize the "service market-provided services for formal firms) with sticks (for paradigm" with policies to enhance the role of the tax example, increasing evenhanded enforcement of regula- administration as a facilitator and a provider of services to tions and, hence, the expected cost of being caught). The taxpayer-citizens. Promoting taxpayer education and devel- correct mix of policies, however, is likely to vary across oping taxpayer services in filing returns and paying taxes, countries and over time, depending on the relative impor- broadcasting advertisements that link taxes with govern- tance of the various determinants of informality. Moreover, ment services, stimulating voluntary compliance by lower- policies aimed at reducing firm informality should be con- ing compliance costs, simplifying taxes and their payment, sidered in conjunction with the labor market and social and promoting a taxpayer--and a tax administrator--"code protection issues associated with the possibility of large of ethics" have proved to be useful complementary measures contingents of previously informal workers shifting to to the punishment paradigm to enhance compliance. The other segments of the labor market. use of semiautonomous revenue authorities has been shown in several countries to improve tax administration with a Simplifying tax laws, facilitating compliance, more service-oriented approach to tax enforcement. and strengthening enforcement The service paradigm fits squarely with the perspective As mentioned above, underreporting of sales and incomes that emphasizes the role of social norms in tax compliance. for tax purposes is common even in large firms, although Experience from other countries shows that a government's the degree of compliance and its relation with firm size vary commitment to evenhandedly enforce the tax laws while widely among countries. To deal successfully with this facilitating taxpayer compliance can have an important dimension of informality, there is also a need for an inte- effect on the pervasive culture of noncompliance found in grated approach consisting of both sticks and carrots. many countries in the region. Disclosure of public expendi- First, there is substantial room for measures that fall ture information, and the participation of and supervision within the traditional monitoring and "punishment para- by citizens--that is, "voice"--in the way taxes will be digm" of tax compliance. In most countries, improvements spent, may also help increase trust in the state and may are needed in the three main aspects of tax administration: contribute to social norms of compliance. These factors taxpayer registration, audit, and collection. Taxpayer regis- have been credited with success in increasing tax 18 I N F O R M A L I T Y: E X I T A N D E X C L U S I O N compliance (and collections) in Chile and Spain, particu- policies emerging from the constitutional exercises that larly through widespread consensus among political parties strove to be more inclusive. Hence, however well meaning about the need for the reform of the tax system, improved or inclusive policies may be, they must be well designed. democratic governance, and highly visible enhancements The inclusive state must also be a competent state. in social and other public services. Informality and the development agenda Enhancing the effectiveness and legitimacy Informality reflects underdevelopment, and this report seeks of the state to tease out some of the interactions and directions of causal- Improving the quality of state policy making is one ele- ity between the phenomena of informality and development. ment in a larger agenda of reducing the culture of infor- There is evidence that, in part, informality is merely a stage mality that also requires increasing the perceived efficiency, in the development process: the ubiquitous microfirm fairness, and accountability of the state. To spin it posi- reflects the unattractive options in the small modern sector tively, a particular policy change (such as greater and more and the traditional reliance on family and community. How- evenhanded enforcement of adequate and equitable taxa- ever, other evidence suggests that, in part, informality is a tion laws and other regulations) may reduce informality by canary in the coal mine--the symptom of poor policies and, more than what would be expected, given individual elas- more profoundly, a lack of confidence in the state and per- ticities, if it induces a change in the social norm of tax and haps in our fellow citizens. To return to Hirschman, con- regulatory compliance that permits the state to improve fronted with a lack of voice in and relevance of the state, the provision of public goods and services and to enforce Latin Americans take their business elsewhere; and, in doing the law. this, they further undermine the region's growth prospects. As with informality more specifically, the task of bol- Hence, redressing the causes of high informality is part and stering the legitimacy of the state involves both carrots and parcel of the broader development agenda. sticks. Identifying the correct set of carrots to foster a sense Note of greater inclusion and responsiveness requires detailed 1. See de Ferranti et al. (2000) for an earlier application of the analysis of specific country circumstances to detect the economics of insurance in the region. "binding constraints" whose removal would cause a shift in the decision of a large number of firms and workers located References close to the three borders between formality and informal- Arias, Omar, and Leonard Lucchetti. 2007. "Informal Employment ity. Adequate sticks require modern enforcement tech- and Self-Rated Welfare: Measuring Compensating Differentials." niques (for example, many countries in Latin America and Photocopy. World Bank, Washington, DC. the Caribbean have not yet developed adequate tax- Bosch, M., and W. F. Maloney. 2006. "Gross Worker Flows in the Presence of Informal Labor Markets. Evidence from Mexico, auditing systems like those common in OECD countries), 1987­2002." Policy Research Working Paper 3833, World political resolve, and evenhanded enforcement of laws and Bank, Washington, DC. regulations that are perceived as fair. Cunningham, W. 2007a. "Life Trajectories of Latin American Again, to improve the legitimacy of the state through Workers." Photocopy. World Bank, Washington, DC. more equitable public policies and programs, it is necessary ------. 2007b. "Minimum Wages and Social Policies: Lessons from to consider carefully the formality incentives or disincen- Developing Countries." Photocopy. World Bank, Washington, DC. Cunningham, W., L. McGinnis, R. G. Berdu, and C. Tesliuc. 2007. tives generated by government programs. Greater inclusion "The Promise of Youth: Policy for Youth at Risk in Latin America or a more participatory social contract may not by them- and the Caribbean." Photocopy. World Bank, Washington, DC. selves reduce informality. Some countries' laudable de Ferranti, D., G. E. Perry, F. H. G. Ferreira, and M. Walton. 2004. attempts to extend free social protection services to infor- Inequality in Latin America and the Caribbean: Breaking with His- mal workers, made without reforming badly designed tory? Washington, DC: World Bank. social security systems that tax many of its formal contrib- de Ferranti, D., G. Perry, I. Gill, J. L. Guasch, W. F. Maloney, C. Sanchez-Paramo, and N. Schady. 2003. Closing the Gap in Educa- utors, have inadvertently reinforced perverse incentives tion and Technology. Washington, DC: World Bank. that could actually lead to higher levels of informality. Fur- de Ferranti, D., G. Perry, I. Gill, L. Servén, F. Ferreira, W. F. Maloney, thermore, the rise in informality in both metropolitan and M. Rama. 2000. Securing Our Future in the Global Economy. Brazil and Colombia results partially from well-intentioned Washington, DC: World Bank. 19 I N F O R M A L I T Y de Ferranti, D., G. Perry, D. Lederman, and W. F. Maloney. 2001. ------. 1981. Essays in Trespassing. Economics to Politics and Beyond. From Natural Resources to the Knowledge Economy. Washington, DC: Cambridge, UK: Cambridge University Press. World Bank. IDB (Inter-American Development Bank). 2006. Good Jobs Wanted: de Soto, H. 1989. The Other Path. The Invisibles Revolution in the Third Labor Markets in Latin America. Washington, DC. World. New York: Basic Books. Loayza, N., and J. Rigolini. 2006. "Informality Trends and Evans, D. S., and L. S. Leighton, 1989. "Some Empirical Aspects of Cycles." Policy Research Working Paper 4078, World Bank, Entrepreneurship." American Economic Review 79 (3): 519­35. Washington, DC. Fajnzylber, P., W. F. Maloney, and G. V. Montes Rojas. 2006. Perry, G., O. Arias, J. H. Lopez, W. F. Maloney, and L. Servén. "Microenterprise Dynamics in Developing Countries: How 2006. Poverty Reduction and Growth: Virtuous and Vicious Circles. Similar Are They to Those in the Industrialized World? Evidence Washington, DC: World Bank. from Mexico." World Bank Economic Review 20 (3): 389­419. Rofman, R., and L. Lucchetti. 2006. "Pension Systems in Latin Gasparini, L., and L. Tornarolli. 2006. "Labor Informality in Latin America: Concepts and Measurements of Coverage." Social Pro- America and the Caribbean: Patterns and Trends from Household tection Discussion Paper 0616, World Bank, Washington, DC. Survey Microdata." Photocopy. World Bank, Washington, DC. Vegas, E., and J. Petrow. 2007. "Raising Student Learning in Latin Gill, I., T. Packard, and J. Yermo. 2004. Keeping the Promise of Social America and the Caribbean: The Challenge for the 21st Century." Security in Latin America. Washington, DC: World Bank. World Bank, Washington, DC. Hirschman, A. O. 1970. Exit, Voice, and Loyalty: Responses to Decline in World Bank. 2006. World Development Indicators. Washington, DC. Firms, Organizations, and States. Cambridge, MA: Harvard Univer- Available at http//:www.worldbank.org. sity Press. ------. 2007. "Job Creation in Latin America and the Caribbean." ------. 1971. A Bias for Hope. New Haven, CT: Yale University Press. Photocopy. Washington, DC. 20 CHAPTER 1 The Informal Sector: What Is It, Why Do We Care, and How Do We Measure It? SUMMARY: This chapter seeks to unpack our understanding of the term informality, why we may care about it, and what dynamics may be driving its elements. The number of phenomena it encompasses and the limitations of its measures are manifold, dictating caution in employing the term. Yet two stylized facts remain: First, however measured, informal- ity is high in Latin America, although not obviously so for the region's level of development; and it remains an important phenomenon. Second, in several countries it has experienced striking increases over the last decades. Whatever adverse reg- ulatory, poverty, growth, or social morale implications "informality" may have, they have become more relevant with time. Introduction: What is informality? the "urban traditional sector," and so on. These terms The term informality means different things to different beggar analysis by assuming what has to be demon- people, but almost always bad things: unprotected workers, strated (Hart 1973, p. 68). excessive regulation, low productivity, unfair competition, With this, Hart cautions us against inquiry with exces- evasion of the rule of law, underpayment or nonpayment of sively well-informed ideas of what we may find. taxes, and work "underground" or in the shadows.1 The Fortunately, the accumulation of rich data sets over the multiplicity of adjectives from very distinct fields of study last decades has cast progressively more light on the realm suggests that we may have a classic "blind men and the ele- of the informal, permitting us to document the great phant problem"--everybody touches part of the animal, heterogeneity of actors and their razón de ser. Among them but understands only the part that they touch. More likely we find the following: still, we are exploring several distinct phenomena as we attempt to describe one ungainly composite "informality." Labor: To further complicate things, Keith Hart, the purported · workers, particularly the old and young, who would coiner of the term informality (and one who did not think prefer a job with standard labor protections, but are the sector was necessarily bad), argued that the source of unable to get one; our blindness--the undocumented nature of the sector-- · workers who have quit formal sector jobs to start a left the sector especially vulnerable to being a tabula rasa on microbusiness to be their own boss, make more money, which analysts projected their particular concerns: and avoid paying social protection taxes; and women leaving formal salaried jobs for the flexibility of bal- Most enterprises run with some measure of bureaucracy ancing home and income-raising responsibilities. are amenable to enumeration by surveys, and, as such, constitute the "modern sector" of the economy. The Microfirms: remainder--that is, those who escape enumeration--are · microentrepreneurs with no intention of or potential variously classified as the "low-productivity urban sector," for growing, and hence no intention of engaging the the "reserve army of the underemployed and unemployed," institutions of civil society; 21 I N F O R M A L I T Y · microentrepreneurs stymied in their expansion by Drag on productivity and growth excessively high barriers to registering with the gov- Rigidities in either the labor or product market that prevent ernment and thereby accessing other inputs offered the optimal allocation of workers among sectors generally by the informal sector. lead to output and welfare losses. Regulatory failures that lead to higher informality may have a direct impact on pro- Firms: ductivity. But beyond this, informality itself has been pos- · firms and individuals avoiding taxation or other tulated to have adverse impacts on productivity. As noted in mandated regulations because everybody else does, Poverty Reduction and Growth: Virtuous and Vicious Circles and because enforcement is weak and uneven; (Perry et al. 2006), workers uninsured against health, old · firms registering only part of their workers and part of age, and other risks may have lower productivity and fewer their sales--or declaring only part of the salary of incentives to invest in human capital accumulation. Firms their workers--due to an excessive regulatory burden. unable to access credit, larger sales/product markets, and Though far from exhaustive, these three pairs are illus- sources of innovation, and those evading taxes may operate trative of the variety of types of agents captured under the at a suboptimal scale. Competition with noncomplying rubric of "informality" and, further, capture three different firms leads to productivity losses at formal firms. At the margins of informality discussed later. They also suggest aggregate level, a large concentration of workers in small the reasons we care about informality. firms rather than larger firms may lead to lower productivity growth. Why do we care about informality? Each example above has a different underlying logic and Erosion of the functioning and legitimacy reason for being and, hence, a reason that we may care of market- and equity-enhancing institutions about its existence or size from a policy perspective. Noncompliance with tax collection and market-supporting regulation erodes the rule of law and the integrity of public Unprotected families institutions, and limits society's ability to address collective As the regional flagship report Securing Our Future (de needs that range from infrastructure to the mitigation of Ferranti et al. 2000) noted, while development is often seen inequality. Noncompliance may become a social norm that as a process of increasing income, in practice we have also increases the costs of enforcing the law, undermines the seen the emergence of institutions to shelter families from legitimacy of societal institutions, and creates horizontal adverse shocks, be they loss of job, illness, or natural and vertical inequities (with better-off insiders and worse- calamity. The presence of a large fraction of the workforce off outsiders). This said, compliance with legal norms may in Latin America that does not count on formal mecha- be endogenous to the perceptions of the current effectiveness nisms to hedge or mitigate these shocks is, hence, of intrin- of public institutions and, more profoundly, to the nature of sic concern. What complicates policy making is that, as the the underlying "social contract." chapters in this volume will show, workers often choose jobs that lack such benefits or they willingly leave jobs that Indicators of other problems offered such benefits, valuing more other characteristics of A sizable body of literature sees informality as arising from informal jobs. In this case, the worker and his/her family poor regulation or other government failures. To the degree must be at least as well-off as before, but may still be vul- that this is the case, unusually high or increasing informal- nerable to some types of misfortune (in particular, health) ity may be suggestive of poor policy regimes. for which informal protections are few. There also may be externalities for society in, for instance, the classic case of Informality and the relationship between families undersaving for retirement. Further, if, as is sug- the individual and the state gested throughout this report, some informality is due to Implicit in each of the examples above is a relationship the low valuation of government-provided services com- between the individual or firm and the state. Economic pared with their implicit or explicit costs to workers, then theory posits a legitimate role of the state in a number of a choice to be unprotected may point to a dysfunctional and areas. The state redresses coordination failures in the provi- inefficient social protection system. sion of public goods (for example, roads, defense, public 22 T H E I N F O R M A L S E C T O R : W H AT I S I T, W H Y D O W E C A R E , A N D H O W D O W E M E A S U R E I T ? security) and in the prevention of social bads (such as on lack of compliance with revenue-raising norms, illegal pollution). Further, it fills in missing markets--establish- when the sector engages in unsanctioned activities, unfairly ing courts, property rights, risk-pooling mechanisms--and competitive by those focusing on how industrial structure is sets the rules in the modern economy. Finally, it concerns affected by such evasion (Capp, Elstrodt, and Jones 2005), itself with distributional issues and power asymmetries-- unprotected by those thinking about why workers in least- redistributing from rich to poor, ensuring that labor­ developed countries (LDCs) are not covered by labor legis- capital relationships are not too one-sided or that no firm or lation (International Labour Organization [ILO]), and group of firms gains too much economic power, and that all subcontracted by those concerned with the potentially citizens receive equal treatment regarding the provision of exploitative dynamics of globalization (Castells and Portes key services. To redress the market failures identified above, 1989). Each of those descriptors can be seen as evasion, the state necessarily requires the power to monitor and broadly construed, of the state's legitimate and efficiently coerce agents to do things that, privately, they would not executed brief.2 Opportunistic evasion is, of course, the pri- do. This view of the state has led to seeing the informal mal form of "opting out," despite the fact that "voice" sector through a lens emphasizing lack of compliance with through the political system may be perfectly adequate. legal norms. Though this is not the only lens (Hart defines Many of the cases above have an exclusionary comple- informality simply as "undocumented," an important dimen- ment. Those firms avoiding labor legislation, for instance, sion taken up in chapter 8), it enjoys currency, particularly may be implicitly creating a dual labor market where their in the economics field, and will be a central organizing employees would prefer to enjoy the full benefits of the theme of this report. social protection system, but find themselves, for at least However, even that fairly narrow definition raises the the present, in inferior jobs. question of why agents are not in compliance with state norms. Among the many lenses through which this ques- Defensive evasion and exclusion: Coping with the tion has been viewed, one of the most influential lenses on imperfect, captured, or informal state the labor and firm side has focused on their exclusion from However, as a large body of literature has documented, the critical state benefits and, concomitantly, the circuits of the state often deviates from the economists' ideal. Simply put, modern economy. However, this report highlights a second the state does its job badly--ranging from poor regulation lens through which to view informality that is more akin to to oppressive or exclusionary measures, forcing agents, who Hirschman's (1970) "exit": many workers, firms, and fami- perhaps are inclined toward compliance under the ideal lies choose their optimal levels of engagement with the state, to cope by defensive evasion. De Soto (1989), mandates and institutions of the state on the basis of their Djankov et al. (2002), Friedman et al. (2000), Loayza, valuation of the net benefits associated with formality and Servén, and Oviedo (2005), and Schneider (2005), among the enforcement effort and capability of the state. That is, many others, have stressed the very high registration costs, they make implicit cost­benefit analyses about whether to the regulatory burden to becoming formal, as well as the cross the relevant margin into formality and frequently high ongoing costs of fully integrating with the state that decide against it. Under this view, high informality results drive firms to stay off the state's radar. from a massive opting out of formal institutions by firms The postulated reasons for this state deviation from and individuals, and offers an indictment of the state's reg- the ideal range widely both in view of the nature of the state ulations and services and of its enforcement capability. and in implications for policy. A large body of literature As a starting point, it is useful to sketch three types of stresses that the bureaucracy may be populated by rent seek- relationships between the individual and the state and, ers and, in principle, defensive evasion in this case could be more generally, the institutions of civil society, that capture largely alleviated by regulatory reform. However, in more these two dimensions. extreme views (Acemoglu, Johnson, and Robinson 2001; North, Wallis, and Weingast 2005), the state is behaving in Opportunistic evasion a deliberately and coherently exclusionary manner, manifest- In the case closest to economic theory's vision of the state, ing an underlying stable political-economy equilibrium the informal sector is seen as evading legal norms that give where incumbent business and labor elites defend their rise to additional adjectives: tax-evading by those focusing rents and will find ways to offset and nullify any tinkering 23 I N F O R M A L I T Y with the costs of doing business. In this spirit, the informal design of its policies and, hence, substantially reducing the firm, as depicted by de Soto (1989), is excluded from the distortions that both exclude and encourage exit. benefits of the state and hobbled in its participation in the market economy; the informal worker is excluded Passive evasion and state irrelevance from the benefits enjoyed by a privileged caste of workers. This discussion of the limits of the Latin American state Further, in the absence of a major shock to the political- brings us back to Hart's (1973) emphasis on how multiple economy equilibrium, they are permanently so. institutional systems coexist within a polity and that Other, more generous views, (for example, Centeno and the state is only one candidate among many. This may par- Portes 2003) see weak Latin American states as assigning ticularly be the case for very rudimentary microenterprises themselves an unmanageable--and usually unenforceable-- that may not consider themselves part of the modern load of regulatory measures. That is, what we see is less a economy/social order, and whose production requires little conspiracy to exclude than overwhelmed and poorly coor- in the way of services from the largely irrelevant state. dinated bureaucracies. However, the exclusionary views do Such firms are described colorfully in Geertz's (1963) semi- touch on a leitmotif in the political science literature that nal Peddlers and Princes, which traces the social evolution stresses the informality of Latin American political systems. from the bazaar economy in Indonesia to that of the more In particular, this literature studies the divergence rational, modern "firm." The premodern or bazaar economy between the formal structures of democracy and the econ- encompasses a vast number of proto-firms that are not con- omists' ideal bureaucracy, on one hand, and how gover- strained by access to the benefits that normally are associ- nance is really done, on the other. O'Donnell (1996) argues ated with formality but, as Hart stresses, operate within that often behind formal elections and alternation of subsystems of institutions that coexist with, substitute for, power lies particularistic access to the state with roots in or compete with the "formal" state institutions. In fact, century-old traditions of patron­client relations. Everyone what is striking in Geertz's description of two Indonesian "understands the model" of particularistic access, distrust towns is the significant discussion of institutions for of the state and its evenhandedness is high, tax morale3 managing credit, risk, and collective issues of irrigation, and the general feeling of social reciprocity are low. but very little about the state. Further, as chapter 7 will discuss, the state is perceived as Here we find a conceptual kinship with the literature on providing little: relatively few citizens are covered by what social capital that deals with "informal" relations of trust, has been called the truncated welfare state--low quality and reciprocity, and the like that exist in the absence of formal coverage of public social services, such as health care or institutions (see Alesina and La Ferrara 2000; Glaeser, education, further erode tax morale and prompt opting Laibson, and Sacerdote 2000; Greif 1993, 1998; and Stiglitz out of the system of taxes and transfers. Hence, a Latin 2000, among others). While, generally, such relations are American citizen weighing working with a state that considered positively--that is, as ways of solving problems diverges substantially from the ideal, or employing other of contracting and market failure--they may, in addition, be "nonformal" ways of solving social problems and market preferable in some dimensions to formal institutions that failures, may not perceive the informal-formal dichotomy may eventually displace them. Local institutions are likely as quite so sharp. to be more closely tailored to the relevant market failure and At a meta level, it may be argued that the underperfor- less subject to moral hazard due to closer monitoring by mance of Latin American states along these dimensions family or village. As Bentolila and Ichino (2000) argue, the partly reveals poorly resolved social tensions and manifests informal safety nets in southern Europe cushion families what we might use as conceptual shorthand--a dysfunc- against employment shocks better than the formal unem- tional underlying "social contract." Beyond high informal- ployment insurance schemes of northern Europe. Studying ity, this can also be seen in the inability of the state to financing in Chicago migrant communities, Bond and redress the long-standing high inequality, in the weak rule Townsend (1996) conclude, "We are inclined to view the of law, or in the recurrent bouts of macroeconomic instabil- small role played by the formal sector as stemming, at least ity. Difficult as these phenomena have been to manage in in part, from community disinterest as opposed to formal the region, the report is generally optimistic about the sector negligence" (p. 24) due to the insufficient flexibility of possibilities of improving the quality of the state and the formal institutions. With some caution, they suggest that 24 T H E I N F O R M A L S E C T O R : W H AT I S I T, W H Y D O W E C A R E , A N D H O W D O W E M E A S U R E I T ? "formal sector institutions attempt to create more flexible FIGURE 1.1 financial instruments by either using or mimicking existing Margins of informality informal and semi-formal structures" (p. 24). In sum, infor- mal institutions cannot be ruled out ex ante as suboptimal, Share of labor force (percent) 35 given the type of enterprises operating within them and the Intersectoral margin (labor) level of development of the state. 30 The demand for formal institutions increases with the 25 sophistication of the firm and, more generally, of society. Microfirms Modern firms 20 Geertz's peddlers become organized firms whose growth will require access to an increasingly sophisticated set of 15 Intersectoral margin (firms) de Soto socially provided inputs. In line with de Soto's anecdote of 10 the Peruvian street vendors who sought to pay taxes so they Intrafirm margin 5 would be granted de facto property rights over their pitches, participation in formal institutions can be seen as a 0 "normal" input, increasing with firm size or sophistication 1 2 3 4 5 9 10 25 50 100 250 500 1,000 (see Levenson and Maloney 1998). This is entirely consis- Firm size (number of workers) tent with the logic, postulated in the social capital litera- Formal Informal ture, that individuals optimize their investment in informal networks with a view toward long-run returns, except that Source: Maloney 2006. here the "networks" include formal institutions. At the economywide level, the reach and density of for- and Maloney 2001; Fields 1990; and Henley, Arabsheibani, mal institutions are almost certainly endogenous to the and Carneiro 2006). But policy demands that we identify the complexity of the society, and, as Stiglitz (2000) suggests, most germane interactions, and this, in turn, requires identi- we may expect a greater need and density for formal institu- fying the critical margins along which individuals and firms tions with development--a natural evolution from informal are making calculations about or facing constraints to becom- to more formal institutions.4 At each point along the course ing formal. Maloney (2006) sketches three such margins and of this process, the state makes its own cost­benefit analysis suggest what types of agent­state interactions are at play (fig- on what size firm is worth monitoring and taxing to finance ure 1.1): (1) the intrafirm margin where firms are partly formal its mandate, and what size is not worth doing so, thus leav- and partly not, (2) the intersectoral margin between informal ing the institutional space free for the kind of institutions and formal firms, and (3) the intersectoral margin of formal and discussed above. Tendler (2002), in fact, describes a "devil's informal workers operating through the labor market. These deal" in Brazil between the state and the informal sector margins are not exhaustive, nor are they unrelated to each implicitly on the boundaries of this space. More generally, other. However, they do capture much of the relevant activity there is an equilibrium where small firms find nonstate solu- covered in the informality discourse and help us isolate the tions to their needs, and the state occupies itself with firms most relevant areas on which to focus. above its enforcement threshold. In this view, informality is neither cause nor result of underdevelopment, nor is it nec- The intrafirm margin essarily pathological; rather it is a normal phase in the Firms across the size spectrum are often partially informal development process--a lack of formalization of enterprises across several dimensions. Underreporting of sales is com- and the dominance of local institutional systems that com- mon globally. As chapter 6 will show, the investment cli- plement, compete with, or substitute for those of the state. mate surveys conducted by the World Bank find that the It is not so much "exiting" as never really "entering." percentage of sales reported by Brazilian firms ranges between 60 percent in microfirms and 80 percent in very Three margins of informality large firms. Survey and anecdotal evidence from Latin In all likelihood, all these agent­state interactions can be America suggests that medium-size to large firms will found among the phenomena lumped under what is commonly have a substantial share of their operations, clearly a very heterogeneous informal sector (see Cunningham including workers, off the books. In Argentina, roughly 25 I N F O R M A L I T Y 15 percent of workers will receive pay partly "en blanco" on close to the margin of becoming formal? Alternatively put, the books and partly "en negro"­off the books­without the how relevant is de Soto's story in explaining informality? corresponding labor taxes paid by either worker or firm. As Second, how binding are the impediments at this margin? discussed later, firms can often be fully compliant in one Chapters 7 and 8 will take up both these questions. dimension--perhaps paying taxes--and not in others-- registering workers in social security. The intersectoral margin (workers) The labor literature has long focused on the relationship The intersectoral margin (firms) and flows between workers in the formal sector, covered by The growing informal firm is on the border of registering labor legislation, and those in the informal microfirm or complying with the labor or tax laws. To formalize the sector who are not covered. The latter are often considered classic de Soto story, we can follow Lucas (1978) and think the most disadvantaged of the urban labor market, as they of steady-state firm size as determined by its underlying are precarious, often termed subsistence, and thought to be cost structure that reflects the ability of the entrepreneur, the rump end of the global value chain. A large informal among other characteristics. The resulting heterogeneity of sector has also been seen as evidence of a labor market seg- costs gives us the distribution of firms across the size spec- mented by acute formal sector rigidities arising from exces- trum, whether a mom-and-pop store or a Wal-Mart or sively high minimum wages or union bargaining (see, for Mexican Elektra. And, as Jovanovic (1982) argues, entre- example, Esfahani and Salehi-Isfahani 1989; Mazumdar preneurs have a rough idea of what their ability is, but only 1976; and Rauch 1991). upon actually opening the business can they make that However, evidence has been mounting that a sizable estimate precise. Some will find out they are unprofitable share of entrants into the informal sector do so because and quit; others will find their profits surprisingly high they will become better-off. The report will use the term and seek to expand. At this point, the latter group may voluntary to denominate entry yielding higher or equal lev- need the services of the state, or of collateral services that els of welfare. This does not imply that they are not poor require being recognized by the state. or that they are happy--only that this is the better of two In the de Soto (1989) view, the costs of becoming formal options, given their lower human capital and the low pro- are too high and firms are effectively excluded from the ductivity of the economy. Returning to Lucas (1978), we formal realm and forced to remain suboptimally small. may argue that, in fact, at low levels of aggregate produc- Banerji and Jain (2006), following Rauch (1991), argue that tivity, the opportunity cost of becoming self-employed is lack of monitoring below a certain size threshold produces such that more workers with a comparative advantage in size dualism, where small firms take advantage of the wage operating a small business will actually do so. This may differential, in the former case, to produce lower-quality explain an important part of self-employment and goods for the poorer section of the consumption market.5 microentrepreneurship in many countries in the region. However, there are also firms whose underlying produc- The idea that entry into informality occurs for various rea- tivity is so low that they will never demand the services of sons is not new. Hart (1973) never saw the informal sector the state. Going back to Geertz's (1963) study of the bazaar as intrinsically bad, and Fields (1990) noted that there is economy in Indonesia, he describes entrepreneurs who, in an "upper tier" to informal employment that does very fact, lack the organizational skills to function as a modern well. The critical empirical question is, What share of the firm, let alone grow to a large size. Very poorly educated sector corresponds to those who would prefer formal jobs workers, many less than a generation away from subsistence versus those who are as well-off as they would be in the farming, would also, on average, have low ability levels in formal sector? Gregory (1986) and Maloney (1999, 2004) running a firm. Alternatively, women may choose to oper- argued that, for Mexico, the evidence of segmentation is ate as independents to better balance home and income- weak, the majority of the sector is "voluntary," and the earning roles, with no plans to expand (Cunningham unprotected/exploited view of informality seems an inap- 2001). Further, we do not know how unsubstitutable for- propriate lens.6 The next three chapters confirm this find- mal inputs are for less formal ones. ing for Mexico and the Dominican Republic, as well as for Hence, there are two central open questions surrounding the majority of the informal self-employed workforce this margin. First, how many informal firms are actually across the region. They also reveal that, in most countries 26 T H E I N F O R M A L S E C T O R : W H AT I S I T, W H Y D O W E C A R E , A N D H O W D O W E M E A S U R E I T ? BOX 1.1 The ILO definition of informality The three margins discussed in the text are fully consistent five employees). As such, the table below shows that total and useful for analyzing informality, both as the ILO tradi- employment in the informal sector includes self-employed tionally defined it, based on what might be called the "pro- (3); own-account workers, with or without family workers ductivity view" (rows in the table below) that focused on (5); microentrepreneurs (4); and their employees (6). the type of production unit (rows); and the newer focus on Under this definition, understanding the logic of the pro- informal employment defined according to the "social protec- duction would have required focusing most on the second tion" or "legalistic" view by job status (columns). In the for- and third margins--how microfirms become formal and mer definition, the informal sector enterprises are defined as the nature of flows between those people working in such production units operated by single individuals or house- firms and those in the "modern" sector of the economy. The holds that are not constituted as separate legal entities more recent shift to a "legal" definition of informality rec- independent of their owners and in which capital accumu- ognizes that "informal employment" can be found both lation and productivity are low. This includes "family within and outside the small-firm sector. Consequently, units" (those operated by nonprofessional own-account informal employment now includes informal contractual workers with or without contributing family workers) and arrangement in firms that are otherwise formal, (1) and (2), "microenterprises" (productive units with no more than and hence would now include the intrafirm margin. ILO conceptual framework: informal employment Job by status in employment Production Contributing Members of Own-account unit by type Employers family Employees producers' workers workers cooperatives Informal Formal Informal Formal Informal Informal Formal Informal Formal Formal sector 1 2 enterprises Informal sector 3 4 5 6 7 8 enterprisesa Householdsb 9 10 Source: Hussmanns 2004. Note: Cells shaded in dark gray refer to jobs, which, by definition, do not exist in the type of production unit in question. Cells shaded in light gray refer to formal jobs. Unshaded cells represent the various types of informal jobs. Informal employment: cells 1­6 and 8­10. Employment in the informal sector: cells 3­8. Informal employment outside the informal sector: cells 1, 2, 9, and 10. a. As defined by the Fifteenth International Conference of Labour Statisticians 1993 (excluding households employing paid domestic workers). b. Households producing goods exclusively for their own final use and households employing paid domestic workers. of the region, informal salaried workers appear to corre- compatible, with the latter spanning all three margins. Con- spond more to the traditional queuing view. cerns with better enforcement of tax codes or "corruption," Framing the informality "decision" as one occurring more generally, are likely to focus primarily on the first mar- across three distinct margins focuses the diagnostic and pol- gin. The World Bank's Doing Business measures focus pri- icy discussion on the relevant set of individuals and consid- marily on the second de Soto margin, as might those erations, and is general enough to encompass most existing concerned with access to credit and informal microfirm pro- frameworks as well. For instance, the ILO definition of infor- ductivity more generally. Traditional concerns with seg- mality, both traditional and more recent (box 1.1), is fully mented labor markets focus primarily on the third margin. 27 I N F O R M A L I T Y Measuring the informal sector FIGURE 1.2 The previous sections have suggested how difficult it may Methods for measuring the informal sector be to present a picture of informality, even if we had the data to see clearly the component elements--data that, Voluntary surveys almost by definition, we do not have. Which aspect are we Direct methods interested in? Large firms evading taxes? Microfirms that Tax audits will manage through family and community mechanisms and not bother with the state? Workers rationed out of for- Discrepancy between aggregate income mal jobs? Or women quitting jobs with benefits to stay and expenditure with their children and work independently? This hetero- geneity is clearly the manifestation of multiple social and Discrepancy between total labor force and economic phenomena that have given rise to a cacophony of Classes of formal employment methods characterizations and measurement attempts. The next sec- Velocity of tions discuss a subset of these measures, what they may or Indirect circulation methods approach may not be capturing conceptually and in practice, and Monetary Transactions methods how they can foreshadow later discussions of the most approach relevant margins of informality on which policy makers Currency demand approach should focus. Kaufmann- By definition, most economic activities that are classi- Physical input Model Kaliberda method fied as informal are not captured by national accounts and (electricity approach consumption) official statistics. One exception is informal employment, Lacko's method which can typically be measured or proxied using questions Source: García-Verdú 2007. from household survey data on affiliation to social security, the mandated benefits workers receive, or the size of the firms they work for (in terms of the number of employees), difference between the income declared in tax returns and or using a combination of those variables. the income actually found after an audit. A potential prob- There are several methods that can be used to obtain lem in extrapolating to the national economy is that audits estimates of the magnitude of the informal sector. These are usually nonrandom and, hence, may not be representa- methods have been described in detail, and their strengths tive. In both cases, the applications of these methods have and weaknesses have been discussed extensively.7 These can been limited to a few developed countries because of the be separated into three classes: (1) direct methods, (2) indi- paucity of the available data. rect methods or "indicator" approaches, and (3) the model For Latin America, we have three principal sources of approach (figure 1.2). data. One is the newly collected investment climate assess- ment that asks firms about the level of underreporting of Direct approaches to measurement income and workers. The World Bank Doing Business The direct methods are microeconomic in nature and use indicators use an analogous definition in their compila- either voluntary survey data or the results from tax audits tions. Figure 1.3 presents the unweighted average of these to construct estimates of total economic activity and its responses across firms in Latin America and the Caribbean official and unofficial (or measured and unmeasured) com- and suggests a wide range of noncompliance, from less than ponents. Voluntary surveys typically ask respondents to 5 percent in Chile (similar to Organisation for Economic declare or reveal their incomes, labor status, or impressions Co-operation and Development [OECD] levels) to over of levels of tax compliance in their industry. This method 40 percent in Panama. has been criticized for its sensitivity to how the questions Household and labor surveys provide the second and far are posed, and its confidence in the respondents' willing- more extensive source of direct data. The issue is finding ness to truthfully reveal their income. Tax audit-based mea- the right definition of informality. As noted in box 1.1, the sures define the magnitude of the informal economy as the ILO has traditionally employed what might be called a 28 FIGURE 1.3 Selected measures of informality a. Percent of sales not reported1 b. Percent informal labor force (legalistic definition)2 Panama Paraguay Nicaragua Bolivia Brazil Peru Honduras Mexico Nicaragua Costa Rica Ecuador El Salvador Guatemala Colombia Mexico Guatemala Colombia Latin America and the Caribbean Ecuador Latin America and the Caribbean Paraguay El Salvador Bolivia Argentina Uruguay Venezuela, R. B. de Argentina Brazil Jamaica Peru Uruguay Chile Chile 0 10 20 30 40 50 10 20 30 40 50 60 70 80 c. Percent informal labor force: Productive definition (% employed)3 d. Self-employment (% of labor force)4 Bolivia Dominican Republic Paraguay Bolivia Colombia Venezuela, R. B. de Peru Honduras Guatemala Paraguay Ecuador Peru Nicaragua Colombia Honduras Guatemala Latin America and the Caribbean Jamaica Jamaica Ecuador El Salvador Latin America and the Caribbean Brazil Mexico El Salvador Venezuela, R. B. de Panama Dominican Republic Brazil Panama Chile Argentina Mexico Uruguay Costa Rica Costa Rica Uruguay Chile Argentina 0 20 40 60 80 100 10 15 20 25 30 35 40 45 50 e. Lack of pensions (% of labor force)5 f. Shadow economy (% of GDP)6 Bolivia Bolivia Paraguay Panama Ecuador Peru Nicaragua Guatemala Guatemala Uruguay Honduras Honduras El Salvador Nicaragua Mexico Latin America and the Caribbean Dominican Republic Brazil Venezuela, R. B. de Colombia Latin America and the Caribbean Argentina Jamaica Brazil Ecuador Colombia Venezuela, R. B. de Jamaica Dominican Republic Uruguay Mexico Panama Costa Rica Chile Argentina Costa Rica Chile 20 30 40 50 60 70 80 90 100 0 10 20 30 40 50 60 70 80 Sources: Gasparini and Tornarolli 2006; investment climate surveys 2006; Loayza and Rigolini 2006; Schneider 2005; World Bank 2006b. Note: 1. Informality is measured by the percentage of sales that businesses do not report for tax purposes (Investment Climate Surveys 2006). 2. "A salaried worker is informal if s(he) does not have the right to a pension linked to employment when retired" (Gasparini and Tornarolli 2006, 10). 3. "An individual is considered an informal worker if (s)he belongs to any of the following categories: (i) unskilled self-employed, (ii) salaried worker in a small private firm, (iii) zero-income worker" (Gasparini and Tornarolli 2006, 8). 4. "Self-employment is measured as the percentage of self-employed workers with respect to the total active population" (Loayza and Rigolini 2006, 15). 5. Share of the labor force not covered by a pension scheme (World Bank 2006b). 6. "The shadow economy includes all market-based legal production of goods and services that are deliberately concealed from public authorities for the following reasons: (1) to avoid payment of income, value added or other taxes, (2) to avoid payment of social security contributions, (3) to avoid having to meet certain legal labor market standards, such as minimum wages, maximum working hours, safety standards, etc., and (4) to avoid complying with certain administrative procedures, such as completing statistical questionnaires or other administrative forms" (Schneider 2005, 600). In all cases, regional figures are unweighted averages. 29 I N F O R M A L I T Y "productive" definition, focused more on informal firms. security but are not. The data suggest that evasion in larger Since these data are available on a global basis, they are firms is a relatively minor issue in Uruguay, for instance-- commonly used in empirical work.8 The weighted average around 10 percent--and quite large in Ecuador, Nicaragua, of 28 percent of the Latin American and Caribbean labor and Peru--more than 30 percent. Column (i)/(i) (ii) cap- force is below those of Africa and South Asia, but above tures a related measure--the percentage of workers classi- those of the OECD and Eastern Europe. Along these lines, fied as formal in the productive definition and as formal in the report tabulates a measure calculated by Gasparini and the legalistic definition. The fit ranges from poor in Bolivia, Tornarolli (2006) that is broadly consistent with the ILO Nicaragua, Paraguay, or Peru (at rates of 30­40 percent) to measure for Latin America (see the annex) but adds in paid reasonably good in Brazil, Chile, Uruguay, and Colombia. workers in those microfirms. For the purposes of this The third column of the next panel suggests that workers report, we will refer to the ILO definition as "self- classified as informal in small firms are, in fact, generally employment" and to the Gasparini-Tornarolli definition as informal in the legalistic definition as well, with rates of the "productive" definition. overlap often greater than 90 percent. A second definition, called by Saavedra and Chong Several points merit mention here. First, the substantial (1999) the "legalistic" or "social protection" definition, mismatch of classification of formal workers in the first focuses more on coverage of workers by mandated labor panel, as captured in column (ii), reflects the different ques- protections. It thus is more concerned with workers' tions underlying the measures; the ILO focus on small welfare per se (or perhaps with job quality) than with the firms in the productivity definition, by design, cannot cap- nature of their employment; and when not including self- ture evasion or coverage in large firms, while the legalistic employed or owners, who are often not required to register definition is more informative in this respect. While, in the with social security administrations, it captures compliance aggregate, the two measures are highly correlated, for par- with labor laws. This is more consistent with the ILO's ticular questions the definition matters. For example, in 7 (2002) more recent emphasis in its "Decent Work" report out of 12 countries in figure 1.4, the relative representation on noncompliance by either enterprises or workers with of women versus men in the informal sector depends on the all or some of the rules and regulations in the body of particular measure used. national or local legislation, commercial, and/or labor leg- Second, informality measured along the labor dimen- islation. This new focus implies expanding the definition to sions in many countries is largely a small-firm phenome- include informal contractual arrangements among other- non. In most countries, the share of uncovered workers in wise formal entities (see box 1.1). Hence, this definition firms with more than 10 workers is a minority; and, as puts a greater emphasis on the division between informal figure 1.5 shows for Argentina, Brazil, and Mexico, in firms salaried workers in any size firm and the informal self- of more than 10 workers, the share of workers plausibly not employed; and, in fact, the report will show that there are covered is small. Though there has been increasing infor- substantial differences in the behavior of these two classes malization of the large-firm labor force in Argentina and of workers that make the bifurcation of informal workers metropolitan Brazil (and the reverse in Mexico), the rele- critical for analysis. Global data do not exist for this vant margins for understanding the razón de ser of the infor- measure, but, again, Gasparini and Tornarolli (2006) have mal laborer seem more along the margin of small firms calculated consistent series for Latin America, and other growing into formality, and the intrasectoral margin of authors providing background papers for this report use worker flows among the formal and informal sectors. some close variant. Hence, along this dimension of formality--that is, compli- Overall, both the productive and the legalistic measures ance with labor laws--the intrafirm margin with large give broadly similar measures of the level of informality in firms in mind, while still employing an important share of Latin America and the Caribbean. However, the individuals informal workers in some countries, does not seem to be under each measure may differ substantially.9 Column (ii) of where the majority of the action is, overall. Understanding table 1.1 tabulates the share of the labor force that, when the decisions that employers and workers in small firms classified as formal under the productive definition, is infor- make on whether to register with the authorities becomes a mal under the legalistic definition. In practice, it is a mea- central question to be taken up in later chapters. sure of what fraction of workers in firms with more than five A final direct measure capturing social protection is an employees probably legally should be covered by social index of pension coverage of the population that considers 30 T H E I N F O R M A L S E C T O R : W H AT I S I T, W H Y D O W E C A R E , A N D H O W D O W E M E A S U R E I T ? TABLE 1.1 Correspondence of the "productive" and "legalistic" definitions of informality Formal productive Informal productive Formal L Informal L Formal P & L Formal L Informal L Informal P & L Total Sample (i) (ii) (i)/(i)+(ii) (iii) (iv) (iv)/(iii)+(iv) (i)+(ii)+(iii)+(iv) (i)+(iv) Argentina 2004 Only salaried workers 50.6 15.6 76.5 6.1 27.8 82.1 100.0 78.4 Bolivia 2002 Only salaried workers 24.6 35.5 40.9 1.1 38.9 97.3 100.0 63.5 All workers 7.6 15.5 33.0 0.8 76.1 98.9 100.0 83.7 Brazil 2003 Only salaried workers 53.3 10.6 83.4 11.8 24.2 67.2 100.0 77.6 All workers 36.2 8.8 80.4 10.2 44.8 81.5 100.0 81.0 Chile 2003 Only salaried workers 67.0 11.6 85.3 10.7 10.8 50.1 100.0 77.7 All workers 51.8 11.4 82.0 11.2 25.6 69.5 100.0 77.4 Colombia 1999 Only salaried workers 86.0 14.0 86.0 100.0 86.0 All workers 13.6 10.6 56.1 2.8 73.0 96.3 100.0 86.6 Ecuador 1998 Only salaried workers 36.9 32.4 53.2 2.6 28.1 91.6 100.0 65.0 El Salvador 2003 Only salaried workers 49.9 20.8 70.6 1.9 27.4 93.6 100.0 77.3 All workers 28.8 16.2 64.0 1.5 53.4 97.2 100.0 82.2 Guatemala 2002 Only salaried workers 37.8 24.1 61.0 2.3 35.7 93.9 100.0 73.5 All workers 15.4 15.1 50.5 1.0 68.5 98.5 100.0 83.9 Mexico 2002 Only salaried workers 37.6 25.7 59.4 3.4 33.2 90.7 100.0 70.9 Nicaragua 2001 Only salaried workers 29.5 30.7 49.0 2.3 37.5 94.1 100.0 67.0 All workers 14.9 20.4 42.2 1.5 63.3 97.7 100.0 78.1 Paraguay 2003 Only salaried workers 23.6 27.4 46.2 2.1 47.0 95.8 100.0 70.6 All workers 10.5 17.1 38.1 1.4 71.0 98.1 100.0 81.5 Peru 2002 Only salaried workers 26.6 36.1 42.4 1.4 35.9 96.2 100.0 62.5 All workers 11.4 21.6 34.6 1.4 65.5 97.9 100.0 77.0 Uruguay 2004 Only salaried workers 64.1 9.9 86.7 8.3 17.7 68.0 100.0 81.8 All workers 49.3 8.4 85.5 10.0 32.3 76.3 100.0 81.6 Venezuela, 2003 Only salaried workers 53.4 19.2 73.6 5.0 22.4 81.8 100.0 75.9 R. B. de Source: Gasparini and Tornarolli 2006. Note: Table shows what fraction of workers categorized as informal under the Òpro ductive" definition are also informal under the "legalistic" definition, and analogously for the formal. Formal L = formal using legalistic definition, informal L = informal using legalistic definition, and formal P&L = formal under both definitions. both labor-related and universal pension schemes. This FIGURE 1.4 measure raises an important issue. If our concern is that Informality and gender: women versus men (ages 25­64) families are covered by certain protections, it need not be % informal women % informal men the case that these protections are linked to the particular 1.5 labor contract. Throughout the region, there has been an 1.3 expansion of social protection programs that aim to provide a minimum safety net for families, regardless of labor mar- 1.1 ket status. These are not captured in the pension measure, 0.9 and cross-country comparisons are not yet available. Finally, all these stock measures of informality obscure 0.7 the fundamentally dynamic nature of the labor market. 0.5 Chapters 2 and 4 document high flows between jobs with and without formal pension programs. Not only does this Peru . de BoliviaBrazilChile Mexico aguay Salvador UruguayR.B Argentina suggest a need to understand workers' choices among sec- El Guatemala Nicaragua Par tors and the constraints they face in making them, but also Venezuela, that the individuals in the informal population change sub- Productive Legalistic stantially from one month to the next. Such high flows have implications for what coverage really means. As chap- Source: Gasparini and Tornarolli 2006. ter 7 will show, in Mexico and Uruguay, poor workers flow 31 I N F O R M A L I T Y Mexico, there is little evidence that either informal salaried FIGURE 1.5 or self-employed workers are less well-off than comparable Informal salaried workers (legal definition) across firm size and time workers in protected jobs, why should policy makers care about the particular bundle of money and benefits in which Argentina (greater Buenos Aires) those workers are paid? The answer is complex but under- % of informal and self-employed workers scores the importance of understanding decisions across the 50 1980 45 margin of worker flows between the two sectors for the 40 35 interpretation of informality measures. 30 2003 25 20 The multidimensional continuum of informality 15 10 As suggested above, formality is multidimensional and 5 0 continuous. At the microfirm level, there is a substantial 1 2­5 6­25 26­100 101­500 501 gray area (see Tokman 1992) where firms comply with cer- Firm size (number of workers) tain norms and not necessarily to the same degree. Gray Brazil (metropolitan areas) areas prevail and, in the extreme, one can argue that a % of informal and self-employed workers realistic--but one that is less easy to operationalize-- 50 description is that of a continuum from complete lack of 45 40 1990 integration with formal institutions to full compliance 35 30 with all. 25 Compliance along any one particular dimension of for- 20 2002 15 mality is not discrete. For example, Robles et al. (2001) 10 5 find that 16 percent of microfirms in Peru do not pay any 0 tax, 83 percent pay some taxes, and 2 percent pay all taxes 1 2­5 6­10 11 Firm size (number of workers) that they are required to pay. Figure 1.6 suggests that, in Mexico, firms of, say, seven workers may well have one Mexico (urban areas) worker registered but not all. Further, the registered­not % of informal and self-employed workers registered dichotomy does not capture the gradations of 50 45 protection that exist across labor contracts. Though not for- 2004 40 35 mally covered, workers nonetheless appear to benefit from 30 some social norms of fairness. Souza and Baltar (1979) 25 1994 20 introduced the concept of the efeito farol, whereby the min- 15 10 imum wage ends up being an indexation mechanism valid 5 also for the informal sector. In fact, Maloney and Nuñez 0 1 2­5 6­10 11­15 16­50 51­100 101­250 250 (2001) and Cunningham (2007) show that, in many coun- Firm size (number of workers) tries of Latin America, the minimum wage is most binding Sources: Encuesta Nacional de Empleo Urbano 1994, 2004; Encuesta among the informal salaried, suggesting that salary norms Permanente de Hogares 1980, 2003; Pesquisa Nacional por Amostra are respected outside the realm of the official work contract. de Domicilios 1990, 2002. Chavez and Chacaltana (1994) show, in a study of microen- terprises in Peru, that a large percentage of workers with- so frequently in and out of covered jobs that, in practice, out a formal contract and without social protection benefits they will never accumulate enough years to gain a pension. do enjoy vacations and the customary December bonus They pay, but are de facto not covered. Further, such salary, much as dictated by labor legislation. Among larger flows also raise the question (discussed at length in chap- firms, such norms may carry over to workers hired illegally ters 2 and 3) of what a worker's choice to be uncovered (bajo la mesa) or subcontracted out, but on the same implies about the social protection and, to a lesser extent, premises. In Argentina, however, there is an almost one-to- pension definitions as measures of worker welfare. That is, one correspondence between pension coverage and other if, as appears to be the case in the Dominican Republic and labor benefits, suggesting substantial variance by country. 32 T H E I N F O R M A L S E C T O R : W H AT I S I T, W H Y D O W E C A R E , A N D H O W D O W E M E A S U R E I T ? FIGURE 1.6 FIGURE 1.7 Distribution of manufacturing sector firms according to worker Distribution of manufacturing sector firms according to tax registration in Mexico compliance in Mexico % of firms % of firms 100 100 90 80 70 90 60 80 70 50 60 40 50 30 40 20 30 10 20 0 10 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 0 Firm size (number of workers) 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 Firm size (number of workers) No workers registered Some workers registered All workers registered Tax compliant Not tax compliant Source: Encuesta Nacional de Micronegocios 1992­2002. Source: Encuesta Nacional de Micronegocios 1992­2002. The point remains, however, that many of those workers across the de Soto margin is likely to be somewhat different who are unprotected in large firms (identified in table 1.1) from that of larger firms partially evading obligations. may, in fact, enjoy other elements of the standard labor contract.10 Indirect and modeling approaches to estimating But it is also the case that the firms may be formal along aggregate informality one dimension, but not along another. Levenson and Maloney Indirect methods are macroeconomic in nature, and com- (1998), taking a broad view of formality as integration with bine various aggregate economic variables and a set of not only government but also civil society more generally, assumptions to produce estimates of total economic activ- find for Mexico that firms may pay taxes, but only pay labor ity (that is, measured and unmeasured, official and unoffi- taxes as they get larger, and engage with business associa- cial). Box 1.2 outlines several popular methods--in tions only when they grow larger still. Figure 1.6, combined particular, those based on unexplained components of with figure 1.7, suggests that while most firms with a labor money demand or electricity consumption--and some force of five workers are registered with tax authorities, com- of their drawbacks. By far the most common method is that pliance with labor law is far less complete. This is consistent of the Multiple Indicator­Multiple Cause (MIMIC) Model with the findings of Robles et al. (2000) from Peru that the that imputes a level of underlying informality from a set of majority of firms pay the value-added tax, just over half pay presumed causes of informality on one hand, and measur- municipal taxes, 45 percent pay income taxes, and 13 per- able consequences of it on the other. This exercise was first cent pay labor taxes. For larger firms, as later chapters will undertaken by Loayza (1996) for Latin America with a show, there tends to be partial formality that is relatively relatively tight, theoretically motivated set of input and well synchronized across the tax and labor dimension. The outcome variables. However, recently, a more expansive logic here is clearer: larger, more frequently monitored firms estimate of the "shadow" economy as a share of gross will raise suspicions if they report half of their true product, domestic product (GDP) by Schneider and Enste (2000, but all of their workers producing it. Hence, the logic dri- 2002) has been tabulated globally and is presented in ving less easily monitored microfirms operating fuzzily panel (f) of figure 1.3. As box 1.2 suggests, these have been 33 BOX 1.2 Indirect methods of estimating informality One indirect method of estimating informality is to economy are assumed to occur only in cash. Third, it also attribute the discrepancy between aggregate income and assumes a base year in which the magnitude of the infor- expenditure from the National Income and Product mal economy is zero or negligible--again, an unrealistic Accounts, which capture economic activity, to the infor- assumption for most countries. mal sector. For this method to work, it is necessary to Thus, both the physical input and the currency demand have measures of gross domestic product (GDP) obtained methods--two of the most widely used approaches--are independently through the expenditure and the income somewhat arbitrary in the following sense. Depending on approaches. Given that only one independent measure of the assumption made about the base year in which the GDP is typically available for most countries, in practice magnitude of the shadow economy is zero or negligible, the application of this approach has been limited to a few one can obtain widely different estimates of the magnitude developed countries. of the shadow economy.1 Another indirect method commonly employed is the The third group of methods is the model approach. physical input (electricity consumption) approach. This The most popular among these is the Multiple Indicator­ method assumes that electricity consumption is the Multiple Cause (MIMIC) or structural equation model.2 "single best physical indicator of overall [official and The MIMIC approach postulates that magnitude of the unofficial] economic activity" (p. 27). The method then unofficial economy can be modeled as a latent or index defines the growth rate of the shadow economy as "the variable. While this variable is unobservable, its causes difference between the growth of official or measured (for example, an increase in the tax burden) and effects GDP and the growth rate of electricity consumption" (such as an increase in the demand for currency) can be (p. 28). This method has been criticized on several observed directly. grounds, all related to the assumption of a constant coeffi- A system of equations forms the basis of this model: cient of use per unit of GDP. First, it does not consider one set models the effects (or indicators) as a function of technological progress, which reduces the amount of elec- the latent variable; the other group models the magni- tricity consumption per unit of output. Second, it needs tude of unofficial economy as a function of the causal to assume a base year in which the magnitude of the variables. The parameters in this system of equations are informal economy is zero or negligible--an unrealistic estimated simultaneously, typically using maximum assumption for most countries. Third, it does not consider likelihood. The fitted values of the latent or index vari- the incorporation over time of new households to the elec- able obtained from the reduced form equation are then tric grid, a fact that explains a large fraction of the increase used to produce an estimate of the unofficial economy. in electricity consumption in developing countries. The model approach has been criticized (see Breusch A third indirect method that has also been commonly 2005) since it has been shown that its results are sensitive employed is the currency demand approach. This method to transformations of the data, to the units of measurement, begins by estimating a form of money demand equation and to the sample used. Another criticism is that no theory in which the dependent variable is the ratio of cash hold- is used in order to determine which variables to include as ings to current and deposit accounts (M0/M2). The equa- indicators or as causes. Moreover, the shadow estimates, tion controls for most known determinants of money while relying on the MIMIC Model to generate trends over demand; it also includes as covariates variables that are time, appear to rely on traditional currency demand or the thought to be determinants of the shadow economy (for physical input methods for the initial levels, which makes example, the tax burden). It then defines the growth rate it vulnerable to the criticisms of these two methods. of the shadow economy as the difference between the fit- Notes ted values obtained using the estimated model and the 1. See Thomas (1993, 1999) for a more detailed description observed values from actual data. Just like in the case of and criticism of these two methods. the physical input approach, this method has been criti- 2. For a detailed description and critique of these models, see cized on several grounds. First, it assumes a common Breusch (2005). velocity of circulation of money between the official and unofficial economies. Second, transactions in the shadow Source: García-Verdú 2007. 34 T H E I N F O R M A L S E C T O R : W H AT I S I T, W H Y D O W E C A R E , A N D H O W D O W E M E A S U R E I T ? BOX 1.3 Schneider and Enste in the New World: Checking MIMIC estimates against Mexican data Mexico is one of the few countries that calculate official Official and unofficial estimates of the informal economy, statistics on the contribution of the informal sector to Mexico 1993­2003 total value added. The National Statistical Institute 13.5 31.0 (INEGI) employs the International Labour Organisa- tion's official definition of the informal sector, and 30.0 imputes its value added using a variety of sources. Over 13.0 29.0 the period 1993­2003, the INEGI calculates that the informal sector averaged 12.4 percent of gross domestic 12.5 28.0 product (GDP)--a share that remained relatively con- stant across the period (figure at right). This conflicts 27.0 with the estimates from Schneider (2005) and Schneider 12.0 and Enste (2000, 2002), which suggest that over the 26.0 periods 1990­91 and 1999­2000, the shadow economy 11.5 25.0 in Mexico increased from 24.1 to 30.1 percent of GDP. 3 4 5 7 8 9 0 2 3 199 199 199 1996 199 199 199 200 2001 200 200 INEGI1 (left) Schneider2 (right) Loayza3 (right) Sources: 1. Cuenta Satélite del Subsector Informal de los Hogares. Sistema de Cuentas Nacionales de México. Instituto Nacional de Estadística, Geografía e informática (INEGI). 2. Schneider (2005), using DYMIMIC and currency-demand methods. 3. Loayza (1996), using MIMIC Method. subject to substantial criticism for, among other reasons, explanations, some of which will be touched on in this vol- their relatively atheoretical combination of different causal ume. Again, since most independent workers in Latin factors and indicators, and the difficulty in assuming that America are not covered by pensions, and many informal informality is the only thing linking the two. Both suggest salaried workers are found working in these very small that the shadow economy runs the risk of being a measure firms, we may expect lack of pension coverage to follow of an agglomeration of known size but unclear content. An GDP closely as well. Globally, the shadow economy measure exercise comparing MIMIC estimates with official esti- is moderately correlated with self-employment and pension mates is presented for Mexico in box 1.3. coverage.11 The limited correlation of the tax compliance measure with both GDP and the other measures is sugges- Correlations among measures and trends tive that, if it is reliable, it may be measuring a different over time phenomenon. As chapter 8 will show, tax evasion or elusion As table 1.2 indicates, globally, the pensions, self- may be a relevant phenomenon across the firm-size spec- employment, and shadow measures show a modest degree of trum and be related to levels of social norms or collective correlation generally of the expected sign. All are impor- responsibility, and less to income levels. Labor informality tantly negatively correlated with GDP. This makes certain measured as lack of compliance with legislation, however, sense since, as figure 1.8 shows, self-employment decreases may fundamentally be a small, low-productivity firm issue sharply with development, from high levels in Latin rather than a compliance issue per se. America and the Caribbean region (60 percent in Peru) to In Latin America, the productive measure, including near single-digit levels in the OECD. The close connection self-employed and all employees, and the legalistic/social between self-employment and GDP per capita has already protection measure are highly correlated (.8­.9) with been documented by, among others, Blau (1987), Loayza each other and, not surprisingly, with the pension measure. and Rigolini (2006), and Maloney (2001) with a variety of It is perhaps not unexpected that all are somewhat less 35 I N F O R M A L I T Y TABLE 1.2 Correlations across measures of informality All countries Shadow economy Self-employment Sales nonreported Lack of pensions GDP pc PPP 05 Shadow economy 1 Self-employment 0.58 1 Sales nonreported 0.13 0.17 1 Lack of pensions 0.60 0.81 0.43 1 GDP per capita (PPP 05) 0.69 0.76 0.30 0.85 1 Shadow Self- Lack of Informality Informality Latin America economy employment Sales nonreported pensions (productive) (legalistic) GDP pc PPP 05 Shadow economy 1 Self-employment 0.35 1 Sales nonreported 0.29 0.06 1 Lack of pensions 0.43 0.62 0.04 1 Informality (productive) 0.60 0.70 0.11 0.80 1 Informality (legalistic) 0.58 0.68 0.32 0.89 0.90 1 GDP per capita (PPP 05) 0.58 0.75 0.24 0.66 0.83 0.75 1 Sources: Gasparini and Tornarolli 2006; investment climate surveys 2006; Loayza and Rigolini 2006; Schneider 2005; World Bank 2006b. Note: GDP = gross domestic product; pc = per capita; PPP = purchasing power parity; PPP05 = data correspond to year 2005. correlated with the ILO self-employment measure that in the productive measure. As will be documented in chap- omits salaried workers. The underreporting of sales mea- ter 4, this arises from the fact that the increase happened sures is capturing something else and is poorly correlated through increasing informalization of large firms, not the with other measures. Finally, the shadow economy is not emergence of new informal microfirms. Some anomalies-- well correlated with the self-employment or underreport- such as why Ecuador should have the largest increase in lack ing of sales measure or pensions measure, but is moderately of coverage in pensions, but apparently little increase in legal correlated with the productivity and legalistic measures. Its informality--are likely due to the time spans covered. Also poor performance on the self-employment measure may be important to keep in mind are the subnational differences in due to some odd values of the measure. For instance, trends. Though the legal definition for Brazil as a whole Uruguay has relatively low levels of self-employment, but shows little increase, there has been a dramatic increase in is among the highest in the size of its simulated shadow metropolitan informality across the same period, and it will be economy. explored in chapter 4. The shadow measure somewhat strangely shows global increases in informality for all coun- Trends across time tries of the world from 1990 to 2000--increases from 30 to The underreporting of sales is available for only a handful of 36 percent of GDP, an increase of 24 percent--and substan- countries; however, we report the productive, legal/social tial increases in Latin America.12 Perhaps the estimates are a protection, shadow economy, lack of pensions, and ILO self- bit too substantial, showing an average increase over the employment for Latin America and the Caribbean where period 1990­2000 of 7.4 percentage points or approxi- available (figure 1.9). With the exception of the self- mately 21 percent. This is far above every other measure and employed definition, all measures suggest increases across is reasonable only if we assume that productivity in the the available sample period, although there are suggestive informal sector rose substantially more than that in the for- differences among them and even within individual country mal sector across the period.13 Because of these odd results, experiences. For instance, Argentina shows important and the theoretical concerns discussed above, the report does increases in the legal and pension coverage variable, but not not rely significantly on this measure in its analysis. 36 T H E I N F O R M A L S E C T O R : W H AT I S I T, W H Y D O W E C A R E , A N D H O W D O W E M E A S U R E I T ? FIGURE 1.8 Global correlation of measures of informality with GDP Underreporting of taxes (% of sales)1 Self-employment (% of labor force)2 100 100 90 90 80 80 70 70 60 60 50 50 40 40 30 30 20 20 10 10 0 0 0 10,000 20,000 30,000 40,000 0 10,000 20,000 30,000 40,000 2005 GDP per capita, PPP adjusted 2005 GDP per capita, PPP adjusted Lack of pensions (% of labor force)3 Shadow economy (% of GDP)4 100 100 90 90 80 80 70 70 60 60 50 50 40 40 30 30 20 20 10 10 0 0 0 10,000 20,000 30,000 40,000 0 10,000 20,000 30,000 40,000 2005 GDP per capita, PPP adjusted 2005 GDP per capita, PPP adjusted Latin America Advanced countries Rest of the world Sources: Investment climate surveys 2006; Loayza and Rigolini 2006; Schneider 2005; World Bank 2006b. Note: 1. Informality is measured by the percentage of sales that businesses do not report for tax purposes (Investment Climate Surveys 2006). 2. "Self-employment is measured as the percentage of self-employed workers with respect to the total active population" (Loayza and Rigolini 2006, 15). 3. Share of the labor force not covered by a pension scheme (World Bank 2006b). 4. "The shadow economy includes all market-based legal production of goods and services that are deliberately concealed from public authorities for the following reasons: (1) to avoid payment of income, value added or other taxes, (2) to avoid payment of social security contributions, (3) to avoid having to meet certain legal labor market standards, such as minimum wages, maximum working hours, safety standards, etc., and (4) to avoid complying with certain administrative procedures, such as completing statistical questionnaires or other administrative forms" (Schneider 2005, 600). Nonetheless, the overall perception from the various mea- number of phenomena it encompasses and the limitations of sures and other sources, as well as the general "feeling" of the its measures are manifold, raising some serious doubt about region, is that informality has risen. What will be dealt with how useful the term really is. Yet two stylized facts remain: in the next chapters are some of the forces driving the move- First, however measured, Latin America ranks high in its ments over the last decades and across which margins. degree of informality. Not especially high for its level of devel- opment, but informality remains an important phenomenon. Conclusions Second, several countries have experienced striking increases This chapter has sought to unpack our understanding of the in informality across the last decades. Whatever adverse regu- term informality, why we may care about it, and, broadly latory, poverty, growth, or social morale implications infor- speaking, what dynamics may be driving its elements. The mality may have, they have become more relevant with time. 37 FIGURE 1.9 Trends in informality by various definitions Informal labor force: Productive definition (% of employed workers)1 Informal labor force: Legalistic definition (% of salaried workers)2 Venezuela, R. B. de 1989­2003 18.5 Argentina 1992­2005 11.9 Uruguay 1992­2004 7.0 Honduras 1992­2002 6.8 Venezuela, R. B. de 1995­2003 7.7 Panama 1995­2003 5.4 Brazil (metropolitan) 1992­2002* 6.2 Colombia 1996­2004 4.8 Peru 1997­2003 4.6 Nicaragua 1993­2001 5.9 Jamaica 1996­2002 3.4 Peru (metropolitan Luna) 1990­2000 4.5 Dominican Republic 1996­2004 3.1 ** Paraguay 1997­2003 2.6 Mexico 1990­2004a 4.0 Mexico 1996­2002 2.4 Chile 1990­2003 1.1 El Salvador 1991­2003 2.2 Argentina 1995­2005 1.9 Ecuador 1994­98 0.7 Bolivia 1997­2002 1.5 Brazil 1990­2003 0.8 Ecuador 1994­2003 1.1 Costa Rica 1992­2003 0.4 Paraguay 1997­2003 0.9 Nicaragua 1993­2001 0.9 Colombia 1996­99 2.4 Brazil 1992­2003 3.3 Chile 1990­2003 5.2 El Salvador 1991­2003 12.0 10 5 0 5 10 15 20 15 10 5 0 5 10 15 Percent Percent Lack of pensions (% of labor force)3 Self-employed: ILO (% of employed workers)4 Ecuador 1990­2004 12.75 Colombia 1992­2005 17.3 Argentina 1992­2004 10.26 Bolivia 1990­2000 6.9 Dominican Republic 1991­2005 4.7 Nicaragua 1993­2001 8.08 Chile 1996­2005 0.1 Costa Rica 1990­2004 3.83 Costa Rica 1990­2005 1.4 Venezuela, R. B. de 1995­2004 3.82 Jamaica 1997­2005 2.3 Chile 1990­2003 2.75 Panama 1991­2005 2.4 El Salvador 1995­2004 2.7 Uruguay 1995­2004 2.19 Honduras 1996­2005 3.4 Colombia 1996­99 0.64 Argentina 1996­2005 3.7 Brazil 1992­2002 0.54 Peru 1996­2005 6.9 El Salvador 1995­2003 8.6 4.33 Mexico 1991­2005 10 5 0 5 10 15 20 15 10 5 0 5 10 15 Percent Percent Shadow economy 1990­2000 (% of GDP)5 Peru 12.8 Panama 12.7 Bolivia 11.7 Guatemala 10.1 Uruguay 9.8 Honduras 8.9 Average 7.3 Brazil 7.3 Chile 6.2 Venezuela, R. B. de 6.2 Mexico 6.0 Colombia 5.7 Ecuador 5.5 Nicaragua 5.1 Jamaica 5.0 Costa Rica 4.2 Dominican Republic 3.8 Argentina 3.3 0 2 4 6 8 10 12 14 Percent of GDP Sources: Gasparini and Tornarolli 2006; ILO 2006; Loayza and Rigolini 2006; Schneider 2005; World Bank 2006b. Note: 1. "An individual is considered an informal worker if (s)he belongs to any of the following categories: (i) unskilled self-employed, (ii) salaried worker in a small private firm, (iii) zero-income worker" (Gasparini and Tornarolli 2006, 8). 2. "A salaried worker is informal if s(he) does not have the right to a pension linked to employment when retired" (Gasparini and Tornarolli 2006, 10). 3. Share of the labor force not covered by a pension scheme (World Development Indicators 2006). 4. "Self-employment is measured as the percentage of self-employed workers (employers, own account workers) and contributing family workers with respect to the employed workers" (ILO 2006, 15). 5. "The shadow economy includes all market-based legal production of goods and services that are deliberately concealed from public authorities for the following reasons: (1) to avoid payment of income, value added or other taxes, (2) to avoid payment of social security contributions, (3) to avoid having to meet certain legal labor market standards, such as minimum wages, maximum working hours, safety standards, etc., and (4) to avoid complying with certain administrative procedures, such as completing statistical questionnaires or other administrative forms" (Schneider 2005, 600). *% of workers without carteira (work card). **Based on the balanced panel sample (common municipalities) for the period 1990­2004. a. Author's calculations based on ENEU. 38 T H E I N F O R M A L S E C T O R : W H AT I S I T, W H Y D O W E C A R E , A N D H O W D O W E M E A S U R E I T ? TABLE 1A.1 Comparisons of ILO and Gasparini-Tornarolli measures of self-employment ILO Table: Proxied with Gasparini Data vs. ILO Data Informal sector Independent workers Domestic workers Microfirms Total Country Year Gasparini ILO Gasparini ILO Gasparini ILO Gasparini ILO Difference Argentina 2004 20.7 17.9 -- 7.4 24.6 19.0 45.3 44.3 1.0 Bolivia 2002 47.4 44.6 -- 4.3 18.2 17.8 65.6 66.7 1.1 Brazil 2003 25.4 21.0 -- 9.3 22.2 14.3 47.6 44.6 3.0 Chile 2003 21.3 21.5 -- 6.2 15.7 11.1 37.0 38.8 1.8 Colombia 2004 39.6 37.6 -- 5.8 27.9 16.6 67.5 59.9 7.6 Costa Rica 2003 22.0 18.1 -- 5.3 19.5 20.2 41.5 43.6 2.1 Ecuador 2003 35.1 31.9 -- 5.2 20.6 19.4 55.7 56.5 0.8 El Salvador 2003 36.7 32.1 -- 5.7 20.3 16.4 57.0 54.2 2.8 Honduras 2003 45.4 40.8 -- 4.8 18.3 13.8 63.7 59.4 4.3 Mexico 2002 24.8 19.5 -- 4.4 23.0 17.9 47.8 41.8 6.0 Panama 2003 34.8 24.7 -- 7.1 15.4 10.7 50.2 42.5 7.7 Paraguay 2003 33.9 30.0 -- 11.8 26.6 19.9 60.5 61.7 1.2 Peru 2003 40.3 34.5 -- 5.7 19.0 15.8 59.3 56.0 3.3 Rep. Dominicana 2004 39.1 32.5 -- 5.7 12.2 11.3 51.3 49.5 1.8 Uruguay 2004 24.1 17.2 -- 9.1 18.3 11.5 42.4 37.8 4.6 Venezuela, R. B. de 2003 39.4 33.0 -- 3.0 14.7 17.6 54.1 53.6 0.5 Sources: Gasparini and Tornarolli 2006; ILO 2006. Note: -- = not available. Annex 4. We also know that, within the formal sector, individuals choose among degrees of formal sector protection (for example, how Formostofthereport,werequiremoredetailthanprovidedby much insurance and what kind to buy). the ILO tables. Hence, Gasparini and Tornarolli (2006) have 5. See Livingstone (1991), who argues that many of these goods replicated the ILO data with identical data and definitions. and services are "appropriate" for lower-income consumers. In fact, Although in this chapter we use measures that include infor- these consumers are willing--given their preferences, information, mal salaried workers in our definition, we check here to be rates of discount, and income--to eat in a cheaper, less hygienic restau- sure thatthesubcomponentofourmeasurethatcorrespondsto rant in any Latin American capital or ride a less safe taxi or mototaxi. 6. Sethuraman (1981), in another ILO report, mentioned that self-employment is similar to that of the ILO. The right-hand there was no evidence that informal workers were en route to or column of table 1A.1 suggests that, with some exceptions-- queuing for a formal sector job, but that their activities were a per- Colombia, Mexico, Panama--the two series are quite close. manent source of income. This may testify more to extreme barriers to entry than to voluntariness in the sector. Notes 7. For two comprehensive surveys of these methods, see Thomas 1. This section draws heavily on Maloney (2006). (1993) and Schneider and Enste (2000, 2002). 2. Without abandoning the central principle, a certain amount 8. The version of the original ILO definition presented here con- of state-sanctioned evasion may be optimal. Undertaking a cost­benefit siders an individual as an informal worker if she or he belongs to any analysis of monitoring and enforcement, the state may decide to leave of the following categories: (1) unskilled self-employed, (2) salaried its coverage incomplete. As an example, many countries collect no worker in a small private firm, or (3) zero-income worker. It is impor- taxes below a certain level of income or have streamlined labor regu- tant to note that labor market­related definitions include within the lations for microfirms. In this case, the "informal" becomes simply informal sector at least two types of workers, self-employed people the population that it wasn't socially optimal to force to be formal. and informal wage earners for whom the micro determinants and 3. The tax morale literature departs from the finding that, under motivations to participate in formal or informal economic arrange- normal estimates of individual risk aversion, the existing penalties ment vary, as will be discussed in chapter 2. for cheating and the probability of being caught are simply too low 9. For a discussion of Brazil, see Henley, Arabsheibani, and to explain the high rates of compliance in the advanced countries. Carneiro (2006). See, for example, Graetz and Wilde (1985); Alm, McClelland, and 10. The gray area of informality may also exist in the public sector Schulze (1992); and Frey and Feld (2002). where rigid hiring and firing laws generate the need to hire public 39 I N F O R M A L I T Y employees through diverse contractual arrangements, ranging from Capp, J., H. Elstrodt, and W. Jones, Jr. 2005. "Reining In Brazil's low-skilled workers who work for a subcontractor to highly skilled Informal Economy." McKinsey Quarterly; available at http:// professionals who work permanently in public institutions but are www.mckinseyquarterly.com. paid as consultants. In most of these cases, these workers are formal in Castells, M., and A. Portes. 1989. "World Underneath: The Origins, terms of tax compliance, but informal from the labor viewpoint. Dynamics and Effects of the Informal Economy." In The Informal 11. The documentation is not always clear on what variables Economy: Studies in Advanced and Less Developed Economies, ed. A. are used as causes and indicators, although, in some articles by Portes, M. Castells, and L. A. Benton, 11­37. Baltimore: Johns Schneider and Enste (2000, 2002), self-employment is mentioned as Hopkins University Press. a cause. Centeno, M. A., and A. Portes. 2003. "The Informal Economy in the 12. In particular, the estimate of the shadow economy as a share Shadow of the State." Photocopy. Princeton University, NJ. of GDP for the group of African countries is approximately the same Chavez, E., and J. Chacaltana. 1994. "Cómo se Financian las Microem- as the estimate for Latin America and the Caribbean, despite the fact presas y el Agro." CEDEP, Ed. Stilo Novo SRL, Lima, Peru. that the group's average GDP per capita is only 38 percent of the Cunningham, W. 2001. "Breadwinner Versus Caregiver: Labor Force average GDP per capita in Latin America and the Caribbean. In the Participation and Sectoral Choice over the Mexican Business same way, the group of countries in Eastern Europe and Central Cycle." In The Economics of Gender in Mexico: World, Family, State, Asia has a significantly lower estimate of the shadow economy as a and the Market, ed. E. G. Katz and M. C. Correia, 85­132. share of GDP than does Latin America and the Caribbean, despite Washington, DC: World Bank. the fact that the group's average GDP per capita is only 9 percent ------. 2007. "Minimum Wages and Social Policies: Lessons from higher than the average GDP per capita in Latin America and the Developing Countries." World Bank, Washington, DC. Caribbean. Cunningham, W., and W. F. Maloney. 2001. "Heterogeneity among 13. For operational purposes, the ILO defines informal employ- Mexico's Microenterprises: An Application of Factor and Cluster ment as consisting of self-employed or own-account workers (exclud- Analysis." Economic Development and Cultural Change 50 (1): 131­56. ing administrative workers, professionals, and technicians), unpaid de Ferranti, D., G. Perry, I. Gill, L. Servén, F. Ferreira, W. F. Maloney, family workers, and employers and employees working in establish- and M. Rama. 2000. Securing Our Future in the Global Economy. ments with less than 5 or 10 persons employed, and excludes paid Washington, DC: World Bank. domestic workers. For Latin America, ILO self-employment suggests de Soto, H. 1989. The Other Path. New York: Basic Books. that the informality rate increased from 42.8 to 47.4, a rise of 11 per- Djankov, S., R. La Porta, F. Lopez-de-Silanes, and A. Shleifer. 2002. cent. Using the productive definition of the ILO (2002), Gasparini "The Regulation of Entry." Quarterly Journal of Economics 117 (1): and Tornarolli (2006) and Rofman and Luchetti (2006) suggest that 1­37. labor-defined informality has indeed increased over time in the Esfahani, H., and D. Salehi-Isfahani. 1989. "Effort Observability and region. Gasparini and Tornarolli argue that informality increased in Worker Productivity: Towards an Explanation of Economic 13 countries, fell in 2, and remained constant in 3. Dualism." Economic Journal 99 (June): 818­36. Fields, G. S. 1990. "Labor Market Modelling and the Urban Infor- mal Sector: Theory and Evidence." In The Informal Sector References Revisited. Paris: Organisation for Economic Co-operation and Acemoglu, D., S. Johnson, and J. A. Robinson. 2001. "The Colonial Development. Origins of Development." American Economic Review 9 (5): Frey, B., and L. Feld. 2002. "Deterrence and Morale in Taxation: An 1369­98. Empirical Analysis." Working Paper 760, Center for Economic Alesina, A., and E. La Ferrara. 2000. "Participation in Heterogeneous Studies and the Ifo Institute, Munich. Communities." Quarterly Journal of Economics 114 (3): 847­904. Friedman, E., S. Johnson,. D. Kaufmann, and P. Zoido-Lobatón. 2000. Alm, J., G. McClelland, and W. D. Schulze. 1992. "Why Do People "Dodging the Grabbing Hand: The Determinants of Unofficial Pay Taxes?" Journal of Public Economics 48: 21­48. Activity in 69 Countries." Journal of Public Economics 76: 459­93. Banerji, A., and S. Jain. 2006. "Quality Dualism." Photocopy. García-Verdú, R. 2007. "Measurement of the Shadow Economy or Shad- Department of Economics, University of Virginia, Charlottesville. owy Measurement?" Photocopy. World Bank, Washington, DC. Bentolila, S., and A. Ichino. 2000. "Unemployment and Consumption: Gasparini, L., and L. Tornarolli. 2006. "Labor Informality in Latin Are Job Losses Less Painful near the Mediterranean?" Discussion America and the Caribbean: Patterns and Trends from Household Paper 2539, Centre for Economic Policy Research, London. Survey Microdata." Photocopy. World Bank, Washington, DC. Blau, D. 1987. "A Time-Series Analysis of Self-Employment in the Geertz, C. 1963. Peddlers and Princes: Social Development and Economic United States." Journal of Political Economy 95 (3): 445­67. Change in Two Indonesian Towns. Chicago: University of Chicago Bond, P., and R. Townsend. 1996. "Formal and Informal Financing Press. in a Chicago Ethnic Neighborhood." Federal Reserve Bank of Glaeser, E. L., D. Laibson, and B. Sacerdote. 2000. "The Economic Chicago Economic Perspectives 3­27. Approach to Social Capital." Working Paper 7728, National Breusch, T. 2005. "Estimating the Underground Economy using Bureau of Economic Research, Cambridge, MA. MIMIC Models." Photocopy. School of Economics, Faculty of Eco- Graetz, M., and L. Wilde. 1985. "The Economics of Tax Compliance: nomics and Commerce, Australian National University, Canberra. Facts and Fantasy." National Tax Journal 38: 355­63. 40 T H E I N F O R M A L S E C T O R : W H AT I S I T, W H Y D O W E C A R E , A N D H O W D O W E M E A S U R E I T ? Gregory, P. 1986. The Myth of Market Failure: Employment and the Labor Maloney, W. F., and J. N. Nuñez. 2001. "Measuring the Impact Market in Mexico. Baltimore: Johns Hopkins University Press. of Minimum Wages: Evidence from Latin America." Policy Greif, A. 1993. "Contract Enforceability and Economic Institutions Research Working Paper 2597, World Bank, Washington, DC. in Early Trade: The Maghribi Traders' Coalition." American Mazumdar, D. 1976. "The Urban Informal Sector." World Development Economic Review 83: 525­48. 4: 655­79. ------. 1998. "The Historical and Comparative Institutional Analy- North, D., J. Wallis, and B. Weingast. 2005. "The Natural State: The sis." American Economic Review Papers and Proceedings 83: 80­84. Political-Economy of Non-Development." Burke Center for Inter- Hart, K. 1973. "Informal Income Opportunities and Urban Employ- national Relations, University of California, Los Angeles. ment in Ghana." Journal of Modern African Studies 11: 61­89. O'Donnell, G. 1996. "Illusions about Consolidation." Journal of Henley, A., G. R. Arabsheibani, and F. G. Carneiro. 2006. "On Democracy 7 (2): 34­51. Defining and Measuring the Informal Sector." Policy Research Perry, G., O. Arias, H. López, W. Maloney, and L. Servén. 2006. Working Paper 3866, World Bank, Washington, DC. Poverty Reduction and Growth: Virtuous and Vicious Circles. Hirschman, A. O. 1970. Exit, Voice, and Loyalty: Responses to Decline in Washington, DC: World Bank. Firms, Organizations, and States. Cambridge, MA: Harvard Univer- Portes, A., M. Castells, and L. A. Benton, eds. 1989. The Informal sity Press. Economy: Studies in Advanced and Less Developed Economies. Hussmanns,R.2004."DefiningandMeasuringInformalEmployment." Baltimore: Johns Hopkins University Press. Bureau of Statistics, International Labour Organization, Geneva. Rauch, J. E. 1991. "Modelling the Informal Sector Formally." Journal ILO (International Labour Organization). 2002. "Decent Work and of Development Economics 35: 33­47. the Informal Economy." Report VI for the 90th International Robles, M., M. Torero, J. Saavedra, N. Valdivia, and J. Chacaltana. 2000. Labour Conference, Geneva. "EstrategiasyRacionalidaddelaPequeñaEmpresas." ILO, Lima. ------. 2006. LABORSTA (labor statistics). ILO. Available at Rofman, R., and L. Luchetti. 2006. "Pension Systems in Latin http://laborsta.ilo.orgl. America: Concepts and Measurements of Coverage." Social INEGI (Instituto Nacional de Estadística, Geografía e Informática). Protection Discussion Paper 0616, World Bank, Washington, DC. 2004. Cuenta Satélite del Subsector Informal de los Hogares Saavedra, J., and A. Chong. 1999. "Structural Reforms, Institutions 1998­2003. Cuentas por Sectores Institucionales, Sistema de and Earnings: Evidence from the Formal and Informal Sectors in Cuentas Nacionales de México. Urban Peru." Journal of Development Studies 35 (4): 95­116. Jovanovic, B. 1982. "Selection and Evolution of Industry." Economet- Schneider, F. 2005. "Shadow Economies around the World: What Do rica 50 (3): 649­70. We Really Know?" European Journal of Political Economy 21 (3): Levenson, A., and W. Maloney. 1998. "The Informal Sector, Firm 598­642. Dynamics and Institutional Participation." Policy Research Schneider, F., and D. H. Enste. 2000. "Shadow Economies: Size, Working Paper, 1988, World Bank, Washington, DC. Causes, and Consequences." Journal of Economic Literature 38 (1): Livingstone, I. 1991. "A Reassessment of Kenya's Rural and Urban 77­114. Informal Sector." World Development 19 (6): 651­70. ------. 2002. The Shadow Economy: An International Survey. Loayza, N. 1996. "The Economics of the Informal Sector: A Simple Cambridge, UK: Cambridge University Press. Model and Some Evidence from Latin America." Carnegie- Sethuraman, S. V. 1981. "The Urban Informal Sector in Developing Rochester Conference Series on Public Policy 45: 129­62. Countries: Employment, Poverty and Environment." Interna- Loayza,N.,andJ.Rigolini.2006."InformalityTrendsandCycles."Pol- tional Labour Organization, Geneva. icy Research Working Paper 4078, World Bank, Washington, DC. Souza, P., and P. Baltar. 1979. "Salario minimo e taxa de salarios no Loayza, N., L. Servén, and A. M. Oviedo. 2005. "The Impact of Brasil." Pesquisa e Planejamento Economico 9: 629­60. Regulation on Growth and Informality--Cross-Country Evidence." Stiglitz, J. E. 2000. "Formal and Informal Institutions." In Social World Bank Policy Research Working Paper, World Bank, Capital: A Multifaceted Perspective, ed. P. Dasgupta and I. Washington, DC. Serageldin. Washington, DC: World Bank. Lucas, R. E., Jr. 1978. "On the Size Distribution of Business Firms." Tendler, J. 2002. "Small Firms, the Informal Sector, and the Devil's Bell Journal of Economics 9 (2): 508­23. Deal." Institute of Development Studies Bulletin 33 (3): 98­104. Maloney, W. F., 1999. "Does Informality Imply Segmentation in Thomas, J. J. 1993. Informal Economic Activity. Ann Arbor: University Urban Labor Markets? Evidence from Sectoral Transitions in of Michigan Press. Mexico." World Bank Economic Review 13: 275­302. ------. 1999. "Quantifying the Black Economy: `Measurement ------. 2001. "Self-employment and Labor Turnover in Developing Without Theory' Yet Again?" Economic Journal 109 (456): 381­89. Countries: Cross-country Evidence." In World Bank Economists' Tokman, V. 1992. "The Informal Sector in Latin America: From Forum, ed. S. Devarajan, F. H. Rogers, and L. Squire. Washington, Underground to Legality." In Beyond Regulation: The Informal Econ- DC: World Bank. omy in Latin America, ed. V. Tokman. Boulder, CO: Lynne Rienner. ------. 2004. "Informality Revisited." World Development 32 (7): World Bank. 2006a. Doing Business Indicators. Available at http:// 1159­78. www.doingbusiness.org. ------. 2006. "Informality Deconstructed." Photocopy. World ------. 2006b. World Development Indicators. Available at http:// Bank, Washington, DC. www.worldbank.org. 41 CHAPTER 2 The Razón de Ser of the Informal Worker SUMMARY: This chapter brings to bear a variety of conceptual and empirical approaches to flesh out the rationale (razón de ser) for the existence of the informal worker. The chapter argues that the traditional segmentation/exclusion view as the sole explanation of informality is seriously incomplete and offers evidence of a large voluntary entrance into informal work. Two types of informal employment emerge: one component corresponds to workers excluded from more desirable for- mal jobs, and the other to those driven by voluntary motives. The majority of independent or self-employed workers are voluntary, attaching significant value to the nonpecuniary benefits of autonomous work and choosing to "exit" formal social protection systems. In contrast, the majority of informal salaried workers appear to be excluded from more desirable jobs both as formal salaried and independent workers, although voluntary motives are significant for many workers (for example, youth) and predominant in some country contexts (such as the Dominican Republic and Mexico). In both cases there is substantial heterogeneity of motives and demographic characteristics with gradients and gray areas that cloud universal conclusions about what determines the relative size and welfare implications of each group. C HAPTER 1 SUGGESTED THAT INFORMALITY with a range of motivations and constraints that spans the in the labor market spans all three possible traditional view of workers excluded from the formal sector, margins: firms of all sizes contracting some as well as those stressing the voluntary "exit" of many part of their workforce without mandated workers from state-sponsored labor and social protection labor benefits; owners of small firms contem- systems. plating registering their workers with the labor ministries; and informal and formal workers weighing the advantages Informal work: Adding exit to exclusion and disadvantages of jobs in the two sectors. What deter- As noted in chapter 1, much of the debate over the nature mines whether workers are formal or informal at each of and implications of a large informal sector can be traced these margins has been the focus of intense debate over the broadly to two views of whether workers are driven to or last four decades and the policy implications critically pushed toward informal work. In the "exclusion" view, depend on the diagnosis. This and the next three chapters informal workers, either self-employed or salaried, are the bring to bear a variety of analytical tools--motivational and disadvantaged class of a segmented labor market arising self-rated welfare responses, labor market transitions, earn- from economic dualism and institutional rigidities. ings comparisons, and labor force dynamics across the busi- Workers would prefer the presumed higher wages and ben- ness cycle--to shed light on the significant drivers of efits of formal work, but are rationed out. In contrast, the decisions at these margins. Although a crisp, stylized image integrated view emphasizes the voluntary character of infor- of the sector would be desirable, the picture that emerges is mal employment. Workers may choose informal jobs, given a complex one of a heterogeneous sector containing workers their preferences, skills, other means of social protection, 43 I N F O R M A L I T Y and their valuation of costs and the characteristics of infor- FIGURE 2.1 mal and formal employment. The main rationale behind Relative sector sizes and wages under a nominal wage rigidity each view is discussed below. Since the traditional exclusion view is better known, there is a more detailed discussion of Wage I Wage F the integrated view. Wage rigidity in the formal sector WM The exclusion view: Segmentation and beyond An extensive body of literature with its roots in Harris and WI WF Todaro (1970) has equated the informal sector with under- employment or disguised unemployment--the disadvan- DF DI taged or excluded sector of a labor market segmented by rigidities in the formal or covered sector of the economy.1 This view has mainly focused on the second margin, or LI LF labor flows between informality and the formal salaried sec- tor: above-market-clearing wages force workers to queue LF for preferred formal jobs while subsisting in the informal Source: Authors' illustration. microfirm sector, which is characterized by little capacity for capital accumulation (thus low productivity), an absence of labor benefits, irregular work conditions, high Business Group at the World Bank, microfirms would like turnover, and lower rates of remuneration. The large pro- to formally register their businesses and workers, but bur- portion of women in the sector with lower earnings raises densome or costly registration precludes their doing so, and additional suspicions of discrimination in the allocation of mandated non-wage benefits make hiring/firing too costly. "good" jobs. The steady rise in informal salaried work in Thus, a size dualism is created in which small firms are forced larger firms in several countries, such as Argentina (see to operate informally. More generally, if the state makes chapter 1), has also led to focus on the within-firm margin. compliance with taxes and regulations too costly, even larger Related to both is the Portes, Castells, and Benton (1989) firms will be forced to operate as partially informal. In gen- structural articulation view that informal work is a means eral, regulations that affect the profitability of firms in com- by which large firms resort to informal labor (directly or petitive markets can induce size and labor market dualism through subcontracting) to sidestep labor regulations and (Fortin, Marceau, and Savard 1997; Rauch 1991). unions in order to adjust to increased global competition. However, labor market dualism can result from factors Figure 2.1 depicts one mechanism of how labor market other than rigidities in labor institutions or badly designed segmentation may happen. A nominal wage floor, arising regulations. The high levels of quitting may induce formal from a minimum wage (Wm) or union power, will expand firms to pay higher-than-market-clearing "efficiency wages" informal employment and drive an earnings wedge to secure the more productive employees (see Shapiro and between the two sectors, with informal workers occupying Stiglitz 1984; Stiglitz 2000).2 Since informal, small firms inferior (lower-pay) jobs. Thus, segmentation yields two are able to monitor workers' efforts almost without cost, key predictions: identical workers receive higher earnings they do not need to pay the efficiency wage (Esfahani and in formal jobs; and, given the opportunity to move, infor- Salehi-Isfahani 1989). In his interviews with hundreds of mal workers would happily take a formal job. This and the business executives, labor leaders, and human resource per- next two chapters document numerous findings gauging sonnel, Bewley (1999) found that there was an aversion to these predictions, although, as argued below and in chap- cutting wages of either current employees or new hires, ter 3, informal-formal earnings gaps cannot offer unam- even during economic downturns, under the concern that biguous tests of segmentation. this would hurt workers' morale and their support for Poorly designed regulations may also impede firms to internalizing the firm's needs to enhance its competitive- engage in as much formality as they may wish and, as a ness. Because of information asymmetries, young workers result, generate higher informal employment (Loayza 1996). without a track record may find formal employers reluctant Along the de Soto margin, since elaborated by the Doing to hire them and hence may queue in the informal sector 44 T H E R A Z Ó N D E S E R O F T H E I N F O R M A L W O R K E R while they gain experience. Finally, in many countries, posited the idea of a two-tier sector--one of which is vol- social security contributions are collected with income untary and prosperous--and, in fact, recent studies for taxes. Even when these are collected separately, the pay- numerous Latin American countries conclude that the ment of social security contributions provides public infor- majority of informal self-employed workers have chosen mation so that firms tacitly colluding to underreport their occupation after undertaking a cost­benefit analysis output and tax liabilities will also resort to informal hiring. of whether to be self-employed or salaried and then That is, part of labor informality may be a result of pure whether to be formal or informal.4 The literature on infor- evasion of income and sales taxes. mal salaried workers is scant to date, but the same pair of The persistence of exclusionary factors is often associated considerations may enter both sectors' calculations: com- with weak institutions and state capture by both elites and parative advantage, and, as will be argued in chapter 7, the organized segments of middle classes, or, in Hirschman's perceived inadequacy of the benefits of formality or their (1970) terms, "the lack of voice of those excluded." The high opportunity cost and alternative self-insurance mech- implications of this view can be disturbing. First, over half anisms or free social protection programs. As chapters 5 of Latin America's workers, far more in some cases, are and 6 will elaborate, firms also appear, more generally, to without desired social protections. Second, the increase in choose the degree of participation in formal institutions informality is driven by either increasingly onerous labor, according to their business needs. Finally, as will be posited tax, or business regulations or increased global competition in chapter 8, an important leitmotif underlying both that is causing a race to the bottom in workers' rights and worker and firm decisions is a concern with the overall protections. Both informal workers and firms lack "voice" competency and legitimacy of the state. in the political system to change the exclusionary rules and As will be discussed in more detail in chapter 3, work- processes that pushed them to informality. The next sec- ers' comparative advantage is central to their occupational tions and chapters of the report show evidence that "exclu- choice. This arises from the matching of individual abilities sion" factors are indeed important, document their or tastes to the different demands and characteristics of negative impacts on productivity and welfare, and discuss jobs. Lucas (1978) argued that individuals choose between some of the reforms necessary to mitigate them. salaried work and self-employment, depending on whether they are relatively more talented as an entrepreneur or as a The integrated market view: Opting out salaried employee. Rosen (1981), Heckman and Sedlacek of or "exiting from" formality (1985), and, recently, Carneiro, Heckman, and Vytlacil However, there is a second and equally important lens (2005) argue that a worker's comparative advantage-- through which to view informality that is more akin to emerging from a variety of work-related skills and Hirschman's voluntary "exit": workers and firms make characteristics--drives him or her to choose among distinct implicit cost­benefit analyses about whether to cross the jobs the one that best matches his or her talents and tastes. margin into formality, and they frequently decide against Informal and formal jobs differ by more than labor protec- it.3 In this lens, much of the informal sector, in fact, offers tions, and formal benefits are just one ingredient in jobs that are equally valued by workers to those they could workers' calculations. In figure 2.1, workers equilibrate get in the formal sector. Contrary to the predictions of the utilities--not just earnings--in choosing between jobs in exclusion view, this implies that many informal workers are the two sectors. equally well-off (in broad welfare terms) as in other formal In addition to Lucas's (1978) discussion of comparative jobs fit to their skills; and, being "voluntarily" informal, advantage based on relative entrepreneurial skill, informal they can move to the formal sector but choose not to. It is jobs may offer an entry point to the labor market for youth imperative to highlight that this does not imply that these and unskilled middle-age workers that partially remedies workers are prosperous or happy--only that they would deficient or obsolete skills through on-the-job training not be better-off in the formal sector jobs that workers with unavailable to them in formal salaried jobs. For married similar characteristics occupy. women, informal jobs offer flexibility to better balance Again, when Keith Hart coined the term informality, he work and child rearing. For some talented workers, infor- surely never assumed the sector was necessarily bad, and mality may offer better prospects for upward mobility than neither did several subsequent authors. Some authors have the formal sector.5 45 I N F O R M A L I T Y In addition, as Maloney (1999, 2004) notes, it is often consider saving for their retirement unaffordable or to critically overlooked that social protections are not free. resort to other informal arrangements, such as investing in Workers pay for them, either explicitly (through contribu- a microenterprise that may be sold upon retirement. Youth, tions) or implicitly in terms of lower wages, as has been in particular, tend to have a high discount rate (or myopia) documented, for example, in Ecuador, Peru, and Colombia and be less concerned about having jobs with pension con- (Kugler and Kugler 2003; MacIsaac and Rama 1997, tributions. Further, poor workers who face temporary 2001). Drawing on the industrialized countries' experience financial hardship will curtail savings to preserve current of predominantly formal salaried labor, "decent" jobs are consumption so that incentives for opting out of benefits generally considered to be those covered by labor legisla- programs would be higher during crises. The work on tion. However, the low quality of many formal services and Chile by Barr and Packard (2000) and Packard (2002) cor- high administrative overhead costs may cause many workers roborates the significance of some of these considerations. to see mandatory contributions to benefits programs as a They find that participation in the government's voluntary disadvantage of formal salaried work. As the sociologist pension scheme, a private individual account in which Bryan Roberts (1991) notes, based on his interviews with workers receive what they contribute, is barely around informal workers in Guadalajara, Mexico: "The absence of 4 percent. Thus, independent workers are choosing to be welfare coverage is a drawback, but, on the other hand, "unprotected" even by the scheme that arguably best aligns many informants cited the deductions made for welfare as a costs and benefits in the region.6 disadvantage of formal employment, particularly since the Moreover, informal support networks may be able to services they received were poor" (p. 50). partially substitute for unemployment and health insur- This misalignment of costs and benefits can lead work- ance or retirement funds at lower transaction costs and with ers to regard their contribution as a tax and to opt out of greater ease of monitoring those who intend to abuse the the system. In fact, a microentrepreneur, faced with bor- system. Morduch (1999) reviews numerous studies that rowing constraints to expand a business, may be reluctant find that informal insurance has a significant ability to off- to hand over current resources for an uncertain promise of set income or other shocks, although it is far from perfect. an old-age payment in the distant future. At worst, it Bentolila and Ichino (2000) argue that, in Italy and Spain, means throwing money away if pension funds are raided or household consumption fell only 20 percent as much in inflated away to finance the fiscal deficit. But even with sta- response to unemployment of the household head as it did ble macroeconomic times, Levy (2006) finds that the sub- in Germany, where formal credit and social protections sys- stantial rates of transition of many Mexican workers tems are more advanced. They attribute this to the greater between jobs with formal pension contributions and those influence of (informal) interhousehold transfers.7 Some with none would, in all likelihood, leave them with insuffi- workers may find settling for informal mechanisms justi- cient months of benefits to qualify for a pension. As chap- fied by the other benefits of informality--for instance, flex- ter 7 will discuss, this situation is relevant even in ibility. This is akin to strains of the social capital literature countries with low levels of informal employment, such as that see individuals optimizing their investment in infor- Uruguay. mal networks with a view toward long-run returns broadly Even in the case where benefits are well aligned with conceived. costs, workers may still opt out if they prefer higher cash In fact, not being formally enrolled with the social secu- compensation or if an alternative exists at lower cost or is rity system does not preclude being de facto covered by other better suited for their needs. The clearest example is that labor protections. Recent work (Cunningham 2007; Maloney often an entire family is covered by medical benefits when and Nuñez 2002) finds that minimum wages in several Latin any one member is formally employed, so the second for- American countries are actually most binding among salaried mal sector worker sees zero return to paying labor taxes, workers in informal firms. This "lighthouse" effect, as it is explicitly or implicitly. Using cross-country employment called in Brazil, suggests that there are norms on level of pay surveys for the region, Galiani and Weinschelbaum (2006) that informal employers follow even if they do not register find that secondary workers are indeed more likely to be their workers with social security administrations. informal if the household's primary worker has a formal Similar considerations apply to firms approaching the job. Liquidity constraints may lead very poor workers to de Soto margin contemplating registering their business or 46 T H E R A Z Ó N D E S E R O F T H E I N F O R M A L W O R K E R workers with the authorities. Saavedra and Chong (1999) later in chapter 8)--also enters here. In any case, the opti- argue that, in deciding to register, firms consider whether mal choice for firms, workers, and the state itself would not the increased expansion of scale and access to other institu- obviously be the full formal sector package for all. tional supports that facilitate growth makes the registra- In sum, workers, in making their multidimensional tion expenses worthwhile. Levenson and Maloney (1996) evaluation of sectoral choice along the third margin--as show that formality along a broad set of dimensions (regis- well as microfirms contemplating a jump into formality tering with tax and labor authorities, or joining business along the de Soto margin--will weigh the costs and bene- associations) can be considered "normal goods" in the pro- fits of being formal, and we cannot assume, immediately, duction function. Entrepreneurs with the underlying that all who are without protections would desire to be oth- ability to expand, eventually, need and seek these inputs. erwise. As Maloney (1999) argues, and chapter 3 will inves- This is consistent with numerous studies showing that tigate further, this insight also compounds the difficulties formality increases with firm size and time in business (see of establishing segmentation, and hence the degree of chapters 5 and 6). involuntary informality, by the traditional method of com- In essence, low productivity may be the main reason why paring earnings adjusted for human capital. In a market many small firms, and their employees, remain informal. with no rigidities, informal earnings should be higher to The fact that most small firms operate in the informal sector compensate workers for the lost value of benefits and what- (that is, size dualism) may just be a natural result of their ever risk they may be facing. On the other hand, they may lower managerial ability (Amaral and Quintin 2006; de be lower to compensate for taxes evaded, greater indepen- Paula and Scheinkman 2006; Straub 2005; and Levenson dence and flexibility, or, perhaps for young workers, on-the- and Maloney 1996). The observation that size and labor job training. market dualism go hand in hand (that is, unskilled workers tend to be employed in informal small firms) can simply Avoiding the extremes in the debate reflect the efficient allocation of more productive labor and The exaggerated size of the informal sector in the region entrepreneurship (Galiani and Weinschelbaum 2006). Sort- raises the stakes of the debate dramatically: if over 50 per- ing also leads to matching of employees and employers with cent of Latin American labor markets are thought of as dis- higher (lower) propensities to evade. Thus, part of informal guised unemployment, then the implied labor market and salaried employment could be a result of microfirms' sole regulatory distortions are indeed large. But if informality is decisions to opt out of formal institutions for reasons unre- largely opting out, as discussed above, then the concern with lated to the presence of labor or other regulatory distortions, segmentation born of minimum wages, excessive union and some may reflect explicit opting-out collusions between power, or excessive business regulations may be overstated. owners and employees. In fact, however, there is no need to be so Manichean in Given the very uneven enforcement of legislation in the approaching the issue: the two views, exclusion and exit, are region, workers and firms are able to choose the degree of complementary rather than competing analytical frame- participation in benefits programs or formal institutions. works. First, individual countries differ greatly in history, Of course, those operating informally need to weigh the institutions, and legal framework; hence, exclusionary mech- benefits of being informal against the expected value of any anisms may be more important in some, and exit mecha- sanctions if they are caught evading (Dávila Capalleja nisms in others. Second, the informal sector is tremendously 1994; Hibbs and Piculescu 2006). But, in fact, a share of heterogeneous, and arguably, therefore, there is a continuum informality could be effectively mutually tolerated; the in the relative importance of exit and exclusion within state may not find it cost-effective or equitable to levy countries. Third, in some cases the two can be virtually penalties or force compliance on small enterprises, particu- indistinguishable. A microentrepreneur concluding, through larly if this could lead to bankruptcy (employment without a cost­benefit analysis, that formality is not worth the exag- social security contributions is deemed better than unem- gerated registration costs may be explicitly excluded or self- ployment). The strength of the sense of reciprocity or, alter- excluded, but the effect is much the same. A poor worker, natively, the level of tax morale--that one should pay excluded from health services by virtue of living in a because others pay, or conversely that evasion is rampant remote, poor rural area or urban neighborhood, may see little and considered an acceptable business practice (discussed point in being formal and paying labor taxes for services 47 I N F O R M A L I T Y to which he or she has no access. Finally, informality is a FIGURE 2.2 multidimensional concept in which agents interact with the Relative sector sizes and wages in the presence of a labor tax state along some dimensions and not others, creating a large gray area between the extremes of full compliance and non- Wage I Wage F Increase in taxes on compliance. Exit and exclusion can play different roles formal wages across different dimensions: a microfirm owner unregistered due to the high costs of registration may feel de facto excluded from desired formal credit, while opting out from WI WF contributing to poorly designed state pension funds on WI WF behalf of his or her workers. The evidence presented below DF DI and across the next chapters suggests that each viewpoint DF has validity in some sectors, in some countries, at some points in time; both lenses can help understand and address LI LF the causes and consequences of informality in the region. Furthermore, it is important to stress that saying that LI LF workers voluntarily choose to be unprotected does not imply Source: Authors' illustration. that this is the best outcome for society as a whole. Ideally, Note: L and W, respectively, are the labor allocated to and the wages paid in the informal (I) and the formal (F) sectors. we would like all families to be covered by at least the basic protections against income and health shocks, even if they, The next sections will bring to bear a variety of tools-- myopically, may not value, for instance, pensions, in the short motivational responses, sociological studies, and analyses of run. That such protections should be linked to a characteris- worker flows--to characterize the informal sector and tic of a job is not obvious, however, as will be discussed in sketch out the underlying motivations for being in it. chapter 7. Even further, saying that a worker is "voluntarily" informal means only that the worker would not be better-off The sectors of informal labor: Characteristics in the jobs that comparable workers have in the formal sector, and dynamics not that he or she is prosperous and happy, or would not Informal labor can broadly be broken into two subsectors: prefer the kinds of formal jobs found in more advanced coun- informal salaried and independent work. Table 2.1 shows tries. However, it does have important ramifications for what the relative importance of the two sectors in the region using we mean by "decent" work: if a worker is indifferent between the social protection/legal definition where data permit. a job with protections and one without, the jobs must, there- Formal salaried urban employment by this definition ranges fore, be equally decent in that worker's judgment. Thus, the from nearly 20 percent in Bolivia, Paraguay, and Peru, to issue of increasing job quality concerns both formal and roughly 60 percent in Chile--reaching roughly 40 percent informal jobs, not just the latter. for the region overall. Salaried informal work, including Finally, a finding of equal levels of welfare across sectors or unpaid and domestic workers--mostly women--comprises voluntary informality does not preclude the presence of large roughly 33 percent of urban employment in the region, labor or business climate distortions. More straightforwardly, ranging from 17 percent in Chile to over 45 percent in regulation-stunted growth prevents the generation of firms Ecuador, Nicaragua Paraguay, and Peru. Informal indepen- with potential for generating higher-paying jobs over the dent (self-employed) workers, comprising single-person long run that would draw workers from the informal firms or owners employing other workers, represent roughly microfirm sector. Second, figure 2.2 shows how labor 24 percent of the regional urban workforce, ranging from distortions--restrictions on firing, labor taxes--can shift in 18 percent in Chile to over 35 percent in Bolivia, Colombia, the demand curve for formal labor, leading to a larger share of theDominicanRepublic,Peru, andthe República Bolivariana informal workers without creating segmentation. Both argu- de Venezuela. Clearly, the relative proportions depend on ments allow us to reconcile the findings that many informal conventions for allocating workers--for instance, the Inter- workers are as well-off as their formal counterparts with those national Labour Organization (ILO) categorizes domestic stressing how distortions affect the size of the informal sector workers as independent--or whether unpaid workers are (Botero et al. 2004; Djankov et al. 2002; Loayza 1996). counted as "salaried." However, the overall picture is one in 48 TABLE 2.1 Distribution of formal salaried, informal salaried (social protection/legal definition), and self-employed in urban areas of Latin America (percent) Formal Salaried Informal Salaried Independent workers Public Domestic Unpaid Formal) Informal Country Year Subtotal Large firms Small firms workers Subtotal Large firms Small firms workers workers Subtotal (professional) (unskilled) Argentina (GBA) 2005 45.33 27.96 4.03 13.34 30.36 10.83 10.87 7.65 1.01 24.32 9.31 15.01 Bolivia 2005 18.50 8.32 0.38 9.81 40.94 16.41 12.19 3.82 8.52 40.56 4.36 36.20 Brazil 2003 48.16 30.25 5.52 12.39 28.52 6.69 9.43 6.56 5.84 23.32 4.05 19.27 Chile 2003 60.81 43.94 4.60 12.27 17.20 7.64 4.26 4.03 1.27 21.99 4.45 17.54 Colombia 2006 30.64 21.25 1.80 7.59 28.01 7.99 12.06 4.65 3.31 41.35 2.85 38.50 Dominican Republic 2006 33.24 18.84 0.60 13.80 27.26 13.24 7.06 5.10 1.86 39.50 2.10 37.40 49 Ecuadora 1998 23.93 12.82 1.68 9.43 49.06 18.00 13.70 5.09 12.26 27.01 1.68 25.33 El Salvador 2003 38.78 26.54 1.13 11.10 33.78 11.25 11.43 4.31 6.79 27.45 0.72 26.73 Guatemala 2002 31.10 23.16 1.50 6.44 40.68 12.63 12.96 3.64 11.44 28.22 0.31 27.92 Mexico 2002 35.33 22.23 2.93 10.16 43.79 14.87 17.61 4.30 7.01 20.88 0.14 20.74 Nicaragua 2001 24.19 14.92 1.68 7.59 47.12 16.15 14.74 5.98 10.26 28.70 0.22 28.47 Paraguay 2003 17.93 6.15 1.05 10.74 48.06 15.74 15.79 11.79 4.74 34.01 0.59 33.42 Peru 2002 18.15 9.72 0.89 7.54 45.66 16.57 14.86 5.44 8.80 36.19 1.40 34.79 Uruguay 2004 51.85 29.26 3.98 18.61 21.81 7.08 6.69 6.65 1.40 26.34 5.75 20.59 Venezuela, R. B. dea 2003 38.77 21.80 2.37 14.60 22.55 8.24 9.47 2.06 2.77 38.68 3.46 35.22 Latin America 40.18 25.32 3.75 11.11 33.24 10.10 11.86 5.52 5.76 26.58 2.94 23.64 Sources: Gasparini and Tornarolli 2006; authorsÕ estimates. Note: GBA denotes Greater Buenos Aires. Formality among independent workers proxied by having a university degree. a I N F O R M A L I T Y which the informal salaried account for almost 60 percent of focus largely on salaried workers. Among these, formal the informal, and independent workers slightly less. employees are generally found in larger firms. The informal As chapter 1 noted, firm size has been an important crite- salaried appear to be found more or less equally in small and rion in the past for categorizing the informal, and table 2.1 large firms, although table 2.2 suggests that, as discussed in further breaks out each category by small firms of less than chapter 1, most are concentrated in relatively small enter- the traditional ILO definition of five workers and the rest as prises, if not necessarily in those employing fewer than five "large." The characteristics of independent workers, partic- workers. In Mexico, 73 percent of paid informal salaried ularly as entrepreneurs or firm owners, will be discussed at workers and 97 percent of unpaid workers are found in firms length in chapters 5 and 6 on firm dynamics, so here we of fewer than 15, with the majority of those firms having TABLE 2.2 Distribution of informal salaried, self-employed, and unpaid workers, by firm size (percent) Argentina, 2003 Informal salaried Self-employed Unpaid workers Firm size Share of firm workers Distribution Share of firm workers Distribution Share of firm workers Distribution 1 0.00 0.00 99.56 72.95 0.44 8.36 2 to 5 32.54 53.12 25.13 27.05 1.79 50.53 6 to 25 47.27 29.57 0.00 0.00 1.26 22.99 26 to 100 20.15 9.98 0.00 0.00 0.29 4.17 101 to 500 16.87 4.95 0.00 0.00 1.03 8.82 501+ 13.35 2.38 0.00 0.00 0.99 5.13 Source: Encuesta Permanente de Hogares. Note: Sample constrained to Gran Buenos Aires. Brazil, 2002 Informal salaried Self-employed Unpaid workers Firm size Share of firm workers Distribution Share of firm workers Distribution Share of firm workers Distribution 1 0.00 0.00 100.00 59.39 -- -- 2 to 5 47.82 41.79 24.40 28.29 -- -- 6 to 10 35.03 17.01 10.00 6.44 -- -- 11+ 17.25 41.20 1.85 5.87 -- -- Source: Pesquisa Nacional Amostra Domicilios. Note: -- = not available. Sample constrained to the metropolitan areas. Mexico, 2004 Informal salaried Self-employed Unpaid workers Firm size Share of firm workers Distribution Share of firm workers Distribution Share of firm workers Distribution 1 0.00 0.00 100.00 64.64 0.00 0.00 2 to 5 53.49 56.78 25.02 30.65 13.65 93.03 6 to 10 50.53 11.55 10.02 2.64 2.63 3.87 11 to 15 40.88 4.58 6.29 0.81 0.79 0.57 16 to 50 28.76 10.90 2.48 1.08 0.43 1.05 51 to 100 16.29 3.11 0.54 0.12 0.07 0.09 101 to 250 8.59 1.01 0.32 0.04 0.22 0.17 251+ 10.47 12.06 0.01 0.01 0.17 1.23 Source: Encuesta Nacional de Empleo Urbano. Note: Second quarter 2004. 50 T H E R A Z Ó N D E S E R O F T H E I N F O R M A L W O R K E R TABLE 2.3 Characteristics of informal salaried employees in Mexican microfirms Workers Paid workers Unpaid workers Variable Relatives Nonrelatives Relatives Nonrelatives Relatives Nonrelatives Observations (n) 1,317,330 994,045 372,032 964,275 945,298 29,770 % of Workers 56.99 43.01 27.84 72.16 96.95 3.05 Spouse 23.62 3.45 31.47 Children 50.66 35.77 55.63 Other relatives 25.72 60.78 12.89 % Male 51.62 65.8 72.4 66.65 43.44 38.18 Average hours 33.17 44.09 42.47 44.58 29.51 27.97 Average age 28.15 30.17 28.64 30.38 27.96 23.03 % IMSS 4.31 51.7 13.02 53.28 0.90 0.30 % written contract 0.91 27.46 3.23 28.31 0.00 0.00 Source: Encuesta Nacional de Micronegocios, 1994. Note: IMSS = Mexican Social Security Institute. fewer than five workers. In Argentina, this is somewhat less women and is sometimes thought to be particularly the case, although 83 percent are found in firms of under 25 exploitative, may also emerge from this dynamic (box 2.1). workers, which also encompass 80 percent of unpaid work- Third, among family members, the verbal open contract ers. Hence, understanding the nature of informal salaried is the rule (96 percent), while written contracts are reserved work is, to an important degree, about understanding for the 28 percent of Mexican workers who are paid nonrel- worker and firm decisions to become formal (the de Soto atives. This surprisingly small number suggests that most margin) and the determinants of worker flows between large employment relations are ruled by informal mechanisms formal firms and small informal firms. The evasion margin without recourse to any sort of formal employment con- among large firms is more important in Argentina, where tracts. As other dimensions of informality, this is very roughly 15 percent of salaried workers in firms of over 100 much a question of size of the firm, as noted in chapter 1 are informal, compared to roughly 10 percent in Mexico. and will be further discussed in chapters 5 and 6 on formal- Therefore, we are dealing with modestly differing mixes of ity decisions of firms. Relatively few firms with fewer phenomena across countries. than 3 workers have registered with the social security The Mexican microenterprise survey offers additional administration, while most firms of 10 or more have at insight into the salaried and unpaid workers in small firms least 1 worker registered. Plausibly, more formal contract- (under 16 workers). Several striking findings emerge from ing relationships will become more important as workers table 2.3. First, the microenterprise appears to be very with fewer personal ties are hired. much a family affair. In Mexico, over half of all workers and almost 30 percent of paid workers are directly related to the Life-cycle patterns and dynamics owner of the firm. This clearly complicates earnings com- The informal self-employed and salaried labor show very dif- parisons, as family workers are likely to receive payment in ferent patterns of behavior both across the life cycle and in kind--either food or shelter. Second, essentially all unpaid their flows through the labor market. To begin, figure 2.3 workers (97 percent) are family workers, and so probably illustrates employment trajectories across the life cycle, plot- should be seen as earning pay within the context of the ting the share of the working-age urban employed popula- family unit rather than truly "unpaid." Whether their mar- tion in the three categories. Young workers are most strongly ginal product to the household is as high as it would be in represented in informal salaried employment in each of the the formal sector, or whether many of these workers are, in three countries, peaking at around 20 percent for those in the fact, underemployed, the data cannot tell us. Finally, even a workforce at around the age urban of 20. Thereafter, it falls large part of home-based work, which largely involves steadily, although middle-aged and older workers still make 51 BOX 2.1 Home-based work: Exploitation or flexible work arrangement? A particular modality of the informal microfirm that As can be seen in figure 2B.1, home-based workers do has received increasing attention is home-based work. earn less than other workers. Controlling for lower skill Numerous authors (Arriagada 1998; ILO 1995; Prugl and levels in the sector, home-based workers earn 22.8, 28.9, Tinker 1997; Tomei 2000; WIEGO 2000) tend to see the and 39.6 percent less than do workers outside the home in subcontracting out of these workers as a way for large firms Mexico, Brazil, and Ecuador, respectively (Cunningham to maintain flexibility, quality, and global profitability by and Ramos 2001). But we cannot rule out that these lower both avoiding benefits and transferring the risks of demand wages are reflecting unobserved characteristics or the price volatility to workers. The absence of an internationally of the flexibility that allows juggling other household accepted definition of home-based work, among other responsibilities, not having to travel to the worksite, or problems, means that the reported shares of the workforce, other benefits that accrue to home care providers. This is ranging from 1.5 to 20.0 percent (see table 2B.1), are prob- partially borne out by the fact that home-based workers ably not comparable (Chen, Sebstad, and O'Connell 1999). spend an average of 30 hours in productive activities each Although some very famous firms, such as Italy's week, compared to more than 40 hours weekly among Beneton, started as a cluster of home-based workers, it is nonhome-based workers (Cunningham and Ramos 2001; questionable how important international links really are. Tomei 2000). This result is driven by women, especially As with informal microenterprises more generally, the those with young children and/ or spouses. In Brazil, fraction that reports working as contract workers, perhaps Ecuador, and Mexico, home-based female workers spend for large companies, rather than selling their work one-third fewer hours on the job than do women who work directly in the local market, is generally small (0.7 per- outside the home. Home-based work may thus be a pre- cent in Brazil, 1.2 percent in Ecuador, and 1.6 percent in ferred work arrangement for those who have both home Mexico). Yet the sector is often over 75 percent women and and market duties. Women with young children and/or those whose personal characteristics make outside work who are married are more likely to engage in home-based inaccessible (Carr, Chen, and Tate 2000; Cunningham work than are those without such household constraints, and Ramos 2001; ILO 1995; Tomei 2000). The concern and less than half are household heads (Cunningham and is that employers may take advantage of the situation of Ramos 2001). Interviews with female home-based workers those whose work options are limited to the home by sub- with children reveal that these women hope to work out- jecting home-based workers to lower remuneration and side the home once their children have left home (Jelin, labor standards than workers who can compete in the Mercado, and Wyczykier 2001). labor market (Krawczyk 1993; WIEGO 2000). Despite suggestive anecdotal evidence, statistical studies have yet to document it as a widespread phenomenon. FIGURE 2B.1 Hourly wage premium of outside home work to home-based work Percent TABLE 2B.1 40 Home-based workers' participation (percent) 35 Estimated Proportion Year of 30 Country proportion female estimate 25 Algeria 3.3 97.0 1989 Australia 2.9 -- 1989 20 Brazil 5.5 74.8 1999 Ecuador 17.3 73.6 1999 15 India 2.5 -- 1981 Japan 1.6 93.5 1988 10 Mexico 4.4 64.4 1999 Philippines 23.0 -- 1980s 5 Peru 10.5 -- 1987 0 United Kingdom 2.3 70.0 1981 Brazil Ecuador Mexico United States 7.53 -- 1985 Men Women Sources: Cunningham and Ramos 2001; ILO 1995. Note: -- = not available. Source: Cunningham and Ramos 2001. 52 T H E R A Z Ó N D E S E R O F T H E I N F O R M A L W O R K E R up around 40 percent (almost half in Argentina) of all infor- FIGURE 2.3 mal employees. By contrast, self-employment is almost the Rate of urban employment across sectors, by age mirror image, with virtually no representation among the Argentina (2001) very young, but then climbing steadily until late middle Percent 50 age. The formal sector is largest among prime-age workers. 45 Older workers are as likely to be found in the formal sector or 40 self-employment. 35 The commonalities across all three countries highlight a 30 suggestive, distinct role for the two informal sectors. First, 25 informalsalariedworkisapointofentrytothelabormarketfor 20 manyoftheyoung.Thisisconsistentwiththeaccumulationof 15 skills enabling young workers to eventually find a job in the 10 formal sector or realize any desire to be self-employed. Second, 5 informal salaried work remains a source of jobs for older work- 0 15 19 23 27 31 35 39 43 47 51 55 59 63 67 ers, around 10­15 percent of prime-age workers. They may lack the skills or capital to become self-employed or get a for- Brazil (2002) Percent mal salaried job, or they may opt out of formality since they 50 cannot accrue sufficient years to secure a meaningful pension. 45 Finally, there is no inexorable evolution from informality to 40 formality with age; informal self-employment and formal 35 salaried work are equally common at the end of work life. 30 Further insight can be gained into the differences between 25 the sectors and, more fundamentally, their respective roles in 20 the labor market by looking at patterns of mobility of work- 15 ers through types of work. As Maloney (1999) argues in the 10 first analysis of this kind for Latin America, the segmentation 5 0 or exclusion view would predict a predominantly unidirec- 15 19 23 27 31 35 39 43 47 51 55 59 63 67 tional move toward formality: on average, workers leave school, enter informality to queue until they find a formal Mexico (2004) Percent job, and eventually retire. In an integrated market, however, 50 workers search among formal and informal jobs, treating 45 them as different in type but not necessarily in quality, so 40 that flows among sectors are more symmetrical. 35 Reliable panel data that permit constructing transition 30 25 probabilities from actual movements of the same workers 20 across sectors are available for three countries in the region: 15 Argentina, Brazil, and Mexico (see box 2.2). Bosch and 10 Maloney (2005) follow them through the three sectors of 5 paid work--informal self-employed, informal salaried, and 0 formal salaried--as well as unemployment and being out of 15 19 23 27 31 35 39 43 47 51 55 59 63 67 the labor force (OLF). Such transitions represent "reduced Self-employed Formal Informal forms," capturing disposition as well as ease of entering a sector. Without further relating them to the dynamics of Source: Cunningham et al. 2007. earnings and the business cycle, they need to be seen as more Note: Considers all self-employed and informal salaried uncovered by social protections in Argentina and Mexico, and the availability suggestive than conclusive. Such structured analysis of the of a carteria de trabajo (work card) in Brazil. determinants of entry into sectors, as well as the analysis of cyclical labor force dynamics, is taken up in chapter 4. 53 I N F O R M A L I T Y FIGURE 2.4a Probabilities of transition among sectors of employment Males Raw intensities Adjusted propensities Argentina, 1993­2001 Argentina, 1993­2001 0.5 2.0 0.4 1.5 0.3 1.0 0.2 0.5 0.1 0 0 Formal, Formal, Informal, Formal, Formal, Informal, self-employment informal self-employment self-employment informal self-employment Brazil, 1982­2001 Brazil, 1982­2001 0.5 2.0 0.4 1.5 0.3 1.0 0.2 0.5 0.1 0 0 Formal, Formal, Informal, Formal, Formal, Informal, self-employment informal self-employment self-employment informal self-employment Mexico, 1987­2004 Mexico, 1987­2004 0.5 2.0 0.4 1.5 0.3 1.0 0.2 0.5 0.1 0 0 Formal, Formal, Informal, Formal, Formal, Informal, self-employment informal self-employment self-employment informal self-employment From/to To/from Sources: Bosch and Maloney 2005, 2007. Note: Probabilities in the second column are standardized by final sector job turnover and destination sector size. The first column in each panel of figures 2.4a and b pre- in both Mexico and Brazil, the probability that a worker sents the probability of moving from one sector of work to found in informal salaried work moves to formal work another,8 and two points emerge as noteworthy. First, the across a one-year period is 40­50 percent. The relatively three labor markets show a high degree of commonality smaller reverse probabilities would seem consistent with in most transitions. The relative mobility between any the conventional one-way flows out of informality. How- given pairing of sectors is broadly similar across the three ever, as Maloney (1999) argues, such observations need to countries, although with some notable exceptions. Sec- take into account both the likelihood of a worker separating ond, the level of mobility is relatively high. For example, from the previous job, regardless of destination sector, and 54 T H E R A Z Ó N D E S E R O F T H E I N F O R M A L W O R K E R FIGURE 2.4b Probabilities of transition among sectors of employment Females Raw intensities Adjusted propensities Argentina, 1993­2001 Argentina, 1993­2001 0.5 2.0 0.4 1.5 0.3 1.0 0.2 0.5 0.1 0 0 Formal, Formal, Informal, Formal, Formal, Informal, self-employment informal self-employment self-employment informal self-employment Brazil, 1982­2001 Brazil, 1982­2001 0.5 2.0 0.4 1.5 0.3 1.0 0.2 0.5 0.1 0 0 Formal, Formal, Informal, Formal, Formal, Informal, self-employment informal self-employment self-employment informal self-employment Mexico, 1987­2004 Mexico, 1987­2004 0.5 2.0 0.4 1.5 0.3 1.0 0.2 0.5 0.1 0 0 Formal, Formal, Informal, Formal, Formal, Informal, self-employment informal self-employment self-employment informal self-employment From/to To/from Sources: Bosch and Maloney 2005, 2007. Note: Probabilities in the second column are standardized by final sector job turnover and destination sector size. the availability of positions to move into the destination Given the low rates of turnover in formal salaried and self- sector. The first differs greatly across sectors. Figure 2.5 employment, it is not surprising that flows between these shows the inverse of the probability of leaving--the average sectors should be the lowest in figure 2.5. This turnover, amount of time spent or duration in a sector. Again, the combined with the size of the destination sector, is similarities across countries are far more striking than also related to the number of positions into which a worker the differences: durations are shortest in unemployment, can move. followed by informal salaried, self-employment, or out of The second column in each panel of figures 2.4a and b the labor force; and longest in formal salaried employment. standardizes, by both turnover and destination, available 55 I N F O R M A L I T Y BOX 2.2 Data from rotating panels in Argentina, Brazil, and Mexico In order to construct continuous time transition matrices covering approximately two years through the period to study worker flows across sectors, Bosch and Maloney 1982­2001. The total number of transitions is 2,520,000 (2005) employ three different surveys that compile infor- (1,190,000 for men, 1,330,000 for women). mation about the labor status of workers and other rele- Mexico vant information. They employ one year as the time unit The Encuesta Nacional de Empleo Urbano (ENEU, National to analyze labor mobility dynamics, mainly as a common Urban Employment Survey) conducts extensive quarterly sampling interval for the three countries. household interviews in the 16 major metropolitan areas. The questionnaire is extensive in its coverage of partici- Argentina pation in the labor market, wages, hours worked, and so For Argentina, in the Encuesta Permanente de Hogares (Per- on that are traditionally found in such employment sur- manent Household Survey) a panel covers the area of the veys. The ENEU is structured so as to track a fifth of each Federal District and surroundings (Gran Buenos sample across a five-quarter period. The authors concate- Aires), which accounts for approximately 60 percent of nated panels from the first quarter of 1987 to the fourth Argentina's urban employment. Until 2003, the survey quarter of 1999. Each individual contributed two transi- was conducted every six months (April/May and Octo- tion pairs (from the first quarter to the fourth, and from ber) with a 25 percent rotation of the panel. As a conse- the second to the fifth), giving rise to approximately quence, each household is followed for two years at 1,785,000 transitions (810,000 for men, 975,000 for sampling intervals of six months. The authors employ women). panels from May 1993 to October 2001. The sample is "Informal" refers here to self-employed (including notably smaller than the Mexican and Brazilian surveys, small firm owners) and informal salaried workers in firms and only 29,000 transitions (13,900 for men, 15,100 for with fewer than 16 employees who lack social security or women) can be studied. medical benefits. Formal salaried workers are defined as Brazil those enjoying labor protections. To focus on the tradi- The Pesquisa Mensual de Emprego (Monthly Employment tional first margin of informality, Bosch and Maloney Survey) follows monthly employment indicators. House- (2005) do not consider unprotected workers in large holds are interviewed four months in a row, and then firms. The remainder of the sample is divided into two reinterviewed eight months later, so that 25 percent of groups: those out of the labor force and the unemployed. the sample is renewed every month. Given this panel Unfortunately, the Brazilian survey lacks information on structure, we can construct four yearly employment sta- firm size, and informal status is given by whether the tus transitions for each individual. The authors put worker holds a signed work card guaranteeing access to together nine consecutive panels starting in February benefits in Brazil. Mexican and Argentine surveys con- 1982. Each panel consists of 12 consecutive cohorts tain very similar questions about benefits and firm sizes. positions in the final sector. This gives a measure of whether What is striking is that there is a reasonably high degree a separating worker from one particular sector is more or less of symmetry in these adjusted probabilities, with some likely to move into an available job in another sector than are exceptions: flows from formal salaried to informal salaried are all workers entering that sector from all other sectors (see not so different from the reverse--or, put differently, if an Duryea et al. 2006). Bosch and Maloney (2007) also show informal salaried worker has a comparative advantage in for- that this standardization can be interpreted as a measure of mal salaried work, the reverse is also likely to be true. Fur- revealed comparative advantage of a worker: in the absence of ther, the symmetry of the flows also suggests that, overall, barriers to entry, what sectors do the individual characteris- there is not a one-way flow from informality into formality, tics and personal constraints predispose workers to enter? with some exceptions. For instance, Brazilian formal salaried 56 T H E R A Z Ó N D E S E R O F T H E I N F O R M A L W O R K E R FIGURE 2.5 Absolute mean duration of labor force status Males Females Years 7 6 5 4 3 2 1 0 Out of Unemployed Self- Informal Formal Out of Unemployed Self- Informal Formal labor force employed labor force employed Argentina Brazil Mexico Source: Bosch and Maloney 2005. men move relatively more into informal salaried work than tually leave behind--there may be many instances of subse- the reverse. The same is true for Mexican females. In the case quent return to informal salaried work in later transitions. of Brazil, this may suggest the reduction in the availability of However, what is clear is that, first, we are dealing more formal salaried jobs across this period, described in detail in with "quantum" informality where there is a large amount chapter 4. It is also notable that, for all cases except Mexico, of movement among sectors, rather than patient queuing by formal workers do not seem to have a comparative advantage most of the informal salaried to enter a formal salaried job to enter self-employed work and vice versa, although these permanently. Second, the high frequency of job changes, flows are generally symmetrical, except for Argentine females. even for formal workers, is consistent with the Levy (2006) As chapter 4 will show, in fact, these averages hide sub- observation that the rotation in and out of formal pension stantial changes in relative fluidity across the business schemes implies that many "quantum formal" workers will cycle. Further, as discussed below, this rough symmetry never accrue enough years in the system to earn a pension, hides important demographic differences, consistent with thus creating incentives for opting out. Exploring the the previous patterns of participation across the life cycle. dynamic demographic signature of the two informal sectors However, it seems safe to say that, on average, workers do over the next sections will further help in understanding not graduate unidirectionally from informal work to avail- their role. able jobs in the formal sector. More generally, the exercise suggests that static summary Patterns of mobility of the self-employed statistics of the allocation of workers obscure important The patterns of transition into informal self-employment by aspects of the dynamism of worker flows among sectors. On age and education suggest that this sector behaves broadly average, informal salaried workers do not remain so for long similarly to its counterpart in advanced economies and rather periods of time; their durations are very short. As is argued less like a queue for formal sector jobs (figure 2.6). Evans and later, this does not necessarily mean that informal salaried Jovanovic (1989) argue that the observed increasing proba- labor is a transitory state that the majority of workers even- bility of entry with age, despite a presumably lower level of 57 I N F O R M A L I T Y risk aversion among the young, is consistent with the exis- Mexican workers leaving the informal (formal) salaried sector tence of credit constraints that dictate that potential entre- become independent, compared with the 57 percent (44 per- preneurs must accumulate the necessary physical and cent) of their older counterparts. Similar results are found in working capital. These credit constraints may be exacerbated Argentina and Brazil, where roughly 20 percent (10 percent) in developing countries, where not only do credit markets of exiting young informal (formal) salaried workers find a function poorly, but weak educational systems make the self-employment opportunity, compared to 40 percent accumulation of human capital difficult as well. (20 percent) among older workers.9 These low rates of entry Strikingly consistent with this view, and with our earlier among the young confirm the pattern found in figure 2.4 that examination of life-cycle employment trajectories, self- self-employment is the least viable source of employment for employment is not a port of entry into work. In all three coun- the young in all three countries. tries, the probability of entry for young workers from OLF or Again, the evidence is consistent with a life-cycle model unemployment is a mere fraction of that for older workers. For in which many workers enter formal employment to accu- less than 10 percent of young workers (aged 16­24) leaving mulate both human and financial capital, and then become the OLF sector and unemployment, self-employment appears self-employed or open a small business--a pattern identi- as the entry point to the labor market, approximately three fied in the OECD literature (see Blanchflower and Oswald times less than for middle-aged workers. The propensity to 1998; and Evans and Jovanovic 1989). The sociologists enterthesectorfrominformalorformalsalariedworkisalso,in Balán, Browning, and Jelin (1973) argue precisely for such all cases, often over double for older workers than for younger a life-cycle model in Mexico, where workers with entrepre- workers. For instance, 35 percent (20 percent) of younger neurial motives enter into salaried work; accumulate FIGURE 2.6 Propensity to move to self-employment from different sectors Out of labor force to self-employment Unemployment to self-employment Probability, conditional on separation Probability, conditional on separation 0.6 0.6 Age Education Age Education 0.4 0.4 0.2 0.2 0 0 16­24 24­40 40­65 Low High 16­24 24­40 40­65 Low High Informal to self-employment Formal to self-employment Probability, conditional on separation Probability, conditional on separation 0.6 0.6 0.4 0.4 0.2 0.2 Age Education Age Education 0 0 16­24 24­40 40­65 Low High 16­24 24­40 40­65 Low High Argentina Brazil Mexico Source: Bosch and Maloney 2005. 58 T H E R A Z Ó N D E S E R O F T H E I N F O R M A L W O R K E R knowledge, capital, and contacts; and then quit to open States by Evans and Leighton (1989), so the considerations their own informal businesses (cited in Maloney 1999).10 of these potential entrepreneurs are similar to those found A further disaggregation of figure 2.6 (not shown) sug- everywhere.12 Nonetheless, since small firms have high gests that older and better-educated workers spend longer mortality rates, both their owners and their workers in the in self-employment (Bosch and Maloney 2005). This is region are more likely to find themselves without employ- consistent with the mainstream firm dynamics literature ment and searching for a new job, without the broad cover- that suggests that young firms, which, ceteris paribus, are age of safety nets common in most developed countries. more likely to be opened by younger workers, have very Further insights are gained by examining transitions by high failure rates (Evans and Leighton 1989; Jovanovic gender. Figure 2.7 suggests a more dynamic corridor of 1982). The higher tenure for better-educated workers is mobility between being out of the labor force and informal similar to that found in the formal sector and the opposite self-employment for women as compared to men. While at of the pattern for the informal salaried. This topic will be first this may suggest discrimination against women in analyzed further in chapter 5. formal sector jobs, further examination reveals that this is It may be argued that both the pattern of late entry and largely driven by married women. Figure 2.8 shows longer duration with age are also consistent with exclusion: that while it is true that single mothers have high levels of a middle-aged worker who loses his or her formal sector participation in informal self-employment and wives are job is unable to find a new one, and therefore is forced into overrepresented in the unpaid sector, single women with- self-employment as a safety net. This dynamic may have out children have the highest rate of participation in formal been of particular relevance during the economic restruc- jobs of any group, male or female. Moreover, replicating turings of the 1990s, when older displaced workers may this analysis for a cohort of single women in Argentina and have found their skills obsolete and of little demand in the Mexico (the two surveys with a marital status variable), emerging sectors. The features of the observed transitions Bosch and Maloney (2005) show that the transitions of sin- mitigate against this view as being the entire or even a gle women are very similar to those of men, and this simi- large part of the story. The rate of transition into the sector larity holds up when disaggregated by duration and in all three countries seems broadly linear in age until mid- propensity. The high mobility is partly due to the fact that dle age. That is, there is not a sudden spike in propensities women have far lower tenure in self-employment, suggest- among older workers, but rather a gradual increase with an ing that these jobs are not intended to be careers, but rather inflection point at prime age when the propensity to enter transitory supplements to family income. often decreases. Overall, this pattern seems more consistent In a view with lineage to Becker's (1991) work stress- with the life-cycle pattern found in more advanced ing structural determinants of employment patterns, economies, although coexisting with some older workers Cunningham (2001a) argues that Mexican women's patterns with little access to the formal sector market once fired.11 of participation--and particularly their gravitation toward Even if the life cycle well describes the patterns of entry self-employment--are driven by their need to balance their into self-employment, it is still the case that this is a risky other responsibilities in the household; child-raising business. Not only are most independent workers out of the requires greater job flexibility than the salaried sector formal circuit of employment protection, but also operat- offers. Women with young children are often more prone to ing a small business is intrinsically riskier everywhere, with be self-employed rather than formal sector employees many firms not lasting through their first year (box 2.3). (Cunningham 2000; Cunningham and Ramos 2001; see Again, the sociological literature provides striking confir- also chapter 3), and interview data from Geldstein (2000) mation of this insight when Balán, Browning, and Jelin for Argentina and Chant (1991) from Mexico suggest that (1973) argue that "although self-employment is a goal for women may more easily balance their productive (market) many Mexican workers . . . [b]ecoming self-employed and reproductive (home care) roles if they work for them- involves a large risk, especially for those men who had sta- selves than if they are employees.13 ble and secure jobs" (p. 216). In fact, chapter 5 will show that comparisons of patterns of entry and exit derived from Patterns of mobility of the informal salaried the Mexican and Nicaraguan microenterprise surveys The patterns of transitions for informal salaried workers in appear to be broadly similar to those derived for the United small firms offer a more nuanced picture. As noted earlier, the 59 I N F O R M A L I T Y BOX 2.3 Informal self-employment: Risky and informal, or risky because informal? Many of the "precarious" characteristics associated with as participation in the numerous formal institutions: fed- informality are natural by-products of the fact that the eral and local treasuries, governmental programs such as informal self-employed resemble the fundamental charac- social security, the legal system, the banking system, teristics of the owners of a small firm. The industrial coun- health inspection, firm censuses, trade organizations, and try literature on firm behavior offers two important civic organizations. These, of course, have costs in terms findings about such firms. First, there is a wide range of of compliance with legal norms, which very small firms sizes among long-standing firms determined by factors can easily avoid in developing countries. Small firms are such as how efficient or hardworking an entrepreneur is, anchored in social networks of family and immediate how well placed is his or her firm, and the logic of the pro- neighborhood that allow them to enforce implicit con- duction process. This means that the existence of many tracts and insure against risks. However, as firms grow, small firms does not necessarily imply failure of either labor they increasingly need to secure property rights or permit or credit markets. The reason that 80 percent of microfirms formal contracting mechanisms, pool risk, and gain access have only one or two employees and tend to be family based to credit. De Soto (1989) offers a striking example in may reflect a logic rooted in the tradition of the family which informal street vendors in Peru tried to pay their farm, or possibly reflects the sustainable reach of informal taxes since this would guarantee them some property contracting relations. rights over their pitch and hence offer some security to A second finding about small firms everywhere is their investments they wanted to make. Chapters 5 and 6 will extraordinarily high rate of failure. Seeking to explain the show that firms do become more formal with age and size. U.S. case, Jovanovic (1982) argues that this is due to the Combining the two characteristics of microfirms and fact that entrepreneurs cannot know how good their loca- this notion of formality implies that small firms will have tion is or how competent they will be as entrepreneurs higher costs, are likely to be informal, and will have very until they actually start the business. Very soon after high failure rates. Although this corresponds exactly to starting, many enterprises find that they are not viable the standard picture of the stagnant, precarious, unpro- and fail. Rough calculations from the Mexican microen- tected informal worker familiar in the literature, it terprise survey suggest that these firms show high failure emerges naturally from potential entrepreneurs trying rates, but not particularly higher than those in the their luck (risk taking), often failing, and not engaging United States. in the formal institutions until they grow. Thus, the exis- Thus, most of the characteristics of the sector may be tence of the microfirm informal sector may be largely intrinsic and not imply any inferiority or undesired pre- unrelated to questions of labor market dualism or even cariousness. Levenson and Maloney (1996) treat "formality" credit market distortions. age trajectories of informal salaried workers sketched in fig- except in Argentina. Thus, the sector may represent both an ure 2.4 are almost the mirror opposite of the self-employed; it important element of disguised unemployment and a port of is the sector with the highest share of very young workers entry into the workforce, particularly for youth, older and shows a constant decline across all subsequent ages. In unskilled workers, and some women. Mexico, for instance, this translates into the average age of As noted before, workers appear to spend relatively little the informal salaried being 29, five years below that of formal time in this sector--durations are just over a year. The salaried workers, and 14 years below that of the self- brevity of tenure is similar to that found for young workers employed. Still, 40 percent to almost half of the sector is in Brazil by Sedlacek et al. (1995), and is similar to the composed of older workers in the three countries. Further- United States, where the median tenure for young workers more, figure 2.9 shows that in contrast to self-employment, 16­24 years of age is only 1.4 years, and for workers aged entry into the sector from either unemployment or OLF 25­34, the median tenure is 3.4 years (BLS 1991). More decreases with age, except in Brazil, as well as education, detailed studies of transitions also show a high degree of 60 T H E R A Z Ó N D E S E R O F T H E I N F O R M A L W O R K E R FIGURE 2.7 Transition between self-employment and out of labor force, by gender Males Females 0.003 0.008 0.007 0.003 0.006 0.002 0.005 0.002 0.004 0.003 0.001 0.002 0.001 0.001 0.000 0.000 Argentina Brazil Mexico Argentina Brazil Mexico Self-employment to out of labor force Out of labor force to self-employment Source: Bosch and Maloney 2005. FIGURE 2.8 Employment sector allocation by gender, marital status, and parental status in Mexico Females Males Percent Percent 70 70 60 60 50 50 40 40 30 30 20 20 10 10 0 0 / / aid er ormal salary wage salary wage Unpaid owner own Inf al Contract Unp mployed Informal Contract Firm Firm Unemployed orm piecework Formal Une mployedormal piecework Formal Inf Inf self-employed self-e All Wives Single mothers All Husbands Single Single, no children Source: Cunningham 2001a. mobility between school, unpaid work, and, to a lesser of this are numerous, including the relative size of the pool of extent, unemployment that suggests a pool of generally incoming young people, the difficulty of signaling the poorer workers not yet tracked into regular employment appropriateness of an applicant for a particular job, the diffi- (see World Bank 2007).14 culty of dismissing a new worker if the match appears poor, In this sense, informal salaried work has an important or exclusionary factors related to deficient schooling or dis- component that corresponds to the very high levels of crimination of unskilled and minority youth. These factors unemployment found among youth in the OECD that are translate to the Latin American context; thus, even in the often double or triple those of prime-aged males 25­54 years absence of nominal rigidities that might segment the of age (Jimeno and Rodriguez Palenzuela 2003). The causes market, both very restrictive labor legislation and poor 61 I N F O R M A L I T Y For the sizable fraction of the informal salaried who are FIGURE 2.9 prime age or older, both exclusionary and exit rationales Propensity to move to informal salaried status from different sectors apply. While the arguments of greater flexibility and higher earnings are less applicable, as noted before, the informal Out of labor force to informal salaried salaried in several countries do earn the minimum wage and Probability, conditional on separation often receive other benefits (such as vacation and extra 1.0 Age Education salaries), and may prefer not to contribute to social security, 0.8 especially if they are past the age at which they cannot accrue 0.6 enough years in the system to earn a pension. However, the 0.4 case for exclusion is also strong, particularly for the unskilled and those fired former formal sector workers whose skills 0.2 have been rendered obsolete. As chapter 4 will show, much 0 of the trend in informality in Argentina, the metropolitan 16­24 24­40 40­65 Low High area of Brazil, and over the last five years in Mexico, is due to Unemployed to informal salaried the increase in informal salaried work affecting, in particu- Probability, conditional on separation lar, groups that previously were covered by formal work con- 1.0 Age Education tracts, including prime-aged and older workers. 0.8 Finally, the low tenures observed among informal 0.6 salaried workers do not mean that this is a transitory state that the majority of workers eventually overcome. The 0.4 actual spells of informal salaried work experienced by many 0.2 workers (especially low-skilled youth and older workers) 0 may be much longer, since many spells end with unemploy- 16­24 24­40 40­65 Low High ment or dropping out of the labor force, and may revert to Argentina Brazil Mexico informal salaried employment.16 The actual time these workers can take to find a formal salaried job, if desired, may Source: Bosch and Maloney 2005. be quite long. In fact, the lifetime occupational history data in table 2.4 show that around half and 84 percent of prime- mechanisms for resolving information asymmetries can aged and older informal salaried workers in Argentina and impede young workers' entry into the formal sector. As chap- the Dominican Republic, respectively, have never held for- ter 4 will show, in Argentina and Brazil, the rise in informal mal salaried employment. In fact, one-third of 46­55-year- salaried workers had a disproportionate impact on young old informal salaried workers report being informal salaried entrants to the workforce. their entire work-life in Argentina, and nearly 60 percent in One scenario is that while in school--and just after leaving the Dominican Republic. Whether this is voluntary or school--many students work at a family business, with or involuntary, the data alone cannot tell; in fact, both the without pay, before moving on to formal salaried jobs.15 As motivational and job satisfaction responses analyzed below noted further below, in the case of Mexico, fully half of those and in chapter 3 show consistency with both hypotheses.17 working for microenterprises are children of the owner. Even for those who are not family members, informal small enter- Motivations for participation in informal work prises may train more apprentices and workers than the As noted above, informal employment may reflect workers formal education system and the mostly government job- opting out of benefits programs voluntarily or exclusion to training schemes together (Hemmer and Mannel 1989). The employment of last resort given the lack of better alterna- informal salaried experience may constitute continued school- tives. This section introduces another tool for distinguish- ing with an attendant lower wage. Still, the education­ ing between the two: asking workers directly what drives occupational profiles indicate that graduation to formal their occupational choices. Employment surveys in a hand- salaried work is unlikely for youth who drop out of school ful of countries of the region have done this in the 1990s, before completing at least a full course of secondary education. although until recently only Mexico and Brazil did so with 62 T H E R A Z Ó N D E S E R O F T H E I N F O R M A L W O R K E R TABLE 2.4 Informal employment and work-life occupational history, Dominican Republic and Argentina (percent of workers) Dominican Republic (2006) Argentina (2005) Status in current job is . . . Have ever held a job as . . . Formal salaried Informal salaried Independent Formal salaried Informal salaried Independent Workers aged 15­25 Formal salaried only 76.5 5.1 7.8 66.3 7.5 11.3 Informal salaried only 15.9 88.7 16.9 25.1 84.4 23.3 Independent only 5.9 5.1 72.8 2.2 7.2 62.4 More than one category 1.7 1.0 2.5 6.4 0.9 3.0 Total 100.0 100.0 100.0 100.0 100.0 100.0 Workers aged 26­45 Formal salaried only 65.9 12.8 12.2 56.0 27.9 29.5 Informal salaried only 17.9 69.4 23.2 25.7 50.5 19.2 Independent only 11.3 14.1 57.4 9.3 9.8 31.0 More than one category 4.9 3.7 7.2 9.1 11.8 20.4 Total 100.0 100.0 100.0 100.0 100.0 100.0 Workers aged 46­65 Formal salaried only 64.9 12.5 13.3 58.2 37.5 50.7 Informal salaried only 14.6 59.1 21.1 15.4 33.2 8.8 Independent only 16.0 24.1 61.8 16.4 7.8 23.9 More than one category 4.5 4.3 3.8 10.1 21.6 16.7 Total 100.0 100.0 100.0 100.0 100.0 100.0 Sources: Arias and Bustelo (2007), Arias, Landa, and Yanez (2007), based on household survey, 2005, 2006 data. Note: Each cell gives the percentage of workers currently in each occupational group who have held employment in the other occupational groups some time during their work-lives. broad coverage. Below we report the findings from studies results;thesedataofferfreshprimafacieevidenceonthesignifi- using these surveys. canceofthevoluntaryorinvoluntarynatureofinformalemploy- Inpreparationofthisreport,newspecialsurveysoninformal mentinavarietyofcountrycontexts. employment were collected in the main urban areas of Bolivia, Colombia, and the Dominican Republic, building on a 2005 Self-employment survey conducted in Greater Buenos Aires (see box 2.4, World Both the sociological and economics literatures offer Bank 2007b).Thesurveysincluded,amongotherthings,ques- reasons why many workers might actually prefer self- tions on the motivations or reasons for workers to be salaried or employment over salaried work. In addition to the already self-employed and to participate or not in labor benefits discussed greater flexibility, autonomy, and entrepreneurship programs (box 2.4). Workers were asked questions such as, "If motives, the literature also talks about risk taking, family you were able to choose, would you rather be a salaried or inde- tradition, and mobility opportunities. Blau (1987) and pendentworker?";"Whatis(are)themainreason(s)whyyouare Maloney (1999) argue that in poor countries, lack of asalariedemployeeratherthanindependent?"(andtheconverse human capital specific to an occupation or firm and the low question); and "What is the main reason why you do not have a formal sector productivity mean that the opportunity cost job with contributions to pension and health insurance bene- of self-employment is low. Contrary to these "exit" factors fits?" The responses were structured around key proposed dri- are, of course, the view of the sector as employment of last vers of informality in the literature. Despite some variations in resort for workers unemployable in the formal sector. questionnaire design, the responses are largely comparable The evidence below suggests that the majority of the sec- across the four countries. Therefore, below we discuss the evi- tor corresponds more to exit motives and the entrepreneurial dencearisingfromtheBrazilianandMexicansurveys,followed view in the advanced country literature, although a signifi- bytheotherfourcountries.Tables2.5and2.6 presentthemain cant component corresponds to involuntary workers.18 63 I N F O R M A L I T Y BOX 2.4 Special informal employment surveys: What can we learn from them? The World Bank, in partnership with line ministries and to the informality of the productive units and the moti- the national institutes of statistics in Argentina, Bolivia, vations to participate or not participate in formal institu- Colombia, and the Dominican Republic, supported the tions; (3) questions on informal-formal occupational implementation of special modules on informality transitions through all other categories except the current attached to regular household surveys in these countries. one ("When was the last time you've been . . . indepen- These modules were collected in the fall of 2005 in dent worker/registered worker/unregistered worker?");1 Argentina and Bolivia, and in the fall of 2006 in Colombia and (4) questions on access to social protection mecha- and the Dominican Republic. The analysis presented in nisms through private means or publicly provided bene- this volume is just a launching point for new work that fits programs other than social security. can be carried out with the rich data in these surveys. Depending on the country, the surveys covered four Note main sets of issues: (1) questions on motivations for 1. The third set of questions will allow analyzing long-term employment category choice and participation in social labor market dynamics, which has been impossible with available security and health insurance plans; (2) questions related short-term panel data. Table 2.5 shows the data from the most recent surveys in of the sociological literature. For instance, more than two- Greater Buenos Aires, and urban areas of Bolivia, Colombia, thirds of the Brazilian informal self-employed in the early and the Dominican Republic. Except for Colombia, over 1990s reported that they would not take a formal salaried 70 percent of independent workers are voluntary, in the sense job, and less than 20 percent in Mexico reported involuntary that they would rather be independent if they were able to reasons (see below).19 Moreover, over half of salaried workers choose their job. In Colombia, by a different measure, only in Bolivia and the Dominican Republic and close to one- 41 percent of urban independent workers can be considered third in Argentina and Colombia have intrinsic preferences voluntary; they reportedly would not take a formal salaried for independent work, consistent with Blanchflower and job with earnings equal to the earnings in their current job. Oswald's (1998) findings for the United Kingdom, the When asked if they would take the same formal job but with United States, and Germany. The informal salaried show lower earnings (a stricter standard), 71 percent of the Colom- somewhat stronger preferences for independent work than bian informal self-employed said they would not. These find- the formal salaried, especially in Bolivia. ings are also remarkably in line with those for Mexico and A comparison of the reported desires of self-employment Brazil based on very different surveys, as well as the findings with comparable international data shown in figure 2.10 TABLE 2.5 Preference for independent employment (percent of workers) Country All salaried Formal salaried Informal salaried All independent Formal independent Informal independent Argentina 37 38 43 70 86 60 Bolivia 55 41 62 73 65 74 Colombia 36 34 40 41 46 41 Dominican Republic 53 51 57 75 85 75 Sources: Arias and Bustelo (2007), Arias, Landa, and Yanez (2007), based on household survey, 2005, 2006 data. Note: Independent workers include self-employed and the small fraction of microfirm owners in each country. Except for Colombia, percentages are based on responses to the question, "If you were able to choose, would you rather be a salaried or independent worker?" For Colombia, the figures are based on (1) independent workers reporting they would not take a salaried job with bene- fits and the same earnings they get as independent; and (2) salaried workers reporting the main reason for being in their current job was their inability to become independent. 64 T H E R A Z Ó N D E S E R O F T H E I N F O R M A L W O R K E R gives further credence to these findings.20 First, the rates few individual socioeconomic and demographic character- of desired self-employment are not out of line with the istics to be good predictors of self-employment desires. average in this sample of countries. Second, the four Latin This suggests that entrepreneurship motives have a large American and Caribbean countries considered (the only ones idiosyncratic component, such as the reasons enumerated with such data) are among those with the lowest divergence above, and that most workers found in this sector are likely between desired and actual self-employment rates. That is, to show strong preferences for its characteristics. self-employment rates in these four Latin American and Nevertheless, table 2.5 shows that a significant fraction Caribbean countries are largely in line with workers' inclina- of the self-employed is involuntary and conforms to the tra- tions to independent work. In a sense, many more workers ditional view of the sector. In particular, the fraction of become self-employed than what one would expect, given the informal independent that would rather be employees is credit and human capital constraints of these countries. This roughly 40 percent in Argentina and 25 percent in Bolivia may be a result of a host of factors, including the low oppor- and the Dominican Republic, while a sizable 59 percent of tunity cost of salaried work (formal or informal) and rigidities the Colombian informal self-employed would rather take that create rationing of salaried jobs. In countries such as a salaried job with benefits over their current job. Again, Colombia and Bolivia, this may lead many workers who lack by similar measures, in Brazil roughly one-third of the the qualities to succeed as entrepreneurs to end up in self- informal self-employed (45 percent among women) are employment; as noted earlier, the importance of tradition and involuntary.21 the family business enters as an important consideration here. The results are further corroborated by the reported A noteworthy pattern of the data is the lack of a clear motivations for actually engaging in independent employ- correlation between desires of self-employment and their ment. Table 2.6 first presents the aggregate responses of for- materialization with level of development. While the mal and informal independent workers in the most recent United States has one of the highest rates of want-to-be- surveys. Where workers were asked to report more than one but-immaterialized entrepreneurship, Japan appears in the reason, the table also shows the second-most important reason other extreme. In fact, using these data, Blanchflower or the relative frequency of total responses. The fraction of (2004) and Blanchflower, Oswald, and Stutzer (2001) find informal independent workers who report loss or difficulties FIGURE 2.10 International comparison of desired and actual self-employment rates Percent 90 60 People preferring self- Gap between desired and 80 employment (left scale) actual (right scale) 50 70 60 40 50 30 40 30 20 20 10 10 0 0 na Colombia BoliviaJapan Spain Italy gary rland many States Norway IsraelSweden gdom France Hun Argenti Denmark Kin CanadaZealandBulgariaPortugalSloveniaPoland Germany Ger ted Netherlands n FederationRepublicRepublic Switze Czech New East Uni West United RussiaDominican Sources: Based on data from Blanchflower 2004; Blanchflower, Oswald, and Stutzer 2001; and author's estimates for Latin American and Caribbean countries, using labor survey data, 2005 and 2006. 65 I N F O R M A L I T Y in obtaining a salaried job as their primary motivation is differences in motivations by gender and age. Most workers 25 percent in Bolivia, and 44 percent in the Dominican in the two countries report reasons that imply the sector is a Republic, although it emerged as the second-most impor- source of valued jobs, including higher earnings, greater flex- tant reason for only 8 percent of cases in the latter. Lack of ibility, or family and personal motivations. In Mexico, 44 per- salaried employment was cited among the main reasons by cent of males said they entered the sector for higher earnings, 59 percent of the informal self-employed in Argentina and and just under 17 percent said they entered because they by 55 percent in Colombia, and accounted for 43 and either lost a previous job or could not find a salaried formal 35 percent of their responses, respectively. All the remain- job. Only 7 percent of women appear to be rationed out of ing responses correspond to voluntary reasons in nature, salaried jobs. In Brazil, almost two-thirds of men and 44 per- such as higher earnings or mobility opportunities, greater cent of women reported that they were happily in the sector. autonomy, flexible hours, family tradition, or having the These findings are consistent with smaller sociological opportunity to become independent, and a variety of other surveys that track workers across time. For instance, in reasons (for example, marriage, studying, the job being Monterrey, Mexico, Balán, Browning, and Jelin (1973) find one's trade), and age (in Colombia). The rates of involuntary that being one's own boss was well regarded and that move- motivations among the formal self-employed compared to ments into self-employment from salaried positions often the informal are much lower in Argentina, somewhat lower represented an improvement in job status. Figure 2.11 in the Dominican Republic, and similar in Bolivia lendscredencetotheseresponses.TrackingMexicanworkers and Colombia. from formal salaried jobs into informal self-employed jobs, The results are further corroborated by the earlier sur- on average voluntary movers gained around 15 percent veys in Mexico and Brazil. Tables 2.7 and 2.8 show the while those who entered because of losing a formal job did responses in these countries further disaggregated to gauge not experience a statistically significant change in income. TABLE 2.6 Distribution of the motivations/reasons for being in the current job as an independent worker (percent) Argentina Bolivia Colombia Dominican Republic Cited as Percent of all Main Cited as Percent of all Main Secondary Motivation/reason relevant responsesa reason relevant responsesa reason reason Formal independent Could not find job as salaried 19.6 9.6 30.8 41.4 35.0 32.5 9.0 Autonomy/no boss 29.5 6.6 13.5 20.3 10.9 16.1 31.5 Flexible hours/less responsibility 34.3 7.0 8.3 23.9 17.6 10.7 24.5 Family tradition 21.5 11.3 6.8 2.3 2.5 3.7 4.1 Earns more n.a. n.a. n.a. 20.3 5.2 26.4 5.3 Better mobility/benefits/prospects 59.4 31.3 n.a. 14.9 4.3 8.4 20.4 Able to become independent 66.8 25.8 24.9 n.a. n.a. n.a. n.a. Age n.a. n.a. n.a. 21.1 16.5 n.a. n.a. Other 10.2 8.5 15.8 7.7 8.0 2.3 5.3 Informal independent Could not find job as salaried 58.8 43.0 25.3 55.3 34.5 44.3 8.1 Autonomy/no boss 16.5 3.2 9.5 22.2 16.4 16.2 33.5 Flexible hours/less responsibility 30.1 6.1 13.3 21.7 11.5 10.2 22.0 Family tradition 11.7 4.2 17.4 4.4 2.8 3.8 9.2 Earns more n.a. n.a. n.a. 11.5 9.7 14.0 7.8 Better mobility/benefits/prospects 29.6 15.0 n.a. 7.4 5.7 7.8 12.4 Able to become independent 43.0 15.9 26.0 n.a. n.a. n.a. n.a. Age n.a. n.a. n.a. 23.2 13.5 n.a. n.a. Other 15.5 12.6 8.5 10.8 5.8 3.7 6.9 Sources: Arias and Bustelo (2007), Arias, Landa, and Yanez (2007), based on household survey, 2005, 2006 data. Note: n.a. = not applicable. In Argentina, individuals could mark all relevant response options; in Colombia, they were asked for the two most important reasons, but without distinguishing first and second, unlike the case of the Dominican Republic. a. Computed as the ratio of the frequency with which each reason is reported to the total number of valid responses in the sample. 66 T H E R A Z Ó N D E S E R O F T H E I N F O R M A L W O R K E R TABLE 2.7 Reported reasons to be informal self-employed in Mexico (percent of workers) Male Female Reason Alla 15­18 years 19­24 years 25­54 years 55­70 years Alla 15­18 years 19­24 years 25­54 years 55­70 years Family tradition 8.9 6.0 17.4 8.3 10.0 3.6 9.8 8.4 2.8 4.5 Complement family income 18.0 40.2 13.9 17.1 20.9 65.1 21.3 59.4 64.5 74.7 Higher income than salaried 43.7 15.2 41.8 47.0 31.5 12.2 7.5 9.9 13.1 6.9 No salaried job 12.4 17.9 8.3 11.2 18.3 6.1 22.4 5.0 5.7 8.0 Flexible hours 3.8 9.4 9.1 3.4 3.3 4.2 11.5 9.0 4.6 1.2 Fired or lost job 4.0 0.3 0.8 4.2 5.5 0.6 0.0 0.3 0.8 0.2 Other 9.1 11.1 8.7 8.8 10.6 8.1 27.5 8.0 8.6 4.5 Percent of sample 68.1 0.7 4.6 49.4 11.3 31.9 0.2 2 23.6 5.1 Percent of gender 100.0 1.0 6.7 72.6 16.5 100.0 0.7 6.2 73.8 16.0 Source: Author's estimates, based on Encuesta Nacional de Micronegocios 1994. a. Considers all self-employed (including people under 15 years of age and above 70) who respond to this question. TABLE 2.8 Reported reasons to be informal self-employed in Brazil (percent of workers) Would you like to leave [your current] job for a job with a signed work contract? (Self-employed) Male Female Response Alla 15­18 years 19­24 years 25­54 years 55­70 years Alla 15­18 years 19­24 years 25­54 years 55­70 years No 67.9 29.9 52.6 68.3 80.6 55.5 24.6 41.8 55.2 71.2 Motivations to prefer an unprotected job (Self-employed) Male Female Reason Alla 15­18 years 19­24 years 25­54 years 55­70 years Alla 15­18 years 19­24 years 25­54 years 55­70 years Earn more in current job 18.0 13.4 17.6 21.0 9.6 10.6 5.1 13.3 12.1 3.9 Needed to care for home 0.2 0.0 0.0 0.1 0.4 26.9 15.3 22.8 27.5 28.8 Need time for other 2.9 7.5 3.2 2.5 3.2 6.7 6.8 7.9 6.6 6.8 activities Happy in current job 64.9 69.5 68.3 64.0 67.6 44.1 59.3 47.7 44.8 39.0 Did not want the 10.1 8.0 9.8 10.0 10.6 7.5 8.5 7.6 6.9 9.8 commitment No answer 4.0 1.6 1.1 2.2 8.6 4.1 5.1 0.8 2.2 11.6 Percent of sample 73.5 0.8 5.0 51.3 13.7 26.5 0.3 1.7 20.0 3.9 Percent of gender 100.0 1.1 6.8 69.8 18.6 100.0 1.0 6.3 75.4 14.8 Source: Author's estimates, based on Pesquisa Nacional por Amostra de Domicílios 1990. a. Considers all self-employed (including people under 15 years of age and above 70) who respond to this question. The results also confirm the advantages of self-employ- 19, above 60 percent compared to roughly 15­20 percent for ment for many women. In Mexico, almost two-thirds of men. In Brazil, one-third of women alluded to the need for women in self-employment reported the need to supplement flexibility to attend to family and personal activities. family income, and only 7 percent appear to be rationed out Roughly 25 percent of Brazilian women over the age of 19 of salaried jobs. Unlike men, Mexican women show a sharp say they entered self-employment to care for the home, com- increase in responding that they have entered self- pared to virtually none of the men. This correlation of rising employment to complement family income after the age of self-employment with moving into marriage age is also 67 I N F O R M A L I T Y could not find a formal job. But this steadily rises with age FIGURE 2.11 to 15 percent for males aged 25­54 and finally to 24 percent Earnings gain from voluntary movement to informality in Mexico for male workers aged 55­70. Involuntary entrants con- % hourly wage gain for males entering informal self-employment tinue to represent a minority of entrants, but there is some from salaried 40 evidence of the sector serving increasingly as a sector of last 35 resort for older workers. In Brazil the trend is exactly the 30 reverse, however; the oldest age group (55­70) reports 25 being least interested in finding a formal salaried job at 81 per- 20 cent. This is consistent with many of these workers not 15 interested in having a covered job if they can no longer 10 assure sufficient years of formal experience for a pension. 5 0 The age and gender breakdowns of the survey responses for 5 the other four countries (not shown) studied are broadly 10 consistent with the results for Mexico and Brazil, with dif- More Higher pay Involuntary ferent gradients across countries. independence Share of respondents Median earnings differential Informal salaried work Although many of the considerations for why workers Sources: Encuesta Nacional de Micronegocios, linked to Encuesta Nacional de Empleo Urbano, 1992; Maloney 1999. might prefer to be informal apply to the informal salaried as well, overall this sector shows a much larger component of involuntary entry consistent with the exclusion view. consistent with the effect of having coverage through their Until now, the information on the informal salaried was spouses, as suggested by the results of Galiani and much scarcer; the informality modules conducted in the Weinschelbaum (2006). Again, a large part of home-based four countries above included special components designed work may also emerge from this dynamic. In no way does the precisely for them. evidence here intend to validate the division of labor in the Table 2.9 suggests that, in contrast to the self-employed, household, but simply to suggest that women's patterns of the majority of informal salaried workers appear to be invol- employment are consistent with their pressing need to bal- untarily in their jobs, although not necessarily queuing for ance family and work-life--whether or not this is largely a formal salaried employment. The inability to find a better socially ascribed gender role. job constitutes a much higher fraction of the reported rea- Furthermore, tables 2.7 and 2.8 offer supporting evi- sons for being in informal salaried jobs than for the formal dence to the life-cycle nature of self-employment. Of the salaried: 48.4 versus 22.4 percent in Argentina; 64 versus very few (1 percent) of male self-employed in the age range 32 percent in Bolivia; 43 versus 16 percent in Colombia; of 15­18 in Mexico, most are entering to complement fam- and 40 versus 22 percent in the Dominican Republic. These ily income and almost 30 percent cite lack of salaried jobs. are consistent with responses from the Brazilian informal For Brazil, of the similarly small fraction of self-employed salaried in 1990 (table 2.10) that roughly 70 percent would in the same age group, about 70 percent would prefer a for- rather have had a formal salaried job, although some care is mal salaried job, unlike the older age groups that increas- in order since, as chapter 4 will show, this was a time of ingly prefer self-employment.22 In sum, self-employment extreme segmentation in the Brazilian labor market. is not a viable or good alternative for most youth to the Although the traditional queuing view enjoys substantial extent they still lack the experience and capital to have a support, some clarifications are in order. While higher than chance to succeed as entrepreneurs. among the formal salaried, the inability to find a better for- Finally, the motivations of Brazilian and Mexican self- mal salaried job is not the overwhelming majority response employed offer mixed evidence on the rationing out of mid- among informal salaried workers. For instance, in the case of dle-aged and older workers from formal salaried jobs. In the Dominican Republic, only 17 percent cite this when Mexico, only 9 percent of prime-age males 19­24 years asked about the second main reason. This explains why, old report entering the sector because they either lost or as will be shown in chapter 3, the 18-percentage-point 68 T H E R A Z Ó N D E S E R O F T H E I N F O R M A L W O R K E R TABLE 2.9 Distribution of the motivations/reasons for being in the current salaried job (percent) Argentina Bolivia Colombia Dominican Republic Cited as Percent of all Main Main Main Secondary Motivation/reason relevant responsesa reason reason reason reason Formal salaried Could not find another job 50.7 22.4 31.7 16.2 21.5 9.4 Independent work is unstable 42.7 18.9 11.3 13.3 24.4 15.7 Less responsibility n.a. n.a. n.a. 0.7 1.0 5.0 Earns more than as independent 29.3 12.9 15.1 24.8 16.8 26.9 Better mobility/benefits/prospects 69.3 30.6 n.a. 8.5 16.7 22.8 Unable to become independent 25.5 11.2 15.0 33.7 17.4 17.3 Other 8.9 3.9 26.9 2.9 2.3 3.0 Informal salaried Could not find another job 87.3 48.4 64.2 43.0 39.8 16.8 Independent work is unstable 23.5 13.0 4.5 6.4 20.0 13.1 Less responsibility n.a. n.a. n.a. 0.8 1.6 8.6 Earns more than as independent 17.9 9.9 4.8 3.9 5.9 13.7 Better mobility/benefits/prospects 17.4 9.6 n.a. 2.3 7.5 13.9 Unable to become independent 26.1 14.5 15.7 40.0 22.3 29.1 Other 8.1 4.5 10.8 3.7 3.0 4.9 Sources: Arias and Bustelo (2007), Arias, Landa, and Yanez (2007), based on household survey, 2005, 2006 data. Note: n. a. = not applicable. a. Computed as the ratio of the frequency with which each reason is reported to the total number of valid responses in the sample. TABLE 2.10 Reported reasons to be informal salaried in Brazil (percent of workers) Would you like to leave [your current] job for a job with a signed work contract? (Informal salaried) Male Female Response Alla 15­18 years 19­24 years 25­54 years 55­70 years Alla 15­18 years 19­24 years 25­54 years 55­70 years No 30.3 22.8 25.0 31.8 55.3 37.4 25.5 28.5 43.5 70.1 Motivations to prefer an unprotected job (Informal salaried) Male Female Reason Alla 15­18 years 19­24 years 25­54 years 55­70 years Alla 15­18 years 19­24 years 25­54 years 55­70 years Earn more in current job 6.2 3.1 7.6 10.6 2.4 2.2 1.7 3.0 2.6 0.8 Needed to care for home 0.2 0.2 0.0 0.1 0.0 18.2 4.7 9.5 25.7 24.3 Need time for other 7.1 8.5 5.1 3.5 4.4 6.7 8.7 5.3 5.4 3.7 activities Happy in current job 67.1 74.1 74.7 68.7 57.1 60.3 73.6 71.6 55.6 47.6 Did not want the 9.8 6.6 8.3 10.8 16.6 7.8 6.7 7.7 7.1 13.7 commitment No answer 9.7 7.6 4.3 6.2 19.5 4.8 4.6 2.9 3.6 9.8 Percent of sample 56.0 9.7 9.6 20.3 5.9 44.0 6.1 6.8 22.3 4.2 Percent of gender 100.0 17.3 17.2 36.2 10.5 100.0 13.8 15.4 50.8 9.4 Source: Author's estimates, based on Pesquisa Nacional por Amostra de Domicílios 1990. a. Considers all informal workers (including people under 15 years of age and above 70) who respond to this question. 69 I N F O R M A L I T Y difference does not prevent these workers from reporting that may be better paid to enter (or return to) domestic similar levels of welfare and job satisfaction than formal service, citing flexibility, in terms of work hours and pay employees. Further, while similar responses are not available schedules, and security in terms of a source of emergency for Mexico, rates of on-the-job search, a measure of dissatis- income, networks to other sources of employment, and a faction, are only somewhat higher than for the largely general feeling of connection with the employer (Geldstein voluntary self-employed.23 2000). However, it is also reported that this is a vulnerable More important, there are numerous indications that job, in which employer abuse, isolation, few chances for while a majority of informal salaried workers in Argentina organization, and limited opportunities for career advance- may be in fact queuing for formal salaried employment, in ment are a reality. the other countries many lean more toward independent The dissatisfaction of younger informal salaried workers employment, most of which is informal. As shown in does reveal that few see the sector as a career job. Only 2 per- table 2.9, the inability to become independent accounts for cent of informal salaried workers in Colombia, 5 percent in 15 percent of their reported reasons for being in their cur- Bolivia, roughly 10 percent in the Dominican Republic, and rent job in Argentina and Bolivia, similar to the formal 17 percent in Argentina consider their job as preparation for salaried. In contrast, in Colombia and the Dominican formal salaried employment or as offering better opportuni- Republic, these represent about 40 percent of all responses, ties for social mobility. These are broadly corroborated by the about seven points higher than for the formal salaried. responses from the Dominican Republic of unpaid workers Moreover, table 2.5 shows that the informal salaried have where roughly 22 percent see their unpaid work, for 73 per- higher rates of overall preference for self-employment than cent their first job, as preparation for the future. This is only the formal salaried: in Bolivia (62 versus 41 percent); the partially consistent with Hemmer and Mamel's (1989) Dominican Republic (57 versus 51 percent); and in asserted role of training in the sector. In any case, consistent Argentina (62 versus 57 percent). Finally, the majority of with previous results for Mexican enterprises, table 2.11 "other reasons" reported by the informal salaried for being shows that most unpaid workers are related to the owner and in their current job are related to the exercise of their trade, paid in kind, and that helping out the family is the over- apprenticeships, and flexibility. whelming motivation for working without pay. Finally, there are clear differences across demographic In sum, while the informal salaried have a majority com- characteristics in the view of being in the sector. Table 2.10 ponent of discontented workers corresponding to the more breaks down motivations by age and gender for the Brazilian classic queuing view of the sector, it is also a heterogeneous informal salaried. Younger workers show the higher discon- sector, and the magnitude of each subcomponent would tent with being informal salaried at about 80 percent. Mean- vary across countries and time. All the observations made while, half of male workers and 70 percent of females close earlier are relevant here. The sector comprises young workers to retirement (55 and above) show no inclination to move to a formal salaried job. This may arise because of quality of life issues--formal jobs being excessively demanding for older TABLE 2.11 individuals (Gonzalez de la Rocha 1994; Lorenzen, Selby, Principal reason for working without monetary compensation in the Dominican Republic (percent of workers) and Murphy 1990), the realization of the limited use in further contributing to pensions, or perhaps now being able to take advantage of the family benefits enjoyed by their Reason Distribution prime-age children in the formal sector. Learning a profession 8.5 Even among prime-age workers, 44 percent of females Helping family 62.4 age 25­54 are inclined to stay as informal salaried and Getting experience to get a better job 13.4 Paying a debt to the owner of the firm 1.3 among these, again flexibility to attend family life gains Compensated with food, housing, or apparel 5.1 importance in marriage age (25 percent). Another logic Compensated with partial or full payment 2.8 for studies applies to domestic workers, a main component of this Other 6.5 group, although with nagging tradeoffs. Interviews with poor working-class Argentine women reveal that it is not Source: Authors' estimates, based on employment survey uncommon for unskilled women to leave formal sector jobs data, 2006. 70 T H E R A Z Ó N D E S E R O F T H E I N F O R M A L W O R K E R possibly rationed out by labor and regulatory distortions, 42 percent of the informal salaried cited their unawareness and those entering the labor force, shopping around for jobs about how private pensions work as the prime reason for and being frustrated by the lack of a track record and infor- being out of the social security system; one-quarter said mation asymmetries in the labor market, and some women their employers did not offer the benefit, and another quar- who find more flexibility in microfirm salaried jobs. Again, ter said their income was too low. In Colombia, 30 percent labor or regulatory distortions that curtail formal salaried reported that their employers did not offer the benefit or employment are not the only factors that may be behind the that most available jobs are like this; 56 percent said their largely involuntary nature of the sector. The final section of income was too low, and only 10 percent reported reasons this chapter examines more closely the reasons that cause that reflect an explicit opting out. In the Dominican workers in both informal sectors to be without old-age and Republic, 42 percent reported similar employer or labor health protections in their jobs. market constraints (including 5 percent unaware of their pension rights); 23 percent alleged insufficient incomes, but Workers' motivations to lack old-age one-third actually cited voluntary reasons. In all four coun- and health protection tries, only a small fraction (5­7 percent) said that the main When asked explicitly about the reasons for not having pen- reason for not being affiliated was low expected benefits or sion and health insurance benefits, a large fraction of infor- lack of trust in the social security system. mal salaried workers report relatively external constraints. Likewise, most of the informal self-employed alluded to low As shown in table 2.12, when the informal salaried in incomes or lack of knowledge as the main reason for not con- Argentina were asked whether they lack social security ben- tributing to the pension system. Insufficient income to afford a efits because their employers did not offer them or because pension plan is reported by three-fourths of independent they reached a consensual agreement to obtain higher earn- workers in Argentina and Colombia, half in the Dominican ings, 95 percent said it was the former reason. In Bolivia, Republic and one-quarter in Bolivia. Meanwhile, 55 and 20 faced with a larger number of response options, about percent of the informal self-employed in Bolivia and the TABLE 2.12 Main reasons why the informal do not contribute to social security (percent of workers) Argentina Bolivia Colombia Dominican Republic Salaried Salaried Salaried Salaried Reason worker Independent worker Independent worker Independent worker Independent Too young, too n.a. n.a. n.a. n.a. 3.8 2.4 5.0 3.4 far in the future Prefers higher earnings 4.5 1.7 n.a. n.a. 1.7 1.3 11.8 1.8 Income too low n.a. 76.0 25.4 26.0 55.8 74.8 22.9 51.2 Not worth it n.a. 3.0 n.a. n.a. n.a. n.a. 2.9 1.9 Employer only 95.5 n.a. 23.7 n.a. 11.8 0.6 21.0 n.a. offered this job Lack of trust n.a. 4.3 6.9 15.4 2.0 6.2 2.4 5.6 Jobs are like that n.a. n.a. n.a. n.a. 4.6 1.8 15.7 n.a. Unaware of right/ n.a. n.a. n.a. n.a. n.a. n.a. 5.2 19.4 could contribute Employer does not n.a. n.a. n.a. n.a. 13.4 1.3 n.a. n.a. require it Children, spouse n.a. n.a. n.a. n.a. 0.3 2.1 0.6 10.1 will look after Prefers own savings n.a. n.a. n.a. n.a. 0.8 2.5 4.3 2.3 Don't know how n.a. n.a. 42.1 55.3 n.a. n.a. n.a. n.a. system works Other n.a. 15.0 1.9 3.3 5.8 6.9 8.3 4.5 Sources: Arias and Bustelo (2007), Arias, Landa, and Yanez (2007), based on household survey, 2005, 2006 data. Note: n.a. = not applicable. 71 I N F O R M A L I T Y Dominican Republic, respectively, alleged lack of knowledge of the benefits and services mandated in formal labor con- about the functioning of private pensions as the prime reason tracts. As will be discussed in chapter 8, other survey evi- for not contributing to social security. Among the 20­30 per- dence reveals that trust and satisfaction with public cent of the informal self-employed who reported explicit vol- institutions are very low, and in some cases are declining, in untary reasons, only 6­8 percent in Colombia and the these and other countries in the region, and that this issue Dominican Republic and 15 percent in Bolivia alluded to low may cultivate a social norm of noncompliance with taxes expected benefits or lack of trust in the social security system and regulations. While the survey responses indicate that a as the main reason not to be affiliated. Most voluntary reasons perceived inadequacy of benefits is not a first-order driver of reported in Colombia and the Dominican Republic were workers' informal status, they cannot tell us whether this relying on their spouses or children to cover their living factor could be a secondary but significant driver of infor- expenses in old age, being too young to be concerned about mality (for example, if income and informational con- this, and a preference to earn more now or save on their own. straints were not binding). However, a small fraction (10­30 percent) of those in jobs Table 2.13 shows that the reported reasons why the infor- not contributing to pensions in all four countries reported mal salaried lack health insurance coverage in Colombia and that they were using other means (e.g., savings, investment the Dominican Republic (the only two countries that col- in assets) to prepare for old age. lected such data) reflect a much larger degree of voluntary Some caution is in order in approaching these responses. choice. About 82 percent of salaried workers without pen- First, it is hard to distinguish true credit constraints from sion benefits in Colombia and three-fourths in the Domini- classic myopia or inertia in the mainstream pension litera- can Republic do not have health insurance coverage on their ture. Turner and Verma (2007) show that of those who were own, in contrast to the almost full coverage of the formal eligible to participate in 401K plans in the United States but salaried. In the two countries a small fraction of informal did not, 40 percent responded that they could not afford to. salaried workers (20 percent in Colombia, 11 percent in the Yet these are workers earning several orders of magnitude the Dominican Republic) reported the employer not offering income of Latin American workers. Further, automatic the benefit as the main reason, and again, a significant enrollment substantially increased their participation; many fraction (23 and 46 percent, respectively) cited very low allegedly income-constrained workers did not opt out, thus incomes as the main reason. suggesting that for many the contributions were, in fact, Among those reporting clear opting-out reasons, 44 per- quite manageable. The explanation for this is some mixture cent in Colombia said they have the benefit through of inertia and myopia. Whether, in fact, informal workers in another family member or the public subsidized regime, the countries surveyed here or elsewhere in the region would and 30 percent in the Dominican Republic reported a host do the same, or whether they would still opt out into infor- of reasons ranging from preferring to cover health costs on mality, we cannot know. However, in several countries of the their own (10 percent), having the benefit through another region, including Chile, Uruguay, and to a lesser extent family member (6 percent), and relying on the public today Argentina, many low-income formal salaried workers health system (7 percent). In both countries, less than are able to contribute to pension plans. In fact, as will be 5 percent reported low coverage or quality of health ser- shown in chapter 3, the Dominican self-employed have aver- vices as the main reason not to be affiliated. age earnings similar to those of formal salaried workers, sug- Low incomes and the preference for other alternatives gesting that, in principle, contributions to the system should also emerged as the main reported reasons of independent be as affordable. Thus, the response of "lacking income" in workers in Colombia and the Dominican Republic to these surveys cannot be taken at face value and should be lack health insurance coverage. Between 12 and 14 per- plausibly viewed as voluntarily opting out of old-age sav- cent of independent workers without pension plans in ings, combining a mixture of unaffordability due to eco- Colombia and the Dominican Republic had other types nomic hardship, inertia, and myopia. of health insurance coverage, compared to three-fourths Second, although low expected returns and trust in the and two-thirds of the formal self-employed, respectively. pension system do not appear prominently as main reasons, Around half of the informal self-employed in the this does not negate that many workers in these countries Dominican Republic and one-third in Colombia said might in fact have concerns with the quality and reliability they do not have health insurance because they could not 72 T H E R A Z Ó N D E S E R O F T H E I N F O R M A L W O R K E R TABLE 2.13 Main reasons why workers do not contribute to health insurance (percent of workers) Colombia Dominican Republic Informal Formal Informal Informal Formal Informal Reason salaried independent independent salaried independent independent Health service inefficient/ 0.4 1.1 1.0 1.1 1.0 1.0 low coverage Health service low quality 0.4 4.3 0.6 1.6 0.0 1.1 Beneficiary through relative 44.0 57.9 53.7 6.0 19.0 5.6 Prefers higher earnings 1.3 0.5 0.5 2.6 0.0 1.1 Employer does not offer it 9.9 0.4 0.6 6.7 1.1 0.4 Employer does not require it 10.0 0.5 1.2 4.7 0.0 0.8 Don't know how to enroll 0.6 0.1 1.0 4.0 11.5 7.7 Too low incomes 23.4 11.7 33.2 45.5 41.5 49.1 Temporary situation 2.3 0.4 0.9 4.6 1.8 2.8 Prefers self-coverage 1.1 7.6 1.9 10.2 12.4 14.2 Has private insurance 0.3 4.8 0.7 n.a. n.a. n.a. Attend public hospitals n.a. n.a. n.a. 6.7 7.1 11.5 Insurance companies rejection n.a. n.a. n.a. 0.5 0.0 0.5 Other 6.2 10.9 4.6 5.8 4.6 4.2 Sources: Arias and Bustelo (2007), Arias, Landa, and Yanez (2007), based on household survey, 2005, 2006 data. Note: n.a. = not applicable. afford it, compared to 42 and 12 percent of formal self- age does not prevent informal workers in the Dominican employed, respectively. Republic from reporting levels of job satisfaction equal to About 54 percent of the Colombian informal self- those of formal salaried employees. employed and 58 percent among formal cited having the The policy implications of the reported significance of benefit through another family member or the public sub- low incomes for low pension and health insurance coverage sidized regime. The Dominican informal self-employed depend on the alternative interpretations. If workers are reported a host of reasons, including preferring to cover unable to save for their retirement because they are forced to health costs on their own (14 percent), reliance on the pub- spend their incomes entirely on necessities, forcing them to lic health system (12 percent), lack of information on how save (through mandatory savings programs such as social to enroll (8 percent), and being a beneficiary of another security) may reduce their welfare during working years, family member (6 percent). These are very similar for the and, therefore, well-targeted social insurance pillars would formal self-employed, except for their heavier reliance on be more advisable. However, to the extent that this behavior coverage through relatives. In both countries barely 2 per- partially reflects myopia or inertia in the savings behavior of cent said they were not affiliated due to low coverage or some workers for retirement, there is an important role for quality of health services. programs that mandate or provide incentives for voluntary Again, considerations similar to the above apply to the savings. In any case, whether workers end up with insuffi- interpretation of the "low income" responses and the low cient savings for old age because they consume "excessively" fraction of workers citing overriding concerns with the during their earnings years or because they have low long- quality of health services. In this case, risk aversion behav- term income potential, there is some justification for some ior becomes relevant. The fact that many workers resort to government intervention to influence individuals' savings self-coverage or are happy with the protection offered by and retirement decisions or to provide minimum protection universal health services suggests that these are judged from poverty at old age by pooling efforts through the tax to be adequate as compared to privately provided health system. These issues will be discussed in detail in chapter 7. insurance plans considering the cost of the latter. In fact, as Furthermore, the significance of information failures and chapter 3 will show, the lack of pension and health cover- the availability of alternatives to substitute for mandated 73 I N F O R M A L I T Y benefits, either publicly provided or through private means, salaried in the other countries leans more toward indepen- require policy attention. It is imperative to ensure a good dent employment, most of which is informal, as a desired design of social protection and social assistance systems to destination. minimize distorting individuals' employment behavior, as The next chapter revisits this question and the more gen- will also be discussed in chapter 7. eral question of the role of the different sectors by studying Finally, to the extent that the decisions of firms to oper- whether workers' stated voluntary or involuntary motiva- ate informally are important drivers of the informality of tions to be in informal employment conform to their earn- salaried workers, as will be discussed in chapters 4 through ings performance and, ultimately, their welfare. Chapter 4 6, improved design and enforcement of labor, tax and busi- on labor dynamics will introduce additional evidence ness regulations, and incentives for firms to find benefits in exploiting transitions and patterns across the business cycle participating in formal institutions have an important role to more fully flesh out the nature of the informal sector. to play in addressing informal employment. Notes Conclusions 1. The Harris-Todaro (1970) model is perhaps the traditional This chapter first laid out several conceptual and empirical statement of this view. Also see Fields 1990; Peattie 1987; Portes and approaches to understanding the razón de ser of the informal Schauffler 1993; Tokman 1992; and Turnham and Eröcal 1990. worker. It then brought to bear two sets of empirical tools 2. See also Krebs and Maloney (1999) for an efficiency wage to begin to distinguish among these approaches. The first model applied to Mexico. 3. Schneider and Enste (2000) linked Hirschman's "exit" and approach moved beyond simple tabulations of stocks of "voice" to the causes of the informal economy, but their emphasis was workers in different sectors to studying the dynamics of on opting out driven by the desire to avoid dealing with excessive their movements through the labor market, and what these state burdens, rather than on the perceived inadequacy of the benefits sectoral transitions can tell us about why workers are where that formality entails. they are. The second analysis relied on the reasons workers 4. Fields (1990) argued for two tiers, with a minority upper tier report for being in their current job with or without social that is voluntary and prosperous. Yamada (1996), Maloney (1999), Saavedra and Chong (1999), and Mondino and Montoya (2002) have protections. argued with evidence from Peru, Mexico, Brazil, and Argentina that In the cases of both the informal salaried and self- the majority of the informal self-employed are voluntary. employed, there is evidence of substantial heterogeneity of 5. The interviews by sociologists Balán, Browning, and Jelin motives and demographic characteristics. However, the (1973) in Mexico suggest that the paucity of openings for promotion characters of the two sectors emerge as very distinct, lend- on the rigid escalafón (seniority-driven job ladder) can make informal ing credence to both the exclusion and voluntary nature of self-employment the remaining outlet for further advancement. 6. In fact, even in the United States, some workers leave large informal employment. firms to open their own, taking on the responsibility for health care Evidence from workers' patterns of mobility and reported and pensions through private or independent means. motivations to be in their current occupations lends support 7. The fraction of households receiving financial help from rela- to the view that the majority of independent workers are tives in other households is 38 percent in Spain and 11 percent in largely voluntary and attach significant value to the nonpe- Italy versus 1 percent in the United Kingdom and 6 percent in the cuniary benefits of autonomous work (although an important United States. 8. Maloney (1999) standardized probabilities both on initial fraction conforms to the traditional view of the sector as turnover and on final sector size. Bosch and Maloney (2005) standard- employment of last resort). Self-employment is concentrated ized on initial separations. among older workers and is not an entry point of work. 9. As one example, broadly replicated in the other surveys, the Meanwhile, the majority of informal salaried workers National Urban Employment Survey (ENEU) from Mexico shows appear to be involuntary, although there is also significant that transitions into self-employment from the other paid sectors heterogeneity of characteristics (most are young) and occur four to six years later than transitions into formal or informal salaried work, leaving the mean age eight years higher than the next motives. This may not necessarily imply that they are closest sector (see Maloney 1999; Maloney and Aroca 1999). queuing for preferred, better-quality formal salaried 10. See Lopez-Castaño (1990) in Colombia, and Fields (1990) employment; although this seems to be the case for a and Peattie (1982), who find a tendency for employees of large firms majority in Argentina, a significant fraction of the informal to leave and open their own businesses. Aroca and Maloney (1997) 74 T H E R A Z Ó N D E S E R O F T H E I N F O R M A L W O R K E R find confirmation in a logit analysis for Mexico with rotating 18. These results are very close to Gottschalk and Maloney's panel data. (1985) finding that roughly 70 percent of U.S. job changes are vol- 11. This mixed story is put forward by Guadalajaran sociologist untary. Put differently, the implied rates of involuntary entry into Gonzalez de la Rocha (1994), who argues that, for many older workers, self-employment would be normal by the standards of a flexible self-employment provides a safety net by offering insecure occupa- industrial country market. tions (such as the services) in which their age is not a limitation after 19. Survey data for other countries are also broadly consistent. For they have been kicked out of the formal manufacturing or formal instance, Itzigsohn (2000) reports that most informal entrepreneurs in services, while also suggesting some degree of voluntary movement Costa Rica and the Dominican Republic preferred their jobs to being a in that older men may also find the pace of industrial (formal) work formal employee in a special export zone. In Paraguay, only 28 percent too arduous and leave such jobs. The voluntary take is stressed by of those in the informal sector stated a desire to change occupations anthropologists Selby, Murphy, and Lorenzen (1990) who note the (DGE 1998). In Guatemala, only 31 percent of informal independent "surprising desirability of informal sector employment as the basis workers were willing to become formal employees (CIEN 2006). for a household earning strategy, particularly for poorer, older house- 20. These data come from surveys with the same or very similar holds with lower educational qualifications" (p. 144). questions on preferences for self-employment over salaried work to 12. Fajnzylber, Maloney, and Rojas (2006) find that Mexican firms the adult (employed) population of the countries. have mortality rates of the same order of magnitude as those found in 21. Interestingly, the fraction of formal independent workers who the United States by Evans and Leighton (1989); the 18 percent exit are involuntary is only 15 percent in Argentina and the Dominican to wage work (measured as the fraction of self-employed workers Republic, but 35 percent in Bolivia and 54 percent in Colombia. moving to wage work) compares to 14 percent across a year. This is 22. This disproportionately high level of youth reporting comple- consistent with much higher entry rates as a fraction of wage work- menting income makes sense, given that around 2.5 percent of ers, averaging 8 percent in Mexico compared to the 4 percent found 15­18-year-old males are heads of their households, while 30 percent by Evans and Leighton (1989) for the United States. Among young are in the next category. workers 20­28 years old, failure rates are roughly equivalent at 23. Likewise among Paraguayan informal salaried workers, the around 50 percent, while for prime-age workers, 31­35, failure rates percent looking for a better job was only slightly higher than the self- are roughly equivalent at 20 percent. See also Levenson and Maloney employed, at 32 percent (DGE 1998). (1996). 13. While the above findings do not preclude discrimination References against married women or those with children if, for example, an Akerlof, G., and B. Main. 1980. "Unemployment Spells and Unem- employer fears that the women may be likely to be absent for work ployment Experience." American Economic Review 71 (5): 885­93. for long periods (Chant 1992), further evidence favors the idea that Amaral, P., and E. Quintin. 2006. "A Competitive Model of the many of them choose self-employment on the basis of its desirable Informal Sector." Journal of Monetary Economics 53 (7): 1541­53. characteristics. In fact, evidence from Argentina, Mexico, Costa Rica, Arriagada, I. 1998. "The Urban Female Labour Market in Latin and Brazil shows that, when disaggregating the self-employment America: The Myth and the Reality." Report LC/L 1034-P/I, sector, a clear subgroup of young mothers emerges, along with the Economic Commission for Latin America and the Caribbean, subgroups of new entrepreneurs, family business owners, and those Santiago, Chile. who cannot find work elsewhere. The motivational responses dis- Arias, Omar, and Monserrat Bustelo. 2007. "Profiles and Dynamics cussed later lend more credence to these conclusions. of Informal Employment in Latin America." Photocopy. World 14. Balán, Browning, and Jelin (1973) note that, in Mexico, this Bank, Washington, DC. period of life for young workers is one of "shopping around" and try- Arias, Omar, Fernando Landa, and Patricia Yáñez. 2007. "Movilidad ing out various possible life choices. Laboral e Ingresos en el Sector Formal e Informal de Bolivia." 15. This is consistent with findings that the largest determinant Documento de Trabajo, UDAPE, La Paz, Bolivia. of whether a girl reports beginning to work in Mexico is whether her Aroca, Patricio, and William Maloney. 1997. "Logit Analysis in a mother opened a microenterprise (Cunningham and Maloney 2001). Rotating Panel Context and an Application to Self-employment 16. This issue is very similar to the arguments and findings of the Decisions." World Bank Working Paper 2069, World Bank, literature on the importance of re-incidence of unemployment for Washington, DC. unemployment durations. See, for example, Clark and Summers Balán, J., H. L. Browning, and E. Jelin. 1973. "Men in a Developing (1979) and Akerlof and Main (1980) for the United States and Society." Photocopy. Institute of Latin American Studies, Galiani and Hopenhayn (2003) for Argentina. University of Texas, Austin. 17. Note that the proportion figures in table 2.4 are not directly Barr, A., and T. Packard. 2000. "Revealed and Concealed Preferences comparable to the "raw" transition probabilities presented in the first in the Chilean Pension System: An Experimental Investigation." bar on each pair of bars in figure 2.4 because the former are computed Discussion Paper 53, Department of Economics, Oxford Uni- taking the total number of workers in each occupational and age versity, UK. group as denominators, whereas the latter divides by the total num- Becker, G. S. 1991. A Treatise on the Family. Enlarged ed. Cambridge, ber of workers. MA: Harvard University Press. 75 I N F O R M A L I T Y Bentolila, S., and A. Ichino. 2000. "Unemployment and Consumption: ed. E. G. Katz and M. C. Correia, 85­132. Washington, DC: Are Job Losses Less Painful near the Mediterranean?" Discussion World Bank. Paper 2539, Centre for Economic Policy Research, London. ------. 2001b. "Sectoral Allocation by Gender of Latin American Bewley, T. F. 1999. Why Wages Don't Fall during a Recession. Workers over the Liberalization Period of the 1990s." Cambridge, MA: Harvard University Press. Policy Research Working Paper 2742, World Bank, Blanchflower, D. G. 2004. "Self-employment: More May Not Be Washington, DC. Better." Swedish Economic Policy Review 11 (2): 15­74. Cunningham, W., and W. F. Maloney. 2001. "Heterogeneity among Blanchflower, D. G., and A. J. Oswald. 1998. "What Makes an Mexico's Microenterprises: An Application of Factor and Cluster Entrepreneur?" Journal of Labor Economics 16 (1): 26­60. Analysis." Economic Development and Cultural Change 50 (1): 131­56. Blanchflower, D. G., A. J. Oswald, and A. Stutzer. 2001. "Latent Entre- Cunningham, W., and C. Ramos. 2001. "The Home as the Factory preneurship across Nations." European Economic Review 45: 680­91. Floor: Home-based Work in Latin America." Photocopy. LCSPR- Blau, D. 1987. "A Time-Series Analysis of Self-Employment in the Gender, World Bank, Washington, DC. United States." Journal of Political Economy 95 (3): 445­67. Dávila Capalleja, E. 1994. "Regulaciones laborales y mercado de tra- BLS (Bureau of Labor Statistics, U.S. Department of Labor). 1991. bajo en México." In Regulacion del Mercado de Trabajo en America Bureau of Labor Statistics News 92­386. Latina, ed. G. Marquez. San Francisco: Centro Internacional para Bosch, M., and W. F. Maloney. 2005. "Labor Market Dynamics in el Desarrollo Económico. Developing Countries: Comparative Labor Markets Analysis de Paula, A., and J. Scheinkman. 2006. "The Informal Sector." Using Continuous Time Markov Processes." Policy Research Photocopy. University of Pennsylvania, Philadelphia. Working Paper 3583, World Bank, Washington, DC. de Soto, H. 1989. The Other Path. The Invisible Revolution in the Third ------. 2007. "Comparative Analysis of Labor Market Dynamics World. New York: Basic Books. and Informality in Developing Countries." Photocopy. World DGE (Direccion General de Estadistica, Encuestas y Censos). 1998. Bank, Washington, DC. "Sector Informal." Encuesta Integrada de Hogares, 1997/1998, Botero, J., S. Djankov, R. La Porta, and F. Lopez-de-Silanes. 2004. Paraguay. "The Regulation of Labor." Quarterly Journal of Economics 119 (4): Djankov, S., R. La Porta, F. Lopez-de-Silanes, and A. Shleifer. 2002. 1339­82. "The Regulation of Entry." Quarterly Journal of Economics 117 (1): Carneiro, P., J. Heckman, and E. Vytlacil. 2005. "Understanding 1­37. What Instrumental Variables Estimate: Estimating Marginal and Duryea, S., G. Marquez, C. Pages, and S. Scarpetta. 2006. "For Better Average Returns to Education." Photocopy. University of Chicago. or for Worse? Job and Earnings Mobility in Nine Developing and Carr, M., M. Chen, and J. Tate. 2000. "Globalization and Home- Emerging Economies." Photocopy. Inter-American Development based Workers." Feminist Economics 6 (3): 123­42. Bank, Washington, DC. Chant, S. 1991. Women and Survival in Mexican Cities: Perspectives on Esfahani, H., and D. Salehi-Isfahani. 1989. "Effort Observability and Gender, Labour Markets and Low-Income Households. Manchester, Worker Productivity: Towards an Explanation of Economic Dual- UK: Manchester University Press. ism." Economic Journal 99 (June): 818­36. ------, ed. 1992. Gender and Migration in Developing Countries. Evans, D. S., and B. Jovanovic. 1989. "An Estimated Model of Entre- London: Belhaven Press. preneurial Choice under Liquidity Constraints." Journal of Politi- Chen, M., J. Sebstad, and L. O'Connell. 1999. "Counting the Invisi- cal Economy 97 (4): 808­27. ble Workforce: The Case of Home-based Workers." World Devel- Evans, D. S., and L. S. Leighton. 1989. "Some Empirical Aspects of opment 27 (3): 603­10. Entrepreneurship." American Economic Review 79 (3): 519­35. CIEN (Centro de Investigaciones Económicas Nacionales). 2006. Fajnzylber, P., W. F. Maloney, and G. V. Montes Rojas. 2006. Economía Informal: Superando las barreras de un Estado excluyente. "Microenterprise Dynamics in Developing Countries: How Similar Guatemala City, Guatemala. Are They to Those in the Industrialized World? Evidence from Clark, K., and L. H. Summers. 1979. "Labour Market Dynamics and Mexico." World Bank Economic Review 20 (3): 389­419. Unemployment: A Reconsideration." Brookings Papers on Economic Fields, G. S. 1990. "Labor Market Modelling and the Urban Informal Activity 1: 13­60. Sector: Theory and Evidence." In The Informal Sector Revisited, ed. Cunningham, W. 2000. "Mexican Female Small Firm Ownership: D. Turnham, B. Salome, and A. Schwartz. Paris: Organisation for Motivations, Returns, and Gender." Photocopy. LCSPR-Gender, Economic Co-operation and Development. World Bank, Washington, DC. Fortin, B., N. Marceau, and L. Savard. 1997. "Taxation, Wage Con- ------. 2007. "Minimum Wages and Social Policy: Lessons from trols and the Informal Sector." Journal of Public Economics 66: Developing Countries." Photocopy. World Bank, Washington, DC. 293­312. Cunningham, W., R. García-Verdú, L. McGinnis, and C. Tesliuc. Galiani, S., and H. Hopenhayn. 2003. "Duration and Risk of Unem- 2007. The Promise of Youth: Policy for Youth at Risk in Latin America ployment in Argentina." Journal of Development Economics 71 (1): and the Caribbean. Washington, DC.: World Bank. 199­212. ------. 2001a. "Breadwinner Versus Caregiver: Labor Force Partici- Galiani, S., and F. Weinschelbaum. 2006. "Modeling Informality pation and Sectoral Choice over the Mexican Business Cycle." In Formally: Households and Firms." Photocopy. World Bank, The Economics of Gender in Mexico: Family, Work, State, and Market, Washington, DC. 76 T H E R A Z Ó N D E S E R O F T H E I N F O R M A L W O R K E R Gasparini, L., and L. Tornarolli. 2006. "Labor Informality in Latin Levenson, A. R., and W. Maloney. 1996. "Modeling the Informal America and the Caribbean: Patterns and Trends from Household Sector: Theory and Empirical Evidence from Mexico." Photocopy. Survey Microdata." Photocopy. World Bank, Washington, DC. University of Illinois, Chicago. Geldstein, R. 2000. "Non-labor Market Coping Strategies in ------. 1998. "The Informal Sector, Firm Dynamics and Institu- Argentina." Photocopy. LCSPR-Gender, World Bank, tional Participation." Policy Research Working Paper 1988, Washington, DC. World Bank, Washington, DC. Gonzalez de la Rocha, M. 1994. The Resources of Poverty: Women and Levy, S. 2006. "Productividad, Crecimiento y Pobreza en México: Survival in a Mexican City. Oxford, UK: Blackwell. ¿Qué sigue después de Progresa-Oportunidades?" Photocopy. Gottschalk, P., and T. Maloney. 1985. "Involuntary Terminations, World Bank, Washington, DC. Unemployment, and Job Matching: A Test of Job Search Theory." Loayza, N. 1994. "Labor Regulations and the Informal Economy." Journal of Labor Economics 3 (2): 109­23. Policy Research Working Paper 1335, World Bank, Washington, Harris, J. R., and M. P. Todaro. 1970. "Migration, Unemployment, DC. and Development: A Two-Sector Analysis." American Economic ------. 1996. "The Economics of the Informal Sector: A Simple Review 60 (1): 126­42. Model and Some Evidence from Latin America." Carnegie- Hart, K. 1972. Employment, Income and Inequality: A Strategy for Rochester Conference Series on Public Policy 45: 129­62. Increasing Productive Employment in Kenya. Geneva: International Lopez-Castaño, H. 1990. "Inestabilidad Laboral y Ciclo de Vida en Labour Organization. Colombia." Coyuntura Economica 20 (1): 173­91. Heckman, J., and G. Sedlacek. 1985. "Heterogeneity, Aggregation, Lucas, R. E., Jr. 1978. "On the Size Distribution of Business Firms." and Market Wage Functions: An Empirical Model of Self- Bell Journal of Economics 9 (2): 508­23. selection in the Labor Market." Journal of Political Economy 93 (6): MacIsaac, D., and M. Rama. 1997. "Do Labor Market Regulations 1077­125. Affect Labor Earnings in Ecuador?" Journal of Labor Economics 15 Hemmer, H., and C. Mannel. 1989. "On the Economic Analysis of (3): 36­65. the Urban Informal Sector." World Development 17 (10): ------. 2001. "Mandatory Severance Pay: An Assessment of Its 1543­52. Coverage and Effects in Peru." Policy Research Working Paper Hibbs, D. A. Jr., and V. Piculescu. 2006. "Tax Toleration and Tax 2626, World Bank, Washington, DC. Evasion: Why Firms Enter the Unofficial Economy." Working Maloney, W. F. 1999. "Does Informality Imply Segmentation in Paper in Economics 173, Department of Economics, Göteborg Urban Labor Markets? Evidence from Sectoral Transitions in University, Sweden. Mexico." World Bank Economic Review 13: 275­302. Hirschman, A. O. 1970. Exit, Voice, and Loyalty: Responses to Decline in ------. 2004. "Informality Revisited." World Development 32 (7): Firms, Organizations, and States. Cambridge, MA: Harvard Univer- 1159­78. sity Press. Maloney, W. F., and P. Aroca. 1999. "Logit Analysis in a Rotating ILO (International Labour Organization). 1995. World Employment. Panel Context and an Application to Self-employment Geneva. Decisions." Working Paper 2069, World Bank, Washington, DC. Itzigsohn, J. 2000. "Immigration and the Boundaries of Citizenship: ------. 2007. "Labor Market Dynamics in Latin America: Evidence The Institutions of Immigrants' Political Transnationalism." from Argentina, Brazil and Mexico." Photocopy. World Bank, International Migration Review 34 (4): 1126­55. Washington, DC. Jelin, E., M. Mercado, and G. Wyczykier. 2001. "Home Work in Maloney, W. F., and J. N. Nuñez. 2002. "Measuring the Impact of Argentina: Boosting Employment Through Small Enterprise Minimum Wages, Evidence from Latin America." In Law and Development." Working Paper 6, Series on Homeworkers in the Labor Markets, ed. J. Heckman and C. Pages. Chicago: University Global Economy, International Labour Organization, Geneva. of Chicago Press. Jimeno, J., and D. Rodriguez Palenzuela. 2003. "Youth Unemploy- Mondino, G., and S. Montoya. 2002. "The Effects of Labor Market ment in the OECD: Demographic Shifts, Labour Market Institu- Regulations on Employment Decisions by Firms. Empirical Evi- tions and Macroeconomics Shocks." Working Paper 19, European dence for Argentina." In Law and Employment: Lessons from Latin Network of Economic Policy Research Institutes, Brussels. America and the Caribbean, ed. J. Heckman and C. Pages. Chicago: Jovanovic, B. 1982. "Selection and Evolution of Industry." Economet- University of Chicago Press. rica 50 (3): 649­70. Morduch, J. 1999. "Between the Market and State: Can Informal Krawczyk, M. 1993. "Women in the Region: Major Changes." Insurance Patch the Safety Net?" World Bank Research Observer 14: ECLAC Review 49: 7­19. 187­207. Krebs, T., and W. F. Maloney. 1999. "Quitting and Labor Turnover: Packard, T. 2002. "Pooling, Saving and Prevention: Mitigating Old Microeconomic Evidence and Macroeconomic Consequences." Pol- Age Poverty in Chile." Policy Research Working Paper 2849, icy Research Working Paper 2068, World Bank, Washington, DC. World Bank, Washington, DC. Kugler, A., and M. Kugler. 2003. "The Labor Market Effects of Pay- Peattie, L. 1982. "What Is to Be Done with the `Informal Sector'? A roll Taxes in a Middle-Income Country: Evidence from Case Study of Shoe Manufacturers in Colombia." In Towards a Colombia." IZA Discussion Paper 852, Institute for the Study of Political Economy of Urbanization in Third World Countries, ed. Labor, Bonn, Germany. H. Safa. Delhi: Oxford University. 77 I N F O R M A L I T Y ------. 1987. "An Idea in Good Currency and How It Grew: The Selby, H., A. Murphy, and S. Lorenzen. 1990. The Mexican Urban House- Informal Sector." World Development 15 (7): 851­60. hold, Organizing for Self-Defense. Austin: University of Texas Press. Portes, A., M. Castells, and L. A. Benton, eds. 1989. The Informal Shapiro, C., and J. Stiglitz. 1984. "Equilibrium Unemployment As a Economy: Studies in Advanced and Less Developed Economies. Baltimore: Worker Discipline Device." American Economic Review 74: 433­44. Johns Hopkins University Press. Stiglitz, J. E. 2000. "Formal and Informal Institutions." In Social Portes, R., and R. Schauffler. 1993. "Competing Perspective on the Capital: A Multifaceted Perspective, ed. P. Dasgupta and I. Latin American Informal Sector." Population and Development Serageldin. Washington, DC: World Bank. Review 19 (1): 33­59. Straub, S. 2005. "Informal Sector: The Credit Market Channel." Jour- Prugl, E., and I. Tinker. 1997. "Microentrepreneurs and Homeworkers: nal of Development Economics 78: 299­321. Convergent Categories." World Development 25 (9): 1471­82. Tokman, V. 1992. Beyond Regulation: The Informal Economy in Latin Rauch, J. E. 1991. "Modelling the Informal Sector Formally." Journal America. Boulder, CO: Lynne Rienner. of Development Economics 35: 33­47. Tomei, M. 2000. "Home Work in Selected Latin American Coun- Roberts, B. D. 1991. "Employment Structure, Life Cycle, and Life tries: A Comparative Overview." Photocopy. International Labour Chances: Formal and Informal Sectors in Guadalajara." In The Organization, Geneva. Informal Economy: Studies in Advanced and Less Developed Economies, Turner, J., and S. Verma. 2007. "Why Some Workers Don't Take ed. A. Portes, M. Castells, and L. A. Benton. Baltimore: Johns 401(K) Plan Offers: Inertia Versus Economics." Working Paper Hopkins University Press. 56/07, Center for Research on Pensions and Welfare Policies, Rosen, S. 1981. "The Economics of Superstars." American Economic Moncalieri, Italy. Review 71: 845­58. Turnham, D., and D. Eröcal. 1990. "Unemployment in Developing Saavedra, J., and A. Chong. 1999. "Structural Reform, Institutions Countries, New Light on an Old Problem." Technical Paper, and Earnings: Evidence from the Formal and Informal Sectors in Development Center, Organisation for Economic Co-operation Urban Peru." Journal of Development Studies 35 (4): 95­116. and Development, Paris. Schneider, F., and D. H. Enste. 2000. "Shadow Economies: Size, WIEGO (Women in Informal Employment: Globalizing and Orga- Causes, and Consequences." Journal of Economic Literature 38 (1): nizing). 2000. "Women in the Informal Economy," available at 77­114. http://www.wiego.org. Sedlacek, G. L., R. Paes de Barros, and S. Varandas. 1990. World Bank. 2007. Argentina: Poverty Assessment. Informal Employment "Segmentacao e Mobilidade No Mercado de Trabalho: A Cartéira in Argentina: Toward Understanding Its Causes and Consequences. de Trabalho em São Paulo." In Pesquisa e Planejameto Econômico, Washington, DC: World Bank. 20(1): 87­104. Yamada, G. 1996. "Urban Informal Employment and Self- ------. 1995. "Segmentacao e Mobilidade no Mercado de Trabalho employment in Developing Countries: Theory and Evidence." Brasilerio." Unpublished manuscript. Economic Development and Cultural Change 44 (2): 289­314. 78 CHAPTER 3 Informality, Earnings, and Welfare SUMMARY: Based on an examination of differences in earnings and self-rated welfare, this chapter lends credence to both the "exclusion" and the "voluntary" nature of informal employment suggested in chapter 2 by workers' reported motiva- tions to be informal. Earnings and welfare assessments support the view that the majority of independent workers are largely voluntary and attach significant value to the nonpecuniary benefits of autonomous work, while the majority of informal salaried workers tend to be excluded from more desirable jobs both in the formal salaried and independent sectors. However, the data caution against simple generalizations. Both groups exhibit important heterogeneity, and data for the Dominican Republic show that both the informal self-employed and informal salaried are as well-off as formal salaried workers, while in Colombia both informal groups exhibit much lower levels of satisfaction with current jobs. While there are seemingly two tiers of informal employment, gradients and gray areas cloud universal conclusions about who they are, what determines their relative size, and the welfare implications of each tier. C HAPTER 1 ILLUSTRATED THE (for example, flexibility, autonomy) characteristics of infor- considerable heterogeneity of informal mal jobs partially compensate workers for any lower earn- workers in the region and how, on average, ings? This chapter examines these questions, relying heavily they tend to face a significant earnings dis- on analyses of household surveys containing the relevant advantage. Chapter 2 discussed the various data from Argentina, Bolivia, Colombia, and the Domini- hypotheses that aim to explain the razón de ser of the infor- can Republic to illustrate the issues in a variety of country mal worker and their implications regarding the relative contexts. quality of formal and informal jobs. This chapter delves The chapter first describes the characteristics of workers, into a more in-depth investigation of the characteristics their jobs, and the firms that affect the propensity to infor- associated with the propensity of workers to be informal, mal employment (proxied by lack of pension coverage). It and the variation in the earnings and welfare differentials then examines whether informal workers receive equal pay associated with informal work, making a clear distinction for equal work along the entire remuneration scale, and between informal salaried and independent workers. what fraction of informal-formal earnings gaps result from Who are the informal (salaried and independent) workers? differences in worker characteristics and in how these are Does informal employment in itself imply lower earnings-- rewarded in the labor market, including differences in the that is, do informal workers receive lower pay for equal returns to human capital (schooling or work experience). work? Are all informal jobs "bad jobs" in terms of carrying Finally, the chapter examines workers' self-rated welfare lower earnings for a worker's skills, or do some individuals (propensity to consider themselves poor) and job satisfac- find niches in informal employment that are more suitable for tion responses as more complete indicators to assess welfare their skills, preferences, and situations? Does informal work differences between workers in formal and informal jobs, as lead to lower individual welfare, or do the nonpecuniary well as the significance of nonpecuniary job characteristics 79 I N F O R M A L I T Y (that is, flexibility, autonomy, stability/mobility) in the wages that makes workers indifferent toward the various sectoral participation of various workers (for example, types of jobs. The differences in these implicit wages are women/men, the skilled/unskilled). called compensating differentials. Given heterogeneity in worker preferences and skills, both supply and demand for Compensating differentials, comparative particular jobs determine the size of the compensating dif- advantage, and informal work ferential between jobs with different working conditions. As discussed in chapter 2, the nature of informal employ- Thus, the labor market comprises a set of interrelated mar- ment has been examined mainly through two lenses: the kets for different labor types whose wages are set by supply "exclusion" and "voluntary" views. In the exclusion view, and demand. workers would prefer the benefits of formal jobs but are The consideration of major labor-demand constraints in rationed out due to segmentation of the labor market. The the framework can bring to bear the concerns of the exclu- latter is traditionally linked to institutional rigidities, sion view of informality. In particular, with high unem- but--again--can arise from economic dualism, barriers to ployment, an oversupply of some types of labor, or in a labor mobility (geographical or informational), efficiency situation when workers are in "noncompeting groups," the wages, or coordinated evasion of corporate, sales, and payroll market pressure to pay compensating differentials can van- taxes. In contrast, the voluntary view suggests that infor- ish. In this case, more workers would be willing to take the mal jobs largely reflect workers' implicit choices given their less desirable jobs at wages below those in the more desir- preferences, skills, the costs and benefits of formality, and able-but-scarce jobs. Therefore, when the demand of some the availability of other means of social protection, the last labor types is very constrained for whatever reason, includ- in turn being a function of the quality of existing country ing market segmentation, the room for workers' choice is policies and institutions. As noted before, both views have greatly diminished and their wages are set primarily by the potential explanatory power to understand the nature of employing firms. informality in the Latin American and Caribbean region. This framework has three key implications for the The labor literature on compensating differentials and nature of informal employment. First, in choosing occupational choice based on workers' comparative advan- between informal and formal employment, workers weigh tage provides a framework that encompasses these two seem- the advantages and disadvantages of each potential job, ingly irreconcilable views. The basic idea, first advanced by subject to the availability of jobs with their desired attrib- Adam Smith (1876, p. 111), is that the wages paid to vari- utes; therefore, comparative advantage could make the ous types of labor must, in general, equalize total advantages informal sector a better match for many of them. As noted and disadvantages, pecuniary and nonpecuniary, and that in chapter 2, informal and formal jobs tend to require workers select occupations that yield the highest net varying degrees of skills, and the various amenities of advantage for their tastes and skills. This is the cornerstone informal and formal jobs may be valued differently by of the modern literature on comparative advantage in the some groups. Some workers might find advantageous labor market as a determinant of occupational choice, human niches to their observed and unobserved skills in occupa- capital investments, and earnings performance (Carneiro, tions with a higher propensity to be informal (such as Heckman, and Vytlacil 2001; Heckman and Sedlacek 1985; those in construction). Rosen 1981). Others may be willing to forgo some of the benefits of for- According to this, individuals choose the jobs that mality in exchange for the nonpecuniary benefits of informal better fit their preferences and specific ranges of talents, jobs. As an example, for low-skilled youth and older work- including cognitive, social, and mechanical skills. Jobs that ers, informal salaried jobs may offer an entry point to the are more desirable (due to amenities such as fringe benefits, labor market that partially allows them to remedy deficient stability, safety, autonomy, and flexibility) or that require schooling or the obsolescence of skills through on-the-job relatively abundant skills should have lower-than-average training unavailable to them in formal salaried jobs. For wages while jobs that are less desirable or demand scarce women, the demands for flexibility to balance work and skills should pay higher wages. A competitive labor market family responsibilities that arise with child rearing may determines an implicit (hedonic) wage for each type of labor render the greater flexibility and autonomy of informal and equilibrates when labor mobility leads to a set of relative jobs a better match. Furthermore, employers could attract 80 I N F O R M A L I T Y, E A R N I N G S , A N D W E L F A R E enough workers without offering a compensating wage dif- The profile of participation in informal employment ferential if a large subset of workers does not regard a given As discussed in chapters 1 and 2, informal employment job attribute (such as not having social security) as a disad- encompasses a diverse range of people. Using contributions to vantage. In some cases, amenities and disadvantages may social security as a criterion to define informal employment, cancel each other out without giving rise to compensating we can distinguish four employment categories: informal and differentials. formal independent, informal and formal salaried. The infor- Second, there is no clear a priori presumption that infor- mal independent generally comprise relatively better-paid mal employment should carry lower, equal, or higher earn- small firm owners, self-employed professionals (for example, ings than formal jobs. On the one hand, informal jobs doctors, lawyers, teachers), semiskilled workers with some should command higher earnings to compensate workers for technical competencies (for example, artisans, handymen, the value of lost fringe benefits such as social protection construction laborers, taxi drivers), and unskilled workers in for health risks and old age (net of payroll contributions), precarious employment or under a semi-dependent work rela- and their greater risk of unemployment and income shocks. tionship (street vendors, small artisans under subcontract). On the other hand, earnings may be lower to adjust for the The informal salaried largely comprise domestic employees, a value of nonpecuniary amenities such as flexible hours, variety of young apprentices, and older unskilled workers in autonomous work, training, and tax savings, particularly for small enterprises (for example, sales clerks, beauticians, independent workers but much less so for the informal salaried artisans), but also the workers in larger firms who are salaried workers. Thus, differences in earnings cannot be under informal labor arrangements. used to test segmentation in the labor market (Magnac Regression analysis can be used to isolate the character- 1991; Maloney 1999). istics that best predict the propensity that a given worker Moreover, since workers may sort into formal and infor- engages in the four employment categories. The analysis is mal jobs acting on the returns to their skills and competen- carried out for Argentina, Bolivia, and the Dominican cies in each sector, direct average earnings comparisons are Republic, considering a host of characteristics of workers, flawed. Frequently, high-paying skilled jobs also offer bet- their jobs, and the firms for which they work. Figures 3.1 ter fringe benefits. It is necessary to control for differences to 3.3 present the odds ratios or relative propensities that a in observed and unobserved skill level (leading to produc- worker with a given characteristic belongs to each of the tivity differences reflected in wages) in order to measure two informal groups rather than to the formal salaried, correctly informal-formal compensating differentials. Fur- holding other characteristics constant. An odds ratio of 1 thermore, average earnings gaps do not fully characterize represents no effect of a variable; a ratio greater than 1 indi- the earnings gains or losses from formal employment for cates that the characteristic increases the odds of that any given worker. Some workers could emerge as top earn- employment category compared to the formal salaried ings performers for their skills in informal jobs, but if they group, and a ratio less than 1 indicates that it diminishes were to move to a formal job they may actually lose out in the odds. For instance, an odds ratio of 0.75 signifies that the earnings rankings. Section 3 of the chapter presents the chance that a worker with the given characteristic gets empirical results in this vein. a given job type is 75 percent of the chance of being formal Finally, more generally, the relative qualities of informal salaried. An odds ratio of 1.5 implies a 1.5 times greater and formal jobs and the ensuing differences in workers' wel- chance of being in the given category than formal salaried. fare have to be assessed more broadly to take into account The larger the odds ratio, the stronger the relationship, so nonpecuniary job attributes. The conditional informal- these provide an implicit ordering of workers with differ- formal earnings comparisons could be seen as providing ent education levels, age groups, sectors, firm sizes, and so orders of magnitude of the potential net value of the ameni- on across the four occupational groups. ties and disadvantages of informal jobs. Establishing whether Adjusting for workers' personal characteristics, firm amenities fully compensate for any lower informal earnings so size has the strongest negative association with informal that workers are equally well-off in terms of welfare faces the work, and sector of employment also has an important challenge of limited data. The last section of the chapter pre- independent effect. The odds of being informal salaried sents new analyses of subjective welfare and job satisfaction in microenterprises relative to large firms are over 40 to data that attempt to provide some evidence on this key issue. 1 in Bolivia and the Dominican Republic, and over 11 to 81 I N F O R M A L I T Y 1 in Argentina.1 In these countries, the odds fall signifi- FIGURE 3.1 Propensity to informal employment by firm size and economic cantly for employees of medium-sized companies, although sector, 2005­06 remaining at 12 to 1 in the Dominican Republic. Propensity to informal salaried work and firm size Employment industry also significantly affects informal- Odds ratio compared to formal salaried, firm size: large firm ity propensities, with no universal patterns across coun- Ratio tries. Adjusting for firm size and other characteristics, 45 40 informal independent workers are more likely to be 35 found in construction and transport as well as in agricul- 30 tural activities and commerce in the Dominican Repub- 25 lic, and in construction and commerce in Argentina. 20 Informal salaried employment is more likely in the same 15 sectors in the Dominican Republic, while it is more 10 5 prevalent in transport, social, and personal services in 0 Argentina. Informal employment propensities are quite Dominican Republic Argentina Bolivia similar across sectors in Bolivia, except for greater odds (urban) (Greater Buenos Aires) (urban) of independent work in agriculture and informal salaried Micro firm Small and medium size employment in transport. These differences across and within countries in sectoral propensities to informality Propensity to informal self-employment and economic sector Odds ratio compared to formal salaried, manufacturing may reflect differences in the incentives entailed by exist- Ratio ing regulations and in their enforcement across sectors 60 and countries, as well as an intrinsic inclination toward 14 informality of workers with occupational traits fit to sec- 12 tors such as construction and commerce. 10 Among personal characteristics, education and tenure 8 are the strongest predictors of informal employment (figure 6 3.2). The informal employment propensities decline pro- 4 gressively with education, especially with full completion 2 of secondary schooling. Compared to a college graduate, the 0 odds that a worker with incomplete primary or no educa- Dominican Republic Argentina Bolivia (urban) (Greater Buenos Aires) (urban) tion is informal salaried or informal independent are 15 and 8 times greater in Bolivia, respectively, about 2 to 1 in the Propensity to informal salaried work and economic sector Dominican Republic, while in Argentina the odds are 6 to Odds ratio compared to formal salaried, manufacturing Ratio 1 and above 2 to 1, respectively. Holding constant other 10 characteristics, those with a high school diploma face equal 8 chances of informal employment as the college-educated in 6 Argentina, 4-to-1 odds in Bolivia, and still close to a 2-to-1 4 chance in the Dominican Republic. The significant fall in the informality propensity with the completion of high 2 school suggests that this diploma may serve as a signal to 0 Dominican Republic Argentina Bolivia employers that a worker has the minimum skills required (urban) (Greater Buenos Aires) (urban) to justify the cost of a formal labor contract. This may be Construction Transport and communication more important in the presence of job rationing due to Primary sector Commerce labor market rigidities or other sources of job segmenta- Public administration, education, health, and social services tion. This is consistent with the differences observed between Argentina and Bolivia, on one hand, and the less Source: Based on Arias and Bustelo (2007), Arias, Landa, and Yanez (2007), and author's estimates using household survey 2005, 2006 rigid Dominican labor market on the other. Interestingly, data. results (not shown here for brevity) indicate that in all three 82 I N F O R M A L I T Y, E A R N I N G S , A N D W E L F A R E FIGURE 3.2 Propensity to informal employment by education and tenure, 2005­06 Propensity to informal self-employment and education Propensity to informal salaried work and education Odds ratio compared to formal salaried, college graduate Odds ratio compared to formal salaried, college graduate Ratio Ratio 16 16 14 14 12 12 10 10 8 8 6 6 4 4 2 2 0 0 Dominican Republic Argentina Bolivia Dominican Republic Argentina Bolivia (urban) (Greater Buenos Aires) (urban) (urban) (Greater Buenos Aires) (urban) Primary incomplete Primary complete Secondary incomplete Secondary complete College incomplete Propensity to informal self-employment and job tenure Propensity to informal salaried work and job tenure Odds ratio compared to formal salaried, 5 years or more tenure Odds ratio compared to formal salaried, 5 years or more tenure Ratio Ratio 10 10 9 9 8 8 7 7 6 6 5 5 4 4 3 3 2 2 1 1 0 0 Dominican Republic Argentina Bolivia Dominican Republic Argentina Bolivia (urban) (Greater Buenos Aires) (urban) (urban) (Greater Buenos Aires) (urban) Less than 1 year Between 1 and 5 years Source: Based on Arias and Bustelo (2007), Arias, Landa, and Yanez (2007), and author's estimates using household survey 2005, 2006 data. countries the college-educated are generally equally or more propensities (figure 3.3). Although far from universal, likely to be formal independent than formal salaried, some common patterns emerge by age and women's mari- adjusting for other characteristics. tal status. As shown in chapter 2, younger workers experi- Finally, workers with less than one year in their current ence much higher rates of informal salaried work. The occupation have a much larger risk of being informal salaried regression analysis shows that this is largely a result of than those with longer tenures. For instance, compared to youth's much shorter tenures and higher rates of employ- those with five or more years in their occupation, the odds are ment in small firms. In fact, holding these and other fac- 9 to 1 in Argentina, 7 to 1 in Bolivia, and 3 to 1 in the tors constant, the odds ratio of informal to formal salaried Dominican Republic. Short tenures also correlate with greater work is constant across all age groups in Argentina and chances of being informal independent relative to formal the Dominican Republic, although they still fall system- salaried in Argentina and Bolivia (4 to 1 and 2 to 1, respec- atically with age in Bolivia, where youth are 50 percent tively), while the odds are similar in the Dominican Republic. more inclined to informal over formal salaried work Individual and family demographics show an indepen- than middle-aged workers. The age profile of informal dent but second-order correlation with informality independent work is consistent with life-cycle theories of 83 I N F O R M A L I T Y FIGURE 3.3 Propensity to informal employment by age, gender roles, and work preferences, 2005­06 Propensity to informal self-employment and age Propensity to informal salaried work and age Odds ratio compared to formal salaried, 36­45 years old Odds ratio compared to formal salaried, 36­45 years old Ratio Ratio 3.0 3.0 2.5 2.5 2.0 2.0 1.5 1.5 1.0 1.0 0.5 0.5 0 0 Dominican Republic Argentina Bolivia Dominican Republic Argentina Bolivia (urban) (Greater Buenos Aires) (urban) (urban) (Greater Buenos Aires) (urban) 15­25 years 26­35 years 46­55 years 56­65 years Propensity to informal self-employment and gender Propensity to informal salaried work and gender Odds ratio compared to formal salaried, man Odds ratio compared to formal salaried, man Ratio Ratio 5.0 5.0 4.5 4.5 4.0 4.0 3.5 3.5 3.0 3.0 2.5 2.5 2.0 2.0 1.5 1.5 1.0 1.0 0.5 0.5 0 0 Dominican Republic Argentina Bolivia Dominican Republic Argentina Bolivia (urban) (Greater Buenos Aires) (urban) (urban) (Greater Buenos Aires) (urban) Married Single With small children Propensity to informal self-employment and desire to work Propensity to informal salaried work and desire to work independently. Odds ratio compared to formal salaried, independently. Odds ratio compared to formal salaried, prefers salaried work prefers salaried work Ratio Ratio 1.0 1.0 0.8 0.8 0.6 0.6 0.4 0.4 0.2 0.2 0 0 Dominican Republic Argentina Bolivia Dominican Republic Argentina Bolivia (urban) (Greater Buenos Aires) (urban) (urban) (Greater Buenos Aires) (urban) Wants to work independently Source: Based on Arias and Bustelo (2007), Arias, Landa, and Yanez (2007), and author's estimates using household survey 2005, 2006 data. 84 I N F O R M A L I T Y, E A R N I N G S , A N D W E L F A R E self-employment, although it also varies across countries. preference for the job amenities they enjoy--or, conversely, The odds of independent (formal or informal) employment it could be an expression of their higher satisfaction with to formal salaried work increase with age in Argentina and salaried work. Other results (not shown) indicate that, con- the Dominican Republic, but are unrelated to age in trolling for all characteristics, urban migrants (coming Bolivia. The latter finding suggests that the accumulation from rural or other urban areas) in all three countries and of savings and occupation-specific skills may be less foreign migrants in Argentina and the Dominican Repub- important for the ability to engage in the type of indepen- lic have similar employment propensities as nonmigrants. dent employment predominant in very low-income The same is true for indigenous workers in Bolivia. Thus, countries. the observed higher incidence of informal employment in The gender patterns of participation in informal work these groups is largely a result of their having personal are broadly consistent with the findings reported in characteristics or jobs in sectors or firms with a higher chapter 2 for Mexico and other middle-income countries, propensity to informality, rather than to their migrant or although with some country variation. For instance, ethnicity status, per se. Argentine married women are more inclined than men to In summary, the results show that firm size, education be independent (4 to 1 informal, 2 to 1 formal) and infor- levels, tenure, and sector of employment--variables related mal salaried (2 to 1) than formal salaried, regardless of to the productivity of workers and firms and the ease of family structure. The same is true for Dominican married government enforcement--are the most important predic- women with small children. Conversely, single Argentine tors of informal employment in Argentina, Bolivia, and the and Dominican women have greater odds than men of Dominican Republic. There is a similar pattern of impacts engaging in formal salaried employment. These results are of these characteristics on the propensity to be informal consistent with Galiani and Weinschelbaum's (2006) find- salaried or informal self-employed in all three countries. ings; using data for a large sample of Latin American However, the orders of magnitudes differ markedly, which countries, they found that, ceteris paribus, a spouse is suggests that the underlying mechanisms determining more likely to work informally if the head of household informal employment vary with level of economic develop- has a formal job. Meanwhile, Bolivia shows no gender ment, economic structure, and the nature of public poli- differences in informal participation, regardless of the cies and institutions. presence of small children (or elderly) in the family. The next section examines empirically whether the Overall, there is no systematic evidence of discrimination informal-formal characteristics of jobs have an impact on forcing women into informal employment; the pattern of workers' earnings in Argentina, Bolivia, and the Domini- female employment that allows them to balance work and can Republic. The analysis considers the ample spectrum of family life varies with economic and social conditions, remunerations and worker prototypes in the labor market including the availability of protection through other and demonstrates that workers' sectoral participation is family members or publicly provided benefits programs, affected by the expected returns to their skills and the per- in their countries. ceived qualities of informal and formal jobs. With regard to other personal characteristics, there is strong evidence that workers act on their tastes for inde- The question of equal pay for equal work pendent or salaried work, and little evidence of an indepen- in the informal and formal sectors dent correlation of informal employment with migration or The heterogeneity in the prototype of informal workers is ethnicity. The bottom panel of figure 3.3 shows that the reflected in a wide earnings variation. Figure 3.4 illustrates chance that workers who report a preference for salaried this with a comparison of the distributions of hourly earn- over independent work are actually employed as indepen- ings for formal, informal salaried, and independent workers dent is only 30 to 40 percent of their chance of being for- in urban areas of Argentina and Bolivia.2 No distinction is mal salaried. Interestingly, the chances that these workers made between the informal and formal self-employed, as are informal salaried are 50 percent (Bolivia) to 70 percent the latter are a small fraction of employment and their (Argentina and the Dominican Republic). This could separation would complicate the econometric analysis. The reflect the fact that formal salaried workers have a stronger left panels contain the frequency distributions (that is, 85 I N F O R M A L I T Y FIGURE 3.4 Distribution of hourly earnings for workers in urban areas of Argentina and Bolivia, 2005­06 Argentina Argentina Density Percent of workers 0.8 100 80 0.6 60 0.4 40 0.2 20 0 0 6 4 2 0 2 4 6 8 0 1 2 3 4 5 10 15 20 25 Log of hourly labor income Hourly labor income (2005 Argentine pesos) Bolivia Bolivia Density Percent of workers 0.8 100 80 0.6 60 0.4 40 0.2 20 0 0 4 2 0 2 4 6 8 0 5 10 15 20 30 40 50 60 70 80 Log of hourly labor income Hourly labor income (2005 Bolivian pesos) Self-employed Informal salaried Formal salaried Source: Based on Arias and Khamis (2007), Arias, Landa, and Yanez (2007), and author's estimates using household survey 2005, 2006 data. Note: Incomes measured in 2005 local currency. smoothed histograms) showing the earnings where most the bottom). Second, the informal salaried have a slight workers are concentrated in each group (that is, near the earnings advantage over independent workers in the bottom peaks). The right panels present the cumulative distribu- 40 percent of the distribution (their distribution lies below tions of hourly earnings which show the fraction of workers and closer to the formal salaried), but this is reversed in the at or below any given level of hourly earnings (that is, the top of the distribution (the high-earnings jobs). Third, the percentiles). The horizontal distance between the distribu- earnings distribution for independent workers is more dis- tions yields measures of the gap in informal-formal earn- persed, consistent with the international evidence of higher ings at different earnings percentiles. For instance, in earnings variations in independent activities.3 Moreover, Argentina the difference in hourly earnings between the there are two clear tiers of low- and high-earnings indepen- formal and the two informal groups at the 40th percentile dent workers (illustrated in circles for Argentina), consistent is $2.5 ( $5 2.5), while in Bolivia the difference at the with the heterogeneous composition of the group. 80th percentile is $15 ( $20 15). Table 3.1 quantifies the raw (unconditional) gaps in There are three main points worth noting. First, hourly earnings for jobs of low, median, and high remuner- although the formal salaried have an earnings advantage at ation within each occupational category for the three coun- any point of the pay scale (their distribution is further to the tries considered.4 Clearly, there are significant differences in right), informal-formal earnings gaps are larger for workers the pattern of earnings gaps across and within the countries. in low-earnings jobs (the distributions are further apart at On average, informal salaried workers earn between 40 to 86 I N F O R M A L I T Y, E A R N I N G S , A N D W E L F A R E TABLE 3.1 Unconditional hourly earnings gaps (percent difference in earnings) for formal employees, informal employees, and independent workers in urban areas in Argentina, Bolivia, and the Dominican Republic, 2005 Worker status by country Low-earnings jobs Median-earnings jobs High-earnings jobs Average-earnings jobs Argentina Informal/formal salaried 53.3 47.8 45.1 43.1 Independent/formal salaried 58.0 37.3 19.3 22.3 Informal salaried/independent 11.1 16.7 32.0 26.8 Bolivia Informal/formal salaried 60.0 64.0 68.0 67.0 Independent/formal salaried 69.6 61.0 51.0 60.0 Informal salaried/independent 32.3 7.3 31.8 18.0 Dominican Republic Informal/formal salaried 44.0 46.5 46.0 45.9 Independent/formal salaried 9.4 11.0 5.8 5.7 Informal salaried/independent 38.2 51.8 49.0 48.8 Source: Based on Arias and Khamis (2007), Arias, Landa, and Yanez (2007), and author's estimates using household survey 2005, 2006 data. Note: The high, medium, and low pay correspond, respectively, to the 80th, 50th, and 20th percentiles of earnings for each group; that is, they represent the top 20 percent of workers with the highest earnings, the bottom 20 percent with the lowest earnings, and workers with median earnings. Average pay = mean earnings. 66 percent less than formal salaried employees in all coun- having an unregistered job. More generally, as noted earlier tries, while independent workers earn as little as 60 percent in the chapter, differences in earnings can result from labor less in Bolivia and 28 percent less in Argentina but have a market segmentation or barriers that prevent workers from slight advantage in the Dominican Republic. accessing the best possible job for their skills, or can simply The differences in average hourly earnings understate the reflect the compensated earnings differentials from differ- disadvantage of informal salaried workers in low-earnings ences in the amenities of informal and formal jobs. jobs in Argentina, but not as much in Bolivia and the To determine whether informal workers receive equal Dominican Republic. Independent workers face a larger pay for equal skills, one must compare earnings of workers earnings disadvantage in the low-earnings jobs, which falls with similar observed and unobserved characteristics who progressively for workers at the top of their pay scale and differ only in having a formal or an informal job. This is best actually turns into an earnings advantage in the Dominican accomplished through regression analysis to purge earnings Republic. The low-earnings informal salaried are in fact gaps of spurious correlations induced by observed and unob- slightly better off than low-earnings independent workers served worker characteristics that affect earnings and cause in Argentina and Bolivia, but earn less in the top-paying selection (either by choice or rationing) into formal, infor- jobs. In these countries low-earnings workers face a more mal salaried, or independent work. For instance, the most precarious employment situation whether they are salaried talented individuals may be more likely to obtain formal or independent. The distinct patterns across countries salaried employment because of better prospects for mobil- suggest that the relative earnings of formal and informal ity in a career as wage earner. On the contrary, individuals workers probably vary with economic structure, overall pro- with more entrepreneurial ability are more likely to suc- ductivity, and other conditions specific to their countries. ceed as independent workers. Those with low work attach- However, these earnings gaps do not prove that formal ment and little adherence to authority or rigid work sector jobs are superior, per se. As we saw before, informal schedules may be excluded from formal salaried employ- workers tend to be younger and less educated, working in ment or voluntarily seek the flexibility of self-employment smaller firms and sectors such as commerce and construc- even at the cost of lower earnings. Many women may forgo tion, all of which are cause for lower earnings. Their lower the higher earnings of being employees in exchange for earnings may be due to their disadvantage in productive flexibility in informal employment. The end result is selec- attributes (both observed and unobserved) rather than to tion into occupations based on returns and tastes. 87 I N F O R M A L I T Y Even among workers with the same credentials, experi- tional Heckman-selectivity corrections. The extent to which ence, and line of work, some earn more than others due this procedure corrects fully or partially for self-selection to, for example, their innate abilities, or because they biases depends on the strength of variables that affect sector landed jobs in more productive firms. That is, there are participation but not earnings (see the studies by Arias low-paying and high-paying jobs among doctors, teachers, [2007, 2005] and the coauthors for details). carpenters, and salespeople, as well as in construction, man- We decompose the estimated earnings gaps into the ufacturing, and large and small firms. Gaps in average earn- portion due to between-group differences in characteristics ings between informal and formal workers with the same (endowments) and the part due to differences in how these observed characteristics may misrepresent the situation of characteristics are remunerated (returns) across job cate- those with earnings below or above the average pay for gories. The latter differences in returns are more adequate their skills; some workers may be paid differently or find measures of the compensated earnings differentials between advantageous niches across job categories. Informal jobs informal and formal jobs.7 Figure 3.5 shows the results of may carry a lower reward in manufacturing where small- this decomposition for the three countries considered. The scale production tends to be less productive, while pay may predicted earnings gaps (in logs) at low-, median-, and be similar across all categories in service jobs where high-paying jobs are given by the height of the bars. The economies of scale are less important. shaded areas reflect the portion attributed to informal More important, the returns to observed skills (such as workers' less favorable characteristics; the rest represents education or work experience) may be different. Schooling our measures of the compensating earnings differentials returns may be higher for the formal salaried, since education among formal and informal jobs, which also encapsulate is generally more productive in activities employing more any differences in earnings arising from possible labor mar- skills and physical capital. However, informal occupations ket segmentation due to macroinstitutional rigidities. prevalent in small-scale activities require fewer cognitive The results provide a different picture across countries. skills and may offer low-educated workers compensatory In Argentina the bulk of the informal earnings gap reflects practical skills that enhance the productivity of their school- unequal pay for similar skills, especially at the bottom of ing. Returns to education might be higher for independent the earnings scale. Approximately two-thirds of the earn- workers, since they can optimize the use of their credentials in ings difference between formal and informal Argentine their economic activity with the flexibility of small size and employees in low-paying jobs is due to the overall lower lesser division of labor. The returns to work experience can remuneration to the skills of the latter in the labor market. also differ for similar reasons.5 Although many of these differ- This fraction falls to 54 and 42 percent in the median- and ences are hard to ascribe to segmentation or to voluntary sort- high-paying jobs.8 The contribution of unequal pay is sim- ing of workers, they have implications on life-cycle incomes. ilar for independent workers in the low-paying jobs, falling In light of all this, earnings regressions are estimated to 38 percent in median-pay jobs, and turning insignificant separately for each of the three employment categories at at the top. That is, all of the earnings advantage of the best- different percentiles of the group-specific earnings distrib- paid formal salaried workers over the best-paid indepen- utions. This allows measurement of earnings gaps between dent workers in Argentina is due to differences in their formal and informal workers who have the same observed characteristics rather than unequal pay. In fact, the absolute characteristics (for example, education, work experience, level of both informal earnings gaps falls as we move up to gender) and other unmeasured characteristics and in the compared workers in the best-paid jobs of each of the three same positions of the pay scale fit to their skills and job cat- sectors. The results are consistent with high-earnings inde- egories. For instance, we measure the earnings gap between pendent workers having primarily voluntary motivations the best-paid formal salaried and the best-paid self- to be independent in Argentina, to the extent that they employed with the same skills set using quantile earnings cannot get better pay for their skills in formal salaried jobs. analysis.6 To deal with selection biases with cross-section The gaps at the bottom of the earnings distribution are household survey data, we proxy unobserved factors by esti- consistent with compensating earnings differentials in mates of the relative probabilities that individuals work as favor of those in formal salaried jobs (for example, to com- formal, informal salaried, or independent and include these pensate for lower flexibility/autonomy), and/or labor mar- in the earnings regressions. This is a modification of conven- ket segmentation against the self-employed. 88 I N F O R M A L I T Y, E A R N I N G S , A N D W E L F A R E FIGURE 3.5 Earnings cost of informality in urban areas of Argentina, Bolivia, and the Dominican Republic Argentina Argentina Formal-informal salaried Formal salaried-independent Log earnings gap Log earnings gap 1.5 1.5 1.0 1.0 0.5 0.52 0.19 0.55 0.5 0.37 0.26 0.29 0.32 0.35 0 0.08 0.28 0.31 0.35 0 0.5 Low pay Median pay High pay Low pay Median pay High pay Bolivia Bolivia Formal-informal salaried Formal salaried-independent Log earnings gap Log earnings gap 1.5 1.5 1.0 1.0 0.88 0.38 0.46 0.62 1.25 1.25 0.5 0.5 0.68 0.45 0.57 0.66 0 0 0.45 0.35 0.5 0.5 Low pay Median pay High pay Low pay Median pay High pay Dominican Republic Dominican Republic Formal-informal salaried Formal salaried-independent Log earnings gap Log earnings gap 1.5 1.5 1.0 1.0 0.51 0.31 0.5 0.84 0.5 0.77 0.25 0.24 0.51 0.65 0 0.44 0.54 0 0.81 0.05 0.5 0.5 1.0 Low pay Median pay High pay Low pay Median pay High pay Due to characteristics Due to returns Source: Based on Arias and Khamis (2007), Arias, Landa, and Yanez (2007), and author's estimates using household survey 2005, 2006 data. Note: See table 3.1 for definitions. The results for Bolivia show the existence of unequal pay the gaps). The differences in returns actually favor the infor- for informal jobs, but with an opposite pattern for the infor- mal salaried in high-pay jobs. That is, workers in the high- mal salaried and independent workers. Differences in charac- pay jobs of the informal salaried sector receive better teristics account for the bulk of earnings gaps between the remunerations for their skills than workers in the high-pay formal and informal salaried, especially in the low- and high- formal jobs. The pattern is similar for the self-employed pay jobs, while differences in returns are more important in and is close to the results for Argentina. The returns to their the average-pay jobs (accounting for close to 60 percent of skills become increasingly similar to the formal salaried as 89 I N F O R M A L I T Y we move up the earnings scale, and in fact they are remuner- of a relatively flexible labor market. Labor rigidities, how- ated slightly better than the formal salaried at the best- ever, are known to be relatively high in Argentina and paying jobs for their skills set. As in Argentina, all of the Bolivia, although other factors may also be at play--for earnings advantage of the best-paid formal salaried workers example, excessive costs of registration in Bolivia, lax over the best-paid independent workers arises from their enforcement, and coordinated general tax and payroll evasion more favorable characteristics rather than unequal pay, and in Argentina. yet the absolute level of the earnings gaps is less monotonic Again, these "unexplained" earnings gaps cannot be along the pay scale. interpreted as evidence of segmentation in the labor mar- Finally, the Dominican Republic shows a mixed pattern ket. They provide an order of magnitude of the pecuniary for the informal-formal salaried earnings gaps and a very value that informal job amenities may have for informal different situation for independent workers. Close to 60 per- salaried and independent workers net of the implicit value cent of the earnings gaps between formal and informal of the foregone benefits of formal salaried jobs. Yet, they employees in jobs of median pay or above are due to differ- provide inconclusive evidence to discriminate against the ences in their characteristics, but the latter explain little of segmentation and comparative advantage hypotheses. the gap in the low-pay jobs. That is, the lower earnings of Moreover, the results may still suffer from biases arising informal salaried workers in the bottom of the earnings from the failure to fully account for unobserved productiv- distribution are entirely due to lower remunerations to ity differences between workers. In fact, even if it accounts their characteristics. However, contrary to Argentina and for self-selection into the sectors, the analysis still assumes Bolivia, they are not as worse off in an absolute sense, since that workers' positions in the earnings distribution remain the level of the informal-formal salaried earnings gaps is intact were they to switch sectors. For example, it presumes actually much lower at the bottom than at the top of the that an informal worker positioned at the top of the infor- distribution. Meanwhile, consistent with the simple earn- mal earnings distribution would remain at the top of the ings differentials shown above, the results actually show formal salaried distribution were he or she to take a formal that independent workers are in a favorable situation salaried job. However, when comparative advantage is pres- regardless of their position in the earnings scale. In ent, some workers could emerge as top earnings performers fact, their advantage in earnings arises from better remu- for their skills in the informal sector, but if they were to nerations to their characteristics, which fully compensate move to a formal job they may actually lose out in the earn- for their less favorable personal and job characteristics, ings rankings if their unobserved skills are less rewarded in especially in the low-earnings jobs. If they had a similar set that sector. The decision to participate in the formal sector of average characteristics as formal salaried workers, they could depend on the expected return for the individual to would actually earn even more than their formal salaried both observed and unobserved characteristics. In this case, counterparts. the above comparisons of high-earnings workers in the In sum, all three countries show a clear earnings dis- informal and formal sectors would not give a true measure advantage of informal salaried workers that cannot be of the potential change in earnings that either would derive accounted for by lower observed productivity, and is indeed by moving across sectors. This complicates the estimation much larger in the low-paying jobs for any skills set. The through conventional regression methods. situation is similar for independent workers in Argentina Arias and Khamis (2007) apply recently developed and Bolivia, and markedly different in the Dominican econometric methodologies to Argentinean data (Heckman, Republic. There is substantial heterogeneity in earnings Urzua, and Vytlacil 2006; Heckman and Vytlacil 2001, performance, particularly among independent workers. The 2005) to deal with these issues and properly analyze the rel- similarities and differences in performance across countries evance of labor market comparative advantage in the partic- may be related to differences in country policies and institu- ipation and earnings performance of workers in the formal tions as well as the dynamism of the informal economy. and informal sectors. To correct for selection biases, they use For instance, the good performance of the self-employed in variations in the enforcement of labor legislation across the Dominican Republic has been associated with the Argentine provinces and workers' reported preferences for growing and well-performing tourism sector and construc- salaried or independent work as factors that affect employ- tion booms that the country has experienced, in the context ment sector participation, but not the earnings returns to 90 I N F O R M A L I T Y, E A R N I N G S , A N D W E L F A R E TABLE 3.2 Impact of informality on earnings in Argentina, estimated log-earnings differences and treatment parameters Formal salaried versus informal Formal salaried versus Informal salaried versus Treatment type salaried [SE] self-employed [SE] self-employed [SE] Treatment on the treated 2.088* 0.187 0.449 [0.187] [0.443] [0.426] Treatment on the untreated 1.892* 0.122 0.989** [0.204] [0.245] [0.510] Average treatment effect 2.002* 0.105 0.600* [0.105] [0.291] [0.244] Source: Arias and Khamis 2007. Note: SE = standard error. *p < .005. **p < .001. formality or informality. The methods yield a whole range of negative effect for those workers who have more self- earnings differentials as a function of workers' propensity to employed-like characteristics. That is, workers with inde- formal or informal employment, by comparing workers who pendent-like characteristics (observed and unobserved) are at the margin of indifference between jobs fit to their would receive lower earnings were they to move to formal skills and preferences in the three sectors. A summary of the salaried jobs, although the effect is not statistically signifi- results is presented in table 3.2. This presents a distinct set cant. The results are again consistent with workers' primar- of summary parameters to answer different policy questions: ily voluntary motivations to be independent in Argentina. the average treatment effect (ATE), that is, the mean earn- These results are in contrast to the findings from quantile ings gain from formality for a randomly selected worker; the earnings estimation and suggest that the sorting of workers treatment on the treated (TT), that is, the mean earnings based on differences in tastes for various types of work is of gain from formality derived by those with characteristics major importance for those entering self-employment. This similar to workers currently in formal jobs; and the treat- again underscores the importance of considering differences ment on the untreated (TU), that is, the mean earnings gain in the nonpecuniary qualities of independent work as (or loss) for those in informal (salaried or independent) jobs revealed by the motivational analysis of chapter 2. The were they to switch to formal salaried jobs.9 These are alter- absence of compensating differentials suggests that the per- native measures of the mean earnings gain from having a ceived amenities (for instance, independence and flexibility) formal occupation for workers with the same set of observed and disadvantages (for example, lack of benefits and risk) of and unobserved characteristics, who are indifferent between self-employment tend to cancel each other out. a formal and an informal job and are found participating in On the contrary, for informal salaried workers all the different sectors. parameter estimates are positive and large, so that informal The results corroborate the mixed view of the Argentine salaried work implies very high earnings penalties when labor market and support the importance of comparative compared to formal salaried work. Although the full results advantage in workers' selection into formal salaried and self- suggest some heterogeneity in the full comparison of earn- employment but also a role for segmentation. On the one ings of formal and informal salaried workers, the summary hand, the results reveal little difference in the earnings of treatment parameters are very similar, so that workers with formal salaried and independent workers once one fully informal-like characteristics (observed and unobserved) accounts for the sorting of workers based on preferences and would experience roughly similar earnings gains were they the returns to their observed and unobserved skills. All three to move to formal salaried jobs. To the extent that these treatment parameters are statistically insignificant. The are derived from comparing identical workers at the margin ATE is positive; however, this reflects a combination of a of indifference between the two sectors, they provide mea- positive earnings effect for workers whose characteristics sures of differences in earnings arising from nonpecuniary make them more prone to formal salaried work (TT) and a characteristics of jobs across sectors or from labor market 91 I N F O R M A L I T Y disequilibria or segmentation. The magnitude of earnings TABLE 3.3 gaps seems very large to arise from compensating earnings Correlation between income and self-rated poverty differentials and suggests the presence of segmentation between informal and formal salaried employment. This is Income consistent with the evidence cited above on the involuntary Country Nonpoor Poor Total Coincidence nature of individual reasons for being informal salaried in Argentina Argentina. The next section presents an analysis of subjec- Self-rated Nonpoor 45.1 13.8 58.8 tive response data that attempts to establish more definitely Poor 21.5 19.7 41.2 64.8 Total 66.5 33.5 100.0 whether the value of nonpecuniary job characteristics fully Dominican Republic compensates for the lower earnings of informal workers in Self-rated Nonpoor 46.4 19.0 65.4 Argentina as well as other countries. Poor 18.9 15.7 34.6 62.2 Total 65.3 34.7 100.0 Informality and self-rated welfare One way to assess whether the informal-formal earnings Source: Based on Arias and Lucchetti (2007). disparities reflect actual differences in welfare is by consid- ering individuals' perceptions of their own welfare. Some household surveys in a number of countries have collected of coincidence in the classifications: in both countries, data on subjective well-being by asking individuals ques- approximately 65 percent of individuals self-rate their sta- tions such as the following: "Do you consider yourself or a tus similarly to what conventional income poverty mea- member of your family to be: very poor, poor, nonpoor, or sures dictate. rich?" We use available surveys for Argentina, Bolivia, and Figure 3.6 also shows that self-rated poverty measures the Dominican Republic to examine the link between follow fairly closely the pattern of income poverty classifica- informal-formal occupational status and expanded notions tions by education level and labor market status of respon- of welfare beyond incomes. In Argentina and the Dominican dents. In the case of Bolivia, income poverty rates are Republic, the surveys include data that allow alternative calculated with the 1999 national living-conditions survey. measures of informal employment including pension cover- There is a strikingly similar correlation in the case of edu- age, firm size, and the temporary or permanent nature of cation; both poverties are lower among the well-educated. the labor contract.10 Regrettably, the Bolivian survey does There is also considerable coincidence in the case of occu- not allow a clear distinction between informal and formal pational status, although some noteworthy divergences salaried workers, other than through proxies of whether the are already apparent. For example, the gap between the worker is temporary and employed in a blue-collar occupa- self-rated and income poverty rates of the self-employed tion. Workers are classified according to two groups: those is much smaller than among the informal and formal answering "rich" or "nonpoor" and those answering "poor" salaried. Moreover, small-business owners tend to self-rate or "very poor." Except for Bolivia, the surveys contain suffi- less poor than their income would suggest. cient income data that allow classification of workers also in Figure 3.7 and table 3.4 present the main results of terms of their income poverty. Regressions are estimated the multivariate regression analyses for Argentina and relating the subjective responses to a large number of indi- the Dominican Republic. The coefficients in the figure rep- vidual and family characteristics including the job category resent the impact of alternative measures of informal variables, and comparing these with the results from simi- employment on the propensity of individuals to consider lar regressions of income­poverty status of the individual themselves poor (figures in bold indicate that the effect is according to conventional poverty-line analysis. zero at conventional statistical significance), and have a Before discussing the results, it is useful to establish similar interpretation to the odds ratios discussed earlier in some facts on the data. Table 3.3 first presents a compari- the chapter. The main conclusions are summarized below. son of the self-rated and income poverty classifications in The coefficients in the table are standardized for proper Argentina and the Dominican Republic. Despite the comparisons of the effect of the variables on self-rated markedly different processes involved in classifying work- (binary indicator) and income poverty (a continuous ers as income-poor and self-rated poor, there is a high level variable). 92 FIGURE 3.6 Fraction of workers who are income-poor and self-rated poor, by education and occupational group Percent Argentina Argentina 70 Formal salaried Informal salaried 60 Permanent contract 50 Temporary contract Large firm 40 Medium size enterprise Small firm 30 Microfirm Self-employed 20 Small firm employer Employed 10 Out of labor force Unemployed 0 Primary Secondary Tertiary 0 10 20 30 40 50 60 70 Percent Percent Bolivia Bolivia 70 Self-employed 60 Employee Underemployed 50 Private 40 Public White collar 30 Blue collar Employer 20 Employed 10 Out of labor force Unemployed 0 Primary Secondary Tertiary 0 10 20 30 40 50 60 70 Percent Percent Dominican Republic Dominican Republic 70 Formal salaried Informal salaried 60 Permanent contract 50 Temporary contract Large firm 40 Medium size enterprise Small firm 30 Microfirm Self-employed 20 Small firm employer Employed 10 Out of labor force Unemployed 0 Primary Secondary Tertiary 0 10 20 30 40 50 60 70 Percent Self-rated poor Income poor Overall self-rated poor Overall income poor Source: Based on Arias and Lucchetti (2007). Note: Self-rated poverty is based on individual responses. Income poverty is based on national poverty lines and per capita family incomes. 93 I N F O R M A L I T Y wealth, living conditions, and so on). However, the FIGURE 3.7 Dominican informal salaried consider themselves as poor as Impact of informality on self-rated poverty formal salaried workers, despite being more likely to be Argentina income-poor. Thus, incomes are not a good proxy for the Propensity to self-rated poverty actual welfare situation of the Dominican informal salaried Independent workers. As can be seen in figure 3.7 and table 3.4, these large firm results are strikingly robust to different definitions of infor- Independent formal salaried mal salaried employment. In the case of Bolivia, the regres- Independent sion results of Arias and Sosa-Escudero (2005) find that permanent contract employees, temporary salaried workers, and blue-collar Microfirm worker workers, most of whom are likely to be informal, also have large firm lower self-rated welfare than the formal salaried. Informal salaried formal salaried Contrary to the results for the informal salaried, inde- Temporal salaried pendent workers in both countries consider themselves as permanent contract poor as formal salaried workers. This is at odds with their 0 0.5 1.0 1.5 2.0 being more likely to be income-poor in Argentina, all other Odds ratio factors constant, but is consistent with their lower income Dominican Republic poverty in the Dominican Republic. Thus, in these two Propensity to self-rated poverty countries, independent workers report having similar levels of welfare as the formal salaried workers. This was also Independent large firm found to be the case in Bolivia, where the self-employed are Independent as likely to be self-rated poor as formal salaried workers, formal salaried holding other factors constant (Arias and Sosa-Escudero Independent permanent contract 2005). Microfirm worker An important finding also shown in figure 3.7 and large firm table 3.4 is that access to social protection correlates with Informal salaried higher self-rated welfare. In both Argentina and the Domini- formal salaried can Republic, individuals who live in families where some Temporal salaried permanent contract member has social security benefits are about 17 percent less likely to self-rate as poor (see figure 3.8). This result 0 0.5 1.0 1.5 2.0 Odds ratio also holds if access to benefits is proxied by employment in large firms or by having a permanent labor contract. Thus, Source: Based on Arias and Lucchetti (2007). Note: Self-rated poverty is based on individual responses. workers seem to attach an important welfare value to hav- ing access to social protection. The results that Dominican informal workers report the The impact of being an informal salaried employee on same level of welfare as formal salaried employees are self-rated poverty is identical to its impact on income corroborated by the recently collected data on job satisfac- poverty in Argentina, but is quite opposite that in the tion as part of the special informal survey in this country. Dominican Republic. The informal salaried Argentines Table 3.5 presents the responses to the question "How satis- consider themselves poorer than formal salaried workers, fied are you with the following aspects of your current job?" holding other factors constant, even when we control for for the four occupational groups. These questions refer to their own labor income and the incomes of other family workers' general satisfaction with their jobs and a variety of members. Moreover, the informal salaried in Argentina specific working conditions including incomes, working tend to self-rate as poor as their family income indicates, all hours, benefits, flexibility, and mobility opportunities. As other factors held constant. Thus, they have a lower level of can be seen, there is strikingly little difference in the levels self-rated welfare that cannot be explained on the basis of of satisfaction reported by informal salaried and informal their individual and family characteristics (proxies for independent workers relative to formal salaried workers 94 I N F O R M A L I T Y, E A R N I N G S , A N D W E L F A R E TABLE 3.4 Impact on the propensity to self-rate poor and be income poor (normalized logit regression coefficients) Argentina Dominican Republic Informality proxy Self-rated Income Self-rated Income Informality proxied by firm size Independent worker 0.06 0.27 0.07 0.12 Microfirm worker 0.17 0.23 0.13 0.06 Family member in large firm 0.11 0.34 0.02 0.39 Informality proxied by pension Independent worker 0.06 0.24 0.01 0.06 Informal salaried worker 0.26 0.27 0.04 0.10 Family member formal salaried 0.16 0.41 0.11 0.45 Informality proxied by type of contract Independent worker 0.04 0.23 0.04 0.06 Temporal salaried worker 0.27 0.29 0.00 0.09 Family member with permanent contract 0.15 0.41 0.16 0.35 Source: Based on Arias and Lucchetti (2007). Note: Coefficients are normalized so that they measure the relative weight of each explanatory variable on the total variation explained. Coefficients in bold are not statistically significant at the 5 percent confidence level. Regressions control for other individual and family characteristics. the informal is remarkable given that the differences in FIGURE 3.8 characteristics between the groups are not controlled. The Direct impact of having access to social protection through a family member on self-rated poverty only exception is with regard to the benefits offered by the job which, as can be expected, are regarded as inferior by Propensity to self-rated poverty the informal, as well as mobility opportunities (which may Odds ratio 1.4 reflect between-group differences in skills). The lower satis- faction with respect to incomes among the informal salaried 1.2 is exclusively due to family workers being included in this 1.0 group, when the latter are excluded, the rates become 0.8 similar to those for the formal salaried. Moreover, in 0.6 terms of flexible work schedules, the informal self-employed Dominicans report higher levels of job satisfaction. 0.4 Regrettably, the recent Argentine informal survey did 0.2 not include questions on job satisfaction; however, a 0 comparison with the responses to a similar question in Argentina Dominican Republic Colombia offers a useful polar case to benchmark. These are Family member with permanent contract presented in table 3.6. In sharp contrast to the Dominican Family member with formal salaried job Republic, the informal salaried and informal self-employed Family member in large firm workers in Colombia are clearly more dissatisfied with their jobs in all of the surveyed dimensions. In several cases, the Source: Based on Arias and Lucchetti (2007). Note: Self-rated poverty is based on individual responses. Income differences are fairly large, although still only a quarter of poverty is based on national poverty lines and per capita family incomes. informal workers report being unsatisfied with their jobs. The similarities in responses between the informal salaried (informality proxied by pension contributions). Interest- and the informal self-employed in Colombia are quite ingly, the formal self-employed report the higher levels of remarkable. The results are consistent with the overall job satisfaction. The fact that the overall job satisfaction rate lower degree of voluntariness directly revealed and implicit is only 5­7 percent higher for the formal salaried than for in the reasons to be in the current occupation for both 95 I N F O R M A L I T Y TABLE 3.5 Job satisfaction and informal employment in the Dominican Republic What is your overall level of satisfaction Formal Informal Formal Informal with your economic activity or main job? salaried salaried independent independent Very satisfied 9.3 5.7 25.2 8.8 Satisfied 69.3 67.6 55.0 62.4 Unsatisfied 20.0 25.0 14.7 25.0 Very unsatisfied 1.4 1.7 5.1 3.9 Total 100.0 100.0 100.0 100.0 What is your overall level of satisfaction Formal Informal Formal Informal with the following aspects of your main job? salaried salaried independent independent The number of hours you work Very satisfied 5.3 4.5 12.7 5.9 Satisfied 73.2 70.7 75.4 71.6 Unsatisfied 19.1 22.4 10.4 20.0 Very unsatisfied 2.5 2.4 1.5 2.5 Total 100.0 100.0 100.0 100.0 The level of income received (earnings and monetary pay) Very satisfied 3.9 2.4 7.2 3.7 Satisfied 48.5 40.8 61.9 49.2 Unsatisfied 40.1 39.9 22.3 36.8 Very unsatisfied 7.5 8.5 8.6 10.4 Not applicable/don't have 0.0 8.4 0.0 0.0 Total 100.0 100.0 100.0 100.0 Additional benefits received (health insurance, pension, paid vacations, and so on) Very satisfied 5.0 1.0 1.9 0.1 Satisfied 70.7 26.1 1.3 1.0 Unsatisfied 18.9 33.8 0.5 0.3 Very unsatisfied 5.5 18.9 0.0 0.1 Not applicable/don't have 0.0 20.2 96.3 98.5 Total 100.0 100.0 100.0 100.0 Flexible work schedules Very satisfied 6.1 5.4 15.1 8.8 Satisfied 70.2 72.7 80.1 78.8 Unsatisfied 21.7 19.5 4.1 10.9 Very unsatisfied 2.0 2.3 0.7 1.5 Total 100.0 100.0 100.0 100.0 Opportunities for economic mobility or to be promoted Very satisfied 6.1 3.6 9.5 5.2 Satisfied 58.0 46.7 61.7 46.4 Unsatisfied 32.2 39.4 15.3 35.3 Very unsatisfied 3.8 10.3 13.5 13.1 Total 100.0 100.0 100.0 100.0 Source: Author's estimates, based on household survey data, 2006. Note: Numbers are rounded. groups presented in chapter 2. These results for Colombia formal work, figure 3.9 presents some calculations derived and the Dominican Republic caution against simple- from the self-rated poverty regressions for Argentina. These minded generalizations across countries and highlight that reflect the difference between the actual average labor voluntariness or levels of job satisfaction can vary among incomes of various groups of workers and the income level both the informal self-employed and the informal salaried required to make them borderline of self-rating poor. That sectors across and within countries. is, these are measures of the "excess income" above the Finally, in order to have a quantitative notion of the amount individuals in each group need to reach a level of potential value attached to the amenities of informal and welfare that puts them with a 50:50 chance of self-rating 96 TABLE 3.6 Job satisfaction and informal employment in Colombia What is your overall level of satisfaction Formal Informal Formal Informal with our economic activity or main job? salaried salaried independent independent Very satisfied 13.2 3.0 14.9 4.2 Satisfied 79.3 69.9 71.1 70.2 Unsatisfied 7.4 27.1 14.0 25.6 What is your overall level of satisfaction Formal Informal Formal Informal with the following aspects of your main job? salaried salaried independent independent The number of hours you work Very satisfied 8.0 2.0 10.0 2.3 Satisfied 78.8 67.4 69.6 67.4 Unsatisfied 13.2 30.6 20.4 30.3 The level of income received (earnings and monetary pay) Very satisfied 6.8 1.3 8.9 1.8 Satisfied 62.5 45.3 55.4 45.3 Unsatisfied 30.7 53.3 35.7 53.0 Additional benefits received (such as health insurance, pension, and paid vacations) Very satisfied 8.3 0.9 5.1 1.0 Satisfied 73.7 28.7 43.5 29.4 Unsatisfied 18.0 70.4 51.4 69.6 Flexible work schedules Very satisfied 5.9 1.2 8.5 1.7 Satisfied 81.1 68.1 71.7 68.3 Unsatisfied 13.0 30.7 19.8 30.0 The application of your knowledge at work Very satisfied 17.6 6.4 22.3 8.7 Satisfied 75.7 76.1 66.8 75.3 Unsatisfied 6.7 17.5 10.9 16.0 Source: Author's estimates, based on household survey data, 2006. Note: Numbers are rounded. FIGURE 3.9 Differences between actual labor incomes and the level of income needed to avoid self-rating poverty in Argentina Income (1997 Argentine pesos) 400 By sex By marital status By level of education 300 200 100 0 100 200 300 400 500 479 600 Men Women Married Unmarried Primary Secondary Tertiary Informal Self-employed Formal Source: Based on Arias and Lucchetti (2007). Note: Estimates based on the logit regression of self-rated poverty. Positive (negative) values indicate that average labor incomes exceed (fall short of) the self-rated poverty threshold of individuals. 97 I N F O R M A L I T Y poor. The negative values for the informal salaried indicate to address the barriers to more desirable formal salaried that they actually have an income deficit--that is, they earn jobs are of first-order importance, and may include removal less than what is needed to be equally likely to self-rate of labor market frictions, tighter enforcement and improve- poor or nonpoor. The fact that independent workers have ment of labor and tax laws, and improvement of the role of lower values than the formal salaried means that they derive unions in achieving better overall employment, wages, and a similar level of self-rated welfare with less income. productivity growth. Since the regressions hold other observed individual As noted in chapter 2, there is a two-tier divide among and family characteristics constant, these differences may independent workers: a majority of workers who choose partially reflect the significance of the above-mentioned this sector voluntarily and conform closer to entrepreneur- nonpecuniary aspects of employment for the two informal ship motives, and an important but lower fraction who use groups. Although the data do not provide sufficient grounds it as a safety net. The lower tier faces significantly higher to estimate the imputed value of formal benefits in relation earnings disadvantage in the low-paying jobs, but the best- to the cost of contributions and of the nonpecuniary amenities paid independent workers enjoy remunerations similar to of informal work, the figures are suggestive of the relative the best-paid formal salaried workers with similar charac- significance of these factors for different groups of workers. teristics. The evidence for Argentina is consistent with The implications for antipoverty and social protection workers sorting into formal salaried and self-employment policies will be discussed in detail in chapter 7. occupations according to labor market comparative advan- tage. That is, some workers find advantageous niches for Conclusions and policy implications their observed and unobserved skills and tasks in sectors or This chapter lends credence to both the "exclusion" and occupations where jobs have a different propensity to be "voluntary" nature of informal employment suggested by exercised as formal salaried or independent. Moreover, the workers' reported motivations to be in their current occu- results on self-rated welfare differentials suggest that non- pation presented in chapter 2. Evidence from workers' pecuniary factors may be to a large extent compensating reported motivations to be in their current occupation, the independent workers for any lower earnings. This is consis- sources of earnings differentials, and self-rated welfare tent with chapter 2's survey-reported motivations for being assessments lends support to the view that the majority of independent and not contributing to social security. independent workers are largely voluntary and attach sig- However, the Dominican Republic (as well as Mexico, as nificant value to the nonpecuniary benefits of autonomous will be argued in chapter 4) and Colombia provide excep- work. Meanwhile, informal salaried workers tend to be tions that caution that this distinction between the infor- excluded from more desirable jobs in either formal salaried mal salaried and the informal self-employed need not be or self-employment. true in every country. Overall, the nature and determinants The existence of a sizable earnings differential between of informal employment would depend on country-specific informal and formal salaried workers, unrelated to compen- contexts, particularly on cross-country and over time varia- sating differentials, like those found in Argentina and tion in formal sector productivity, the demographic and Bolivia, has implications for the functioning of labor mar- skills composition of the labor force, and the incentives to kets. This can reflect "queues" for formal salaried sector comply with tax and labor regulations (including partici- jobs, given that they are comparatively better paid across pation in the social security system). the spectrum of low- and high-paid jobs in the labor mar- The fact that a fraction of independent workers enjoys ket and have social benefits. It may, for instance, be a prod- higher--although presumably more variable--incomes uct of the labor market not being flexible enough to but are not contributing to social security poses the ques- equalize earnings through arbitrage. However, as discussed tion: How much will they be willing to pay for the social in chapter 2, it may be related to other sources of labor seg- protection benefits that formal wage earners enjoy (which, mentation. For instance, it may reflect a serious problem of as indicated above, are positively valued by workers)? general evasion of income and value-added taxes that must The issue of employment protection and old-age security be addressed (at least in part) with tighter enforcement and merits an analysis that considers the incentives to par- improvements in the structure of these taxes. Thus, policies ticipate in the social security system of workers with 98 I N F O R M A L I T Y, E A R N I N G S , A N D W E L F A R E different preferences regarding job flexibility, with differ- methods used here are different, the results might reflect the differ- ent concerns as to their futures, with different intertempo- ence in time periods. The informal-formal salaried earnings gap has widened considerably in the last 10 years. ral discount rates, and with respect to who derives different 9. As shown by Heckman and Vytlacil (2001, 2005), these para- levels of welfare from a particular benefit package. Workers meters can be derived from an estimate of the marginal treatment may have a differing willingness to pay or accept lower effect using local instrumental variables (LIVs) and other approaches. take-home earnings in exchange for such benefits depend- The results shown here are based on semiparametric estimation but ing on their individual preferences, the cost and quality of also robust to different empirical specifications and alternative esti- the services (real and perceived) provided by the public and mation approaches. (See Arias and Khamis [2007]). 10. The surveys used are the 1997 Encuesta de Desarrollo Social private sectors, and the characteristics of alternative sources in Argentina, the 1999 Encuesta Nacional de Aspiraciones y Priori- of services and benefits not related to the labor contract (for dades de Desarrollo Humano in Bolivia, and the Encuesta de Condi- example, informal insurance, social networks, and such). ciones de Vida in the Dominican Republic. There also may be a role for flexible benefits plans. Chapter 7 analyzes these issues in more detail, while the next chap- References ter summarizes the main lessons for labor market policies Arias, O., and M. Khamis. 2007. "Comparative Advantage and Infor- emerging from the existing literature for Latin America mal Employment." Photocopy. World Bank, Washington, DC. and this report. Arias, O., and W. Sosa-Escudero. 2005. "Subjective and Objective Poverty in Bolivia." Background paper for the 2005 World Bank Notes Bolivia Poverty Assessment. World Bank, Washington, DC. 1. This effect cannot be identified for independent workers, Arias, Omar, Fernando Landa, and Patricia Yáñez. 2007. "Movilidad because the majority of them are in microfirms and one-person Laboral e Ingresos en el Sector Formal e Informal de Bolivia." enterprises. Documento de Trabajo, UDAPE, La Paz, Bolivia. 2. Earnings of formal salaried are net of labor tax contributions, Arias, Omar, and Monserrat Bustelo. 2007. "Perfiles y Dinámicas del while independent earnings are computed netting out the costs and Empleo Informal en América Latina." Photocopy. World Bank, returns to capital as much as labor surveys allow. Washington, DC. 3. Maloney and Mendez (2003) find evidence that the influence Arias, Omar, and Leonardo Lucchetti. 2007. "Informal Employment of minimum wage norms is far stronger on the more concentrated and Self-Rated Welfare: Measuring Compensating Differentials." informal earnings distribution in Argentina. Photocopy. World Bank, Washington, DC. 4. A low-earnings formal job gets the earnings ceiling of the bot- Carneiro, P., J. Heckman, and E. Vytlacil. 2001. "Estimating the tom 20 percent formal workers (the 20th percentile of the Rate of Return to Education When It Varies among Individuals." formal salaried distribution), while a high-earnings formal job gets the Paper presented at the Royal Economic Society annual meeting. earningsfloorofthetop20percentformalworkers(the80thpercentile). Durham, UK, April. 5. The economics literature offers various rationales for lower Galiani, S., and F. Weinschelbaum. 2006. "Modelling Informality returns to experience in independent work. While pay raises based on Formally: Households and Firms." Photocopy. World Bank, seniority are used in salaried work to induce the best workers to Washington, DC. remain in the firm, the self-employed have less incentives to shirk in Heckman, J., and G. Sedlacek. 1985. "Heterogeneity, Aggregation, the job (or quit). Also, since independent workers often have a and Market Wage Functions: An Empirical Model of Self- sunken investment at the start-up of their businesses, they may not selection in the Labor Market." Journal of Political Economy 93 be able to move out quickly from a poor-performing activity. (6): 1077­125. 6. The chapter loosely uses the term "low- and high-paying jobs" Heckman, J., and E. Vytlacil. 2001. "Policy-relevant Treatment to refer to the conditional percentiles of the earnings distribution. Effects." American Economic Review 91 (2): 107­11. 7. For more on the empirical testing of labor market segmenta- ------. 2005. "Econometric Evaluations of Social Programs." In tion and compensating differentials theories, see Magnac (1991). The Handbook of Econometrics, Vol. 5, ed. J. Heckman and E. Leamer. earnings gap decomposition technique was first presented by Oaxaca Amsterdam: North-Holland. (1973). For more on earnings decompositions, see Oaxaca and Heckman, J., S. Urzua, and E. Vytlacil. 2006. "Understanding Ramson (1994). Instrumental Variables in Models with Essential Heterogeneity." 8. The results here differ from the findings of Pratap and Review of Economics and Statistics 88(33): 389­432. Quintin (2006), who found no evidence of systematic differences in Magnac, T. 1991. "Segmented or Competitive Labor Markets?" earnings between formal and informal salaried workers. They use Econometrica 59 (1): 165­87. 1993­95 Encuesta Permanente de Hogares (EPH) data and propensity Maloney, W. F. 1999. "Does Informality Imply Segmentation in scores methods to deal with the problem of bias due to self-selection Urban Labor Markets? Evidence from Sectoral Transitions in of low productivity workers into the informal sector. While the Mexico." World Bank Economic Review 13: 275­302. 99 I N F O R M A L I T Y Maloney, W., and J. N. Mendez. 2003. "Measuring the Impact of Pratap, S., and E. Quintin. 2006. "Are Labor Markets Segmented in Minimum Wages: Evidence from Latin America." Working Paper Argentina? A Semiparametric Approach." European Economic Review 9800, National Bureau of Economic Research, Cambridge, MA. 50 (7): 1817­41. Oaxaca, R. 1973. "Male-Female Wage Differentials in Urban Labor Rosen, S. 1981. "The Economics of Superstars." American Economic Markets." International Economic Review 14: 693­709. Review 71: 845­58. Oaxaca, R., and M. Ramson. 1994. "Discrimination and Wage Smith, A. 1876. The Wealth of Nations. Chicago: University of Decomposition." Journal of Econometrics 61 (1): 5­21. Chicago Press. 100 CHAPTER 4 The Informal Labor Market in Motion: Dynamics, Cycles, and Trends SUMMARY: This chapter takes an aggregate view of the informal sector, asking what determines the size of the sector, both across the business cycle and across longer periods of time. In the process, we shed more light on the razón de ser of the informal sector, the overall functioning of labor markets in least-developed countries (LDCs), and what legal and regulatory factors affect the allocation of workers across sectors. The chapter begins with a discussion of gross labor flows--the movements of workers across types of work--to establish the dynamic relationships among sectors. The chapter then takes a step back and locates the labor market in the context of a standard small-country macroeconomy model to explain pro-cyclical movements in informality. Finally, it examines determinants of the longer-term trends observed in LDC labor markets. C HAPTER 3 DISCUSSED THE DECISION- updated view of how the labor market adjusts across the making process of workers choosing business cycle. Overall, the gross labor flow analysis offers between the formal and informal sectors. further evidence that, as a first approximation, informal This chapter takes a more aggregate view, jobs--particularly those in the informal self-employed asking what determines the size of the sec- sector--are not obviously worse than formal sector jobs for tor, both across the business cycle and across longer periods many workers. This does not imply that there are no invol- of time. In the process, we shed more light on the razón de untary components, or that there are no distortions in the ser of the informal sector, the overall functioning of LDC business climate, but it does offer an importantly different labor markets, and what legal and regulatory factors affect lens through which to view these issues. the allocation of workers across sectors between the formal The chapter then takes a step back and locates the and informal sectors. labor market in the context of a standard small-country We begin the chapter with a discussion of gross labor macroeconomy model. This approach allows for a richer flows--the movements of workers across sectors of work exploration of the interaction of macroeconomic shocks and and unemployment--to establish the dynamic relation- the labor market. More specifically, it offers an explanation ships among sectors. This approach has two advantages. for the pro-cyclical movements we sometimes see. In addi- First, it offers a complement to the traditional comparisons tion, it shows how looking at the relative size and relative of wages adjusted for human capital that, because of their earnings of the informal sector can offer a diagnostic of the inability to account for unobserved job characteristics and state of the labor market. nonpecuniary welfare effects, turn out to be faux amis in the Finally, in the last section we explore the long-run quest to establish segmentation in the market or relative determinants of the size of the informal sector and, in the inferiority or job quality of sectors. Second, it allows us to process, examine some explanations for the expansion of draw on the recent advanced country literature to offer an informality in some countries over the 1990s. 101 I N F O R M A L I T Y Informality through the lens of gross labor primarily an employment sector of last resort. More gener- force dynamics ally, studies of labor market dynamics have moved to the The recent availability of panel data sets that permit follow- center of the debates about how advanced-country markets ing workers across employment states allows us to revisit adjust and offer important lessons for Latin America. In several long-standing issues in the study of developing- the United States, recent work by Davis and Haltiwanger country labor markets and, in particular, the role of the (1992, 1999), Hall (2005), and Shimer (2005a, 2005b, informal sector. As noted in the previous chapter, a sub- 2005c) have documented the huge amount of churning of stantial body of literature with intellectual roots in the workers among sectors. For instance, Davis, Faberman, Harris-Todaro (1970) model sees informal workers as con- and Haltiwanger (2005) show that 10 percent of U.S. stituting the disadvantaged component of a labor market workers separate from their employers each quarter, some segmented by wage rigidities. This general view has impli- moving directly to a new job with a new employer, some cations for the cyclical adjustment of the labor market dur- becoming unemployed, some exiting the workforce. This ing recessions: downward rigidities prevent wages from literature has set off a debate about the sources of unem- adjusting to adverse shocks to the formal sector, leaving the ployment during downturns, whether caused by the informal sector to absorb workers who would be unem- destruction of jobs (Davis and Haltiwanger 1992, 1999) or ployed in societies where workers could afford to be so. As by the cessation of hiring (job finding) (Hall 2005; Shimer will be discussed in the next section, on average, it does 2005a). It has also introduced several new tools and appear, although with important and frequent exceptions, uncovered striking findings that, as we will show, are that the informal sector shows a countercyclical behavior directly applicable to the developing-country context and, consistent with this view and especially during crises. Fig- in particular, explain the cyclical movements of the infor- ure 4.1 suggests that, especially during the Tequila crisis of mal sector. 1995 in Mexico and the 1999 Brazilian crisis, formality fell Bosch and Maloney (2006) and Bosch, Goñi, and Maloney along with the increase in unemployment. (2006) draw on two rotating panels constructed from Worker flows provide the movie to the snapshot pro- employment surveys that permit estimating the probabili- vided by the simple stock indicators presented in chap- ties of transiting among labor market states and how these ters 1 and 2 and further support the idea that much of the change across time in Brazil and Mexico. Box 4.1 details informal sector, and particularly the self-employed, is not the theory and some relevant empirical context. FIGURE 4.1 Formal share of the labor force and unemployment, Brazil and Mexico Brazil Mexico % formal Unemployment rate (%) % formal Unemployment rate (%) 65 8 65 7 7 6 60 60 % formal 6 5 Unemployment rate 5 4 55 55 4 3 % formal Unemployment 3 rate 50 50 2 1983m1 1988m1 1993m1 1998m1 2003m1 1987q1 1991q1 1995q1 1999q1 2003q1 Sources: Bosch, Goñi, and Maloney 2006; Bosch and Maloney 2006. 102 T H E I N F O R M A L L A B O R M A R K E T I N M O T I O N : D Y N A M I C S , C Y C L E S , A N D T R E N D S BOX 4.1 Conceptual issues in gross worker flows Over the last 20 years, the study of the labor market has question. It does, however, highlight potentially impor- departed from the Marshallian view of instantaneous tant directions for the modeling of labor markets with adjustments, where prices (wages) are the only allocation informal sectors. mechanism, toward a view of trade in the labor market as First, empirical evidence has shown that informal work- an uncoordinated, time-consuming, and costly process ers concentrate their activities in the provision of nontrad- for both firms and workers (Pissarides 2000). Agents in ables, whereas formal firms operate mainly with tradables. the labor market have to spend time and resources Hence, it would be plausible to assume an economy where in order to achieve the desired outcomes. Workers are in there are formal firms posting vacancies and, in parallel, search of jobs offered by the firms while firms are looking there are informal firms posting vacancies in a different for workers to fill their vacancies in order to start sector (or, equivalently, self-employment opportunities production--and even if they agree on the wage, there is arise). Workers may then direct their search toward one a chance that they may not find each other. of the sectors or search randomly. The first case gives rise to The process by which firms and workers find each two different market tightness conditions and to an arbi- other has been the subject of researchers trying to under- tragecondition.Thatis,theexpectedreturnfromthesearch stand the functioning of the labor market. One of the must be equal in both sectors. This actually implies that if dominant views is that this process is governed by a wages are higher in the formal sector, the average waiting "matching function," a sort of production function that time must be shorter in the informal sector. If workers brings together vacancies and workers, and whose inputs search randomly, then the probability of finding a formal are the number of vacancies and the number of job job is given by the ratio of formal to informal vacancies. searchers, m m(u,v). This matching function is nor- Albrecht, Navarro, and Vroman (2006) propose even mally considered to be increasing in both arguments and another mechanism to generate an informal sector in a has constant returns to scale. The ratio of vacancies searching and matching framework. Assume that workers to workers determines what the literature calls market are heterogeneous. The highly educated workers always tightness, v/u. work in the formal sector, whereas the uneducated This matching function is the workhorse of an always work in the informal sector. The middle-class immense amount of research aimed at understanding the workers shift between the formal and the informal sec- flows between employment and unemployment in an tors, depending on the incentives provided by policy equilibrium framework. From it, one can obtain what is makers. Similarly, Bosch (2006) also employs worker het- the probability of an unemployed worker finding a job by erogeneity to generate an informal sector. Firms post dividing the number of matches by the unemployment vacancies, initially undefined; when the match between rate m(u,v)/u. the firm and the worker has been established, the firm has The relevant question is how this process works in a to decide what type of contract it wants to sign (formal or country with a division between protected or formal and informal). This decision depends on the idiosyncratic unprotected or informal jobs. This question is actually productivity of the match and on labor cost in the formal not very different from the distinction between skilled sector. This approach implies that firms decide a thresh- and unskilled jobs that has been widely studied in a old level of formality. If the match is sufficiently good, developed-country framework (Acemoglu 2003; Albrecht the firm is willing to bear the higher labor cost of a for- and Vroman 2002; and Dolado, Jansen, and Jimeno mal job. If not, it will hire the worker informally. 2002). It requires making a number of assumptions Labor markets in developing countries most likely about how the labor market works. Do firms post both contain some or all elements discussed here. Understand- types of vacancies? Do workers look for both types of jobs? ing how searching and matching occur in the presence of The theoretical discussion of how the search process informal labor markets is essential to determine the effect works in developed economies is by no means a settled of public policies in the labor market. 103 I N F O R M A L I T Y Cyclical patterns in gross labor flows countries, flows among the formal sector and the two Transitions between formality and informality informal sectors are remarkably symmetrical and appear Figure 4.2 offers additional evidence in favor of a more inte- highly correlated across the business cycle, rising in upturns grated view of the labor market where informal jobs offer and falling in downturns. That is, it is not the case that, as the distinct, but not necessarily inferior, employment opportu- economy recovers, we see fewer individuals being thrown out nities, particularly self-employment. The figure plots the of formality into the informal sector and more informal indi- probabilities of transition among formal and informal sec- viduals being able to find jobs in the formal sector. Such tors, the raw transitions in Mexico and in Brazil, in the latter asymmetric behavior does characterize flows between case detrended to account for the longer-term rise in infor- employment and unemployment in the Organisation for mality across the period that will be discussed later. In both Economic Co-operation and Development (OECD) and, in FIGURE 4.2a Probability of transition between formal salaried and self-employment, Brazil and Mexico Brazil Mexico Probability F to SE Probability SE to F Probability F to SE Probability SE to F 0.4 1.5 18 7 0.3 Formal to self-employment 1.0 16 0.2 6 0.1 0.5 Formal to 14 0 self-employment 5 0.1 0 12 Self-employment 0.2 4 0.5 to formal 10 0.3 Self-employment to formal 0.4 1.0 3 08 1987q1 1991q1 1995q1 1999q1 2003q1 1983 1985 1987 1989 1991 1993 1995 1997 1999 2001 2003 Sources: Bosch, Goñi, and Maloney 2006; Bosch and Maloney 2006. FIGURE 4.2b Probability of transition between formal salaried and informal salaried, Brazil and Mexico Brazil Mexico Probability F to IS Probability IS to F Probability F to IS Probability IS to F 0.8 5 16 45 0.6 Informal salaried to formal 4 3 0.4 2 14 0.2 1 40 Formal to 0 0 informal salaried 0.2 1 12 35 2 0.4 Informal salaried 3 Formal to informal salaried to formal 0.6 4 30 0.8 5 10 1987q1 1991q1 1995q1 1999q1 2003q1 1983 1985 1987 1989 1991 1993 1995 1997 1999 2001 2003 Sources: Bosch, Goñi, and Maloney 2006; Bosch and Maloney 2006. Note: Magnitudes differ between Brazil and Mexico due to technique used to detrend Brazil. 104 T H E I N F O R M A L L A B O R M A R K E T I N M O T I O N : D Y N A M I C S , C Y C L E S , A N D T R E N D S Brazil and Mexico, between all sectors of employment both informal sectors are substantially higher than the formal, formal and informal, and unemployment. But, among sectors of but relatively close to each other at a remarkably low 6 per- employment, what we see are flows in both directions cent. Broadly speaking, then, in Mexico, sentiments about increased during upturns and decreased during downturns, job quality seem similar among the self-employed and exactly consistent with the pro-cyclical patterns in job-to- informal salaried, consistent with the high correlation job flows observed in U.S. literature (Shimer 2005b) that are across the cycle of gross flows across all sectors. generally attributed to workers searching for and finding However, as chapter 3 notes, this differs strikingly better jobs in tighter job markets. across countries. In the Dominican Republic, self-reported The results are most striking for Mexico, but the story is happiness among informal salaried does not suggest that broadly supported with some caveats in Brazil as well. In either is worse-off than formal salaried. However, in several both countries, the gross flows among the informal, particu- countries, the informal salaried clearly are. The motiva- larly the self-employed, and formal sectors do not behave as tional responses of Brazilian informal salaried workers sug- if the former were a kind of unemployment, but rather as an gest that a large fraction are queuing there as well. This alternative job. This is consistent with response data pre- may account for the much lower cyclical correlation sented in chapters 2 and 3 that most of the self-employed observed (0.26 for I-F/F-I transitions versus 0.85 for SE-F/ choose to enter the sector and show equal levels of welfare F-SE transitions). as those in the formal sector. For Mexico, figure 4.3 shows Perhaps paradoxically, a high degree of voluntary entry that 75 percent report voluntary entry from both formal and does not rule out some degree of segmentation for certain informal salaried work, even during the crisis, although, workers, nor distortions more generally. First, at any time, predictably, the share declines substantially then. there are workers seeking jobs in each sector. Introducing a An absence of complementary data for workers entering binding minimum wage or other formal sector wage rigid- informal salaried work leaves us to rely on less direct evi- ity will reduce the probability of finding a job in the formal dence. If we assume that employed workers searching for sector and make the available informal jobs relatively less another job ("Have you been looking for a job over the last attractive, thereby introducing a progressive asymmetry to two months?") are "less voluntarily" employed, then it the flows and a lower degree of voluntary entry, as rigidities appears from figure 4.4 that in the early 1990s and early become more severe, especially during downturns. 2000s, effectively, job satisfaction was relatively similar Second, a broad class of distortions--excessive labor across sectors. In the peak of the recession in 1996, both taxes or firing restrictions, for example--may not induce FIGURE 4.3 FIGURE 4.4 Involuntary transition to self-employment in Mexico Searching while employed, Mexico Percent Percent 25 80 Searching Searching (informal salaried) (self-employed) 20 60 15 Searching (formal) 40 10 20 5 0 0 1992 1994 1996 1998 2002 1987q1 1991q1 1995q1 1999q1 2003q1 Source: Quarterly data from the National Urban Labor Survey Formal self Informal self All 1987:Q1 to 2002:Q4. Note: Searching refers to the proportion of employed workers in Source: Author's estimations, based on Encuesta Nacional de Micro the sector who claim to be looking for a new job and have not Negocios, several years. changed employment status in the previous quarter. 105 I N F O R M A L I T Y segmentation, but will nonetheless reduce the vacancies FIGURE 4.5 opened in the formal sector. Effectively, this will be analo- Decreased availability of formal sector jobs without segmentation gous to a shift in the formal sector demand for labor, and will lead to lower earnings in both sectors; but the vast Wage I Wage F Increase in taxes on majority of workers entering informality may still declare formal wages that they do so voluntarily, given earnings in the formal sector (see figure 4.5). This point is important: it is entirely possible for all workers found in the informal sector to be as WI WF well-off as they would be in the formal sector, in spite of WI WF very high, nonsegmenting distortions in the formal sector. DF DI At the limit, we could imagine an oppressive business DF climate that prevented growth and the creation of new, modern sector jobs, but where workers freely chose LI LF between the extant menu of vacancies. A discussion of models of the impacts of regulations on worker flows and LI LF the size of the informal sector is found in box 4.2. Source: Authors' calculations. BOX 4.2 Simulated effects of labor market legislation on the size of the informal sector Several models have been developed to examine the plau- Table 4B.1 shows the simulations of three changes in sible impact of policies on labor market variables. These labor market policies: a 15 percent rise in minimum wage, are generally calibrated with a generic economy in mind the introduction of dismissal costs worth three months' and, hence, should be taken only as suggestive. wages, and an increase of unemployment insurance bene- Albrecht, Navarro, and Vroman (2006) and Pries and fits equivalent to 20 percent of the wage funded by a Rogerson (2005) calibrate matching models and simulate 15 percent tax on output. In all cases, we see a reduction in the effects of labor market regulation on labor force turn- the probability that workers will find a job vacancy and over and composition. Although Pries and Rogerson work substantial changes in worker turnover (rather than job only with employment and unemployment, Albrecht, destruction) that drive the results. All policies lead to a Navarro, and Vroman (2006) explicitly model the informal reduction in the employment rate of between 1 percent- sector as well. age point (dismissal costs) and 4 percentage points (mini- The Pries-Rogerson model combines variants of two mum wages) and similar levels of "formal" sector output. benchmark models from the literature: the Jovanovic The welfare costs are high only for the minimum wage. (1979) learning model and the Pissarides (1985) match- Under a combination of policies, the interactions roughly ing model. Job flows are driven by idiosyncratic shocks to double the impact of all single policies. job productivity, and worker turnover (in excess of job Albrecht, Navarro, and Vroman (2006) extend turnover) is driven by a process of accumulation of infor- Mortensen and Pissarides (1994) by adding an informal mation about the quality of the match. Both parties to a sector and allowing for heterogeneity among workers. match observe a signal about the match's true quality They assume that workers differ in their maximum pro- prior to deciding whether to form a match, and matches ductivities (low, medium, or high productivity) in formal form only if they judge that the match quality exceeds a sector jobs. The decision about whether to accept an threshold value. True quality is revealed over time, but informal sector job thus depends on a worker's type. All only if a match is formed. Labor market regulations have workers have the option to take up informal sector oppor- an impact by influencing hiring practices--specifically, tunities as these come along, and all workers are equally the level of the threshold. productive in that sector, but workers who are most 106 T H E I N F O R M A L L A B O R M A R K E T I N M O T I O N : D Y N A M I C S , C Y C L E S , A N D T R E N D S TABLE 4B.1 Effects of varying minimum wages, dismissal costs, unemployment insurance, and taxes Differences with respect to benchmark of equilibrium Policy Affected variable Minimum wage Dismissal cost Unemployment insurance Taxes Probability to meet a vacancy 0.13 0.04 0.14 0.13 Unemployment duration 3.02 1.32 1.50 1.06 Annual job destruction (%) 0.20 0.00 0.30 0.30 Annual worker turnover (%) 6.00 4.3 1.80 0.50 Employment rate 0.04 0.01 0.02 0.02 Output 0.04 0.01 0.03 0.02 Welfare loss (%) 1.40 0.18 0.29 0.29 Source: Pries and Rogerson 2005. TABLE 4B.2 Effects of varying severance taxes and payroll taxes Differences with respect to benchmark of equilibrium Policy Affected variable Severance tax Payroll tax Both taxes Labor tightness (vacancies/unemployment) 0.360 0.090 0.230 Low productivity (informal)--medium productivity cut-off 0.053 0.063 0.065 Medium productivity--high productivity (formal) cut-off 0.048 0.101 0.085 Unemployment 0.018 0.003 0.005 Productivityformal 0.061 0.036 0.003 WageFormal 0.069 0.070 0.070 Output 0.019 0.004 0.008 Source: Albrecht, Navarro, and Vroman 2006. productive in the formal sector will reject informal work matches, reducing formal productivity and pushing the in order to wait for a formal job. Similarly, the least pro- unemployment rate down. In this case, the latter effect ductive workers are not hired in the formal sector. The dominates and unemployment falls. Overall, as fewer for- cut-off values that determine how the different types of mal vacancies are open, more workers are forced into workers are allocated across sectors are endogenous and accepting informal jobs. A rise in payroll taxes from 0 to are influenced by formal sector labor market policy. A 20 percent has similar effects on the reduction of vacancy policy change can disqualify some workers from formal creation and the compositional shift toward informal jobs. sector employment; similarly, some workers accept infor- However, contrary to the severance tax, the introduction mal sector work, although they would not have done so of payroll taxes makes the firms keep only highly produc- earlier. tive matches. This increases average formal productivity Effects of raising severance taxes and payroll taxes but also increases the unemployment rate. The last col- under this framework are reported in table 4B.2. The first umn of the table reports the effects of applying both column shows that an increase in severance payments policies simultaneously (each tax is set to be equal to 10 from 0 to 20 percent of the wage makes creating vacancies percent). Both taxes make vacancy creation less attractive, less attractive for formal sector employers. This should, in employment duration in the formal sector increases (that principle, increase unemployment. However, a sever- is, the severance tax effect dominates), productivity falls ance tax also encourages firms to keep low-productivity in the formal sector, and net output decreases. 107 I N F O R M A L I T Y FIGURE 4.6 Probability of transition to unemployment (separation rate), Brazil and Mexico Brazil Mexico 5 20 4 15 3 10 2 5 1 0 1983m1 1988m1 1993m1 1998m1 2003m1 1987q1 1991q1 1995q1 1999q1 2003q1 Formal Informal salaried Informal self-employed Sources: Bosch, Goñi, and Maloney 2006; Bosch and Maloney 2006. The cause of unemployment: Separations driven by the fact that most workers in formal firms are in from the informal sectors large firms. Smaller and medium-size formal firms are A second provocative finding emerges from the flows in extremely countercyclical in their separations. This is, per- and out of unemployment that suggest some rethinking haps, consistent with smaller firms being more likely to fail about the informal sector serving as sort of disguised in a bad economy (Jovanovic 1982; Hopenhayn 1988) or unemployment. Figure 4.6 suggests first that, at any time, perhaps it is because a lack of access to credit to smooth over the probability of separation from the informal salaried difficult times leads to smaller firms being unable to hang on sector is much higher than from the formal sector. In both to valuable workers while larger firms can.2 However, it is Mexico and Brazil, informal salaried workers are roughly also the case that informal workers in firms of all sizes in three times more likely to transit to unemployment, com- Mexico show extraordinarily high separation probabilities. pared with their formal counterparts. Second, job separa- This would seem to suggest that, again, it is the type of con- tion probabilities present much higher volatility in the tractual relationship that is determining firing behavior. informal sector, especially for the informal salaried. Simu- However, an alternative possibility is that, though these lations for Mexico (Bosch and Maloney 2006) and Brazil workers report that they are working in large firms, in fact (Bosch, Goñi, and Maloney 2006) suggest that, in fact, the they are contracted by intermediaries that are smaller firms in-creases in unemployment during recessions are gener- and that hire and fire as such. ated primarily by workers separating from the informal sectors. Why the informal sector generally A number of possible mechanisms may be at play here. expands in downturns First, workers in the informal sector may be occupying low- That said, why does figure 4.1 suggest that the informal sec- skill jobs for which training or experience is not excessively tor expands during downturns, much as traditional views important. During a downturn, these workers are easily dis- would suggest? Indeed, as Gasparini, Haimovich, and posed of since they are not protected by labor regulations. Olivieri (2006) and Loayza and Rigolini (2006) have shown, Further, informal workers are employed normally by small this appears, on average, to be the case. firms, which may suffer disproportionately during reces- The advanced-country literature on understanding un- sions and, hence, destroy more jobs. Bosch and Maloney employment through the lens of gross worker flows offers (2006) explore this issue for Mexico and find that job separa- important insights and suggests that we focus on two sets of tions are, in fact, largely driven by firm size, with the aggre- flows: those into sectors of employment--the job-finding gate finding of acyclical firings in the formal sector being rate--and those out of those sectors--the separation rate. 108 T H E I N F O R M A L L A B O R M A R K E T I N M O T I O N : D Y N A M I C S , C Y C L E S , A N D T R E N D S FIGURE 4.7 Probability of transition from unemployment (job-finding rate), Brazil and Mexico Brazil Mexico Percent Percent 20 8 6 15 4 10 2 5 0 1983m1 1988m1 1993m1 1998m1 2003m1 1987q1 1991q1 1995q1 1999q1 2003q1 Formal Informal salaried Informal self-employed Sources: Bosch, Goñi, and Maloney 2006; Bosch and Maloney 2006. Figure 4.7 shows the job-finding rates of the formal United States, the job-finding rate is also strongly pro- and informal sectors for Brazil and Mexico over the period cyclical, and Shimer (2005a) and Hall (2005) argue that 1983­2003. The main stylized fact emerging from these the magnitude of these fluctuations cannot be well figures is the strong pro-cyclicality of the job-finding rate explained by state-of-the-art search and matching models. in the formal sector, compared with the relative stability of Both Shimer and Hall argue that one explanation for the the job-finding rate in the informal sector. The 1995 crisis excess volatility arises from wage rigidities. In most match- in Mexico made the formal job-finding probability oscillate ing models (see box 4.2), an adverse shock to the economy between 30 and 15 percent, while the job-finding rate in will lead to a decline in the productivity of potentially the two informal sectors fluctuated only around 4 percent- hired workers and to their wages. However, if wages cannot age points. Evidence from Brazil corroborates this finding. fall, then the likely profitability of hiring a new worker The 1984 crises decreased formal job-finding rates dramat- falls even more and, as a result, job vacancies will fall much ically, while the informal counterparts were relatively sta- more than if wages could fall. ble. Moreover, after the reforms at the end of the 1980s, the The relevance to the LDC case is clear. If formal sector job-finding rate saw major decreases from 15 to 5 percent. wages cannot fall while informal earnings can, then this Thus, the job-finding probability in the formal sector might well explain why formal job finding fluctuates so appears to be an important adjustment variable.3 much more than informal job finding does. In a sense, then, These differential rates of job finding across the cycle are the focus on gross labor flows could be seen as simply critical to explaining the frequent expansion of the infor- putting the Harris and Todaro (1970) vintage insight in mal sector in downturns. As the economy slows, formal sec- new bottles. However, the evidence to date suggests that tor hiring falls sharply but informal hiring falls much less, the question should be left open. First, Bosch and Maloney and algebra dictates that the relative size of the informal (2006) for Mexico and Bosch, Goñi, and Maloney (2006) sector is likely to rise. Though we previously showed that for Brazil find that certainly for self-employment and, to separations from informal salaried were the most volatile, a lesser degree, for informal salaried work, wages are net flows into the sector during crises are positive. more flexible than in the formal sector, but not always This raises the question of why the job-finding rates are compellingly so. For example, in the 1988­92 recovery so distinct between the two sectors, an issue with striking where formal hiring showed a disproportionately large resonance in the recent puzzle about the high volatility gain, formal salaried and informal salaried earnings stayed of job-finding rates in the developed economies. In the effectively equal. 109 I N F O R M A L I T Y More fundamentally, the mainstream literature has not turn, their productivity is affected by their access to the converged on wage rigidities as being responsible for the benefits of formality. The size of the informal sector is a common pro-cyclical patterns in job finding that we see function of the productivity differential between formal in both the Brazilian and Mexican formal sectors and in and informal workers among sectors. This differential is, in the United States. Mortensen and Nagypal (2005), for turn, determined by the cost of becoming and remaining instance, even adding wage rigidities and several other formal and the distribution of skills in the workforce. modifications to the Mortensen-Pissarides (1994) model, Specifically, the productivity differential has a worker-driven can simulate only 40 percent of the volatility of the job- component, given by the worker's individual skills, and a finding rate, and they conclude that "in sum, the dilemma sector-related component, given by the relative informal- [of high employment volatility] persists" (p. 24). This sug- formal regulatory burden, the strength of enforcement, and gests that the distinctive job creation behavior in the two the access to productivity augmenting public services. The sectors may involve far more than the traditional focus on size of informal employment is then given by the propor- formal sector rigidities. Further, as discussed in chapter 2, tion of workers whose skills fall below a threshold level there is substantial evidence of binding minimum wages in where the worker is indifferent between the two sectors. the informal sector, although this may be less the case dur- When regulation decreases or enforcement increases, the ing downturns where "norms" may prove more flexible formal sector becomes relatively more attractive for them than legal dictates. and more firms join it. Moreover, regulation is a fixed cost Second, even if wage rigidities turn out to be the critical that all formal firms have to bear. Therefore, when overall difference, the mechanism driving them may or may not productivity increases, the cost of regulation becomes pro- include the usual considerations of unions, minimum portionally smaller so that more firms join the formal wages, or other distortions. Kennan (2005) and Menzio sector, and the reverse is true in downturns. (2005) both stress asymmetric information in the wage- Empirically, Loayza and Rigolini (2006) find that, on bargaining process as inducing rigidities in wages with average, gross domestic product (GDP) growth is nega- respect to productivity shocks. More generally, in the tively correlated with the size of the informal sector-- United States, for instance, Bewley (1999) argues that, for defined as the share of self-employed--indicating that the reasons of staff morale, firms would prefer to fire workers, informal sector is, on average, countercyclical. Further, the the collective memory of whom will soon fade, rather than degree of countercyclicality appears smallest for the coun- reduce all workers' wages, the resentment about which will tries with the very largest informal sectors, Peru and not drop. In short, wage rigidities, the traditional driver of Bolivia, behaving relatively acyclicality (see figure 4.8). segmentation, are part of the story but only one candidate Consistent with their model, they find improvements in for explaining the sharp fall in formal hiring rates.4 Each of law and order strongly reduce the cyclicality of informal the elements of this debate surrounding "the dilemma" of employment. excess volatility in the job-finding rate in the United States applies to understanding the behavior of the formal sector job-finding rate in Latin America. Explaining pro-cyclical movements in informality Both of the previous models offer explanations for counter- cyclical patterns of informality. However, a closer look at The broader macro-context figure 4.1 suggests that this is not the whole story. In Working more in the de Soto (1989) tradition of firms Mexico, from 1988 to 2001, unemployment fell at the deciding to be formal or informal, Loayza and Rigolini same time that informality rose. That is, informality is pro- (2006) develop a model based on Rauch (1991) that offers cyclical across this period. While gross flows increased in an alternative perspective on possible drivers of cyclical both directions, net flows were toward self-employment. behavior of the informal, in this case measured as the self- Fiess, Fugazza, and Maloney (2002, 2006) offer an expla- employed. Here, the cyclical movements are not driven by nation by introducing distinct shocks to the formal and the net effects of worker flows among sectors, but rather informal sectors. More specifically, they argue that because across the de Soto margin. Effectively, individual firms informality is largely concentrated in the nontradables sec- weigh the benefits of being formal against the costs and, in tor and formal jobs in the more tradables sector, standard 110 T H E I N F O R M A L L A B O R M A R K E T I N M O T I O N : D Y N A M I C S , C Y C L E S , A N D T R E N D S FIGURE 4.8 Informal employment reaction to the business cycle Change in self-employment growth rate .01 Peru Honduras 0 Jamaica Bolivia Colombia Ecuador .01 Brazil Mexico El Salvador Uruguay Panama Costa Rica Chile Argentina .02 Trinidad .03 .04 .1 .2 .3 .4 Average self-employment rate Source: Loayza and Rigolini 2006. Note: The graph simulates the change in informal employment growth due to a 5 percentage point change in the GDP per capita growth rate. The simulation is based on a short-run regression like the one shown in table 4, col. 2, but with a sample resulting from applying the constraint of min. obs. 2. The dashed lines are the 90% confidence bands. Only countries in Latin America and the Caribbean are highlighted. From left to right: Trinidad and Tobago, Costa Rica, Argentina, Uruguay, Chile, Mexico, Panama, Brazil, El Salvador, Colombia, Ecuador, Jamaica, Bolivia, Honduras, and Peru. small-economy models that focus on the response of these movements in the labor market and, in the reverse two sectors to cyclical shocks (see Obstfeld and Rogoff way, movements in the labor market can inform what is 1996) offer insights into the cyclical behavior of the infor- driving movements in the macroeconomy. mal sector. A boom in construction, for instance, opens Figure 4.9 plots these three variables for Argentina, opportunities for informal contractors and others, and could Brazil, Colombia, and Mexico. Again, the early Mexican lead to a pro-cyclical expansion of the sector. Loosely link- period shows a relative increase in the size of the informal ing the discussion back to the matching models in the pre- self-employed sector and relative self-employed earnings, at vious section, although increases in job-finding rates in the the same time that the real exchange rate appreciated formal sector may generally exceed those in the informal sharply. More generally, we find three types of regimes. First, sector, a strong enough stimulus to the nontradables/infor- co-integration tests show that, in Brazil during 1994­97 mal sector can outweigh those increases, leading to pro- and in Mexico in the period 1987­91, we find a positive co- cyclical net job creation in the informal sector. On the movement of the two labor series and an appreciation of the other hand, a negative shock to the formal sector, perhaps exchange rate. Even in Argentina, from 1991 to 1996, and due to a major crisis, combined with downward rigidities in Colombia, from roughly 1990 to 1996, visual inspection in earnings, can lead to countercyclical behavior of infor- suggests a similar pattern. This suggests that the expansion mality identified previously, on average. The nature of of informality across this period may have been due to either shocks and the degree of rigidities of earnings, simultane- a productivity or a demand shock to the nontradables sector. ously determine movements of the exchange rate. The A very probable candidate is that the liberalization of the correlation of these variables informs what is driving capital account and other reforms taking place around these 111 I N F O R M A L I T Y FIGURE 4.9 Relative sector shares and earnings, real exchange rate Argentina Brazil Relative earnings, real exchange rate Formal Self-employed Relative earnings, real exchange rate Formal Self-employed 1.6 2.2 1.8 5.0 Integrated (I) Integrated (I) not significant significant 4.5 1.4 2.0 1.6 4.0 1.2 1.8 1.4 3.5 1.0 1.6 1.2 3.0 2.5 0.8 1.4 1.0 2.0 Segmented (II) 0.6 1.2 0.8 significant 1.5 0.4 1.0 0.6 1.0 1985 1987 1989 1991 1993 1995 1997 1999 2001 2003 1983 1985 1987 1989 1991 1993 1995 1997 1999 2001 Colombia Mexico Relative earnings, real exchange rate Formal Self-employed Relative earnings, real exchange rate Formal Self-employed 1.6 Integrated (I) 2.4 1.4 2.6 not significant Integrated (I) Segmented (II) Integrated (III) 1.3 significant significant significant 2.4 1.4 2.2 1.2 2.2 2.0 1.1 1.2 1.0 2.0 1.8 1.0 0.9 1.8 1.6 0.8 1.6 0.8 1.4 0.7 1.4 0.6 0.6 1.2 0.5 1.2 0.4 1.0 0.4 1.0 1985 1987 1989 1991 1993 1995 1997 1999 2001 2003 1987 1989 1991 1993 1995 1997 1999 2001 2003 Unconditioned earnings Formal Self-employed Real exchange rate Formal Self-employed Sources: Fiess, Fugazza, and Maloney 2006; and author's estimates, based on Pesquisa Mensal de Emprego, Instituto de Pesquisa Econômica Aplicada, Encuesta Nacional de Empleo Urbano, and International Financial Statistics. Note: Formal/self-employed earnings capture relative reported earnings in the formal and informal sectors. Formal/self-employed captures the relative size of these sectors as a ratio of employed population. "Significant" means a significant cointegrating relationship found. I, II, and III refer to regime (see text). periods led to an increased demand for nontradables and an four countries experienced deep recessions across these expansion of the informal sector. periods. In a second regime, Brazil (1983­89 and 1998­2002) Finally, in the third regime, Mexico (1997­2003), we and Mexico (1992­96) correspond to the case of a negative find a negative shock to the formal sector in the absence of shock to the formal/traded sector in the presence of some downward formal sector rigidities that leads to an expan- barriers to adjustment of formal sector wages. This is also sion of the relative size and remuneration in the informal clearly the case in Colombia after 1996 and for much of sector and to a depreciation of the currency. the period in Argentina, although the small sample sizes do not permit testing subperiods. The pattern suggests the Drivers of the increase in informality classic informal/nontradable sector adjusting to take in As chapter 1 (figure 1.10) noted, several measures of infor- labor no longer absorbed in the formal sector, discussed mality suggest substantial increases in informality across previously. This is historically plausible, given that all the last decades in Latin America. As noted, using a social 112 T H E I N F O R M A L L A B O R M A R K E T I N M O T I O N : D Y N A M I C S , C Y C L E S , A N D T R E N D S protection definition, urban salaried informality increased in deterioration, and the lack of flexibility of nominal formal Argentina from 1992 to 2003 by around 9 percentage points wages and the exchange rate, led to a more classic pattern and by 17 percentage points in Greater Buenos Aries from of relative contraction of formal sector employment, expan- 1980­2003. Although, nationally, there was effectively no sion of relative earnings, and appreciation that is more change in informality in Brazil, in a single decade (1990­ consistent with traditional queuing models. Hence, a 2000), informality in the metropolitan areas rose by 10 per- pretty straightforward story in the context of macroeco- centage points. In Colombia, figure 4.9 suggests that, over a nomic adjustment can explain substantial movement in longer period of time, informality rose by perhaps 6 percent- informality across this period. age points. And Mexico, despite relatively stable rates of However, dramatic as these movements have been-- informality across the 1990s, saw a sharp 4 percentage point terminating in crises in several of the countries--in inflection in informal salaried work from 2000 to 2004. Argentina, Brazil, Colombia, and perhaps, in the early Diagnosing across which margin these changes are 2000s, Mexico, the data suggest longer-term trends that occurring and why is critical to forming appropriate policy seem to go beyond medium-term macroeconomic adjust- toward the sector. This, however, is made difficult by the ments and point to longer-term evolutions in underlying fact that the 1990s were years of far-reaching reforms on determinants of informality. diverse fronts, as well as the final push on the unfinished Teasing out the impact of these distinct effects is agenda of macroeconomic stabilization. Teasing out the extremely difficult. To begin, global consensus on the impact of each policy is perilous and compounded by the impact of many of these reforms on labor markets remains fact that research on the determinants of informality elusive and the Latin American literature remains in its remains relatively inchoate. The aims of this section are infancy. To date, the strongest effort to study the impact of thus relatively modest--to use several country cases to labor legislation has been undertaken by Heckman and highlight viable hypotheses and to present some evidence Pages (2004), although even here, where the direction of that informs the emerging discussion. In turn, we discuss impact has been identified, often the magnitudes are still issues of macroeconomic stabilization, trade liberalization, up for discussion. Relatively little of the work on trade lib- labor market reforms, supply side effects, and, briefly, eralization has focused on the impact of informality, and reforms to facilitate formalization of microfirms. perhaps even less on the impact of stabilization. Second, this report has stressed the potential importance of, partic- Macroeconomic stabilization and capital ularly, informal independent employment as an exit option account opening for workers evading social protection coverage that costs The previous section noted that substantial movements in them far more than the benefits they receive from it. informality could be generated by distinct shocks to the Changes in the relative attractiveness of formal versus informal and formal sectors. At the beginning of the 1990s informal work can lead to shifting the supply curve of labor numerous countries in the region fixed the exchange rate as to the formal sector and, hence, change the allocation of a means of controlling inflationary expectations while labor. Some of the reforms of the period had potentially far- liberalizing the capital and current accounts. Virtually all ranging effects in this area, but the efforts to quantify countries saw sharp appreciations of the exchange rate that, them, presented here, are rudimentary at best. Finally, all in many cases, were seen as arising from backward indexing the reforms happened at once, thus making isolation of one of earnings in the context of falling inflation. However, effect from the others especially difficult. the previous section suggests that these countries experi- enced the Regime I pattern of a rise in relative size and Demographic and structural factors earnings of the informal self-employed along with the A natural place to begin is in demographic and structural exchange rate appreciation. This is consistent with a changes in the economies over the last two decades. Female demand shock to the nontradable informal sector and views labor market participation has risen sharply and, as chapter 2 stressing borrowing against expectations of future income documents, women at marrying age tend to be dispropor- that had been sharply revised upward in light of the tionately informal: Galiani and Weinschelbaum (2006) improving macroeconomic situation and reforms. There- argue that the inflow of secondary workers to the Argentine after, however, in the lead-up to the crises, macroeconomic workforce, particularly of female workers who choose to be 113 I N F O R M A L I T Y informal because their husbands are already covered by ben- characteristic affects the probability of being informal. Fig- efits, has led to a 13 percent increase in informality from ure 4.10 plots how much informality would change if, 1974­76 to 1999. Jimeno and Rodriguez Palenzuela (2003) using the set of estimated country-specific coefficients of argue that trends in youth participation in the labor market the impact of any particular demographic or sectoral char- are partly responsible for the rise in unemployment in the acteristic, we assign to the country the set of characteristics OECD, which, as chapter 2 suggested, also maps to more that tends to lead to both the highest (Nicaragua) and low- informal salaried work. However, youth participation has est (Chile) levels of informality in the sample. As is clear broadly stabilized as students spend more time in school and, using both the social protection and productive definitions, in turn, education appears often as a determinant of infor- characteristics alone can account for a difference of more mality (see chapter 3)--with some models, such as Krebs than 10 percentage points in level of informality. How and Maloney (1999) and Loayza and Rigolini (2006), show- a given set of characteristics translates into informality ing that more educated workers or workforces are less likely appears even more important. Using the parameters that to be informal. In terms of economic structure, numerous create the highest level of informality (Peru) and those that studies have shown that different sectors have predispositions yield the lowest (again, Chile) yields large differences: to informality--services more, manufacturing less (see Peru's structure, using Chile's coefficients, would lead to a Goldberg and Pavcnik 2003, and Bosch, Goñi, and Maloney 25­50 point reduction in the size of the informal sector. 2006). Clearly, it is critical to understand how the economy trans- To explore how important such changes may be, lates demographic characteristics into informality. Gasparini and Tornarolli (2006) undertook two sets of This translation also turns out to have been far more decompositions, identifying for numerous countries the important to explaining the increase in informality than relative impact of various demographic and structural fac- changes in structure or demography. Figure 4.11 shows tors on the level of informality. For each country, they esti- that for Argentina (1995­2003), Brazil, Chile, El mate the probit coefficients that suggest how much a given Salvador, Paraguay, and Uruguay, structural changes FIGURE 4.10a Variation of informality rate (social protection definition) Using parameters of Chile and Peru Using characteristics of Chile and Nicaragua Venezuela, R.B. de Venezuela, R.B. de Uruguay Uruguay Peru Peru Panama Panama Nicaragua Nicaragua Mexico Mexico Guatemala Guatemala El Salvador El Salvador Chile Chile Brazil Brazil Argentina Argentina 50 40 30 20 10 0 10 20 30 40 50 15 10 5 0 5 10 15 Percent Percent Chile Peru Chile Nicaragua Source: Gasparini and Tornarolli 2006. 114 T H E I N F O R M A L L A B O R M A R K E T I N M O T I O N : D Y N A M I C S , C Y C L E S , A N D T R E N D S FIGURE 4.10b Variation of informality rate (productive definition) Using parameters of Chile and Peru Using characteristics of Jamaica and Nicaragua Venezuela, R.B. de Venezuela, R.B. de Uruguay Uruguay Peru Peru Paraguay Paraguay Panama Panama Nicaragua Nicaragua Mexico Mexico Honduras Honduras Jamaica Jamaica Guatemala Guatemala El Salvador El Salvador Dominican Republic Dominican Republic Costa Rica Costa Rica Chile Chile Brazil Brazil Argentina Argentina 25 20 15 10 5 0 5 10 15 20 25 30 20 15 10 5 0 5 10 15 Percent Percent Chile Peru Jamaica Nicaragua Source: Gasparini and Tornarolli 2006. independent decomposition by the Argentina Poverty FIGURE 4.11 Decomposition of changes in informality (social protection Assessment (World Bank 2007a) concludes that changes definition) for urban salaried workers in such structural factors were not, overall, responsible; rather, there was a generalized tendency toward noncom- Percent 15 pliance with labor legislation. Similarly, an independent exercise for Chile (Contreras, Puentes, and Sanhueza 2006) 10 broadly concurs with Gasparini and Tornarolli (2006), arguing that changes in worker characteristics over the last 5 40 years would have led to a much larger decrease in infor- 0 mality than they actually observe. We might argue that the changes in economic structure 5 arising from reforms, away from formality-intensive sectors into those less so, might be critical. At this point, these do 10 de not seem a first-order explanation. Ramos (2002) found aguay R.B. Brazil Chile Uruguay 1­04 that, across the 1990s in Brazil, there was an expansion into Argentina Argentina SalvadorPar 1993­20031990­2003 1995­20031995­2004 El 1997­2003200 1991­2003 services, but that this would explain a maximum of 25 per- Venezuela, 5­2003 199 cent of the increase in informality. Goldberg and Pavcnik (2003) for Brazil and Colombia; Bosch, Goñi, and Maloney Actual change Characteristics Parameters (2006) for Brazil; and Bosch and Maloney (2006) for Source: Gasparini and Tornarolli 2006. Mexico show that most secular changes in informality have arisen within sectors, not across them. would lead to a decrease in informality if their impact had This point also applies to the Galiani and stayed the same. Only for a particular time period for Weinschelbaum (2006) argument about the importance of Argentina and the República Bolivariana de Venezuela the gender composition of the workforce. Looking closely is there an informality-expanding effect. That said, an at Brazil and Colombia (figure 4.12), we see a complex 115 I N F O R M A L I T Y FIGURE 4.12a Labor force participation, rate of formality, and unemployment, Brazil Males Females Percent Percent Percent Percent 60 84 30 48 82 50 80 25 46 78 40 76 20 44 74 30 15 42 72 70 20 10 40 68 10 66 5 38 64 0 62 0 36 1 3 1 3 9 1 1983 . 1985 1987 1989 199 199 1995 1997 1999 2001 1983 1985 1987 1989 199 199 1995 1997 199 200 Jan. Jan Jan. Jan. Jan. Jan. Jan. Jan. Jan. Jan. Jan. Jan. Jan. Jan. Jan. Jan. Jan. Jan. Jan. Jan. Left scale: Formal Informal Unemployed Right scale: Labor force participation Source: Author's estimations, based on Encuesta de Calidad de Vida and Pesquisa Mensal de Emprego. Note: All aggregates are a share of total population and hence sum to unity. FIGURE 4.12b Labor force participation, rate of formality, and unemployment, Colombia Males Females Percent Percent Percent Percent 50 78 30 60 45 76 25 40 74 55 35 72 20 30 50 70 25 15 68 20 45 66 10 15 10 64 40 5 5 62 0 60 0 35 1q1 2q3 4q1 5q3 7q1 8q3 0q1 1q3 3q1 1q1 2q3 4q1 5q3 7q1 8q3 0q1 1q3 3q1 1985q1 1986q31988q1 1989q3199 199 199 199 199 199 200 200 200 1985q1 1986q31988q1 1989q3199 199 199 199 199 199 200 200 200 Left scale: Formal Informal Unemployed Right scale: Labor force participation Source: Author's estimations, based on Encuesta de Calidad de Vida and Pesquisa Mensal de Emprego. Note: All aggregates are a share of total population and hence sum to unity. evolution between and across genders but with a heavy while the male share in informal work rose less strikingly, emphasis on informalization within genders. with the lost formal workers appearing in the sharply rising In Colombia, the share of men in formal employment as unemployment of the time. Female participation rose a share of the population has fallen sharply since 1995 sharply beginning in 1996 and went entirely to informality. 116 T H E I N F O R M A L L A B O R M A R K E T I N M O T I O N : D Y N A M I C S , C Y C L E S , A N D T R E N D S This represents a break with past trends where the gently decline in formal sector employment. The increase in rising female participation from 1985 to 1995 left the share salaried informality could be made manifest through a cou- of females who were informal relatively constant. The appar- ple of channels. The negative shift of the demand curve for ent correlation of female participation with rising unem- formal labor would lead to lower employment and earnings ployment would seem consistent with an added worker in the formal sector. Part of the fall in earnings could occur effect, documented in Mexico and Argentina (Cunningham through lower benefits, an effect that might be exacerbated 2001a, 2001b): women--particularly married women-- if wages were relatively rigid. The same scenario would enter the workforce if their spouses become unemployed. lead to hiring workers without benefits or subcontracting Hence, adverse macroeconomic shocks are driving the tasks to lower-paid external workers. increase in female participation and the rise in informality However, lower tariffs may also foster the import of both overall and within genders. technology and capital from abroad, thereby increasing the Brazil saw much less pronounced increases in unemploy- demand for complementary skilled labor that, in the long ment, female participation, and the share of women who run, tends to greater formality. Generally speaking, indus- were informal. Rather, what emerges there is a replacement tries that are more exposed to trade tend to pay higher of male formal positions with informal positions concomi- wages and be more formal (see de Ferranti et al. 2001), tant with an unexplained fall in male labor market partici- given the human capital of their workers. In addition, the pation. In both cases, the rising informality within gender availability of higher-quality or lower-cost intermediate contributes dominantly to the overall evolution of infor- inputs in essence, constitutes a productivity shock increase mality in the economy. to the formal sector, which, as shown earlier, lends to lower Finally, the region's overall gain in income would seem informality. to mitigate against the rise. As noted in chapter 1, numer- Empirical evidence of openness to trade on levels of ous studies now have shown that informality, measured as informality is mixed, but generally suggests small effects. the share of the workforce, falls with development; and, Goldberg and Pavcnik (2003) find a very modest impact overall, in spite of the crises, the region made some small of trade reforms in Colombia and none in Brazil. Bosch, progress (see Blau 1987; Gollin 2002; Loayza and Rigolini Goñi, and Maloney (2006), revisiting the Brazilian case 2006; and Maloney 2001).5 However, as Loayza and through the lens of job creation and destruction, find a Rigolini (2006) and others have shown, this broad trend positive, but again small, impact. Figure 4.13 plots the with development explains perhaps 60­75 percent of the predicted values using the Shimer (2005c) methodology variance in the size of the self-employed sector. The residual and suggests that, in the absence of trade liberalization, 25­40 percent is broadly consistent with the simulations formal employment may have been 10 percent higher. above that suggest that similar countries can have very dif- The evidence from Mexico does not suggest a huge impact ferent levels of informality. We now look at changes in trade either. As noted by García-Verdú (2007), among others, regimes and labor legislation as possible explanations. given the dramatic unilateral liberalization beginning in Trade reforms The far-reaching trade reforms of the 1990s are a logical FIGURE 4.13 Actual and predicted size of the industrial formal sector, Brazil suspect, although, theoretically, the effects on informality are ambiguous. On the one hand, cheaper imports (or the 90 No constitutional reform appreciation of the currency that accompanied the trade 85 reforms) may introduce pressure on domestic prices, dri- No trade reform 80 ving local firms out of business, reducing their incentives 75 to open new positions, or pushing them toward cheaper Actual means of production in the informal sector. In the Fiess- 70 Fugazza-Maloney (2006) model, this could be seen as a 65 negative productivity shock to the formal/traded sector, the 19831984198519861987198819891990199119921993199419951996199719981999200020012002 adjustment to which would depend on the degrees of rigid- Source: Bosch, Goñi, and Maloney 2006. ity in the formal sector, but in any case would lead to a 117 I N F O R M A L I T Y 1987 and then continuing through the North American Argentina suggests that the impact of trade reform per se Free Trade Agreement, there is little trend in informality. has a magnitude similar to that in Brazil, although there Aleman-Castilla (2006), broadly following the Goldberg may be significant additional impacts from the various and Pavcnik (2003) methodology, finds that those indus- periods of sustained currency overevaluation. tries more exposed to trade saw higher increases in the Recent trends in Mexico also seem plausibly related to rate of formality. The reason, he argues, was that international exposure. The sharp increase in both self- the impact on product prices was minor, while the reduc- employment (shown above) and informal salaried work (see tion in import prices raised the productivity of the trad- figure 4.15) after 2000 has occurred concomitantly with ables sector and, hence, expanded the demand for formal the entry of China as a major competitor in some areas of labor overall. Mexico's comparative advantage. Hanson and Robertson However, suspicions remain. As Mondino and Montoya (2006) argue that, had China's growth in export capacity (2002) and World Bank (2007a) have shown, a very large remained unchanged after 1995, Mexico's annual export increase in the share of informal salaried workers began in growth rate of Chinese-substitutable goods would have the early 1980s (figure 4.14). Though the last round of been 1.5 percentage points higher in the late 1990s and trade liberalization began only in 1990, the reforms begun 3.0 percentage points higher than the 1.9 percent it experi- under Argentina's Minister of the Economy José Alfredo enced going into the new millennium. This does suggest Martínez de Hoz in the late 1970s radically lowered tariffs that international competition is putting a constraint on and led to an appreciated exchange rate. Galiani and the expansion of some export jobs. On the other hand, Sanguinetti (2003) and Porto and Galiani (2006) find that Lederman, Olarreaga, and Soloaga (2006), using estima- the decreased protection had some effect on both the tions of the gravity model of trade, argue that there is little absolute level of wages and the gap between skilled and evidence that Mexican (and Central American) nonfuel unskilled labor. To the degree that part of the downward exports overall were affected. It is also noteworthy that the pressure on unskilled wages came through the reduction sharp increase seems to occur with the relaxation of restric- of benefits, or subcontracting, it seems possible that trade tions on Chinese textiles and apparel imports in the United liberalization had an impact. However, preliminary analy- States, which was not one of Hanson and Robertson's sis replicating the Goldberg-Pavcnik (2003) exercise6 for affected sectors. As an alternative explanation, the overall FIGURE 4.14 FIGURE 4.15 Informality rate for salaried workers in Greater Buenos Aires Informal salaried versus formal salaried, Mexico, relative earnings and sector size 1.4 0.45 0.40 Informal vs. Formal 1.2 0.35 Formal informal Relative earnings 1.0 0.30 2.2 1.70 0.8 0.25 Wformal Winformal 1.65 2.1 0.6 0.20 1.60 0.15 2.0 0.4 1.55 0.10 1.9 0.2 1.50 0.05 0 0 1.8 1.45 1.40 1.7 1974 1976 1978 1981 1985 1987 1989 1991 1993 1995 1997 1999 2001 1.35 1.6 Formal informal Left scale: Right scale: 1.30 1.5 Mean tariff Informality 1.25 Wage premium Import penetration 1.4 1.20 7 8 9 0 1 2 3 4 5 7 8 0 1 2 3 4 Sources: Porto and Galiani (2006) for mean tariff and wage premia; 198 198 198 199 199 199 199 199 199 1996199 199 1999200 200 200 200 200 Gasparini (2002) for informal salaried work; and Olarreaga, Castro, and Saslavsky (2006) for import penetration. Source: Author's calculations. 118 T H E I N F O R M A L L A B O R M A R K E T I N M O T I O N : D Y N A M I C S , C Y C L E S , A N D T R E N D S reduction in exports due to the U.S. recession may have had increase in informal work. In both Argentina and Brazil, a straightforward impact through, effectively, a reduction the striking fact is the shift of informality toward larger in tradables productivity that, in the absence of wage firms, as the rate of informality in those large firms has rigidities, puts us exactly in Regime III--a depreciation of almost doubled. This casts some doubt on the Beccaria, the currency concomitant with a rise in relative sector size Carpio, and Orsatti (1999) conjecture that the rise in infor- and relative earnings (figures 4.9 and 4.15). There may mality in Argentina was driven by increased subcontract- have been nothing more exotic going on than that, given ing to smaller firms. the slowdown in the U.S. economy, the opportunities However, the reverse appears to happen in Mexico where became better in the informal microfirms than in the medium-size and large firms are becoming more formal maquilas across this period. over time. Table 4.2 suggests that this is not due to An additional table and graph comparing Argentina, increased subcontracting relationships with the firms of Brazil, and Mexico are potentially informative here. fewer than 16 employees covered in the microenterprise Table 4.1 shows the change in the allocation of informal survey. Though the 2002 data are not exactly comparable sector workers across sector sizes over the relevant period of with earlier data, there does not appear to have been an important increase in those selling their products or ser- vices to or buying inputs from large clients. Putting TABLE 4.1 together the story, we might argue that the shifts in infor- Evolution of informality across firm size and time: Distribution mality measured here are due to the increased relative of informal and self-employed workforce attractiveness of opening or working for a microfirm over a. Argentina (GBA) the last quinquennium, and not to greater subcontracting Firm size 1980 2003 Variation (%) or within-large-firm informality due to trade opening. Second, figure 4.16 suggests that, for Mexico, the small 1 45.35 34.60 ­10.74 2 to 5 42.11 40.75 ­1.36 changes in the distribution of formal employment across 6 to 25 7.82 15.55 7.72 ages over the last 20 years of trade liberalization have been 26 to 100 2.71 5.25 2.54 101 to 500 1.25 2.60 1.35 minimal, with some loss of formality among prime-age 501+ 0.76 1.25 0.49 males and perhaps older workers in the 1987­96 period that was absorbed both in informal salaried and indepen- Source: Encuesta Permanente de Hogares. dent work. There have been no substantial changes in the b. Brazil 1996­2004 period. In Brazil, however, the 1990­2002 period brought a shifting down of formal employment of Firm size 1990 2002 Variation (%) roughly 10 percentage points across the whole age spec- 1 35.70 25.52 ­10.18 trum, with a fall of 20­30 percentage points for young 2 to 5 42.52 35.99 ­6.53 workers. In Argentina, a similar pattern has prevailed, 6 to 10 7.15 12.47 5.32 11+ 14.63 26.02 11.39 although the similarly dramatic losses of formal jobs Source: Pesquisa Mensal por Amostra de Domicilios. TABLE 4.2 Subcontracting in Mexico, 1992­2002 (percent) c. Mexico Firm size 1994 2004 Variation (%) Input/client 1992 1994 1996 1998 2002a 1 25.9 30.0 4.2 Big inputs 37.78 39.8 41.6 46.99 41.79 2 to 5 41.0 44.7 3.7 Big clients 5.31 3.85 3.77 4.14 3.19 6 to 10 7.4 7.4 0.1 Maquila 2.28 1.42 1.31 2.98 0.07 11 to 15 3.2 2.8 ­0.4 16 to 50 6.1 6.3 0.2 Source: Encuesta Nacional de Micro Negocios, several years. 51 to 100 3.0 1.7 ­1.2 Note: Firms are considered to be subcontracting if there is a big 101 to 250 1.7 0.6 ­1.1 provider or a big client, or if the firm does maquila work. 251+ 11.9 6.5 ­5.4 a. Includes exports and imports as category in inputs/clients; no separate question for maquila, included as category of Source: Encuesta Nacional de Empleo Urbano. input/client. 119 INFORMALITY FIGURE 4.16 Evolution of informality across age groups Formal salaried Informal salaried Self-employed Percent Argentina Percent Argentina Percent Argentina 100 100 100 1980 1995 2003 1980 1995 2003 1980 1995 2003 80 80 80 60 60 60 40 40 40 20 20 20 0 0 0 16 19 22 25 28 31 34 37 40 43 46 49 52 55 58 16 19 22 25 28 31 34 37 40 43 46 49 52 55 58 16 19 22 25 28 31 34 37 40 43 46 49 52 55 58 Age Age Age Percent Brazil Percent Brazil Percent Brazil 100 100 100 1983 1990 2002 1983 1990 2002 1983 1990 2002 80 80 80 60 60 60 40 40 40 20 20 20 0 0 0 120 15 18 21 24 27 30 33 36 39 42 45 48 51 54 57 60 15 18 21 24 27 30 33 36 39 42 45 48 51 54 57 60 15 18 21 24 27 30 33 36 39 42 45 48 51 54 57 60 Age Age Age Percent Mexico Percent Mexico Percent Mexico 100 100 100 1987 1996 2004 1987 1996 2004 1987 1996 2004 80 80 80 60 60 60 40 40 40 20 20 20 0 0 0 15 18 21 24 27 30 33 36 39 42 45 48 51 54 57 60 15 18 21 24 27 30 33 36 39 42 45 48 51 54 57 60 15 18 21 24 27 30 33 36 39 42 45 48 51 54 57 60 Age Age Age Percent Peru Percent Peru Percent Peru 100 100 100 1986 1991 2001 1986 1991 2001 1986 1991 2001 80 80 80 60 60 60 40 40 40 20 20 20 0 0 0 15 18 21 24 27 30 33 36 39 42 45 48 51 54 57 60 15 18 21 24 27 30 33 36 39 42 45 48 51 54 57 60 15 18 21 24 27 30 33 36 39 42 45 48 51 54 57 60 Age Age Age Source: Author's calculations. T H E I N F O R M A L L A B O R M A R K E T I N M O T I O N : D Y N A M I C S , C Y C L E S , A N D T R E N D S among workers in the early part of the life cycle extend to increases in firing costs, labor taxes, and nominal wage those in their late 20s, unlike in Brazil where the really big rigidities can decrease the size of the formal sector by a few losses level out at about 20. Further, there is also a marked percentage points. Ulyssea (2006), modeling both barriers decrease in formalization among workers over 45 years of to entry and labor regulations, found the latter to be of small age that is roughly double that of the prime-age males. importance, although, as he makes clear, the assumption of In light of this, there can be no wonder about the con- perfect wage flexibility means that many distortions are cern in Argentina about the increased informalization of passed on to workers and, hence, have little effect on the the workforce. The richest country in Latin America once demand for labor. A large body of literature (for example, had a constant and high formal employment rate of almost Botero et al. 2003; Heckman and Pages 2004; and Nickell 70 percent, where 17-year-old workers had the same access and Layard 2000) concludes that more stringent labor to formal sector jobs as prime-age males. It now looks market regulations hamper productivity growth. Holmes closer to Mexico, except the large firms in Mexico are rela- (1998) and Besley and Burgess (2004), exploiting regional tively more formal. At the very least, this represents differ- variation for the United States and India, respectively, find ent experiences with trade liberalization. But the summary important impacts of labor regulation on output, and the picture is striking. Mexico's far-reaching trade liberaliza- latter on employment. Following their general approach, tions, begun in 1987, had minimal impact on informality Almeida and Carneiro (2005) find that Brazilian states or its allocation across age groups or firm sizes. Meanwhile, with lower enforcement have higher productivity, wages, Brazil has experienced an evolution in magnitude of infor- and employment. mality and its allocation across both dimensions similar to Documenting the impact of legislation on formal sector that in Argentina. Yet, as figure 4.13 shows, little of this demand in Latin America has been the subject of several can be attributed to its far-reaching trade liberalization. At major efforts in the region. Of particular importance, the this point, we begin to look for other explanations. National Bureau of Economic Research volume edited by Heckman and Pages (2004) collected some of the most Regulation and distortion serious attempts made to that date to quantify the impacts A substantial body of literature sees the size of the informal of various regulatory changes, and the literature is well sector to be determined substantially by regulatory distor- reviewed in the introduction and chapters in that book. tions or corruption (see Djankov et al. 2002; Friedman For the region and several countries, a credible case can be et al. 2000). Looking particularly at self-employment in built that labor legislation had a substantial impact on cross-section, Loayza and Rigolini (2006) find that varia- the size of the formal sector (see Heckman and Pages 2004; tions in business flexibility account for 16 percent of the and Saavedra and Torero 2004). variance; variations in government expenditure (a measure The Mexican case has attracted little academic work of monitoring intensity) account for 7 percent; and enforce- since there were significant changes in neither the size of ment of contracts, prevalence of the rule of law, and the the informal sector nor, perhaps relatedly, labor legislation. efficiency of the policy and judicial system account for However, the fall in formal employment in the municipal 26 percent. GDP still accounts for 61 percent.7 Loayza, areas of Brazil has received formidable attention by the Oviedo, and Servén (2006), using the Schneider and Enste analysts8 and the present discussion can only graze the tip (2000) shadow economy estimates, find similar correlations of the iceberg. However, increased labor market rigidities with regulatory measures, although, as discussed at length are among the prime candidates to explain the fall in in chapter 1, it is unclear what this measure is actually cap- formal employment not accounted for by trade liberaliza- turing. In general, such effects account for about 30 percent tion in figure 4.12. The constitutional reform of 1988 of the variation of informality across countries. included several important changes to the labor legislation A long tradition sees distortionary regulation, particu- (see Paes de Barros and Corseuil 2004 for a summary). larly in the labor market, as leading to increased informality First, several measures increased labor costs and reduced for a variety of reasons and across all three margins. Box 4.2 employer flexibility. Maximum working hours per week discusses three modeling exercises within the context of were reduced from 48 to 44, the maximum daily work day matching type models discussed in the first part of this in selected industries was reduced from 8 to 6 hours, over- chapter. Together, these simulations suggest that modest time remuneration was increased from 1.2 to 1.5 times the 121 I N F O R M A L I T Y normal wage rate, vacation pay was raised from one to one 1990s. In the context of the Fiess-Fugazza-Maloney model, and a third of the monthly wage, and maternity leave this corresponds well to a negative formal sector productiv- increased from 90 to 120 days. Second, union power was ity shock in the presence of a nominal rigidity and, in fact, expanded: unions were no longer required to be registered the empirical evidence supports that interpretation. In addi- and approved by the Ministry of Labor; decisions to strike tion to this and consistent with it, other policy changes may now rested purely on the union's decision, and the required have led to shifting the demand for formal sector workers advance notification to the employer fell from five to two that interacted with the minimum wage. Kugler (2000) days; strikes in certain strategic sectors were no longer argues that the 1990 reforms that reduced firing costs banned. Previous work by Menezes-Filho (1997) docu- induced greater worker flows in and out of unemployment menting the impact of the reduction in union power across and may have contributed to 10 percent of the reduction in the 1980s on firm profitability and earnings suggests that unemployment. However, they may also have deepened the this channel can be important. Finally, firing costs were fall in formal employment by reducing the costs of firing. raised. The penalty levied on employers for "unjustified" Perhaps more directly, Kugler and Kugler (2003) argue that dismissal, a category encompassing most legitimate separa- the sharp rise in payroll taxes in Colombia--over 10 percent tions for economic reasons in the United States, increased from 41 to 51.5 percent--between 1989 and 1996 may also by four times, from 10 to 40 percent of the accumulated have had important employment effects. As they note, both separation account (Fundo de Garantia por Tempo de Serviço). nominal rigidities and the weak link between taxes and ben- To date, the most comprehensive work relating these efits lead to these higher costs being only 1.4­2.3 percent of changes to the functioning of the labor market was under- those taxes shifted to workers, the rest leading to a decline in taken by Paes de Barros and Corseuil (2001) who find that formal employment of 4­5 percent. Cardenas and Bernal separation rates decreased after the constitutional changes for (2003) similarly find high estimated wage elasticities that short employment spells and increased for longer spells, but would suggest that such an increase in labor taxes would find an ambiguous overall effect. However, again, matching have led to a substantial drop in labor demand.9 models discussed in the previous section suggest that several of Peru's large increase in informality of roughly 10 per- these reforms would lead to a reduction in hiring (job-finding) centage points from 1986 to 2001 in metropolitan Lima rates as opposed to the separations that Paes de Barros and also seems partly due to the increased regulatory burden. Corseuil study. By exploiting cross-industry variation in As Saavedra and Chong (1999) note, during the early proxies related to these reforms, Bosch, Goñi, and Maloney 1990s Peru implemented dramatic trade liberalizations, (2006) find suggestive evidence that proxies for restrictions fiscal reforms, deregulation of the goods markets, extensive on hours worked, increases in union power, and increased fir- privatization, and labor market reforms--the respective ing costs appear to have a strong impact on decreasing formal impacts of which are difficult to disentangle. The results job creation across the 1990s. Simulations suggest that, had were the substantial downsizing of many firms, and the there been no change in the constitution, formal hiring rates productivity increase achieved by the reforms was concen- would have been 40 percentage points higher and overall lev- trated in select sectors that created few additional formal els of formality would have fallen negligibly. jobs. On the one hand, the labor reform reduced employ- The impact of minimum wages interacted with other ment rigidities, firing costs were reduced substantially,10 labor reforms plays an important role in explaining the evo- red tape on hiring temporary workers was minimized, and lution of informality in Colombia. Figure 4.9 shows that, the use of training contracts for younger workers was facil- around 1996, the labor market began a period of severe itated. Further, enforcement of labor laws fell as budgets for segmentation captured by the fact that relative earnings in inspections were slashed and labor courts became less pro- the formal sector rose while the relative size of the sector labor than previously. Finally, union bargaining power also fell sharply. As Arango and Pachón (2004), Kristensen and fell dramatically, partly due to changes in labor legislation, Cunningham (2006), Maloney and Nuñez Mendez (2003), the decentralization of the collective bargaining process, as and Santamaria (2000) show, Colombia's minimum wage is well as the increased use of temporary contracts.11 among the highest and most binding in Latin America, and On the other hand, non-wage labor costs increased it was raised in real terms in the depths of the crisis of the late sharply beginning in the early 1990s (see figure 4.17), due 122 T H E I N F O R M A L L A B O R M A R K E T I N M O T I O N : D Y N A M I C S , C Y C L E S , A N D T R E N D S sticky and, hence, effectively eliminated downward flexi- FIGURE 4.17 bility and made operating informally more attractive (see Labor legislation and related variables, Peru World Bank 2007a). The authors put together a year-by- Index 1995 100 year series of tax rates and labor regulations and compare 130 them to the changes that have taken place in rates of 120 informality (for salaried employees). They show that the 110 increase in rates of informal salaried employment during 1980­88 took place during a period of rising taxes on labor 100 and increased regulation. However, the increase in infor- 90 mality in the period 1995­99 was during a period of 80 declining taxes. Arguably, the impact may have been offset by increased segmentation coming from sliding formal 70 sector productivity and downwardly rigid wages. 60 Changes in labor law in 1991 and 1995, which provided 50 for short-term labor contracts, have received particular atten- 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 tion as possible triggers for the growth in informality.12 The reforms introduced fixed-term contracts and provided for Non-wage labor costs Taxes GDP Index of tax compliance special short-term contract regimes for small firms and young workers. Hopenhayn (2004) finds that the 1995 Sources: Saavedra and Maruyama 2000, Peruvian Central Bank reform substantially increased job turnover rates. Olmedo (BCRP), and National Tax Office (SUNAT). Note: Non-wage costs are presented as a percentage of wages in and Murray (2002) credit the changes with promoting an December of each year. The index of income declaration denotes the number of contributors (workers and firms) presenting tax acceptance of lower job protections and hence, greater infor- statements to national authorities. Note that, actually, a mality. They write, "The new labor laws undermine the idea proportion of these contributors may not be necessarily paying any taxes (December of each year). of permanent work with a modicum of economic welfare, social protection (especially in the fields of medical care, old age security and educational opportunities), and job secu- mainly to the elimination of caps on certain types of con- rity" (p. 430). Rates of informality among salaried workers tributions. In addition, changes in the regulation of the rose when short-term contracts were legalized in 1995 and unemployment compensation individual account (Compen- then persisted after such contracts were abolished in 1999. sación por Tiempo de Servicios) now forced accumulated This pattern is consistent with the Olmedo and Murray contributions to be deposited in commercial bank (2002) claim that informality was made more socially accounts rather than simply counted as a contingent liabil- acceptable by labor law reforms and supported by anecdotal ity and the resources retained as working capital. Further, evidence that suggests that, in contrast to Peru, enforcement the 1991 tax reform induced a steady increase in the capa- of the labor law was relaxed. This argument, however, cannot bilities of tax authorities to reduce evasion. Saavedra and explain the large increases in informality that took place dur- Torero (2004) show that these movements in labor costs, ing 1980­88, when labor regulation expanded. An alterna- in fact, have the predicted impact on formal sector labor tive may, again, be that as Perry and Servén (2005) note, demand. The doubling of the non-wage costs led to a rise overvaluation was severe across this 1995­99 period due to in overall labor costs of 10­11 percent which, at their esti- the devaluations of neighboring currencies; and, much like mated wage elasticity, could lead to a 3­4 percent decrease the slowdown in the U.S. economy for Mexico, this may have in formal salaried employment. constituted a negative shock to the tradables sector. Regulatory explanations for the rise of the informal sec- tor in Argentina are both many and relatively inconclusive, Increased incentives to exit and are reviewed in World Bank 2007a. For example, Bour A critical policy conclusion of the analysis in the first section and Susmel (2000) argue that the low inflation under the above--that there is a high degree of mobility among for- Argentine convertibility plan made formal sector wages mal and informal sectors--is that we need to be concerned 123 I N F O R M A L I T Y with the incentives generated for working in the two sec- over time, other than during the crisis. Hence, it is an tors. A nascent literature is emerging precisely around the open question whether the systematically low level of confi- implications for social policy (see Levy 2006; Maloney 2004) dence is related to the time trend toward higher levels of and this will be dealt with in greater degree in the next informal work. chapter. The central point is that any wedge that is driven Gasparini, Haimovich, and Olivieri (2006) attempt to between the contributions a worker makes explicitly in measure whether the large cash transfers to unemployed terms of labor taxes or implicitly in terms of lower equilib- household heads implied a disincentive for the program rium wages, on the one hand, and the perceived benefits, on participants to search for a formal job. They find for the other hand, makes formality less attractive. Such wedges 2003­04 that the effects were significant and large, reduc- may arise for several reasons. ing the share of workers moving into formal jobs by 5 per- In Brazil, the recent establishment of a minimum pen- centage points. However, the effect vanishes later as the sion for all, regardless of contribution, and the lowering of gap between program benefits and the rising wage shrank the eligibility age compound the effect discussed by Levy and would seem an unlikely candidate for explaining the by increasing the net gain from being informal. Further, the effects since the 1980s. constitutional change toward creating universal health More generally, it is difficult to know how much the insurance may have also become an incentive to informality, pure tax--that is, income surrendered without correspond- although research to date is thin on the point (see Bosch, ing benefits--affects the decision to be informal. Regret- Goñi, and Maloney 2006) and will be dealt with in greater tably, research is in its infancy here, and we can offer only detail in chapter 5. Carneiro and Henley (2001) suggest two approaches that give a fairly high level of variance. The that uncovered employment may have risen because first approach is to exploit the now well-established finding employees and employers collude to avoid costly contribu- that self-employment, which we can use as a proxy for tions to a social protection system that is perceived to be informality, decreases with level of development. If, in fact, inappropriate, inefficient, and a poor value for the money.13 as argued above, this reflects the rising opportunity cost of In principle, then, a universalization of health care de- independent work discussed earlier, then we can see the linked from the labor market may have changed the imposition of the tax as, effectively, a reduction in this cost­benefit analysis of being enrolled in, and hence con- opportunity cost. The second approach uses the Krebs- tributing to, formal sector benefits programs. In the end, Maloney (1999) model of efficiency wages that models the they conclude that this is unlikely, not only because public firm decision to invest and hire in response to the rate at health services continued to be thought of as substantially which formal sector workers quit to enter self-employment, worse than the formal sector product,14 but also because the to calibrate the impact of the tax. effective supply of these services was available even for non- The estimated semi-elasticity of self-employment with contributors several years before the reforms took place, and respect to a change in relative formal sector earnings is 0.03 little progress had been made on implementing the mea- (Maloney 1998) to 0.05 (Loayza and Rigolini 2006) in the sures contemplated in the 1991 social security reform. former case compared with the simulated value of 0.3 in the However, the fall in participation of Brazilian males across latter. Just as an idea of the broad orders of magnitude the 1990s suggests some reasonably potent supply side involved, if we assume that 10 percent of formal sector earn- effects. ings is absorbed in unvalued benefits, and perhaps that It has sometimes been suggested that Argentina's high another 10 percent of the value of earnings is transferred to rates of informal employment are a consequence of low informal workers in untied social protection programs, lead- public confidence in public services and the social security ing to a 20 percent decrease in the relative attractiveness of system, which has resulted in part from repeated economic formal labor, then, in the first case, the size of the informal crises. Argentines do have a relatively low level of confi- self-employed sector is increased by 0.6 to 1.2 percent and, dence in government institutions.15 In 2005, only 18 per- in the second, by 6.0 percent of the share of the workforce. cent, 26 percent, and 26 percent of Argentines reported As an alternative approach, Fernandes, Gremaud, and having confidence in political parties, Congress, and the Narita (2006) simulate the impact of replacing the labor tax judiciary, respectively. However, there is no clear evi- on first minimum wage, thereby effectively providing for- dence of a deterioration in confidence in public institutions mal sector benefits for free, while maintaining revenue 124 T H E I N F O R M A L L A B O R M A R K E T I N M O T I O N : D Y N A M I C S , C Y C L E S , A N D T R E N D S neutrality. If the value-added tax on capital goods is point, that changes in labor market regulation and social deductible, then there is a decline in informality of 1.5 per- security taxes had an important impact in Brazil and cent. If not, however, the decline in physical accumulation Colombia. Supply side impacts could plausibly have been decreases the demand for formal workers and leaves infor- important, although the evidence is speculative. Evidence mality relatively unchanged. The first and third approaches from changes in the difficulty of movement across the suggest modest although nontrivial impacts, and the second de Soto margin--of small firms becoming informal-- suggests substantial impacts of these "supply side" concerns suggests minimal effects working in the other direction. on the size of the informal sector. Policy implications Reforms along the de Soto margin The previous three chapters offer findings that suggest that Finally, probably little responsibility can be placed at the labor market policies are potentially important determi- door of policies that seek to make it easier for microfirms to nants of informality and, hence, there is an agenda for register their workers. As the chapter on firm formality reform. Since experience shows that specific policy recom- decisions shows, these had relatively small effects in the mendations have to be sensitive to country context, this Sistema de Apertura Rápida de Empresas program in section outlines some general principles without attempt- Mexico, the Sistema Integrado de Pagamento de Impostos e ing to be comprehensive. Contribucoes as Microempresas e Empresas de Pequeno Porte Labor policies work on informality through three chan- program in Brazil, and the Monotributo in Argentina. Fur- nels. First, excessive labor costs, whether arising through ther, in all cases, the results are suggestive that, in fact, the labor legislation--exaggerated minimum wages, severance small effects were in the direction of lowering informality. costs, labor taxes, or unrealistic union demands--depress the number of jobs in the formal sector. Other recent reports Conclusion from the Bank, "Minimum Wages and Social Policies: The evidence from the analysis of gross worker flows sug- Lessons from Developing Countries" (Cunningham 2007) gests that, in Mexico and, to a lesser extent, Brazil, the for- and "Job Creation in Latin America and the Caribbean" mal and informal sectors behave across the business cycle as (World Bank 2007c), as well as the Inter-American Devel- if they were dynamically integrated. In both countries, the opment Bank publication Good Jobs Wanted; Labor Markets flows support self-employment as being a desirable sector. in Latin America (IDB 2006), investigate in detail the trade- The picture for informal salaried workers is mixed. In offs between offering strong protections to some workers Mexico, pro-cyclical patterns of rematching across sectors at the possible expense of excluding others. This chapter suggest a sector in line with the previous chapter's findings has offered evidence that, in several countries, exaggerated about the Dominican Republic: informal salaried jobs are labor costs appear to have created the classic segmented not particularly inferior to jobs that comparable workers labor market, especially during downturns. Further, the have in the formal sector. However, the Argentine and experience in the OECD suggests that such regulations Brazilian evidence suggests a case with a higher proportion have an especially heavy exclusionary impact on young peo- of salaried workers queuing for formal sector jobs, consistent ple attempting to find jobs and who, in Latin America, are with the tremendous decline in formality across the 1990s. overrepresented among the involuntary informal salaried. A unified explanation for the expansion of informality The evidence for a very high degree of integration of the across the region over the last decades more generally formal and informal sectors found in several countries and remains elusive, although it appears to result from a combi- subsectors does not necessarily imply satisfactory labor nation of forces--some medium-term macroeconomic in codes. Severance payments, which, in Latin America, nature, others related to longer-term structural changes. Heckman and Pages (2004) find to be among the highest in Some of the increase in the early part of the 1990s was due the world, for example, may substantially reduce the job to nontradable/informal booms driven by liberalization of creation arising from growth without necessarily segment- the exchange rate and subsequent inability to adjust formal ing the market. sector wages downward as difficulties arose surrounding Second, legislation can create incentives to voluntary the exchange rate pegs. Trade liberalization probably had informality, which the last chapters have shown to compose a little impact overall. There is substantial evidence, at this substantial fraction of the sector. This implies that the design 125 I N F O R M A L I T Y of social safety nets and labor legislation needs to take a more allowing workers to collect partial indemnities in some integrated view of the labor market, taking into account the cases of voluntary resignation; introducing individual sav- cost­benefit analysis workers and firms make about whether ings accounts to protect against the risk of unemployment, to interact with formal institutions. This analysis incorpo- and/or, in countries with adequate administrative capaci- rates the benefits of such interactions, the availability of ties, unemployment insurance schemes; establishing more informal substitutes, and the weighing of any greater risk flexible rules concerning dismissals for economic reasons exposure against the benefits of being independent and entre- and seasonal work; ensuring fair, transparent, and timely preneurial. Hence, high labor taxes that do not correspond out-of-court mechanisms to resolve labor disputes, possibly to benefits that workers value cause workers to opt out of the including neutral arbitration or mediation councils--all of formal labor market. The difficulties of managing work and these actions would facilitate job creation and productivity children also lead women to opt into informal independent growth. Minimum wages in most countries are not work where there is more flexibility. As chapter 7 will show, extremely binding, but, in some and particularly in Colom- the provision of substitutes through social security policies bia, they seem to be a brutally segmenting force that begs untied to the labor contract may offer substitutes to formal for moderation. The problem of lack of flexibility to protections that further encourage opting out. support child raising poses special challenges since gender- Third, labor market institutions can have an important specific labor legislation (for example, restrictions of work impact on productivity growth. Theory and anecdotal evi- hours and indiscriminate maternity benefits) may actually dence suggest that excessive restrictions on job reallocation create incentives to discriminate against women, rationing or destruction for just cause, or other related state- or union- them out of the formal market. induced inflexibilities, may have a disincentive effect on Simply tightening the enforcement of existing laws the adoption of new technologies and production processes. regarding, particularly, the largely informal microfirm sec- Productivity gains arising from such innovations account tor may just eliminate jobs, many of which these chapters for half of the differences in levels of economic develop- have shown to be of good quality measured by overall wel- ment, the most important determinants of informality, not fare of the worker. At the other extreme, attempting to to mention long-run worker productivity and welfare more reduce the weight of labor legislation by creating special generally. Hence, the labor market and its institutions classes of less protective contracts can be problematic. On must be viewed as a key part of the national innovation sys- the one hand, their strong incentive effects combined with tem. These issues take on a heightened importance in the the weak enforcement capacities of labor ministries can, in context of more open economies. While there appear to be essence, create a parallel, unregulated market for jobs that has few long-run adverse impacts of trade liberalization on resulted in adverse effects, such as destruction/substitution informality, reaping the benefits of greater integration will away from regulated contracts, higher job turnover, and require greater attention than interactions between trade diminished incentives for training. Further, it may be the and labor regulation. case that they contribute to the overall culture of informality Overall, the present constellation of often well-intended (see chapter 8). On the other hand, the limited use of term but heavy-handed labor regulations serves workers and contracts tied to special provisions (minimum required firms poorly, and both could benefit from substantial training investments) to make high-rotation hiring less reform. In particular, rigorous enforcement of a redesigned attractive can help young people, particularly, enter the labor code that combined strengthened safety nets, well- market. Further, provisions to accommodate different non- designed worker protections, and worker representation wage costs for smaller firms, and flexibility to allow for flex- with the flexibility firms need to adapt in a global economy, ible benefits plans (such as simplified health care/pension has the potential to expand formal employment and reduce plans) under mutual agreement between employers and opting out. employees may allow an extension of the overall rubric of The issues of social protection design are discussed in labor protections without prejudicing the viability of these more detail in chapter 7, but here it is worth mentioning firms. only a few pending reforms of labor protections more specif- In addition, however, the substandard education and ically. Reducing severance payment to bring existing training systems in our region both impede the growth nec- negative incentives for hiring closer to other regions, while essary to generate jobs in the more modern sector of the 126 T H E I N F O R M A L L A B O R M A R K E T I N M O T I O N : D Y N A M I C S , C Y C L E S , A N D T R E N D S economy and reduce workers' attachment to it. Informality Finally, institutional strengthening (staffing, training, sharply decreases with education, partly because the oppor- technical assistance) of the labor ministries and coordina- tunity cost of being independent rises. Further, the poor sig- tion of relevant public agencies (social security administra- naling of education quality due to lack of uniform tion, enterprise development agencies, competitiveness certification or accreditation impedes the entry of young councils) are needed so that they can assume their impor- workers into formal jobs. Remedying these failures, perhaps tant and increasingly more complex role of facilitating with an expansion of intermediation services, may reduce labor productivity growth. the information asymmetries that young workers face. Further discussion of necessary reforms in this area is offered Notes in the World Bank regional report "Improving Educational 1. The self-employed­to­formal (SE-F) and formal­to­self- Quality in Latin America and the Caribbean" (2007). employed (F-SE) transition rates and informal-to-formal (I-F) and Ongoing upgrading of the workforce through training, formal-to-informal (F-I) transition rates show correlations of 0.85 particularly in rapidly evolving industries, is a central ele- and 0.26, respectively. The same is emphatically not true in any case between any of the sectors and unemployment that behave as they do ment of the national innovation system, and critical to in the United States. developing skills used in the modern sector of the economy 2. Hall (2005) calls such inefficient separations "Keynesian." and to productivity growth. Training, however, is not a 3. Finally, it can be argued that the pool of unemployed changes substitute for good schools. Firms overwhelmingly choose its composition during downturns. This has been called the hetero- to train skilled (secondary education or above) workers, and geneity hypothesis. Indeed, crises firing rates increase relative to impacts of training programs on older, unskilled workers quits. If fired workers have a greater propensity to work in the infor- mal sector, this could generate the patterns found in the job-finding are very low. Incentives for worker training of youth rates. This possibility is explored by Bosch and Maloney (2006) and and older workers without prior experience or sufficient the data strongly reject it. Workers with different educational levels, skills--incentives such as tax treatment of investment in ages, or reasons to be unemployed show similar cyclical patterns of human capital similar to that of capital investments, co- job-finding rates. investments of workers, and post-training job matching of 4. Another mechanism is also possible. Firms may post vacancies workers and firms--can ease their entry into the formal and decide whether to formalize the relationship, depending on the quality of the match. A bad shock reduces the overall posting of workforce. To capitalize on the efficacy of training, care vacancies, but it also alters the optimal hiring level of specifically for- should be taken to ensure homogeneous quality and effec- mal workers. Lower overall gain from jobs increases the relative cost tive delivery modalities. Optimal government intervention of signing formal contracts and, therefore, firms shift to a "cheaper" in training systems may require separation of finance and way of producing. These two combined effects lower the job-finding provision, with an unregulated private market and a public rate in the formal sector substantially. However, the job-finding rate subsidy, and a possible new role for public training com- in the informal sector is determined by two opposing effects. On the one hand, the rate is reduced due to the decrease in the number of missions/regulators to certify private providers of training; vacancies created; on the other hand, it is positively affected by may necessitate the building of so-called lifelong learning/ change in the formality levels within the firm. competency systems; and may demand subsidizing techni- 5. The mechanism may differ. Blau (1987), documenting this cal assistance to small and medium-size enterprises for tendency across time in the United States, argued that this and simi- developing human resource strategies. lar trends in the OECD were due to the increased opportunity cost of There is also a role for promoting labor mobility and being self-employed. Maloney argues that given that surveys in the United Kingdom, United States, and Germany (Blanchflower and equal employment opportunities by making greater use of Oswald 1998) suggest that roughly half of salaried workers would labor market intermediation services. Particularly, for women, prefer to be self-employed and not have a boss, this lower labor pro- youth, and minorities, employment centers can provide job ductivity in developing countries may simply make this preference search and employment placement assistance that could affordable. And, in fact, roughly 30 percent of entrants into informal greatly enhance labor mobility, job matching, and labor self-employment in Mexico suggest that greater independence was market equity, while better linking regional labor markets. their motivation. Alternatively, Loayza and Rigolini (2006) argue that income is proxy for lower levels of education, rudimentary infra- With the use of automated systems, these mechanisms have structure, and laggard technology, although the precise mechanism is become easier to implement and administer; and while sev- not elaborated. Finally, Banerji and Jain (2006) argue that it may be eral countries in the region are increasingly making use of preferences for higher-quality goods across the development process them, they remain limited in scope and are underfunded. that reduce the market for informally produced goods. 127 I N F O R M A L I T Y 6. Following Goldberg and Pavcnik (2003), the first stage Aleman-Castilla, B. 2006. "The Effect of Trade Liberalization on applies a linear probability model to regress the formal dummy indi- Informality, Wages, and Productivity: Evidence from Mexico." cator on a vector of worker characteristics (gender, age, age2, and edu- Photocopy. Department of Economics, London School of Econom- cation indicators) and on a set of industry indicators representing the ics and Political Science, UK. workers' industry affiliations. The coefficients on industry indicators Almeida, R., and P. Carneiro. 2005. "Enforcement of Regulation, can be considered "industry formality differentials" stripped of Informal Labor and Firm Performance." IZA Discussion Paper worker characteristics that, in the second stage, are pooled over time 1759, Institute for the Study of Labor, Bonn, Germany. and regressed on trade-related industry characteristics (tariffs and Arango, C., and A. Pachón. 2004. "Minimum Wage in Colombia: imports penetration), either in a fixed-effect or first differences speci- Holding the Middle with a Bite on the Poor." Borradores de fication (Goñi and Maloney 2007). Economía, Banco de la República, 280. 7. Because of the nature of calculating the contribution of each Banerji, A., and S. Jain. 2006. "Quality Dualism." Photocopy. Depart- factor from the regressions, the sum may not equal 100. ment of Economics, University of Virginia, Charlottesville. 8. A comprehensive survey of the literature studying the size and Beccaria, L., J. Carpio, and A. Orsatti. 1999. "Argentina: informal- evolution of the Brazilian informal sector in the labor market can be idad laboral en el nuevo modelo económico." In Informalidad y found in Ulyssea (2005), and a summary of stylized facts of this sector Exclusión Social, ed. J. Carpio, E. Kleinm, and I. Novacovsky, covering the 1980s and 1990s is detailed in Ramos (2002), Ramos 143­50. Buenos Aires, Argentina: Fondo de Cultura and Brito (2003), Ramos and Ferreira (2005a, 2005b), Ramos and Económica. Reis (1995), and Veras Soares (2004). Besley, T., and R. Burgess. 2004. "Can Labor Regulation Hinder 9. On the other hand, Cardenas and Gutierrez (1997) find much Economic Performance? Evidence from India." Quarterly Journal smaller implicit elasticities of demand, and argue that the larger of Economics 119 (1): 91­134. effects were due to the fall in output and the overvaluation of the peso Bewley, T. 1999. Why Wages Don't Fall during a Recession. Cambridge, that drove the rise in unemployment. MA: Harvard University Press. 10. Firing costs were reduced in two dimensions. First, the Blanchflower, D. G., and A. J. Oswald. 1998. "What Makes an tenure-related severance payments schedule was made less steep. Sec- Entrepreneur?" Journal of Labor Economics 16 (1): 26­60. ond, and more important, was the elimination of the obligation to Blau, D. 1987. "A Time-Series Analysis of Self-Employment in the reinstate a worker in his of her job if a labor court determined that United States." Journal of Political Economy 95 (3): 445­67. the dismissal had no just cause. Bosch, M. 2006. "Job Creation and Job Destruction in the Presence 11. By law, temporary workers may belong to a union, but firms of Informal Labor Markets." Photocopy. London School of Eco- implicitly threatened workers who registered in a union by not nomics, UK. renewing the contract. Bosch, M., E. Goñi, and W. F. Maloney. 2006. "The Determinants of 12. This and the next paragraph are nearly complete plagiarism Rising Informality in Brazil: Evidence from Gross Worker from the Argentina Poverty Assessment (2007). Flows." Photocopy. World Bank, Washington, DC. 13. Carneiro and Henley (2001) estimate that the earnings pre- Bosch, M., and W. F. Maloney. 2006. "Gross Worker Flows in the mium needed in the marketplace to compensate covered workers for Presence of Informal Labor Markets: The Mexican Experience having to make social security contributions varies between 7.5 and 1987­2002." Photocopy. World Bank, Washington, DC. 12.2 percent of the mean uncovered hourly wage. Botero, J., S. Djankov, R. La Porta, F. Lopez-de-Silanes, and A. 14. The public system acts as a floor, available to all but used pri- Shleifer. 2003. "The Regulation of Labor." Working Paper 9756, marily by the lower classes (Jack 2000). Although evaluation of stan- National Bureau of Economic Research, Cambridge, MA. dards for minimum quality in infrastructure; human resources; and Bour, J., and N. Susmel. 2000. "Los Determinantes de la Informali- ethical, technical, and scientific procedures in hospitals have been dad Laboral." In La Economía Oculta en la Argentina, 225­66. implemented, these practices are far from being universal in the ser- Buenos Aires, Argentina: FIEL. vices network (PAHO 2005). Cardenas, M., and R. Bernal. 2003. "Determinants of Labor Demand 15. This paragraph is taken from the Argentina Poverty Assess- in Colombia: 1976­1996." Working Paper 10077, National ment (World Bank 2007a). Bureau of Economic Research, Cambridge, MA. Cardenas, M., and C. Gutierrez. 1997. "Demanda de Trabajo, References Impuestos a la Nomina y Desempleo en Colombia." Photocopy. Acemoglu, D. 2003. "Patterns of Skill Premia." Review of Economic Fedesarrollo, Bógota, Colombia. Studies 70: 199­230. Carneiro, F. G., and A. Henley. 2001. "Modelling Formal and Infor- Albrecht, J., and S. Vroman. 2002. "A Matching Model with mal Employment and Earnings: Microeconometric Evidence for Endogenous Skill Requirements." International Economic Review Brazil." Proceedings of the 29th Meeting of the Brazilian 43: 283­305. Economic Society, ANPEC (Associação Nacional dos Centros de Albrecht, J., L. Navarro, and S. Vroman. 2006. "The Effects of Labor Pós-Graduação en Economia). Market Policies in an Economy with an Informal Sector." IZA Dis- Contreras, D., E. Puentes, and C. Sanhueza. 2006. "Informality in cussion Paper 2141, Institute for the Study of Labor, Bonn, Germany. Chilean Labor Markets." Photocopy. World Bank, Washington, DC. 128 T H E I N F O R M A L L A B O R M A R K E T I N M O T I O N : D Y N A M I C S , C Y C L E S , A N D T R E N D S Cunningham, W. 2001a."Breadwinner versus Caregiver: Labor Force Gasparini, L., F. Haimovich, and S. Olivieri. 2006. "Labor Informal- Participation and Sectoral Choice over the Mexican Business ity Effects of the Programa Jefes de Hogar." Photocopy. Center for Cycle." In The Economics of Gender in Mexico: Family, Work, State, and Distribution, Labor and Social Studies, National University of la Market, ed. E. G. Katz and M. C. Correia, 85­132. Washington, Plata, Buenos Aires, Argentina. DC: World Bank. Gasparini, L., and L. Tornarolli. 2006. "Labor Informality in Latin ------. 2001b. "Household Risk, Self-Insurance, and Coping America and the Caribbean: Patterns and Trends from Household Strategies in Urban Argentina." Report 22426-AR, World Bank, Survey Microdata." Photocopy. World Bank, Washington, DC. Washington, DC. Goldberg, P., and N. Pavcnik. 2003. "The Response of the Informal ------. 2007. "Minimum Wages and Social Policies: Lessons from Sector to Trade Liberalization." Journal of Development Economics Developing Countries." Photocopy. World Bank, Washington, DC. 72: 463­96. Davis, S., R. J. Faberman, and J. Haltiwanger. 2005. "The Flow Gollin, D. 2002. "Getting Income Shares Right." Journal of Political Approach to Labor Markets: New Data Sources, Micro-Macro Links Economy 110 (2): 458­74. and the Recent Downturn." Photocopy. University of Chicago. Goñi, E., and W. Maloney. 2007. "On the Effects of Trade on Infor- Davis, S., and J. C. Haltiwanger. 1992. "Gross Job Creation, Gross mality. A Note for the Argentinean Case." Photocopy. World Job Destruction and Employment Reallocation." Quarterly Jour- Bank, Washington, DC. nal of Economics 107 (3): 819­63. Hall, R. 2005. "Employment Efficiency and Sticky Wages: Evidence ------. 1999. "On the Driving Forces behind Cyclical Movements Flows in the Labor Market." Working Paper 11183, National in Employment and Job Reallocation." American Economic Review Bureau of Economic Research, Cambridge, MA. 89 (5): 1234­58. Hanson, G. H., and R. Robertson. 2006. "China and the Recent Evo- de Ferranti, D., G. Perry, D. Lederman, and W. Maloney. 2001. From lution of Mexico's Manufacturing Exports." Photocopy. Univer- Natural Resources to the Knowledge Economy. Washington, DC: sity of California, San Diego. World Bank. Harris, J. R., and M. P. Todaro. 1970. "Migration, Unemployment de Soto, H. 1989. The Other Path: The Invisible Revolution in the Third and Development: A Two-Sector Analysis." American Economic World. New York: Harper and Row. Review 60 (1): 126­42. Djankov, S., R. La Porta, F. Lopez-de-Silanes, and A. Shleifer. 2002. Heckman, J., and C. Pages. 2004. Law and Employment: Lessons from "The Regulation of Entry." Quarterly Journal of Economics 117 (1): Latin America and the Caribbean. Chicago: University of Chicago 1­37. Press. Dolado, J. J., M. Jansen, and J. F. Jimeno. 2002. "On-the-Job Search in Holmes, T. 1998. "The Effect of State Policies on the Location of a Matching Model with Heterogenous Jobs and Workers." IZA Dis- Manufacturing: Evidence from State Borders." Journal of Political cussion Paper 886, Institute for the Study of Labor, Bonn, Germany. Economy 106: 667­705. Fernandes, R., A. P. Gremaud, and R. D. T. Narita. 2006. "Tax Hopenhayn, H. 1988. "A Competitive Stochastic Model of Entry and Reform and Formalization of the Economy: Simulating Different Exit to an Industry." Photocopy. University of California, Los Alternatives for Brazil." Photocopy. Department of Economics, Angeles. University of São Paulo, Brazil. ------. 2004. "Labor Market Policies and Employment Duration: Fiess, N., M. Fugazza, and W. Maloney. 2002. "Exchange Rate The Effects of Labor Market Reform in Argentina." In Law and Appreciations, Labor Market Rigidities, and Informality." Policy Employment: Lessons from Latin America and the Caribbean, ed. J. Research Working Paper 2771, World Bank, Washington, DC. Heckman and C. Pages. Chicago: University of Chicago Press. ------. 2006. "Informal Labor Markets and Macroeconomic Fluctu- IDB (Inter-American Development Bank). 2006. Good Jobs Wanted: ations." Photocopy. World Bank, Washington, DC. Labor Markets in Latin America. Washington, DC. Friedman, E., S. Johnson, D. Kaufmann, and P. Zoido-Lobatón. 2000. Jack, W. 2000. "Health Insurance Reform in Four Latin American "Dodging the Grabbing Hand: The Determinants of Unofficial Countries. Theory and Practice." Policy Research Working Paper Activity in 69 Countries." Journal of Public Economics 76: 459­93. 2492, World Bank, Washington, DC. Galiani, S., and P. Sanguinetti. 2003. "The Impact of Trade Liberal- Jimeno, J., and D. Rodriguez Palenzuela. 2003. "Youth Unemploy- ization on Wage Inequality: Evidence from Argentina." Journal of ment in the OECD: Demographic Shifts, Labour Market Institu- Development Economics 72: 497­513. tions and Macroeconomic Shocks." Working Paper 19, European Galiani, S., and F. Weinschelbaum. 2006. "Modeling Informality Network of Economic Policy Research Institutes, Brussels. Formally: Households and Firms." Photocopy. World Bank, Jovanovic, B. 1979. "Job Matching and the Theory of Turnover." Washington, DC. Journal of Political Economy 87 (5): 972­90. García-Verdú, R. 2007. "Measurement of the Shadow Economy or ------. 1982. "Selection and Evolution of Industry." Econometrica 50 Shadowy Measurement?" Photocopy. World Bank, Washington, (3): 649­70. DC. Kennan, J. 2005. "Private Information, Wage Bargaining and Gasparini, L. 2002. "Microeconometric Decompositions of Aggre- Employment Fluctuations." University of Wisconsin, Madison. gate Variables: An Application to Labour Informality in Krebs, T., and W. Maloney. 1999. "Quitting and Labor Turnover: Argentina." Applied Economics 34 (18): 2257­66. Microeconomic Evidence and Macroeconomic Consequences." 129 I N F O R M A L I T Y Policy Research Working Paper 2068, World Bank, Washington, Mortensen, D., and C. Pissarides. 1994. "Job Creation and Job DC. Destruction in the Theory of Unemployment." Review of Economic Kristensen, N., and W. Cunningham. 2006. "Do Minimum Wages Studies 61: 397­415. in Latin America and the Caribbean Matter? Evidence from 19 Nickell, S., and R. Layard. 2000. "Labor Market Institutions and Countries." Policy Research Working Paper 3870, World Bank, Economic Performance." In Handbook of Labor Economics, ed. O. Washington, DC. Ashenfelter and D. Card. Amsterdam: North-Holland. Kugler, A. 2000. "The Incidence of Job Security Regulations on Obstfeld, M., and K. Rogoff. 1996. "Foundations of International Labor Market Flexibility and Compliance in Colombia: Evidence Macroeconomics." Photocopy. Massachusetts Institute of Tech- from the 1990 Reform." Working Paper R-393, Inter-American nology, Cambridge. Development Bank, Washington, DC. Olarreaga, M., L. Castro, and D. Saslavsky. 2006. "The Impact of Kugler, A., and M. Kugler. 2003. "The Labor Market Effects of Pay- Trade with China and India on Argentina's Manufacturing roll Taxes in a Middle-Income Country: Evidence from Colom- Employment." Photocopy. World Bank, Washington, DC. bia" IZA Discussion Paper 852, Institute for the Study of Labor, Olmedo, C., and M. Murray. 2002. "The Formalization of Infor- Bonn, Germany. mal/Precarious Labor in Contemporary Argentina." International Lederman, D., M. Olarreaga, and I. Soloaga. 2006. "The Growth of Sociology 17 (3): 421­43. China and India in World Markets: Opportunity or Threat for Paes de Barros, R., and C. H. Corseuil. 2001. "The Impact of Latin American Exporters?" Background paper for the regional Regulations on Brazilian Labor Market Performance." Research study, Latin America and the Caribbean Respond to the Growth of Paper R-427, Inter-American Development Bank, Washington, China and India. World Bank, Washington, DC. DC. Levy, S. 2006. "Productividad, Crecimiento y Pobreza en México: ------. 2004. "The Impact of Regulations on Brazilian Labor ¿Qué sigue después de Progresa-Oportunidades?" Photocopy. Market Performance." In Law and Employment: Lessons from Latin World Bank, Washington, DC. America and the Caribbean, ed. J. Heckman and C. Pages. Chicago: Loayza, N. V., A. M. Oviedo, and L. Servén. 2006. "The Impact of University of Chicago Press. Regulation on Growth and Informality: Cross-country Evidence." PAHO (Pan American Health Organization). 2005. "Hospital World Institute for Development Economics Research, United Accreditation Manual." Washington, DC. Nations University, Helsinki. Perry, G., and L. Servén. 2005. "Argentina's Macroeconomic Col- Loayza, N., and J. Rigolini. 2006. "Informality Trends and Cycles." lapse: Causes and Lessons." In Managing Economic Volatility and Photocopy. World Bank, Washington, DC. Crises: A Practitioner's Guide. ed. J. Aizenmann and B. Pinto. New Maloney, W. 1998. "Are Labor Markets in Developing Countries York: Cambridge University Press. Dualistic?" Policy Research Working Paper 1941, World Bank, Pissarides, C. 1985. "Short-Run Equilibrium Dynamics of Unem- Washington, DC. ployment, Vacancies, and Real Wages." American Economic Review ------. 2001. "Self-employment and Labor Turnover in Developing 75: 676­90. Countries: Cross-country Evidence." In World Bank Economists' ------. 2000. Equilibrium Unemployment Theory. 2nd ed. Cambridge, Forum, ed. S. Devarajan, F. H. Rogers, and L. Squire. Washington, MA: Massachusetts Institute of Technology Press. DC: World Bank. Porto, G., and S. Galiani. 2006. "Trends in Tariff Reforms and ------. 2004. "Informality Revisited." World Development 32 (7): Trends in Wage Inequality." Policy Research Working Paper 1159­78. 3905, World Bank, Washington, DC. Maloney, W. F., and J. Nuñez Mendez. 2003. "Measuring the Impact Pries, M., and R. Rogerson. 2005. "Hiring Policies, Labor Market of Minimum Wages: Evidence from Latin America." Working Institutions, and Labor Market Flows." Journal of Political Economy Paper 9800, National Bureau of Economic Research, Cambridge, 113 (41): 811­39. MA. Ramos, L. 2002 "A evolução da informalidade no Brasil metropoli- Menezes-Filho, N. A. 1997. "Unions and Profitability over the tano: 1991-2001." Discussion Paper 914, Institute of Applied 1980s: Some Evidence on Union-Firm Bargaining in the United Economic Research, Brasilia. Kingdom." Economic Journal 107 (442): 657­70. Ramos, L., and M. Brito. 2003. "O Funcionamento do mercado de Menzio, G. 2005. "High-Frequency Wage Rigidity." Photocopy. trabalho metropolitano brasileiro no período 1991-2002: tendên- Northwestern University, Chicago. cias, fatos estilizados e mudanças estruturais." Boletim Mercado de Mondino, G., and S. Montoya. 2002. "The Effects of Labor Market Trabalho, Conjuntura e Análise 22: 31­47. Regulations on Employment Decisions by Firms: Empirical Ramos, L., and V. Ferreira. 2005a. "Spatial and Sector Patterns of Evidence for Argentina." In Law and Employment: Lessons from the Evolution of Informality in Brazil: 1991­2003." Discus- Latin America and the Caribbean, ed. J. Heckman and C. Pages. sion Paper 1099, Institute of Applied Economic Research, Chicago: University of Chicago Press. Brasilia. Mortensen, D., and E. Nagypal. 2005. "More on Unemployment and ------. 2005b. "Spatial Pattern of Formal Employment Vacancy Fluctuations." Photocopy. Northwestern University, Evolution--1995­2003" Discussion Paper 1102, Institute of Chicago. Applied Economic Research, Brasilia. 130 T H E I N F O R M A L L A B O R M A R K E T I N M O T I O N : D Y N A M I C S , C Y C L E S , A N D T R E N D S Ramos, L., and J. Reis. 1995. "Salário mínimo, distribuição de renda ------. 2005b. "The Cyclicality of Hires, Separations, and Job- e pobreza no Brasil." Pesquisa e Planejamento Econômico 25 (1): to-Job Transitions." Federal Reserve of St. Louis Review 87 (4): 99­114. 493­507. Rauch, J. E. 1991. "Modelling the Informal Sector Formally." Journal ------. 2005c. "Reassessing the Ins and Outs of Unemployment." of Development Economics 35: 33­47. Photocopy. University of Chicago. Saavedra, J., and A. Chong. 1999. "Structural Reform, Institutions Ulyssea, G. 2006. "Brazilian Labor Market Informality: A Review of and Earnings: Evidence from the Formal and Informal Sectors in the Literature." Revista de Economia Politica Oct./Dec. 2006, 26 Urban Peru." Journal of Development Studies 35 (4): 95­116. (4): 596­618. Saavedra, J., and M. Torero. 2004. "Labor Market Reforms and Their ------. 2006. "Informalidade no mercado de trabalho Impacts over Formal Labor Demand and Job Market Turnover: brasileiro: uma resenha da literatura." Revista de Economia The Case of Peru." In Law and Employment: Lessons from Latin Política 26 (3). America and the Caribbean, ed. J. Heckman and C. Pages. Chicago: Veras Soares, F. 2004. "Some Stylized Facts of the Informal Sector in University of Chicago Press. Brazil in the Last Two Decades." Anais do XXXII Encontro Santamaria, M. 2000. "External Trade, Skill, Technology and the Nacional de Economia, 142, ANPEC. Recent Increase of Income Inequality in Colombia." Dissertation. World Bank. 2007a. Argentina Poverty Assessment. Informal Employment Available at http://econpapers.repec.org/. in Argentina: Towards Understanding Its Causes and Consequences. Schneider, F., and D. H. Enste. 2000. "Shadow Economies: Size, Washington, DC. Causes, and Consequences." Journal of Economic Literature 38 (1): ------. 2007b. "Improving Educational Quality in Latin America 77­114. and the Caribbean." Washington, DC. Shimer, R. 2005a. "The Cyclical Behavior of Equilibrium Unemploy- ------. 2007c. "Job Creation in Latin America and the Caribbean." ment and Vacancies." American Economic Review 95 (1): 25­49. Photocopy. Washington, DC. 131 CHAPTER 5 Microfirm Dynamics and Informality SUMMARY: This chapter describes the patterns of entry, survival, and growth in the self-employment sector of selected Latin American countries. The chapter shows that those patterns match well the predictions of mainstream models of firm dynamics. Entrants into self-employment tend to be found more frequently among workers who have accumulated human and physical capital while working as salaried workers. Higher-productivity firms are more likely to stay in business and create new jobs while becoming increasingly formal in the process. Few firms, however, tend to evolve along this ideal path, with most of them remaining informal and owner-only. While to some extent this can be explained by the low opportunity costs for entry into the sector, policy-induced barriers to formalization may also play a role by impeding microfirm access to technologies and markets. A S ARGUED EARLIER IN THIS REPORT, THE microenterprises, including the determinants of entry, exit, owners and employees of very small, and growth of small firms. mostly informal firms constitute a major After briefly reviewing some of the industrial country lit- share of Latin America's labor force. Those erature on firm behavior, this chapter presents evidence on workers thus play a very important role in the patterns of entry, exit, and growth of microenterprises in the functioning of the region's labor markets, an issue that Latin America. We then explore the personal and enterprise has been discussed at length in previous chapters. However, characteristics associated with microfirm formality. Insofar a full understanding of the dynamics of the informal sector as our results regarding microfirm dynamics are consid- also requires a complementary perspective that focuses on erably similar to those obtained for industrial countries, we the firm rather than on the worker as a unit of analysis. argue that they support the idea that the informal microen- Indeed, to the extent that entry and exit flows in the terprise sector is, to some extent, made up of voluntary microenterprise sector are at least partially driven by volun- entrants who choose to be there, taking into account their tary choices regarding forms of participation in the labor business prospects in the sector as well as the attributes of market, they must also involve calculations related to the the jobs available in other segments of the labor market. We expected profitability of running one's own business. Thus, interpret this as evidence that the mainstream view may for example, to the extent that exit, not exclusion, is behind be well suited to understanding the developing country's the decision of the many formal salaried workers who microenterprise sector. This approach is consistent with the open new microfirms every year in Latin America, under- early work on Kenya by Hart (1972) and the mounting standing their behavior may require treating them as evidence of entrepreneurial dynamism and relative job satis- profit-maximizing entrepreneurs rather than disguised faction of self-employed individuals--see Bhattacharaya unemployed workers queuing for formal jobs. In that (2002), de Soto (1989), and Maloney (1999, 2004). respect, the size and evolution of the sector depend, at least Furthermore, we find evidence that formality increases to some extent, on the factors that affect the dynamics of rapidly with firm size and productivity, and it is higher 133 I N F O R M A L I T Y among those who voluntarily enter self-employment. Thus, growth in the microenterprise sector, as well as theoretical some of the same attributes associated with entry and success frameworks to explain them. However, notwithstanding the in self-employment are also correlated with compliance increasing importance given to the promotion of micro and with government regulations. This suggests that formality small enterprises in development policy circles, there have operates as a normal input into the production process, been few systematic attempts to see how their dynamics with a minimal degree of participation in some institutions approximate those of the mainstream literature.1 This repre- being a necessary input to growth, and participation sents a loss on two fronts. First, if it seems that behavioral increasing with the success of the business. In most cases, differences are not so great, then development policy makers however, the large majority of microfirms remain too small have a wealth of analytical frameworks at their disposal. Sec- to make the benefits of formality overcome its various costs. ond, a finding of kinship with their advanced country coun- This could be due to the presence of policy-induced barriers terparts would provide additional evidence on the debate to formalization--for example, high registration costs, on how we should conceive of the role of the informal stringent tax and labor market regulations, and other factors microfirm in the developing world. reviewed in the next chapter--that would in turn limit the In the Organisation for Economic Co-operation and growth of informal microfirms and contribute to a vicious Development (OECD), opening a business and being one's continuing circle of low growth and high informality. boss is often celebrated as a desirable alternative to salaried This chapter, however, makes the case for a complemen- work; however, in the developing world, the very large tary interpretation of the evidence. In particular, given the unregulated (informal) microenterprise sectors are fre- similarity between microfirm dynamics in developing and quently seen as the disadvantaged segment of a dual labor industrial countries, we argue that the high rates of market in which workers queue for good jobs. The two microfirm failure and the low rates of job creation and views have different implications for entry and firm dynam- growth, which keep Latin American firms below the thresh- ics. In particular, rather than the patterns of entry/exit and old over which formality becomes a necessity, could also be growth associated with entrepreneurship dynamics, the tra- the result of the low levels of human capital and managerial ditional least developed countries (LDC) literature on infor- ability found among many of the region's microentrepre- mality would predict patterns consistent with the dynamics neurs. This, in turn, could be due to relatively low opportu- of disguised unemployment. nity costs of entering self-employment, driven in turn by low productivity and salaries in the formal sector. The prac- The mainstream view of firm dynamics tical implication of this interpretation is that policy makers The dominant view of the role of self-employment in interested in altering the incentives that drive the large industrial countries stresses the risk-taking, entrepreneur- majority of Latin American microfirms into informality ial nature of the sector, with the celebrated Silicon Valley should focus not only on the direct costs and benefits of for- high tech start-up at its apex. In the classic framework pro- mality, but also on the factors that, by keeping formal sector posed by Lucas (1978), individuals are endowed with a productivity low, indirectly increase the number and reduce given--and known--level of entrepreneurial or managerial the average "quality"--in terms of their managerial ability ability, which determines the returns from self-employment. and inherent business prospects--of new entrants into self- Lucas argued that there is a distribution of entrepreneurial employment. In other words, the agenda for reducing ability in the population: those with a sufficiently high microfirm informality should also cover the factors that drive level of proficiency become entrepreneurs, while the rest productivity in the formal sector, encompassing improve- become wage workers. Among those individuals whose ments in the investment climate and policies aimed at entrepreneurial ability leads them to choose self-employment, increasing the quantity and quality of human capital. the more proficient have firms that are larger and/or more successful. This means that the existence of many small Conceptual framework: Firm dynamics firms does not necessarily imply failure of either labor or and institutional development credit markets; it may also reflect the inherent distribution In the industrial countries, the last two decades have seen of managerial ability in the population. the emergence of a set of stylized facts about the personal This mainstream view of microentrepreneurship is sup- and firm characteristics associated with entry, survival, and ported by evidence showing that, other things equal, some 134 M I C R O F I R M D Y N A M I C S A N D I N F O R M A L I T Y individuals may derive a larger utility from entrepreneur- their costs, so it is very unlikely that new information will ship than from wage work, thus reducing the net opportu- be unfavorable enough to induce them to exit. In addition, nity cost of entering self-employment. Evidence of this has larger firms are those that found high levels of profits when been provided by Blanchflower and Oswald (1998b), who they started, and hence grew. Thus, their efficiency level is stress that 63 percent, 48 percent, and 49 percent of salaried further away from the exit-inducing threshold and they are workers in the United States, the United Kingdom, and less likely to fail.2 Germany, respectively, report that they would prefer to be Empirical evidence favoring the prediction of a negative self-employed, a fraction remarkably similar to the over link between time in business and size, on one hand, and 60 percent of Mexican entrants to self-employment who survival and growth on the other, was obtained for the report doing so voluntarily (see Maloney 1999). Further- United States by Dunne, Roberts, and Samuelson (1988, more, Blanchflower (2004) reiterates the robust finding 1989) and Evans (1987a, 1987b), for Germany by Wagner that those who enter self-employment report higher levels (1994), for the United Kingdom by Geroski (1991), for 10 of job satisfaction than employees. Moreover, Hamilton OECD countries by Bartelsman, Scarpetta, and Schivardi (2000) finds that nonpecuniary benefits--such as being (2003), and for 24 industrial and developing countries one's own boss--explain the lower conditional earnings by Bartelsman, Haltiwanger, and Scarpetta (2004). In a generally found in the sector. developing-country context, however, one could argue Jovanovic (1982) added dynamics to Lucas's view by fur- that market and government failures associated with, for ther assuming that managerial ability is uncertain, and that instance, credit rationing or inefficient contract enforce- individuals can gradually learn about their true cost struc- ment mechanisms could limit the growth of small and ture only by opening and operating a business. Thus, entre- newly created firms, which then would not exhibit the rel- preneurs cannot know how good their location is, or how atively higher expansion rates of their industrial-country good their managerial ability is, until they actually start the counterparts. In this scenario, surviving small firms would business. Entry into self-employment involves a fixed cost, maintain abnormally high profits given their size and effi- which only those with high expected ability and profits may ciency, a prediction that is consistent with the high rates of be willing to pay. After entry, entrepreneurs incorporate the return estimated for small enterprises in the developing information from their actual profits, revise their ability world (see McKenzie and Woodruff 2006, and the refer- estimates, and adjust the level of profit-maximizing output ences therein). The next section presents direct evidence on accordingly. Firms with consistently lower-than-expected the subject, aimed at exploring the similarities and differ- profits tend to contract and eventually go out of business, ences in microfirms' patterns of survival and growth in while unexpectedly high profits cause upward revisions of industrial and developing-country contexts. ability estimates and lead to firm expansion. The framework proposed by Jovanovic (1982) helps Determinants of entry into self-employment explain why small firms are generally less "productive" or A model of industry dynamics that generates implications "efficient." Firms that receive favorable cost information that are broadly similar to Jovanovic's, but is especially use- tend to revise upward their optimal size estimates, and thus ful to analyze issues related to entry into self-employment, to grow faster. Firms that remain very small are those that is the one proposed by Hopenhayn (1992). This model is have received negative signals concerning their cost parame- especially notable for its analysis of the effect of the cost of ters. They are less efficient and have relatively high produc- entry, which can be interpreted as the outside opportunity tion costs, which lead them to stay small. This framework costs for some resources (for example, managerial ability) also helps explain why small firms tend to die more often (see used by the firm. Higher costs of entry lead to a lower Ericson and Pakes 1995; and Lippman and Rumelt 1982). turnover rate, because more ex ante selection occurs. This Indeed, there is a threshold level of efficiency below could be particularly relevant in developing-country con- which it is not profitable to stay in business. Small young texts characterized by higher levels of informality and firms are more likely to cross that threshold when negative lower productivity in the salaried sector, which in Hopen- shocks suggest that in reality their efficiency is below what hayn's model would lead to a lower entrepreneurial ability they had expected. On the other hand, firms that have been threshold for entering self-employment and thus to higher in business for a long time will have an accurate estimate of entry--and exit--rates. 135 I N F O R M A L I T Y In regard to the personal characteristics likely to be asso- prone to move into self-employment. However, in the pres- ciated with entry into self-employment, Johnson (1978) ence of credit constraints, workers earning higher wages in and Jovanovic (1979) postulate that since young people are the salaried sector may also be able to accumulate capital less risk-averse, they would be overrepresented among faster and hence be more likely to enter self-employment. entrants. A similar prediction would emanate from the Moreover, there may be a correlation between previous standard queuing view of LDC labor markets, with young productivity in the salaried sector, remuneration, and people being more likely to be rationed out. Empirical evi- entrepreneurial ability--or at least competence in the dence for the United States, however, suggests the reverse chosen field of entrepreneurship. To the degree that this pattern, with entry increasing with age. As an explanation, would imply a higher probability of success in self- Evans and Jovanovic (1989) offer a variation of Lucas's employment, we may expect those with conditionally model where binding liquidity constraints may lead indi- higher earnings in the formal sector to enter self- viduals to delay or forgo profitable business opportunities, employment while those less skilled may choose not to reducing entry rates and increasing exit rates among those take the risks. with low personal assets, including the young. Further- In sum, the impact of personal characteristics of existing more, they argue that since credit-constrained individuals and would-be entrepreneurs could be potentially different are more likely to start small businesses with a suboptimal in developing-country contexts, at least in the hypothesis amount of capital, returns to capital will be higher and that self-employment in LDCs is driven by the presence of smaller firms (with lower capital stocks) will grow faster dualistic labor markets. Indeed, in this case unemployed than firms that entered closer to their steady state. individuals, those out of the labor force, young workers, and Evans and Jovanovic's model of credit constraints to vol- those with lesser schooling and lower wages should all be untary entry is largely inconsistent with the dualistic view more likely to be self-employed, as they would be in a worse of self-employment as an easy entry holding pattern in sev- position for finding formal salaried jobs. In contrast, the eral ways. Indeed, both the dualistic view of informal labor mainstream literature suggests that older, better-educated, markets and the advanced country sociological literature and well-paid workers with experience in the salaried sector that sees the numerous self-employed among certain ethnic should have a higher probability of entering, staying, and minorities as recruited from "misfits"--individuals who growing in the self-employment sector as they should be lack access to salaried employment due to, for instance, lan- more likely to have accumulated the assets required to start guage barriers, a history of unemployment, or limited labor a business, and better positioned to find, assess, and take market experience (see Carrasco 1999; Evans and Leighton advantage of good business opportunities. 1989; and the references therein)--would predict lower Finally, two additional covariates appear in the main- entry rates into self-employment coming from salaried stream literature but so far yield ambiguous predictions work than from unemployment or from out of the labor (see box 5.1 for a summary of empirical findings in indus- market. However, the opposite prediction could be derived trial country contexts). First, we might imagine that workers from the Evans and Jovanovic's view if individuals acquire with more schooling would find better matches as salaried more capital and knowledge of business opportunities-- workers in larger firms that could better utilize their and to some extent their own managerial ability--while specialized skills. On the other hand, Cressy (1996) and working than while unemployed or out of the labor mar- Rees and Shah (1986) argue that more educated individuals ket. In particular, if formal schooling is relatively poor and can have lower costs of assessing business opportunities, most relative human capital is accumulated on the job, and that human capital may be a complement to manager- then we may find that salaried employment is a logical ial ability. This is suggested by Bates (1990), who shows stepping stone to self-employment. that in the United States, the probability of survival of Moreover, the impact of the level of remuneration small businesses is positively related to the level of educa- earned in these previous jobs also offers some insights that tion of their owners. Second, Carrasco (1999) argues that may help distinguish between mainstream and dualistic men who are married could be less willing to take risks. On models. All things equal, we might expect that those the other hand, the Mexican sociologist González de la earning higher wages in salaried work would be less likely Rocha (1994) suggests that the possibility of combining to be "misfits," or unsuited to formal work, and hence less the self-employment earnings of the household head or 136 M I C R O F I R M D Y N A M I C S A N D I N F O R M A L I T Y BOX 5.1 Patterns of entry and exit in industrialized countries Self-employment and age Unemployment and entry into self-employment Evans and Leighton (1989) find that, for U.S. men, entry Carrasco (1999) finds that in Spain, for given personal into self-employment increases only slightly after the late characteristics, the predicted probability of entering self- 20s, exit falls at a decreasing rate with age, and the self- employment is lower for wage workers than for unem- employment rate reaches a plateau when individuals are ployed individuals. However, the latter become less in their 40s. For Spain, Carrasco (1999) finds that entry likely to switch either when they receive unemployment rates are highest for middle-age people, especially for benefits or when aggregate business conditions are unfa- those aged 35­45. In the United States, Bates (1990) vorable. In the United States, Evans and Leighton (1989) finds that rates of exit from self-employment are lowest find that U.S. men who changed jobs frequently, and who for owners of small businesses aged 45­54. Holtz-Eakin, experienced relatively frequent and long spells of unem- Joulfaian, and Rosen (1994a) find that exit rates reach ployment, have a higher probability of entering self- their minimum at the age of 47. In addition, for each of employment. the 18 industrial countries considered by Blanchflower and Oswald (1998b), a robust positive relationship Self-employment and liquidity constraints emerges between age and the probability of being Evans and Leighton (1989), Evans and Jovanovic (1989), self-employed. and Carrasco (1999) find that higher net worth increases Self-employment and education the probability of entering self-employment. Evans and Bates (1990) finds that in the United States the proba- Jovanovic (1989) also find that initial assets before enter- bility of survival of small businesses rises with educa- ing self-employment are positively related to earnings, at tion. In Spain, entry into the sector also rises with least during the initial years in the sector--as firms grow, education, as found by Carrasco (1999), particularly for the importance of the initial liquidity constraint dimin- those who become self-employed with employees. In ishes. Holtz-Eakin, Joulfaian, and Rosen (1994a, 1994b) other industrial countries, Blanchflower (2000) finds find that liquid assets and inheritances increase the that self-employment rates are usually highest for indi- likelihood of entry into self-employment, reduce the viduals with a small number of years of schooling--the probability of exit, and increase the earnings of surviving United Kingdom being the only exception, with the businesses. Blanchflower and Oswald (1998b) provide reverse being the case--followed by those with college similar evidence for the United Kingdom; by relaxing education. liquidity constraints, previous inheritances increase the probability of being self-employed. As additional evidence Profitability and exit from self-employment they report survey data that reveal that most small busi- Holtz-Eakin, Joulfaian, and Rosen (1994a) use net income nesses are started not with bank loans but with one's own as a proxy for entrepreneurial ability and other relevant or family money, and that shortages of capital are the factors, such as education, time in business, and gender, most common reason for not entering self-employment. not available in their data set. They find that net income Bates (1990) finds that small businesses that were able to has a negative effect on the probability of exiting self- secure loans at the time of getting started are more likely employment. to survive. Previous wages and entry into self-employment Both exit rates and employment growth rates of manu- Evans and Leighton (1989) find that higher previous facturing firms decrease with size and time in business. wages lower the likelihood of entering self-employment. Evans (1987a, 1987b) and Dunne, Roberts, and Samuelson However, the statistical significance of that variable is (1988, 1989) find that size and time in business are both reduced considerably when individuals' labor market related negatively to the exit and growth rate of U.S. man- characteristics are controlled for. ufacturing plants. 137 I N F O R M A L I T Y his/her spouse with the salaries of other family members although often overlooked, are numerous and become more could reduce overall household income risk. important as firms get larger. Consider a select few (and see de Soto 1989; and chapter 6 of this volume): Determinants for small firm informality The high levels of informality found among microenter- · Enforceable/impersonal contracts and credible signaling. All prises in developing economies can be interpreted as result- entrepreneurs have access to social relationships to ing from profit-maximizing decisions by private firms subject enforce implicit contracts among their friends and to environmental constraints that increase the costs or reduce family, who form a small number of their potential the benefits of formality--for example, high costs of firm or customers and employees. Participation in the legal property registration, or market and government failures in system is needlessly expensive for transactions with credit markets, and contract enforcement mechanisms. A these individuals. But this mode of operation is vicious cycle would be at work, by which barriers to small constrained by the ability of the entrepreneur to firms' participation in formal market support institutions maintain personal relations with all involved parties, would limit firm growth, which in turn would help main- a task increasingly unmanageable as firms expand. tain those institutions out of the reach of most small firms. Legally recognized, enforceable contracts lend credi- However, a somewhat nuanced alternative view is that bility to arrangements, permit entry into long-term small firms in developing countries have an inherently lim- commitments, diminish risk, and can reduce moni- ited demand for those formal services, due to their low pro- toring costs. Larger investments require that prop- ductivity levels and limited growth prospects. In this erty rights be secured through the legal system. context, a firm's choices regarding the use of formal or · Access to capital. Informal capital markets (Besley informal arrangements would depend on the stage of the 1995) may be sufficient to fulfill the firm's external firm's life cycle. financing needs at low levels of production. However, This view is related to the burgeoning literature on the small scale and undiversified nature of informal social capital and informal networks, which has highlighted capital markets make them unsuitable for satisfying that absence of participation in state-organized or formal the firm's financing needs at larger scales of opera- market institutions does not imply that agents do not tion; growing firms will turn to formal financial acquire similar services through less formal means and, to a intermediaries such as banks. This, of course, implies lesser degree, that informal alternatives are necessarily infe- costs in terms of better record keeping, and transac- rior to formal ones. In this report, we argue that the calcu- tions costs more generally. lations involved in participation decisions apply to all · Access to public risk pooling mechanisms. Demand for risk institutions, both informal and formal. We thus recast the pooling mechanisms can be expected to increase with question of formality as the firm's decision of how much to firm size. Arnott and Stiglitz (1991) show that where participate in the numerous institutions of civil society: "peer monitoring" generates more information than civic organizations, trade organizations, federal and local that available to market insurers, it is an important treasuries, governmental programs such as social security mechanism in reducing moral hazard and makes non- (including pensions and health care), the legal system, the market (informal) insurance provided by friends or banking system, insurance institutions, health inspections, communities an important complement to market firm censuses, and so on. We argue that a minimal degree insurance. Small firms with relatively little capital of participation in some institutions is a necessary input to may insure against small losses within the family or growth for many firms, and that participation increases community rather than pay the premium associated with the success of the business; it is a normal input into with monitoring pooling across larger groups. As a the production process (see Levenson and Maloney 1998). firm grows, we may imagine that both the magni- Firms derive multiple benefits from formality. We can tude of potential losses and the decreasing ability to view being "formal" as lying at the end of a continuum of peer monitor may lead to increased demand for for- possibilities of participation beginning at the household mal market insurance. and extending through communities and networks to the · Access to business information. As a firm grows beyond formal institutions of the state. The benefits of formality, local customers and suppliers, it may also require 138 M I C R O F I R M D Y N A M I C S A N D I N F O R M A L I T Y more information and business contacts to continue general public and other firms should increase, leading to growing. We may, therefore, expect more participa- greater benefits from formality. tion in business associations, among other changes. A natural corollary is that "inefficient" firms are dispropor- We can also consider inscription in training pro- tionately informal. However, in contrast to other formulations, grams and the use of other business development ser- in this case the causality is not necessarily from informality to vices as formal channels for information gathering. inefficiency: high-cost firms choose less formality because they benefit less from it than more efficient firms that produce at Even though formality also has multiple costs--for higher volumes for longer lengths of time. Similarly, young example, reporting requirements and fiscal obligations--as firms are disproportionately informal. This is partly because young firms grow, the relative benefits of formality may eventu- firms are more likely to be small. Moreover, conditional on ally exceed the corresponding costs. De Soto claims that size, the population of young firms contains a disproportionate Peruvian sidewalk vendors sought, not to avoid, but to pay number that have not received enough signals to figure out taxes as a way to establish property rights over their precar- whether paying the costs of formality are worthwhile; many ious business locations. In reality, although the direct eventually will go out of business. private benefit from paying taxes may be zero (again, Another implication is that the relatively high mortality assuming no enforcement penalties), there may be ancillary rates found among informal firms are not necessarily evi- benefits that make compliance worthwhile. dence of the inferiority of informal employment, and could rather reflect the fact that those firms are predominantly Implications for microfirm dynamics "young" and small which, even in an undistorted industrial- and informality country context, would lead to observed high turnover The combination of the above very stylized concept of par- rates. As more efficient firms grow to their equilibrium ticipation in formal institutions and the insights from size, however, both mortality and informality rates would mainstream models of firm dynamics yields a number of naturally decline. implications. First, it is reasonable to expect considerable In sum, combining the insights from the firm dynamics heterogeneity in the degree of formality. The benefits and costs of literature with those derived from the literature on institu- participation undoubtedly vary across societal institutions, tional development provides a new possible interpretation and vary for firms of a different size and age. While there of the stylized facts on microenterprises and informality. are potential complementarities between different societal Because they have not assembled enough information on institutions, a large number of firms will choose to partici- their true efficiency, young firms tend to be small and to pate in only a subset of institutions at any point in time. have high failure rates. On the other hand, formality is not For example, the legal system and bank financing are com- an all-or-nothing decision, but rather an option in a con- plements, but a firm may have to register legally before tinuum of participation possibilities. For young and small seeking external financing. Thus, informality is not an all- firms, the benefits outweigh the costs of participation in or-nothing state, and the degree of formality varies by firm. formal societal institutions. But as more efficient firms sur- A second implication is that small firms are dispropor- vive and grow, their needs for enforceable contracts, formal tionately informal. Small firms benefit least from partici- credit markets, and access to public risk-pooling mecha- pation because of the small scope of their dealings with nisms increase, and so does their degree of formality or the public and hired employees (relative to the total vol- depth of participation in societal institutions. ume of transactions undertaken by the firm). Indeed, Young and small firms do tend to have higher costs and firms of different sizes may have different degrees of high failure rates and, at the same time, they are more likely interaction with the public. Because implicit contracts to be informal. Although this description corresponds over product quality are cheaper and feasible to enforce exactly to the standard picture of the stagnant, precarious, with friends and family, the entrepreneur may find it unprotected informal self-employment sector familiar in the most cost-effective to primarily serve such customers literature, it is, in fact, the opposite. It emerges naturally when faced with small sales volumes. At larger volumes from workers trying their luck at entrepreneurship (risk (later in the firm's life cycle), friends and family cannot taking), often failing, and not engaging in the formal insti- necessarily buy all the firm's output, so sales to the tutions until they grow. In sum, there may be nothing 139 I N F O R M A L I T Y TABLE 5.1 Entry probabilities into Mexico's self-employment sector (percent) Initial/final status Own account 1­4 workers 5­9 workers 10 or more workers Total entrepreneurs Salaried 4.2 1.7 0.2 0.2 6.2 Contract 9.0 3.1 0.1 0.1 12.4 Salaried + contract 4.7 1.8 0.2 0.2 6.9 Unemployment 9.3 1.5 0.1 0.1 11.0 Other work 4.6 3.4 0.2 0.1 8.3 Out of labor force 2.6 0.8 0 0 3.5 Source: Author's calculations, using data on males from the Encuesta Nacional de Empleo Urbano 1998­2001. pathological about the informal microfirm sector, and its they were either "employers" ("patrones") or "own-account existence may be largely unrelated to questions of labor workers" ("trabajador por su cuenta"). This is similar to the market dualism or even credit market distortions. standard definition used in the U.S. self-employment liter- ature. Evans and Leighton (1989), for instance, define as Microfirm dynamics in Latin America self-employed all sole proprietors, partners, and sole owners Collecting evidence on the entry, exit, and growth dynam- of incorporated businesses. ics of Latin American and the Caribbean microfirms is of interest for at least three reasons. First, we would like to Patterns of entry know if microfirm dynamics in the Latin America and Taking the United States as a benchmark, the rates of entry Caribbean (LAC) region are similar to those in the indus- into self-employment in Mexico are of very similar orders trial countries. Second, the very high mortality rates of magnitude, with a slightly higher fraction of Mexican among LDC firms contribute to the perception of the sector wage workers becoming entrepreneurs: 6.2 percent com- as "precarious." If in fact the failure rates are similar to pared to 4 percent per year found by Evans and Leighton those found elsewhere, it is difficult to make the case that (1989).5 Table 5.1 presents the probabilities of becoming we are dealing with unusually high rates of risk. Third, the an enterprise owner for Mexican males previously demographic characteristics associated with entry and exit employed in other segments of the labor market or located provide useful information about the role the sector is play- outside of the labor force. Rows represent individuals' ini- ing and the constraints facing it. tial labor market positions, and columns represent labor Studying the patterns of entry, survival, and growth market statuses a year later. Each cell shows the percentage in the microenterprise sector requires the use of panel data of individuals who start in a given row-category and end in that allow following the same individuals and firms over the corresponding column-category. time. This creates a severe constraint, as most longitudinal New entrants into self-employment are more likely to industrial surveys exclude small and informal firms. For start their businesses without any employees. Table 5.1 that reason, this section focuses on only a few countries for shows that about two-thirds of the salaried workers which we were able to obtain appropriate data. Our main who transit into self-employment each year--4.2 out of focus is on Mexico, using combined data from employment 6.2 percent--do so without hiring any workers, which is and microenterprise surveys.3 Some of the analysis is repli- consistent with the mainstream firm dynamics literature cated using a longitudinal household survey for Nicaragua. view of new entrepreneurs "testing the waters" before mak- Finally, we present evidence on the job creation potential ing significant investment decisions. Table 5.1 also shows of firms of different sizes using the Colombian industrial that individuals who are outside the labor force have a lower survey, for a period during which its coverage included probability of entering self-employment--3.5 percent--and micro- and small firms.4 In what follows, we define the self- that those coming from contract work or unemployment employed category as including all individuals whose main have higher probabilities of entering the sector--respec- job consists in working in their own businesses--for exam- tively, 12.4 and 11 percent. In all these cases, however, three ple, those individuals who report that in their main job out of four new entrants start without hiring any employees. 140 M I C R O F I R M D Y N A M I C S A N D I N F O R M A L I T Y FIGURE 5.1 Entry and exit into self-employment among men Mexico United States Percent Percent 60 60 50 50 40 40 30 30 20 20 10 10 0 0 5 5 ­6 ­6 18­20 21­25 26­30 31­35 36­40 41­45 46­50 51­55 56­60 61 18­20 21­25 26­30 31­35 36­40 41­45 46­50 51­55 56­60 61 Age Age Entry rate Exit rate Self-employment rate Sources: Evans and Leighton 1989; author's calculations, using Encuesta Nacional de Empleo Urbano. There are also striking similarities in age patterns of self- cases of Mexico and Nicaragua.7 Like men, Mexican employment between Mexico and Nicaragua, on one hand, women's entry rates rise until the early 40s and begin to and the United States on the other. The left and right panels decline after the age of 50. Their exit rates are much higher of figure 5.1 show the fraction of the male labor force in self- and decline at a slower rate until the late 30s, stabilizing at employment by age group, respectively, for Mexico and the around 50 percent (for yearly transitions). The overall rate United States. The figure also exhibits the rates of entry into of self-employment is thus much smaller than that of that sector from salaried work and the rates of exit from self- men, although it also peaks at the age of 50. However, employment, also by age group. The self-employment rate among working individuals, the share of self-employed (calculated as a fraction of the labor force) is higher in women increases monotonically with age at an even steeper Mexico for every age group by at least a factor of two. Note, pace than that of men; it evolves from less than 5 percent however, that the overall age patterns of self-employment in the early 20s (10 percent in Nicaragua) to more than rates as well as those of entry and exit are strikingly similar 50 percent after the age of 55 (70 percent in Nicaragua). in the two countries.6 Econometric examination of the effect of individual Both in Mexico and the United States, exit rates demographic and labor market characteristics on the proba- decrease while entry rates slowly increase with age. As a bility of entering self-employment among Mexican males result, the percentage of individuals who are self-employed confirms the age patterns illustrated in the figures above.8 increases with age at a decreasing rate, peaking at the late Thus, keeping other personal characteristics constant, and 40s. As a foreshadowing of the more detailed analyses comparing with individuals aged 15 to 20 (for which the below, the common upward and downward sloping rela- rate of entry into self-employment is 2.4 percent), the prob- tionships of age with the rates of, respectively, entry and ability of entering self-employment is 5.7 percentage points exit are consistent with the view that older entrepreneurs higher for those aged 21 to 35, and 9 percentage get a more precise view of their underlying entrepreneurial points higher for those in the 36­50 age bracket. These capacity and hence are more likely to enter and less likely results are not consistent with the view of the sector as a to fail than younger individuals. point of entry into the Mexican labor market, but they are Women exhibit similar patterns of entry and exit into very consistent with the U.S. data. They also provide sup- self-employment, which is illustrated in figure 5.2 for the port to Evans and Jovanovic's (1989) liquidity constraints 141 I N F O R M A L I T Y FIGURE 5.2 Entry and exit into self-employment among women Mexico Nicaragua Percent Percent 80 80 70 70 60 60 50 50 40 40 30 30 20 20 10 10 0 0 50 55 18­20 21­25 26­30 31­35 36­40 41­45 46­ 51­ 56­60 61­65 12­20 21­30 31­40 41­50 50­65 Age Age Entry rate Exit rate SE ALL SE SE SAL Source: Author's calculations, using Encuesta Nacional de Empleo Urbano and Nicaragua's Living Standards Measurement Study. hypothesis, and the view that older workers have a more self-employment, and that probability increases by about precise measure of their underlying entrepreneurial capacity. 25 percent for their married counterparts. Moreover, when Educational attainment, on the other hand, has a negative the estimation is performed conditioning on individuals albeit quantitatively small effect on the probability of enter- changing their labor market status, the effect of marriage is ing self-employment. However, when we focus only on about four times larger. workers entering self-employment from salaried work and/or The importance of conditioning on changes in labor when we estimate the effect of education conditioning on market positions becomes clearest in looking at the impact individuals' changing labor market status, the negative rela- of initial employment and labor force participation status. tionship between education and entry into self-employment The unconditional results suggest that those out of the levels off and breaks up at higher levels of schooling, sug- labor force and not studying and those unemployed are gesting that college graduates (that is, doctors, consultants, more likely to enter self-employment, consistent with tra- and so on) may find the sector attractive. Moreover, when we ditional dualistic views of the sector as disguised unem- restrict the definition of self-employment to include only ployment. However, all of the above segments of the labor those business owners who employ at least one worker, we market are characterized by very high rates of turnover, so find that both secondary schooling and tertiary schooling are the corresponding workers may well be more likely to positively linked to entry into self-employment. This result move to all sectors at disproportionately high rates. When is broadly consistent with Carrasco (1999), Evans and we condition on changing labor market status, we find that Leighton (1989), and Rees and Shah (1986) who found that those in formal salaried employment are more likely than entry rises monotonically with education. most other groups to enter informal self-employment. Also consistent with Rees and Shah (1986) and Carrasco Among the unemployed, however, those workers with (1999) is the finding that being married is positively asso- longer unemployment spells--especially those who have ciated with entry, which may reflect either that the sector is been unemployed for between four and six months--have not riskier than other alternatives available in the labor higher conditional and unconditional probabilities of enter- market (that is, salaried employment) or that, in fact, being ing self-employment through opening owner-only busi- married helps diversify risk. Quantitatively, unmarried nesses. This suggests that even if the sector does not individuals have a 4.5 percent probability of entry into function predominantly as a holding pattern for misfits or 142 M I C R O F I R M D Y N A M I C S A N D I N F O R M A L I T Y those rationed out of salaried work, it does offer income explanation is that the sector offers more opportunities for opportunities for the long-term unemployed. very small firms, as it comprises mostly individual con- As for the links between entry into self-employment and tractors operating without any employees; in the U.S. the previous job characteristics reported by salaried workers, construction sector, for instance, owner-only businesses we find that those employed in firms with at most represented 75 percent of all firms, while those with 10 employees have a much higher probability of entry into 20-plus workers were just 2 percent, compared to, respec- self-employment than those coming from larger firms: tively, 49 and 13 percent in the manufacturing sector. As 12.8 percent compared to about 9 percent for those in firms argued by Tybout (2000) and further discussed in the with 11 to 250 workers, and 5 percent for those coming next chapter, small firms tend to locate in industries from firms with more than 250 employees. This may be where they have smaller cost disadvantages with respect due to the higher nonpecuniary benefits (such as social to larger incumbents, and which are characterized by security) and job stability offered by larger firms. lower levels of capital and skilled-labor intensity. For a given firm size, on the other hand, higher prior wages also increase the probability of entry into self- employment. While we cannot tell whether this is the Patterns of survival and exit effect of a relaxation of liquidity constraints or if it reflects As shown in table 5.2, during any given year about 15 per- entrepreneurial ability, the fact that overperformers in the cent of the Mexican self-employed move to salaried work-- salaried sector are more likely to enter self-employment is and the same number of individuals migrate to other labor at odds with the predictions of both the dualistic approach market positions, namely, to contract work (6.2 percent) and the sociological view of that sector as a preferred desti- and out of the labor force (5.7 percent). The probability of nation for "misfits." leaving self-employment is higher for those who do not As for the effect of the sectors of economic activity in have employees: 35 percent compared to slightly less than which individuals have been previously engaged, we find 25 percent for those who own larger firms. Moreover, that at least in Mexico, entry into self-employment is among those leaving self-employment, the probability of most likely for those who hold jobs in the construction switching to salaried work is lower for those who own sector and it is least likely for those employed in smaller firms, when compared with owners of larger firms manufacturing. Thus, while a salaried worker in manufac- who are less likely to switch to contract work or out of the turing has a 4.3 percent probability of entering self- labor force. employment, the corresponding probability for a salaried Econometric estimates of the determinants of firm sur- construction worker with similar personal characteristics vival using data on Mexican microfirms confirm the age is 11.8 percent. The higher rate of entry into self-employ- patterns depicted in figures 5.1 and 5.2, with the probabil- ment coming from salaried work in the construction ity of staying in business increasing with age until the sector persists even when one controls for the fact that 36­50 age bracket. Thus, while the average self-employed this sector also has higher turnover rates--workers there individual aged 15­20 has only a 33.3 percent probability are more likely to change labor market status. One possible of staying in self-employment, individuals with similar TABLE 5.2 Exit probabilities from Mexico's self-employment sector (percent) Remain Salaried Other Out of Initial/final status entrepreneur Salaried Contract contract Unemployed work labor force Total Own account 65.0 17.0 7.4 24.5 1.5 1.9 7.2 100 1­4 workers 76.6 11.7 5.0 16.7 0.8 2.0 3.9 100 4­9 workers 78.6 13.5 3.2 16.8 0.6 1.3 2.7 100 10 or more workers 78.3 15.2 2.3 17.5 0.3 1.2 2.7 100 Total entrepreneurs 69.9 15.0 6.3 21.3 1.2 1.9 5.7 100 Source: Author's calculations, using data on males from the Encuesta Nacional de Empleo Urbano 1987­2001. 143 I N F O R M A L I T Y personal characteristics in the 21­35 and 36­50 age brack- a 4.3 percentage point increase in its likelihood of survival, ets have probabilities of survival in self-employment of, while a tripling of its capital stock is related to a 1.4 per- respectively, 52 and 60.5 percent. centage point increase in that probability. Despite the Education has a negative impact on the probability of small magnitude of the effects, these results are consistent staying in self-employment, suggesting that the "pull" effect both with Jovanovic's "noisy selection" view and with the of better employment alternatives for more educated indi- bulk of the mainstream empirical evidence that firms get a viduals tends to dominate over the "push" effect associated more precise estimate of their cost structures with experi- with lower probabilities of business failure observed in the ence, and past measures of success are informative for the United States among more educated entrepreneurs. How- future evolution of the firm. ever, the effect is quantitatively small--a reduction of about 2 percentage points in the survival probability for Do microfirms grow? individuals with secondary or tertiary schooling--and it All the available evidence points to the fact that microfirms becomes positive and significant when the sample is create relatively more jobs, but destroy even more. This is restricted to firms with at least one worker. In other words, illustrated by figure 5.3, which shows rates of job creation when the focus is on larger firms, the evidence for Mexico and destruction among Colombian manufacturing plants confirms that for the United States, with schooling and firm calculated for different employment size categories--the survival exhibiting a positive correlation. rates are calculated as a percentage of employment in the Consistent with the various mainstream models of firm respective size category. Firms with fewer than 20 employees dynamics reviewed above, we find that higher conditional have the highest rates of job creation, but their job destruc- wages of business owners--which we interpret as reflecting tion rates are proportionally even higher, causing net job higher profitability due to unobserved characteristics-- creation rates to be the lowest of any size category. As firm tend to increase the probability of staying in business. size increases, both job creation and destruction tend to Thus, a doubling of the net earnings from self-employment diminish, but the latter at a larger rate, causing net job cre- is associated with a 4.2 percentage point increase in the ation to be higher among larger firms. This is consistent probability of staying in that sector. The effect is quantita- with the findings for the United States, where smaller man- tively larger when the sample is restricted to businesses ufacturing firms and plants exhibit very high gross job cre- with at least one employee besides the owner, for which a ation rates, but not higher net job creation rates (Davis, doubling of self-employment earnings is related to an Haltiwanger, and Schuh 1996). 8.8 percentage point increase in the probability of main- taining that initial status. FIGURE 5.3 We also find that survival in self-employment is between Job creation and destruction in Colombia, 1977­82 5 and 8 percentage points higher for married individuals-- % of employment possibly because they can count on unpaid family workers-- 0.3 and about 6 percentage points lower for individuals with a 0.2 second job. Moreover, workers in the construction sector 0.1 have a probability of staying in self-employment that is between 10 and 15 percentage points lower than for those 0 engaged in agriculture, manufacturing, and services, while 0.1 those engaged in commercial activities have a 5.5 percentage 0.2 point higher survival probability than their peers in the 0.3 above three sectors. The fact that once again the construction 0.4 sector emerges as a more likely source of workers who switch 0 20 50 100 250 Total sectors could be due to its inherent high rate of firm turnover. Number of employees As for the effects of firm size and age on survival in self- Job creation Job destruction Net employment, we find that they are significant although quantitatively small. In particular, a doubling of the time Source: Author's calculations, using the Colombian Manufacturing Survey. that a microfirm has been in business is associated with 144 M I C R O F I R M D Y N A M I C S A N D I N F O R M A L I T Y Data on the evolution of Mexican microfirms over a FIGURE 5.4 12-month period suggest a broadly similar profile (table 5.3). Job creation and destruction by entering and exiting versus continuing firms in Colombia, 1977­82 Over 12 percent of owner-only firms expand to one to four employees across one-year periods. However, 22.1 percent 0.20 of those between one and four employees contract again to 0.15 be owner-only. Thus, in absolute numbers, there are slightly 0.10 less firms that contract from one to four employees than 0.05 firms that expand into that range. Moreover, over a one- 0 year period, only 5.4 percent of the firms with one to four 0.05 employees do actually expand; 49.2 percent stay in the 0.10 same size range and 45.5 percent contract to owner-only or 0.15 go out of business, with the employer moving into wage 0.20 work or unemployment, or leaving the labor force. 0 20 50 100 250 Total Number of employees Given the low rates of employment growth observed among microfirms, it is not surprising that the vast major- Job creation Job destruction Net ity of them will reach their steady state at a very small size. Source: Author's calculations, using the Colombian Thus, among Mexican and Nicaraguan microfirms with at Manufacturing Survey. least three years of time in business and at most 15 workers, 64 percent have no employees, 21 percent have only one, and between 8 and 9 percent have just two workers (fig- Gross and net job creation rates by entering Colombian ure 5.6). In other words, in those two cases, 93 percent of firms are higher than for continuing firms (figure 5.4). As a microfirms do not grow beyond two employees. In fact, the share of the size category, roughly four times as many jobs large majority of microenterprises are owner-only. In Brazil are created by entering firms as by those already existing. and Mexico, for example, respectively, 87 and 80 percent of Job destruction by exiting firms is roughly equal to that by all microenterprises with fewer than five workers have no continuing firms, so that net job creation by entering and paid employees. Moreover, as illustrated in table 5.4, in exiting firms is positive, while it is negative among contin- Argentina, Brazil, and Mexico owner-only firms represent uing firms. Thus, although entering firms more than com- roughly 58­68 percent of all firms. These proportions are pensate for the jobs destroyed by the plants that fail, the smaller than those found in the United States and the Colombian data suggest that the low rate of expansion of United Kingdom, where, respectively, 71.6 and 80 percent smaller continuing firms leads them to provide a negative of all firms have no employees. It is worth noting that the net contribution to employment growth. lower fraction of firms without employees found for LAC is Most of the net job creation among microfirms is from not inconsistent with the fact that, as reported in previous zero to one employee. In the case of Nicaragua, for instance, chapters, those countries exhibit higher rates of self- looking at firms with at most three employees (plus owner) employment. Indeed, the latter are calculated as a propor- in 1998, we find that for those that were still in business by tion of the labor force, while the figures reported in 2001, employment grew by an average of about 20 percent. table 5.4 are based on the size distribution of the number of For those with two and three employees, the losses actually firms--not employment. exceeded the gains, so all the above growth came from Among firms with at least one employee, however, the firms with at most one employee (figure 5.5). In fact, most fraction of microenterprises with between one and four was due to those with no employees at all, which on aver- workers is much larger in the LAC countries for which data age expanded by 40 percent, despite the fact that 73 per- are available: about 90 percent in Argentina and Mexico cent did not grow at all. Firms with one employee at the and almost 80 percent in Brazil, compared to, respectively, beginning of the three-year period also had a positive con- 55 and 65 percent in the United States and the United tribution to employment growth; although 40 percent of Kingdom (table 5.5). This is consistent with the evidence those lost their only worker, this was offset by hires by presented by Bartelsman, Haltiwanger, and Scarpetta other firms in the category. (2004) on the fact that in all countries for which data 145 FIGURE 5.5 Employment growth of microenterprises in Nicaragua Self-employed One employee Frequency (%) Average change in employment Frequency (%) Average change in employment 100 1.0 100 1.0 0.9 0.9 80 0.8 80 0.8 0.7 0.7 60 0.6 60 0.6 0.5 0.5 40 0.4 40 0.4 0.3 0.3 20 0.2 20 0.2 0.1 0.1 0 0 0 0 3 2 1 No 1 2 3 4 5 6 7 Net 3 2 1 No 1 2 3 4 5 6 7 Net change change Two employees Three employees Frequency (%) Average change in employment Frequency (%) Average change in employment 100 1.0 100 1.0 80 0.8 80 0.8 60 0.6 60 0.6 40 0.4 40 0.4 20 0.2 20 0.2 0 0 0 0 20 0.2 20 0.2 40 0.4 40 0.4 60 0.6 60 0.6 80 0.8 80 0.8 100 1.0 100 1.0 3 2 1 No 1 2 3 4 5 6 7 Net 3 2 1 No 1 2 3 4 5 6 7 Net change change Source: Author's calculations, using Nicaragua's Living Standards Measurement Study. TABLE 5.3 Growth and contraction probabilities of Mexican microfirms (percent) Own 10 or more Initial/final status account 1­4 workers 5­9 workers workers Exit Total Own account 51.9 12.4 0.5 0.2 35.00 100 1­4 workers 22.1 49.2 3.9 1.5 23.40 100 4­9 workers 7.8 35.1 22.6 13.1 21.40 100 10 or more workers 4.1 15.2 14.4 44.6 21.70 100 Total entrepreneurs 38.4 26.0 2.9 2.6 30.10 100 Source: Author's calculations, using data on males from the Encuesta Nacional de Empleo Urbano 1987­2001. are available--they consider 19 countries, including 9 much more prevalent in developing than in industrial industrial ones and, from LAC, Argentina, Brazil, and countries. Not surprisingly, Bartelsman et al. (2004) find Mexico--small firms with fewer than 20 employees that the average size of firms is much higher in the account for between 80 and 96 percent of the total firm United States than in the above three LAC countries, population. However, it appears that within the group of which they show is not related to differences in sector firms with fewer than 20 workers, microenterprises are composition--for example, some countries specializing 146 M I C R O F I R M D Y N A M I C S A N D I N F O R M A L I T Y FIGURE 5.6 Size distribution of microenterprises in Mexico and Nicaragua Mexico Nicaragua Percentage of firms Percentage of firms 80 80 60 60 40 40 20 20 0 0 1 2 3 4 5 6 7 8 9 10 11 1 2 3 4 5 6 7 8 9 10 11 Number of employees, including owner Number of employees, including owner Source: Author's calculations, using Encuesta Nacional de Empleo Urbano and Nicaragua's Living Standards Measurement Study. Note: All firms were at least three years old at the time of measurement. Firms pooled 1992­98. in industries with a smaller efficient scale--but rather to is something that we investigate next by exploring the per- within sector differences in average firm size. sonal and firm traits that are associated with firm growth. As argued above, a high prevalence of small firms does To examine what characteristics seem correlated with not necessarily imply the presence of high external barriers microfirm employment growth, we use data on the transition to firm growth. As noted by Lucas (1978, pp. 508­23), the between size-brackets of Mexican microfirms over a one-year size distribution of firms may be determined by the under- period considering only surviving firms.9 Consistent with lying distribution of entrepreneurial ability or the implicit mainstream models, entrepreneurs with conditionally higher costs of operation at different sizes. That is, urban microen- earnings (better performance) seem to show a higher propen- terprises could be optimally small, given the owners' entre- sity to grow. Similar results are obtained for Nicaragua, for preneurial abilities, and the costs of conforming to the both male and female entrepreneurs; those with higher profits requirements of formal contracting arrangements, partici- relative to their peers with similar levels of human and physi- pation in credit markets, and such. Thus, the prevalence of cal capital are more likely to add workers. small firms could simply reflect small opportunity costs of Firm size also appears with the negative sign predicted entry into self-employment. Whether this mainstream view by the mainstream literature: bigger firms are more likely to of the sector is in fact relevant in the case of Latin America have achieved their optimal size. Time in business also has a TABLE 5.4 Employment size distribution of firms, selected countries (percent) Number of employees US (2004) UK (2005) Mexico (2004)a Brazil (2003)b Argentina (2003)a With no employees 80.0 71.6 64.4 57.7 68.4 1­4 10.9 18.6 33.0 33.3 28.8 5­19 6.6 7.6 2.0 4.9 2.0 20 or more 2.5 2.1 0.7 4.1 0.8 Source: Author's calculations, using data from U.S. Small Business Administration, U.K. Small Business Service Analytical Unit, Argentina's Encuesta Permanente de Hogares, Brazil's Pesquisa Nacional por Amostra de Domicílios, and Mexico's Encuesta Nacional de Empleo Urbano. a. Third and fourth size ranges: 5­14 workers and 15+ workers, respectively. b. Third and fourth size ranges: 5­10 workers and 11+ workers, respectively. 147 I N F O R M A L I T Y TABLE 5.5 Employment size distribution of firms with employees, selected countries (percent) Number of employees US (2004) UK (2005) Mexico (2004)a Brazil (2003)b Argentina (2003)a 1­4 54.5 65.7 92.4 78.7 91.1 5­19 33.0 26.9 5.6 11.6 6.3 20 or more 12.5 7.4 2.0 9.7 2.5 Source: Author's calculations, using data from U.S. Small Business Administration, U.K. Small Business Service Analytical Unit, Argentina's Encuesta Permanente de Hogares, Brazil's Pesquisa Nacional por Amostra de Domicílios, and Mexico's Encuesta Nacional de Empleo Urbano. a. Third and fourth size ranges: 5­14 workers and 15 workers, respectively. b. Third and fourth size ranges: 5­10 workers and 11 workers, respectively. negative effect, as expected in the context of the literature many microentrepreneurs choose to be there. In practice, one on developed countries, but its coefficient is not statistically is likely to find some evidence for both views, suggesting significant. Similarly, consistent with Ericson and Pakes that the self-employment sector is quite heterogeneous, (1992b, pp. 53­82)--although less so with Evans and with a relatively well-off entrepreneurial group coexisting Jovanovic (1989, pp. 808­27)--we find that higher capital with those involuntarily informal, respectively, the "upper stocks are positively correlated with employment growth. tiers" and "lower tiers" of the sector, in Gary Fields's (1990) As for the relationships between firm growth and the formulation. However, as chapter 2 suggests, at least in the age and marital status of entrepreneurs, we find that they case of Latin America, the existing evidence suggests that are broadly similar to those obtained in the survival microentrepreneurs who entered the sector voluntarily are analysis: business owners who are married and aged most likely the majority. 36­50 are most likely to expand their firms, exhibiting Not surprisingly, formality rates are higher among better- employment growth rates that are, respectively, 11.1 and performing microentrepreneurs, and for those who entered 9.4 percentage points higher than those of their peers.10 self-employment voluntarily--who arguably constitute the Interestingly, employment growth for female entrepre- upper tier of the sector. Thus, as seen in the left panel of neurs increases with their number of children, suggesting table 5.6, among formal Brazilian microfirm owners the that, as has been noted in the literature, most employees fraction that was unemployed prior to entering the sector is of microfirms are related to the owner. This is also consis- 21 percent, compared to 32 percent for nonlicensed firm tent with the finding that a key determinant of female owners. Similarly, licensed businesses are more likely to children entering the labor force is whether their mother show signs of success, such as having plans to expand-- had a microenterprise (Cunningham and Maloney 2001). 46 percent among licensed firms compared to 37 percent It may also explain why the rising probability of hiring for nonlicensed businesses--and their owners are less peaks for both genders around the period when their likely to plan on going back to salaried jobs--6.5 versus children are old enough to contribute. 13.4 percent, respectively. It must be noted, however, that the positive correlation between good firm performance Informality among microfirms and formality could be driven by causal effects in either In the literature that associates informal microenterprises direction--better-performing firms from the upper tier of with the disadvantaged sector of labor markets segmented the microenterprise sector being more likely to be formal, by government- or union-induced rigidities, workers enter or formality causing improvements in performance. We self-employment involuntarily whereas they queue up for will examine this issue in more detail in the next chapter. salaried jobs. Informality, in this approach, is not a choice, That informality is higher among microentrepreneurs in but rather the option of last resort for otherwise- the so-called upper tier of the sector is also illustrated by unemployed workers. In contrast, the parallel tradition that the fact that, as seen in table 5.7, older business owners--at views self-employment through the lens of the firm stresses least until age 47 in Mexico and 45 in Brazil--are less the entrepreneurial dynamism of the sector and the fact that likely to be informal, and so are those with higher levels of 148 M I C R O F I R M D Y N A M I C S A N D I N F O R M A L I T Y TABLE 5.6 Reasons for starting up, firm prospects, and firm licensing in Brazil Main reason to Firms with Firms without Firms with Firms without start a microfirm license (%) license (%) Plans for future license (%) license (%) Didn't find a job 20.9 32.2 Expand 45.5 36.6 Profitable business 2.2 1.2 Same level 31.2 31.2 Flexible hours 1.6 2.3 Change activity, 9.2 9.5 remain independent Be independent 27.8 17.1 Find a salaried job 6.5 13.4 Family tradition 11.0 8.1 Don't know 7.6 9.3 To help family income 12.2 20.8 Difficulties to % firms (with % firms (with regularize when licenses) licenses) starting up (2003) Accumulated experience 10.7 8.7 Yes 18 5.1 Make good deal 10.7 7.6 No 57.4 10.4 As a secondary job 2.5 2.1 Didn't try 24.7 84.5 Sources: Pooled Pesquisa Economia Informal Urbana 1997 and 2003 (except for "Difficulties to Regularize" from 2003). Note: Sample restricted to entrepreneurs aged at least 20. TABLE 5.7 Informality among Mexican and Brazilian microenterprises by age, education, and previous activity of the owner Mexico Brazil (1) (2) (3) (4) (5) (6) Variable No license No taxes No social security No license No taxes No social security Owner's age 0.019*** 0.014*** 0.012*** 0.009*** 0.001** 0.0001 (0.000) (0.000) (0.008) (0.000) (0.034) (0.979) Owner's age squared 0.0002*** 0.0001*** 0.0001 0.0001*** 0.0000** 0.000 (0.000) (0.000) (0.103) (0.000) (0.042) (0.637) Primary schooling 0.186*** 0.102*** 0.089** 0.048*** 0.008*** 0.0415*** (0.000) (0.000) (0.045) (0.000) (0.001) (0.005) Secondary schooling 0.365*** 0.218*** 0.205*** 0.107*** 0.022*** 0.170*** (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) Tertiary schooling 0.523*** 0.357*** 0.390*** 0.247*** 0.039*** 0.3379*** (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) Female 0.075*** 0.062*** 0.008 0.095*** 0.009*** 0.011 (0.000) (0.000) (0.647) (0.000) (0.000) (0.225) Previously unemployed 0.035*** 0.034*** 0.068*** 0.039*** 0.014*** 0.074*** (0.001) (0.000) (0.005) (0.000) (0.000) (0.000) Observations (n) 28,525 28,525 4,599 45,509 45,509 8,841 Pseudo-R2 0.189 0.159 0.215 0.220 0.159 0.220 Source: Author's calculations, using Pesquisa Economia Informal Urbana and Encuesta Nacional de Micronegocios. Note: Robust p-values appear in parentheses. Firm size, time in business, fixed location, sales to large firms, sector, and state are also controlled. Samples of business owners aged 21 and older, without complete tertiary education, operating agricultural activities, and with at most five workers. **p < .05; ***p < .01. 149 I N F O R M A L I T Y schooling. Thus, for instance, the probability of a micro- FIGURE 5.7 enterprise not being registered is 37 percentage points Informality among Mexican microenterprises by number of paid lower for Mexican businesses owned by individuals with at workers and time in business least some secondary schooling, compared to those who a. Number of paid workers have no schooling at all. The likelihood of not paying taxes 1.0 or social security contributions also diminishes signifi- cantly with the education of the business owner, a result 0.8 that is also obtained for Brazil. Moreover, after controlling for the effect of age and education, female entrepreneurs 0.6 and individuals who opened their businesses because they 0.4 "could not find a job" have a higher probability of being informal. As shown in table 5.7, this is found both for 0.2 Brazil and Mexico, and for different indicators of informality. This result is consistent with both types of microentrepre- 0 0 1 2 3 4 5 neurs--females and previously unemployed workers-- viewing their business activity as more temporary than males b. Years in business and voluntary entrants into the sector, respectively. 1.0 While most microenterprises are informal, formality 0.8 rates increase quite rapidly when firms incorporate paid employees. In Mexico, 63 percent of all urban microenter- 0.6 prises are not registered with the federal treasury, and 72 percent pay no taxes. Nevertheless, among those 0.4 microfirms with at least one paid employee, only 31 percent 0.2 are unregistered and less than half (46 percent) report pay- ing no taxes. As illustrated in figure 5.7, by the time firms 0 have hired five paid workers the fraction that remains 0.17 0.67 1 2 3 4 unregistered has fallen to 13 percent and only 28 percent Not registered No tax payment declare paying no taxes. Moreover, while 86 percent of the No workers registered firms with only one paid employee do not pay social secu- Source: Author's calculations, using Encuesta Nacional de rity contributions for that worker, among those with five Micronegocios. paid workers, 71 percent report paying social security for at least some of their employees. Informality also tends to diminish, although less Similar evidence emerges from data on Brazilian microen- rapidly, with time in business. As suggested by the results terprises: while 76 percent do not have an operating license in table 5.8, the effect of time in business is stronger for and 94 percent do not pay taxes, those rates fall to 51 and younger firms, but it remains negative--older firms being 75 percent, respectively, among firms with at least one paid less likely to be informal--until firms have been in busi- employee. Moreover, among those that employ five paid ness for 18 years (22 in the case of Brazil). This is illus- workers, 67 percent have operating licenses and 23 percent trated for the case of Mexico in figure 5.7b. Depending report paying taxes. As shown in table 5.8, the positive effect on which definition is employed--lack of registration, tax of firm size on formality persists even when one controls, in a payments, and social security contributions for workers-- regression framework, for a number of other firm and business informality rates drop from between 70 and 90 percent owner characteristics. In Colombia, for instance, the probabil- among firms with under a year of existence, to between 60 ity of a microenterprise with one paid worker not being and 71 percent for those that have been in business for at registered or not paying taxes falls by, respectively, 20 and least four years. Similarly, while only 12 percent of Brazil- 15 percentage points in comparison with owner-only enter- ian microenterprises with at most three years of existence prises, and further reductions in informality probabilities are have operating licenses, that fraction is twice as large found among those with more paid workers.11 among older firms. It is worth noting, however, that the 150 M I C R O F I R M D Y N A M I C S A N D I N F O R M A L I T Y TABLE 5.8 Informality among Mexican, Brazilian, and Colombian microenterprises by paid employment and time in business Mexico Brazil Colombia (1) (2) (3) (4) (5) (6) (7) (8) (9) No social No social No social Variable No license No taxes security No license No taxes security No license No taxes security One paid worker 0.322*** 0.255*** 0.122*** 0.031*** 0.202 0.146 0.937 (0.000) (0.000) (0.000) (0.000) (0.000)*** (0.000)*** (0.000)*** Two paid workers 0.406*** 0.311*** 0.162*** 0.195*** 0.064*** 0.123*** 0.323 0.239 0.969 (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000)*** (0.000)*** (0.000)*** Three paid workers 0.506*** 0.426*** 0.347*** 0.253*** 0.080*** 0.187*** 0.344 0.280 0.981 (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000)*** (0.000)*** (0.000)*** Four paid workers 0.528*** 0.434*** 0.469*** 0.273*** 0.108*** 0.260*** 0.349 0.283 0.981 (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000)*** (0.000)*** (0.000)*** Five paid workers 0.506*** 0.447*** 0.574*** 0.272*** 0.086*** 0.307*** 0.354 0.268 0.979 (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000)*** (0.000)*** (0.000)*** Time in business 0.011*** 0.013*** 0.009*** 0.009*** 0.001*** 0.008*** 0.297 0.266 0.017 (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000)*** (0.000)*** (0.000)*** Time in business 0.0003*** 0.0003*** 0.0002*** 0.0002*** 0.0000*** 0.000*** 0.041 0.032 0.002 squared (0.000) (0.000) (0.002) (0.000) (0.000) (0.000) (0.000)*** (0.000)*** (0.000)*** Observations (n) 28,525 28,525 4,599 45,509 45,509 8,841 27,177 27,691 27,690 Pseudo-R2 0.189 0.159 0.215 0.220 0.159 0.220 0.174 0.095 0.665 Sources: Author's calculations using Encuesta Nacional de Micronegocios and Pesquisa Economia Informal Urbana; for Colombia, calculations performed by Carolina Mejia and Mauricio Cardenas. Note: Robust p-values appear in parentheses. In the cases of Mexico and Brazil, age, education, and previous unemployment of the owner, as well as fixed location, sales to large firms, sector, and state, are also controlled. Samples of business owners aged 21 and older, without complete tertiary education, operating in nonagricultural activities, and with at most five workers. For Colombia, other controls include having a fixed location and sector. The time-in-business variable is expressed in years for Brazil and Mexico, and is categorical for Colombia. ***p < .01. decision to operate formally or informally is often made at services and manufacturing appear in an intermediary the time of starting up. As seen in the bottom-right row of position. Other firm characteristics that are strongly asso- table 5.6, almost 85 percent of nonlicensed Brazilian ciated with lower informality rates are the fact of operat- microfirms did not even try to regularize their businesses ing out of a fixed location, and that of selling mainly to when they began operating, compared to 75 percent of for- large companies. For given firm sizes, having a fixed loca- mal businesses that at least attempted to do so. tion is associated with registration probabilities that are With regard to the sectors of economic activity where up to 25 percentage points higher in the three countries informality is more prevalent, at least in the case of for which we have data on microenterprise formality microenterprises, we find that in both Mexico and Brazil (table 5.9). Higher probabilities of paying taxes and social informal firms are most frequently found in the construc- security are also found in those three countries among tion sector, possibly due to lower firm survival rates in firms operating out of fixed locations. One interpretation this sector (see previous section), and they are least com- of this result is that firms that, because their activities do mon in retail trade (table 5.9). In both countries, other not need a fixed locale for their operations, could also be 151 I N F O R M A L I T Y TABLE 5.9 Informality among Mexican, Brazilian, and Colombian microenterprises by age, education, and previous activity of the owner Mexico Brazil Colombia (1) (2) (3) (4) (5) (6) (7) (8) (9) No social No social No social Variable No license No taxes security No license No taxes security No license No taxes security Fixed location 0.170*** 0.111*** 0.076*** 0.202*** 0.033*** 0.136*** 0.254 0.144 0.024 (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000)*** (0.000)*** (0.000)*** Large main 0.175*** 0.114*** 0.168*** 0.017* 0.036*** 0.048** client (0.000) (0.000) (0.000) (0.082) (0.000) (0.019) Services 0.188*** 0.168*** 0.054*** 0.021*** 0.007*** 0.033*** 0.196 0.158 0.018 (0.000) (0.000) (0.000) (0.000) (0.000) (0.001) (0.000)*** (0.000)*** (0.000)*** Manufacturing 0.226*** 0.176*** 0.040** 0.107*** 0.012*** 0.058*** 0.170 0.140 0.023 (0.000) (0.000) (0.035) (0.000) (0.000) (0.000) (0.000)*** (0.000)*** (0.000)*** Construction 0.368*** 0.271*** 0.178*** 0.137*** 0.013*** 0.055*** (0.000) (0.000) (0.000) (0.000) (0.000) (0.002) Observations (n) 28,525 28,525 4,599 45,509 45,509 8,841 27,177 27,691 27,690 Pseudo-R2 0.189 0.159 0.215 0.220 0.159 0.220 0.174 0.095 0.665 Sources: Author's calculations using Encuesta Nacional de Micronegocios and Pesquisa Economia Informal Urbana for, respectively, Mexico and Brazil; for Colombia, calculations performed by Carolina Mejia and Mauricio Cardenas, using data from Colombia National Administrative Department of Statistics. Note: Robust p-values appear in parentheses. Firm size, time in business, as well as age, education, and previous unemployment of the owner, sector, and state, are also controlled. Samples of business owners aged 21 and older, without complete tertiary education, operating in nonagricultural activities, and with at most five workers. For Colombia, other controls include size and time in business. *p < .1. **p < .05. ***p < .01. less motivated to formalize themselves because of the formal.12 These results are consistent with the findings of lower probability of being caught. Moreover, for firms de Paula and Scheinkman (2006), who show that since pur- that could operate either with or without a fixed locale, chases from informal firms are ineligible for tax credits, the desire to remain informal--without being caught-- value-added tax systems create incentives for the formali- could be the deciding factor for choosing the second zation of suppliers located upstream along productive option. Note, however, that this would not be without chains--then again, informality also tends to be conta- costs, as informal firms that operate on an ambulant basis gious, as firms that buy from (or sell to) informal firms also are also, as shown in the next chapter, less likely to accu- have an incentive to operate informally. mulate assets and increase sales and employment. Even after controlling for the presence of large clients As for firms selling to large companies, being formally and a fixed location, as well as for the owners' human capi- registered and paying taxes are likely to be requirements tal and other personal characteristics, firms with higher imposed by their clients, which would explain the higher levels of productivity are less likely to be informal. This level of formality among microenterprises that cater to is illustrated in figure 5.8, which reports the effect of a large companies. The increase in the probability of being doubling of labor productivity--equivalent to about one formal for firms having large companies as clients is between standard deviation in net revenues per worker--on the 11 and 18 percentage points in Mexico. The increase is probability of Mexican and Brazilian microenterprises being smaller, while still significant, in Brazil--between 2 and 5 informal. The effects are larger in the case of Mexico, where additional percentage points in the probability of being a 100 percent increase in labor productivity is associated 152 M I C R O F I R M D Y N A M I C S A N D I N F O R M A L I T Y Conclusions FIGURE 5.8 This chapter has shown that the patterns of entry, survival, Effect of doubling labor productivity on probability of being informal (%) and growth in the self-employment sector in Latin America match well the predictions of mainstream models of firm Add owner, dynamics developed for industrial-country contexts, sug- firms traits Brazil gesting that the behavior of many microfirm owners is dri- Firm size, age, sector ven by exit choices and not just by exclusion factors. Thus, Add owner, entrants into self-employment tend to be found more fre- firms traits Colombia quently among workers who have accumulated human and Firm size, age, sector physical capital while working as salaried workers. Among Add owner, the latter, those with relatively higher wages given their firms traits human capital--the overachievers--are more likely to Mexico Firm size, become firm owners, and those with more schooling are age, sector more likely to open businesses with at least some paid 20 15 10 5 0 employees. Once in the sector, the owners of more prof- Mean of lnypw itable firms are more likely to stay in business and hire No license more workers. Moreover, as firms grow older and larger, No registered workers their growth rates tend to diminish and their degree of No taxes informality tends to gradually decrease. Sources: Author's calculations, using Encuesta Nacional Few firms, however, tend to evolve along this ideal de Micronegocios and Pesquisa Economia Informal Urbana; for path, with the large majority of microfirms remaining Colombia, calculations performed by Carolina Mejia and Mauricio Cardenas, using data from Colombia National Administrative owner-only, and a large fraction of them failing. Thus, at Department of Statistics. Note: Coefficients on labor productivity in regressions where least in the short term, the sector is not likely to be an other regressors and samples are as described in tables 5.7 and 5.9, important source of net job creation--if any--for Latin respectively, (see notes to those tables). American economies. Two complementary explanations can be proposed to explain the high failure rates and lim- with a drop of about 15 percentage points in the probability ited growth and job creation of informal firms. On one of not being registered, and about 10 and 9 fewer percent- hand, informality and the factors that may be behind it-- age points in the probabilities of not paying taxes and social to be discussed in more detail in the next chapter--could security contributions, respectively. In Brazil, increases in be to blame, with policy-induced barriers to formaliza- productivity have a small effect on the probability of pay- tion impeding microfirm access to technologies and mar- ing taxes--between 1 and 2 percentage points higher for a kets, which in turn would keep them small and doubling of labor productivity--but they reduce the prob- unproductive, and in many cases lead them to exit alto- ability of not having an operating license and not paying gether, thus perpetuating a vicious circle of low growth social security contributions by as much as 9 and 10 per- and high informality. On the other hand, the evidence centage points. As for the evidence on Colombia, we presented in this chapter is also consistent with an alter- find that a doubling of labor productivity is associated native explanation for the type of microfirm dynamics with increases of up to 10 percentage points in formality observed in Latin America. In particular, the presence of probabilities. low opportunity costs for entry into the sector would also The apparent negative link between productivity and lead to a predominance of low-productivity businesses with informality can be interpreted, as mentioned above, as evi- low growth prospects and high failure rates. In this con- dence that firms in the upper tier of the microenterprise text, in order to reduce informality, policy makers should sector are more likely to become formal. However, as will focus not only on altering the direct costs and benefits of be discussed in the next chapter, it is also possible that at formality but also on the drivers of formal sector produc- least to some extent causality also runs in the opposite tivity, including measures to improve the investment direction, with access to formality leading to further climate and policies aimed at increasing human capital increases in firm productivity. accumulation. 153 I N F O R M A L I T Y Notes 9. Firm size is divided into owner-only firms, 2­5 employees, 1. This section draws heavily on Fajnzylber, Maloney, and Rojas 6­10 employees, 11­15 employees, 16­50 employees, 51­100 employ- (2006b). Only 6 of the 53 papers mentioned by Blanchflower (2004) ees, 101­250 employees, and 250 and more employees. The depen- in his self-employment literature review concern developing coun- dent variable is the imputed percentage difference between the mean tries, and they focus mostly on the determinants of earnings. To our value of employment in each firm size bracket. By definition, firms knowledge, the only previous evidence on the determinants of entry, that remained in the same size interval have a value of zero. exit, and growth of microenterprises in developing countries are the 10. While the positive sign encountered in the marital status vari- papers on Africa by Goedhuys and Sleuwaegen (2000), Liedholm able may well reflect the use of nonpaid family workers, the results (2002), Liedholm and Mead (1999), McPherson (1995, 1996), and are virtually unchanged when the firms with a majority of nonpaid Mead and Liedholm (1998). Other recent studies on firm dynamics in workers are kept in the sample. developing countries have focused mostly on larger and/or formal 11. The estimations with Colombian data were generously shared firms, including Aw, Chung, and Roberts (2003), Bartelsman, Halti- by Mauricio Cardenas and Carolina Mejia. wanger, and Scarpetta (2004), and Roberts and Tybout (1997). 12. The quantitatively smaller effect obtained for Brazil could 2. Other dynamic models also generate these patterns, although reflect better tax enforcement among smaller firms in Brazil, which with different analytical structures. Ericson and Pakes (1992a, 1995) would then not be significantly affected by their links to larger ones. propose a model of active exploration--as opposed to the passive Alternatively, if large companies also exhibit high tax evasion, then they learning assumption in Jovanovic's model--that incorporates firm- would not necessarily impose strict tax compliance requirements to specific sources of uncertainty derived from stochastic outcomes of their smaller providers. The second hypothesis is favored by the high investments made by firms in order to improve their profitability. level of tax evasion revealed by the 2003 Brazil Investment Climate Favorable outcomes from the firms' own investments--including Assessment survey, in which manufacturing firms reported that one- those that lead to entry into the industry--tend to move them third of the sales of the average enterprise goes unreported for tax pur- toward "better" states, while good outcomes of direct competitors poses, compared to 23 percent for Colombia and 30 percent for move them to less profitable states. As in Jovanovic's model, entry, Mexico--the averages for LAC and the OECD are, respectively, 23 and exit, and investment decisions are made to maximize the expected 6.5 percent. discounted value of future net cash flows conditional on the current information set. References 3. The evidence on Mexico is drawn mostly from Fajnzylber, Arnott, R., and J. E. Stiglitz. 1991. "Moral Hazard and Nonmarket Maloney, and Rojas (2006b). Institutions: Dysfunctional Crowding Out or Peer Monitoring?" 4. The above-mentioned surveys are Encuesta Nacional de Empleo American Economic Review 81: 179­90. Urbano/Encuesta Nacional de Microgenocios (ENEU/ENAMIN) for Mexico, Aw, B. Y., S. Chung, and M. J. Roberts. 2003. "Productivity, the Nicaragua Living Standards Measurement Study (LSMS), and the Output, and Failure: A Comparison of Taiwanese and Korean Colombia Manufacturing Survey. Manufacturers." Economic Journal 113 (491): F485­F510. 5. While this could suggest, as noted by Hopenhayn, that at the Bartelsman, E., J. Haltiwanger, and S. Scarpetta. 2004. "Microeco- margin the process of selection of entrepreneurs is poorer in Mexico, nomic Evidence of Creative Destruction in Industrial and Devel- one must also bear in mind that the higher rate of entry from salaried oping Countries." Policy Research Working Paper 3464, World work found in Mexico is partly a result of the relatively smaller size of Bank, Washington, DC. the wage sector in this country. Bartelsman, E., S. Scarpetta, and F. Schivardi. 2003. "Comparative 6. For comparison purposes with Evans and Leighton's figures for Analysis of Firm Demographics and Survival: Micro Level Evidence the United States, the rates of entry into self-employment reported for the OECD Countries." Economics Department Working Paper for Mexico in figure 5.1 are calculated as the fraction of the number 348, Organisation for Economic Co-operation and Development, of salaried workers who enter self-employment in the course of one Paris. year. Similarly, the self-employment rate is calculated by dividing the Bates, T. 1990. "Entrepreneur Human Capital Inputs and Small number of self-employed in a given age range by the sum of salaried Business Longevity." Review of Economics and Statistics 72 (4): and self-employed individuals of that age. Finally, exit rates are the 551­59. fraction of self-employed workers who move to wage work. Besley, T. 1995. "Savings, Credit, and Insurance." In Handbook 7. Note that in figure 5.2 entry rates are calculated for individu- of Development Economics, Volume II, ed. J. Behrman and T. N. als coming either from salaried work or from out of the labor force, Srinivasan. Amsterdam: Elsevier. which is a much more common position for women than for men. Bhattacharaya, P. C. 2002. "Rural-to-Urban Migration in LDCs: A Moreover, self-employment rates are also calculated as a proportion of Test of Two Rival Models." Journal of International Development the whole population in the same age range. Finally, exit rates in fig- 14: 951­72. ure 5.2 cover all self-employed women who move either out of the Blanchflower, D. G. 2000. "Self-employment in OECD Countries." labor force or to other labor market positions. Labour Economics 7: 471­505. 8. See Fajnzylber, Maloney, and Rojas (2006b) for details on the ------. 2004. "Self-employment: More May Not Be Better." sample and estimation methodology. Swedish Economic Policy Review 11 (2): 15­74. 154 M I C R O F I R M D Y N A M I C S A N D I N F O R M A L I T Y Blanchflower, D. G., and A. J. Oswald. 1998a. "Entrepreneurship and Fields, G. S. 1990. "Labor Market Modelling and the Urban Informal the Youth Labour Market Problem: A Report for the OECD." Sector: Theory and Evidence." In The Informal Sector Revisited, ed. Organisation for Economic Co-operation and Development, Paris. D. Turnham, B. Salome, and A. Schwartz. Paris: Organisation for ------. 1998b. "What Makes an Entrepreneur?" Journal of Labor Economic Co-operation and Development. Economics 16 (1): 26­60. Geroski, P. A. 1991. Market Dynamics and Entry. Oxford, UK: Carrasco, R. 1999. "Transitions to and from Self-employment in Blackwell. Spain: An Empirical Analysis." Oxford Bulletin of Economics and Goedhuys, M., and L. Sleuwaegen. 2000. "Entrepreneurship and Statistics 61: 315­41. Growth of Entrepreneurial Firms in Côte d'Ivoire." Journal of Cressy, R. 1996. "Are Business Startups Debt-Rationed?" Economic Development Studies 36 (3): 123­46. Journal 106: 1253­70. González de la Rocha, M. 1994. The Resources of Poverty: Women and Cunningham, W. V., and W. F. Maloney. 2001. "Heterogeneity Survival in a Mexican City. Oxford, UK: Blackwell. among Mexico's Microenterprises: An Application of Factor and Hamilton, B. H. 2000. "Does Entrepreneurship Pay? An Empirical Cluster Analysis." Economic Development and Cultural Change 50 (1): Analysis of the Returns to Self-employment." Journal of Political 131­56. Economy 108 (3): 604­31. Davis, S. J., J. Haltiwanger, and S. Schuh. 1996. Job Creation and Hart, K. 1972. Employment, Income, and Inequality: A Strategy for Destruction. Cambridge, MA: MIT Press. Increasing Productive Employment in Kenya. Geneva: International de Paula, A., and J. Scheinkman. 2006. "The Informal Sector." Labour Organization. Photocopy. University of Pennsylvania, Philadelphia. ------. 1973. "Informal Income Opportunities and Urban Employ- de Soto, H. 1989. The Other Path: The Invisible Revolution in the Third ment in Ghana." Journal of Modern African Studies 11: 61­89. World. New York: Basic Books. Holtz-Eakin, D., D. Joulfaian, and H. S. Rosen. 1994a. "Entrepre- Dunne, T., M. J. Roberts, and L. Samuelson. 1988. "Patterns of neurial Decisions and Liquidity Constraints." RAND Journal of Firms Entry and Exit in U.S. Manufacturing Industries." RAND Economics vol. 25 (2): 334­47, summer. Journal of Economics 19 (4): 495­515. ------. 1994b. "Sticking It Out: Entrepreneurial Survival and Liq- ------. 1989. "The Growth and Failure of U.S. Manufacturing uidity Constraints." Journal of Political Economy 102 (1): 53­75. Plants." Quarterly Journal of Economics 104 (4): 671­98. Hopenhayn, H. A. 1992. "Entry, Exit, and Firm Dynamics in Long Ericson, R., and A. Pakes. 1992a. "An Alternative Theory of Firm Run Equilibrium." Econometrica 60 (5): 1127­50. and Industry Dynamics." Cowles Foundation Paper 1041, Yale Johnson, W. R. 1978. "A Theory of Job Shopping." Quarterly Journal University, New Haven, CT. of Economics 92: 261­78. ------. 1992b. "Markov-Perfect Industry Dynamics: A Framework Jovanovic, B. 1979. "Job Matching and the Theory of Turnover." for Empirical Work." Review of Economic Studies 62 (1): 53­82. Journal of Political Economy 87: 972­90. ------. 1995. "Markov-Perfect Industry Dynamics: A Framework ------. 1982. "Selection and Evolution of Industry." Econometrica for Empirical Work." Review of Economic Studies 62 (1): 53­82. 50 (3): 649­70. Evans, D. S. 1987a. "The Relationship between Firm Growth, Size, Levenson, A., and W. Maloney. 1998. "The Informal Sector, Firm and Age: Estimates for 100 Manufacturing Industries." Journal of Dynamics, and Institutional Participation." Policy Research Industrial Economics 35 (4): 567­81. Working Paper 1988, World Bank, Washington, DC. ------. 1987b. "Tests of Alternative Theories of Firm Growth." Liedholm, C. 2002. "Small Firm Dynamics: Evidence from Journal of Political Economy 95 (4): 657­74. Africa and Latin America." Small Business Economics 8 (1­3): Evans, D. S., and B. Jovanovic. 1989. "An Estimated Model of Entre- 227­42. preneurial Choice under Liquidity Constraints." Journal of Politi- Liedholm, C., and D. C. Mead. 1999. Small Enterprises and Economic cal Economy 97 (4): 808­27. Development: The Dynamics of Micro and Small Enterprises. London: Evans, D. S., and L. S. Leighton. 1989. "Some Empirical Aspects of Routledge Press. Entrepreneurship." American Economic Review 79 (3): 519­35. Lippman, S. A., and R. P. Rumelt. 1982. "Uncertain Imitability: An Fajnzylber, P., W. F. Maloney, and G. V. Montes Rojas. 2006a. "Does Analysis of Interfirm Differences in Efficiency under Competi- Formality Improve Microfirm Performance? Quasi-experimental tion." Bell Journal of Economics 13 (2): 418­38. Evidence from the Brazilian SIMPLES Program." Photocopy. Loayza, N. 1996. "The Economics of the Informal Sector: A Simple World Bank, Washington, DC. Model and Some Evidence from Latin America." Carnegie- ------. 2006b. "Microenterprise Dynamics in Developing Coun- Rochester Conference Series on Public Policy 45: 129­62. tries: How Similar Are They to Those in the Industrialized Lucas, R. E., Jr. 1978. "On the Size Distribution of Business Firms." World? Evidence from Mexico." World Bank Economic Review 20 (3): Bell Journal of Economics 9 (2): 508­23. 389­419. Maloney, W. F. 1999. "Does Informality Imply Segmentation in ------. 2006c. "Releasing Constraints to Growth or Pushing on a Urban Labor Markets? Evidence from Sectoral Transitions in String? The Impact of Credit, Training, Business Associations Mexico." World Bank Economic Review 13: 275­302. and Taxes on the Performance of Mexican Microfirms." Policy ------. 2001. "Self-employment and Labor Turnover in Developing Research Working Paper 3807, World Bank, Washington, DC. Countries: Cross-country Evidence." In World Bank Economists' 155 I N F O R M A L I T Y Forum, ed. S. Devarajan, F. H. Rogers, and L. Squire. Washington, Pakes, A., and R. Ericson. 1998. "Empirical Implications of Alternative DC: World Bank. Models of Firm Dynamics." Journal of Economic Theory 79 (1): 1­45. ------. 2004. "Informality Revisited." World Development 32 (7): Rees, H., and A. Shah. 1986. "An Empirical Analysis of Self-employ- 1159­78. ment in the UK." Journal of Applied Econometrics 1: 95­108. McKenzie, D. J., and C. Woodruff. 2006. "Do Entry Costs Provide Roberts, M. J., and J. R. Tybout, eds. 1997. Industrial Evolution in an Empirical Basis for Poverty Traps?" Economic Development and Developing Countries: Micro Patterns of Turnover, Productivity, and Cultural Change 55 (1): 3­42. Market Structure. Washington, DC: World Bank. McPherson, M. A. 1995. "The Hazards of Small Firms in Southern Tybout, J. 2000. "Manufacturing Firms in Developing Countries: Africa." Journal of Development Studies 32 (1): 31­54. How Well Do They Do, and Why?" Journal of Economic Literature ------. 1996. "Growth of Micro and Small Enterprises in Southern 38 (1): 11­44. Africa." Journal of Development Economics 48 (2): 253­77. Wagner, J. 1994. "The Post-entry Performance of New Small Firms in Mead, D. M., and C. Liedholm. 1998. "The Dynamics of Micro and German Manufacturing Industries." Journal of Industrial Economics Small Enterprises in Developing Countries." World Development 26 42 (2): 141­54. (1): 61­74. 156 CHAPTER 6 Informality, Productivity, and the Firm SUMMARY: This chapter focuses on both the determinants and the consequences of informality from the perspective of pri- vate firms. The chapter begins by arguing that firm informality encompasses not only the large number of unregistered microfirms described in chapter 5, but also those medium-size and large firms that fail to comply with at least some gov- ernment regulations. This is illustrated by recent estimates of a large incidence of tax and social security evasion among Latin American firms of all sizes. The chapter then reviews the evidence on the factors that may lead private firms to exhibit different degrees of informality, and analyzes the channels through which regulatory noncompliance can affect individual and aggregate productivity. Finally, the chapter lays out the possible approaches that policy makers could adopt to tackle the issue of firm informality. W HILE INFORMALITY IS OFTEN between small and large firms. Finally, we show that recent associated with small, unregis- initiatives to reduce red tape and introduce simplified tax tered microfirms, tax and social and registration systems for microfirms have led to statisti- security evasion is commonplace cally significant increases in the number of new formally among larger Latin American registered firms, although further research is required to firms. This chapter reviews the main firm characteristics establish whether the corresponding effects are permanent associated with informality among those firms, using data or temporary, and to resolve existing controversies on their from recently concluded enterprise surveys in a number of magnitude. In particular, some recent estimates suggest Latin American countries. Not surprisingly, we find that the that the simplification of entry regulation leads more former incidence of tax and social security evasion varies consider- high-ability wage workers to open formal businesses, but it ably across and within countries, and it is generally higher has small or no effects on the formalization of unregistered for small, low-productivity firms that started their opera- businesses. tions without a formal registration. We argue that there are reasons to expect important With regard to the factors that may help explain this overall productivity gains if a larger fraction of firms would variation, we find that informality tends to increase with the formalize and if resources could be shifted away from low- quantity of labor and product markets regulations, and to productivity informal firms toward the formal sector. This decrease with the quality of governance--for example, the could be the case in a context where increasing returns to prevalence of the rule of law and the level of democratic scale are relevant, at least among very small firms, and accountability. Moreover, firm informality is positively informality is accompanied by a preponderance of small related to the incidence of corruption, but it can be curtailed firms. Moreover, unfair competition from informal firms by the improvement of market support institutions--the could slow down the process of creative destruction by courts, financial markets, and the development of links which inefficient firms are replaced by their more efficient 157 I N F O R M A L I T Y competitors, and negatively affect the incentives of formal unregulated firms that avoid most taxes and labor regula- firms to innovate and adopt new technologies. Finally, tions, and do not comply with most government regulations. growth could increase as previously informal firms gain Informality, in this approach, is seen as affecting mostly the increased access to markets and services. These positive very low end of the firm-size spectrum. links between formality and growth are supported by In a more general approach, as argued in chapter 1 of this firm-level evidence presented in this chapter. Indeed, we volume, it is reasonable to define the informal sector as find that firms that started operations without formally encompassing all the firms that, at least to some extent, registering--at least initially--and those located in choose to operate outside of the scope of existing regulations. regions or sectors where informality is more prevalent Thus, medium-size and large firms can be considered infor- exhibit, on average, much lower productivity than their mal even if they are appropriately registered, provided that, peers. Moreover, exogenous increases in formality levels for instance, they underreport their sales for tax purposes; do caused by changes in the costs of microfirm formalization not register all their workers with the social security admin- are also found to lead to improved firm performance. istration; or do not comply with some government regula- In practice, however, policy makers may need to act tions regarding mandatory operating licenses or permits, as on several fronts at the same time in order to tilt the well as product quality and safety regulations. cost­benefit analysis of firms toward formality by combining For analytical purposes, however, it is useful to divide the both positive and negative incentives--respectively, "car- informal sector into two different subsectors. The first is the rots" and "sticks." Thus, the impact of interventions aimed informal microenterprise sector, which we described in at reducing the costs of being formal through the removal of the previous chapter and is composed of mostly informal, regulatory constraints may be larger when accompanied by unregistered microenterprises. The second informal subsec- measures to enhance evenhanded enforcement of regulations tor, which Djankov et al. (2003) denominate the "unofficial and increase the potential benefits of joining the formal economy," is made up of firms that are only partially infor- sector--for example, through improvements in access to mal, in the sense that they are formally registered but keep credit, contract enforcement mechanisms, and technical a fraction of their workers and sales hidden from govern- assistance. In particular, recent evidence from randomized ment regulators, and/or fail to comply with at least some experiments suggests relatively high returns to capital government regulations related, for instance, to mandatory among very small Mexican microenterprises, which implies permits and licenses. While unofficial firms tend to be that considerable increases in income could be obtained mostly small, in some countries and regions this subsector through incentives to the formalization of small businesses can include medium-size and even large companies.1 and expanded access to microcredit. These efforts, however, Data recently collected through surveys of registered may have larger effects on the subset of higher-productivity firms in seven Latin American and Caribbean countries con- informal firms (or high-ability wage earners considering firm that many small, medium-size, and even large firms also entry into the microfirm sector) that may be closer to the exhibit some degrees of informality. This is illustrated in fig- margin separating the formal from the informal sector--for ure 6.1, which shows that sales and employment underre- example, because they have more to win from formalizing or porting (for tax purposes) is commonplace among registered because the opportunity costs of operating informally are medium-size and large firms in selected Latin American and higher for them. Moreover, for very-low-productivity infor- Caribbean countries of different sizes and levels of income.2 mal firms it is possible that, as argued in the previous chap- The high variation in levels of tax and social security ter, a larger impact on overall formality levels could be evasion suggested by this figure, across countries with rela- achieved in the medium-to-long term through actions tively similar levels of income, implies that cross-country aimed at increasing job opportunities in the formal sector. differences in informality are not driven just by levels of economic development. In Brazil and Panama, for instance, Informality among registered firms firms reportedly evade as much as 30 to 40 percent of their The informal sector is often associated with the large number taxes, compared to less than 20 percent in Uruguay and of unregistered small businesses found in most urban centers Peru, and less than 5 percent in Chile.3 As discussed in the of the developing world. Thus, what people often have in next section, possible explanations may be associated with mind when thinking about the informal sector are small, the size of tax and social security burdens, labor legislation, 158 I N F O R M A L I T Y, P R O D U C T I V I T Y, A N D T H E F I R M governance issues related to red tape and corruption, levels security evasion is more prevalent among smaller firms. In of regulatory enforcement, and the value attributed by Bolivia and Mexico, for instance, about 35 percent of sales firms to market- and government-provided services. go underreported among microenterprises, compared to A second finding derived from figure 6.1 is that, for most around 10 to 15 percent among firms with 100 workers or countries, the data suggest, as expected, that tax and social more. However, in Panama, Peru, and Uruguay, the data do not suggest a clear pattern linking underreporting rates to FIGURE 6.1 firm size. Thus, large firms appear to evade taxes and social Unreported sales and workers, by firm size (%) security contributions at rates that are comparable to those of smaller enterprises--between 10 and 15 percent in Peru Number of employees and Uruguay, and above 30 percent in Panama. 5 6­10 The negative relationship between underreporting rates 11­20 Panama 21­50 and employment size in the case of Argentina, Bolivia, 51­99 100 Colombia, and Mexico is maintained when the firms' time 5 in business, location, and sector are controlled for in a 6­10 11­20 regression framework (figure 6.2). Thus, hypothetically Brazil 21­50 51­99 doubling the size of a firm leads to a reduction in underre- 100 porting rates of more than 5 percentage points in Bolivia 5 6­10 and 4 in Mexico, with roughly one-half of those effects 11­20 Mexico 21­50 obtained for the cases of Colombia and Argentina, respec- 51­99 tively. A small effect is also found for Uruguay, where the 100 5 underreporting of sales for tax purposes diminishes by 1.5 6­10 11­20 percentage points as firms hypothetically double in size. The Bolivia 21­50 time that a firm has been in business, however, is not found 51­99 100 to be significantly related to informality levels, with the 5 6­10 only exceptions being Colombia (where for each additional 11­20 Argentina decade since starting up firms appear to reduce their sales 21­50 51­99 underreporting by about 1 percentage point, and Mexico 100 5 (where the opposite effect is obtained--increasing underre- 6­10 porting with time in business, at a rate of 1 percentage point 11­20 Colombia 21­50 for each additional decade since starting up). 51­99 100 5 6­10 11­20 Uruguay 21­50 FIGURE 6.2 51­99 Effect of doubling employment size on underreporting rates (%) 100 5 6­10 11­20 Argentina Peru 21­50 51­99 Bolivia 100 5 6­10 Colombia 11­20 Chile 21­50 51­99 Mexico 100 0 10 20 30 40 50 8 6 4 2 0 Unreported sales Unreported workers Unreported sales Unreported workers Source: Authors' calculations, using the World Bank's Enterprise Source: Authors' calculations, using the World Bank's Enterprise Survey Database. Survey Database. 159 I N F O R M A L I T Y FIGURE 6.3 FIGURE 6.4 Effect of having started informally on underreporting rates (%) Effect of doubling labor productivity on underreporting rates (%) Average Argentina (7 countries) Bolivia Bolivia Colombia Colombia Mexico Mexico Panama Peru Peru Uruguay 0 5 10 15 20 25 4 3 2 1 0 Unreported sales Unreported workers Unreported sales Unreported workers Source: Authors' calculations, using the World Bank's Enterprise Source: Authors' calculations, using the World Bank's Enterprise Survey Database. Survey Database. Another relevant finding common to at least five coun- about one standard deviation in that variable--is associated tries is that firms that start operating in the informal sector with underreporting rates that are, on average, 2 percent- exhibit higher rates of sales and employment underreport- age points lower. ing than other firms of the same size, sector, and location, As suggested above, the negative correlation between even many years after having registered, and after control- productivity and informality is subject to different interpre- ling for sector and location characteristics.4 Indeed, even tations. On one hand, more productive firms may arguably after controlling for the time that firms have been in have more to lose from operating irregularly (a topic that we business--on average, more than 20 years, and about 8 years discuss in the next section while summarizing the main after having registered--starting without a formal registra- firm-level determinants of informality). On the other hand, tion is associated with rates of tax and social security evasion however, it is also possible that productivity is affected by that are between 6 and 25 percentage points higher than for whether firms operate formally or informally, as well as by firms that registered at the time of starting up (figure 6.3). the general level of informality prevailing in their region This suggests that efforts to facilitate early registration of and sector (a topic that we cover in the third section of this new firms have the potential to reduce informality both chapter). through increasing the number of tax-paying firms and through reducing tax evasion levels among registered firms. Firm-level determinants of informality Within given sectors, however, and for given firm size Why do some firms comply with government regulations and time in business, we find that firms with higher levels while others opt for going underground? It is reasonable to of labor productivity exhibit, in general, lower rates of tax assume that private firms voluntarily chose to operate in and social security evasion. This is illustrated in figure 6.4, the formal or the informal sector based on rational profit- where we report the estimated effects on the rates of sales maximizing calculations, not unlike those underlying and employment underreporting of a hypothetical dou- investment and production decisions. In particular, the bling in output per worker (controlling for firm size, time extent to which firms comply with government regulations in business, location, and sector of activity). The effects are is likely to depend on their weighing of the various costs not significant for Panama, nor for sales underreporting in and benefits associated with operating formally or infor- Bolivia and Colombia. However, for all the other cases, we mally. Some of the main factors that firms are likely to take find that doubling firms' labor productivity--a change of into account are the nature of the regulatory framework, 160 I N F O R M A L I T Y, P R O D U C T I V I T Y, A N D T H E F I R M the extent to which regulations are enforced, and the vari- longer-term relationship--and that in turn affects their ous opportunity costs associated with operating in the incentives to invest in training and capital goods. underground economy. High levels of corruption may also play a role in shaping firms' incentives for operating informally. Indeed, when Benefits and costs of informality for private firms caught by government inspectors, irregular firms may have Among the main advantages that firms may consider when the option of bribing those inspectors to evade fines and opting for informality are the possibility of reducing or other hassles, which may reduce the risk associated with eliminating tax payments and social security contributions, operating informally. Second, in countries where formal and the possibility of avoiding costly and burdensome gov- firms face a high risk of being extorted by corrupt officials, ernment regulations, including but not restricted to those entrepreneurs may decide to operate informally exactly to affecting labor markets.5 Some of the main "benefits" from reduce their vulnerability to extortion. In fact, there is evi- informality are thus directly linked to the value of taxes dence that this is the case in several so-called transition and social security contributions that irregular firms are economies where one of the main motivations for firms' able to avoid. going underground is to "dodge the grabbing hand" (see Other indirect "benefits" are related to cost savings Friedman et al. 2000; Johnson et al. 2000; and Johnson, derived from avoiding the often complex administrative Kaufmann, and Shleifer 1997). procedures associated with tax and regulatory compliance, Besides the risk of being caught, informality entails and to the added flexibility enjoyed by informal firms in other private costs. Thus, operating in the underground their employment and production decisions. Thus, for economy eliminates--or at least greatly reduces--access to instance, informal firms arguably enjoy lower hiring and fir- the courts and other formal contract enforcement mecha- ing costs, and they have more degrees of freedom when set- nisms. This may increase the vulnerability of informal ting wages and work hours. Moreover, informal firms may firms in their transactions with other private parties as well be able to reduce their costs--and potentially increase their as with government. As a result, they may be forced to sales--as a result of not having to comply with government- restrict their transactions to the potentially limited set of imposed standards for products and production processes. trading partners that are deemed trustworthy. This has And, last but not least, informal firms may be able to cir- negative implications not only in terms of social welfare, as cumvent the red tape associated with obtaining government it leads to forgoing potential gains from increased trade, permits and licenses. As documented by the World Bank's but also in terms of reinforcing the above-mentioned fac- Doing Business reports, many of those procedures are often tors that tend to limit the expansion of informal firms. costly and time-consuming, which may lead some firms to Another important cost associated with informality is opt for operating informally in order to avoid them. given by the narrower set of formal financing mechanisms Potentially countervailing the above cost savings, infor- available to informal firms. Indeed, bank and other formal mal firms need to deal with the risk of being caught and financial institutions are generally not willing to grant closed down. Since better enforcement of regulations credit to companies that lack proper documentation, increases the expected value of the fines and other losses including that related to government registration and derived from being detected by regulators, it reduces the licensing, tax compliance certificates, and audited financial incentives for operating informally. However, since it is statements, all of which are generally lacking in the case of neither feasible nor efficient for governments to supervise informal firms. Moreover, if to evade taxes companies do all individual firms, enforcement tends to be concentrated not register all assets as belonging to the company, their on larger firms. As a result, informality has the effect of ability to use them as collateral for bank loans may also be limiting firm growth, both because smaller informal firms limited. Similarly, in the case of firms that hide a fraction of are less likely to be caught by government inspectors and their revenues to elude taxes or other regulations, financial because the uncertainty associated with informality dis- statements may misrepresent their financial soundness courages investments in illiquid assets. Moreover, in the and economic prospects, thus reducing their attractiveness particular case of informality with respect to labor regula- to prospective lenders. The same applies to prospective tions, irregular firms are likely to have a harder time investors, which reduces informal firms' ability to raise attracting more educated workers and engaging them in a equity capital. 161 I N F O R M A L I T Y Informality can also forbid firms from benefiting from separating firms with and without employees (beyond the government support programs targeted at small and owner). Among the main advantages of formality, the medium enterprises (SMEs), as those programs are often surveyed firms mention the avoidance of fines and bribes, restricted to registered tax-paying firms. This can be a seri- followed by the possibility of gaining new clients and ous obstacle to the growth of informal firms, at least if one expanding operations--to which, interestingly, firms with at assumes that SME support programs effectively compen- least some employees give more importance than owner-only sate for market failures that limit the ability of those firms firms. As for the main disadvantages associated with formal- to access formal credit markets and acquire the technolo- ization, the factors to which the enterprises interviewed gies and human capital needed for their expansion. How- attribute more importance are the need to renew their ever, if firms do not place a large value on participating in licenses every year and the tax obligations resulting from SME-supporting government programs, or if market sup- formality--the second one being more important for firms port institutions in general are not well developed--for with employees. example, credit markets do not function well or contract Also cited as important advantages of formality, enforcement mechanisms are slow and costly--then the although less frequently than the above-mentioned factors, cost of being excluded from those institutions as a result of are the possibility of improving access to credit and using operating informally is lower, and a larger share of output is contract enforcement mechanisms--the second one being likely to be found in the underground economy. more important for larger firms. However, when asked Evidence on the relative importance of the various costs about the most important advantage of formality, only and benefits of informality--or, equivalently, the main 14 percent of the surveyed firms say that they are motivated advantages and disadvantages from formalization--is by the desire to expand or seek new clients, 8 percent men- presented in figures 6.5 and 6.6, based on information tion access to credit as the top reason for formalizing, and provided by firms surveyed by the International Finance Cor- just 1.5 percent mention access to the courts. In contrast, poration in 65 municipalities in Bolivia, Brazil, Honduras, 47 percent of the firms say that the top reason underlying Nicaragua and Peru.6 The figures report the average degree their decision to register their enterprise is "to comply with of importance of various factors on a scale from 0 to 4, the law" and 29 percent want to avoid fines or bribes. Thus, it appears that the risk of being caught prevails over posi- tive incentives associated with access to markets and FIGURE 6.5 services, a finding that, in general terms, applies both to Advantages of formalization reported by IFC-surveyed firms Avoid paying fines FIGURE 6.6 Compliance Disadvantages of formalization reported by IFC-surveyed firms with the law Avoid Annual paying bribes license renewal Gain new clients Payment of taxes Improved Regular inspection access to credit Operation on a Accounting greater scale Legal power to demand Compliance contracts upheld with labor laws 0 1 2 3 4 0 0.5 1.0 1.5 2.0 2.5 Firms without employees Firms without employees At least one employee At least one employee Source: Authors' calculations, using the World Bank's Enterprise Source: Authors' calculations, using the World Bank's Enterprise Survey Database. Survey Database. 162 I N F O R M A L I T Y, P R O D U C T I V I T Y, A N D T H E F I R M owner-only and to larger firms--the former being slightly FIGURE 6.7 more afraid of fines, the latter being more concerned about Impact of changes in costs and benefits of informality bribes, and both giving limited importance to issues related to access to credits and the courts. Firms remain informal The relative importance of the various private costs and benefits associated with informality is likely to vary with a number of firm characteristics, including size, time in busi- ness, and productivity. Thus, for instance, compared to larger firms, microenterprises may face a smaller risk of being Benefits Costs Benefits Costs caught by government inspectors when operating irregu- larly. However, they are likely to have a harder time Firms become formal amortizing the fixed costs associated with regulatory compliance--for example, the costs of firm registration, per- mits, and licenses--that may be a non-issue for large firms. Moreover, the red tape and monetary costs associated with formalization probably have a larger effect on recently cre- ated firms, which may choose to avoid them until they have Benefits Costs Benefits Costs accumulated sufficient evidence regarding their actual prof- Source: Prepared by authors, based on hypothetical cost and itability and likelihood of staying in business. benefit figures. Similarly, firms with inherently low productivity and/or growth prospects are likely to have a lower demand for credit and business development services, as well as for con- where a reduction in the benefits from informality proves tract enforcement mechanisms, thus being less affected in insufficient to induce firms to formalize because the costs of their informality decisions by the level of development of informality--or, equivalently, the advantages of operating market support institutions--the courts, financial markets, formally--are too low. In those cases, as illustrated in the and the like. This would explain, for instance, why firms lower panel of figure 6.7, for policy reforms to have binding belonging to the upper tier of the microenterprise sector effects on firm behavior, efforts may need to be directed are more likely to operate formally, or why registered firms both at reducing the benefits and at increasing the costs of with relatively low levels of productivity are more likely to operating informally.7 report higher rates of tax and social security evasion. As for- As an example, establishing the existence of very high malized by Rauch (1991), to the extent that taxes and reg- registration costs does not, in itself, imply either that this ulations are enforced mainly among large, formal sector is why firms don't register or that not registering is a fun- firms, entrepreneurs with higher managerial ability are damental determinant of average small firm performance. endogenously allocated to big firms where they are able to De Soto's (1989) telling anecdotes--for example, the side- compensate for the corresponding higher regulatory costs. walk vendor who wishes to pay his/her taxes as a way of In contrast, smaller and informal businesses are more likely securing quasi-property rights to his/her pitch--do sug- to be run by less-talented entrepreneurs, which would be gest that the high costs of formalization may impede infor- compensated by the lower costs of informal operations. mal firms from enforcing their property rights and It is worth noting that, from a policy point of view, pro- accessing public services, and may limit their access to grams aimed at reducing informality by means of affecting markets, thus negatively affecting their performance. How- its private benefits--for instance, by reducing barriers to ever, one must bear in mind that registration costs are only entry into the formal sector--may prove ineffective if the one of the factors that informal firms are likely to consider costs of operating informally are too low--for instance, when assessing whether to enter the formal sector. And, because regulatory enforcement and the odds of getting depending on the importance of other costs and benefits caught are low or because firms place little value on market- associated with formality, registration costs may not be the or government-provided services available to formal firms. binding constraint for most small informal firms, perhaps This point is illustrated in the upper panel of figure 6.7, because enforcement is limited in any case. 163 I N F O R M A L I T Y This view is supported by recent evidence on Mexico It is worth noting, however, that "pushing" all firms reported by McKenzie and Woodruff (2006). Using a sur- into the formal sector may not necessarily be feasible or vey of informal microfirms, they show that the vast major- good social and economic policy. Indeed, if, as argued by ity of them give as the principal reason for not being Levenson and Maloney (1998), formality operates as a nor- registered not that registration is too expensive or time- mal input in the production function of firms, it is possible consuming (respectively, 2 and 8 percent of surveyed that the intrinsic cost structure of many informal microen- firms), nor that the costs of operating as registered busi- terprises may never, in fact, dictate that they grow large nesses are too high (4 percent of firms), but that they are enough to need most of the formal institutions of civil soci- too small to make it worth their while (75 percent). ety. Thus, for instance, given their very restricted markets, A somewhat different picture emerges, however, from many of those microfirms may find it more efficient to use household survey data from Argentina, where most unreg- informal contract enforcement mechanisms and to operate istered microfirm owners report that they have remained on the basis of internal sources of finance. As a result, forc- informal because registration is too expensive (57 percent) ing them to formalize or trying to bring them into formal or complicated (4 percent), compared to 39 percent who credit or capital markets would amount to "pushing on a state that they do not register because it would be useless. string," and it could lead large numbers of self-employed Data from the Dominican Republic suggest a breakdown workers into open unemployment, while pushing formal similar to Mexico's: only 8 percent of informal microfirm sector wages downward. owners report that they remain informal to avoid taxes and Potentially countering these effects, however, one could social security costs, while 20 percent want to save the time argue that overall productivity could increase in the corre- and money involved in formalization. In contrast, almost sponding economies, as surviving microenterprises--the two-thirds indicate that they are too small to make formal- "upper tier" of the sector--become more efficient, thanks to ity worthwhile, they don't need to be formal, or they don't formality, and as the goods and services previously produced register because no business like theirs does. Interestingly, by "lower-tier" microfirms are offered by larger and more 92 percent state that their businesses have not suffered as a productive firms. Which effect prevails depends, however, result of being informal, and 38 percent indicate that they on whether formalization does indeed increase the produc- have actually benefited from their informal status--mainly tivity of some "upper-tier" microfirms, and on whether through lower taxes and the avoidance of government reg- informality does generate considerable negative externali- ulations. Similarly, a survey of Guatemalan informal entre- ties on the rest of the economy (topics that we cover in the preneurs, performed by the National Economic Research next section of this chapter). Center, indicates that the majority of them do not perceive any concrete benefit from complying with government reg- Cross-country evidence on the determinants ulations (CIEN 2006). of informality The above evidence leaves open the hypothesis that, Despite convincing cross-country evidence confirming the depending on each specific country context, registration relevance of several of the above-cited potential costs and costs may or may not be binding for most small firms and, benefits of informality, the data suggest that, to affect the in some cases, they could at best be a marginal contributor size of the underground economy, policy makers may need to informality. If this is indeed the case, further incentives to act on several fronts at the same time. The effects of reg- may be needed to entice small firms to enter the formal sec- ulations, for instance, appear to depend on the quality of tor, including both positive and negative incentives--the governance. Moreover, as argued above, small changes in carrots and sticks mentioned above. While the latter only some of the private costs or benefits of informality may involves increased government enforcement of regulations not have a binding effect on firms' decisions regarding reg- and potentially higher penalties for evaders, positive incen- ulatory compliance. tives range from tax reductions to changes in labor market The presence of a positive relationship between the regu- regulations and improvements in private and public ser- lation of firm entry and labor markets, on one hand, and the vices available to formal firms (for example, credit, contract size of the informal sector, on the other, has been illustrated enforcement, technical assistance, and so forth). by Botero et al. (2003) and Djankov et al. (2002). These 164 I N F O R M A L I T Y, P R O D U C T I V I T Y, A N D T H E F I R M authors were the first to construct large cross-country data- where tax regulations and enforcement are perceived as bases covering, respectively, the legal requirements for reg- being fair--thus increasing "tax morale"--low levels of istering new firms (together with the time and costs tax evasion and informality can be achieved without nec- involved in the corresponding procedures); and data on essarily reducing tax burdens, thereby allowing for an employment, collective bargaining, and social security adequate provision of productivity-increasing public laws.8 Djankov and his coauthors show that, for given levels goods. of per capita income, the informal sector tends to be larger The above findings imply that both the quantity and the in countries where registering a new firm involves a larger quality of regulation matter for explaining cross-country number of procedures or where employment and industrial differences in the size of the informal sector. In particular, relations laws are more rigid. These results suggest that, at reducing the quantity of regulations may be a good way of least on average, entry and labor regulations are not driven diminishing informality in countries characterized by bad mainly by public interests, nor are they means for increasing governance, but it may have a much smaller--or even a the efficiency with which society operates. null or negative--impact where the quality of institutions Using a smaller sample but new estimates of the size of is high. This is illustrated by the findings of Loayza, the informal sector for 14 Latin American and Caribbean Oviedo, and Servén (2005) that labor and product countries, Loayza (1996) shows that informality is posi- markets10 regulations are positively related to the size of tively associated with levels of taxation and labor market the informal sector only for countries with low governance regulations, and negatively correlated to the strength and quality, below a threshold that corresponds roughly to the efficiency of government institutions. Loayza and Rigolini levels of Greece, Japan, and Spain. To measure the quality (2006) confirm these results in a dynamic framework, of governance, Loayza, Oviedo, and Servén use indicators of showing that, in the long run, informality is negatively and the absence of corruption in the political system, prevalence robustly related to the flexibility of business regulations, of the rule of law, and level of democratic accountability. the value of public services associated with law and order, They argue, and their results seem to confirm, that in coun- and the capacity of governments to monitor and enforce tries with better quality of governance, regulations are formal taxes and regulations.9 more likely to be driven by valid social goals, as opposed to One caveat to the above results is that the long-run the interests of particular groups, and their enforcement is links between regulations and informality may apply dif- probably more transparent and less discretionary. In con- ferently in countries characterized by strong or weak insti- trast, where corruption is high, and democracy or the rule tutions, with good or bad governance systems. This is of law is weak, increasing the quantity of regulations is exemplified by the finding of Friedman et al. (2000), in a likely to stimulate informality. Consistent with the find- sample of 69 countries, that higher tax rates are not corre- ings of Friedman et al. (2000), Loayza, Oviedo, and Servén lated with a larger unofficial economy, and may, in fact, be (2005) also find that higher levels of fiscal regulations are linked to a smaller informal sector. They interpret this associated with smaller informal sectors in countries with result by suggesting that, across the countries in their good governance, but fiscal regulations are unrelated to the sample, the incentive to evade high tax rates is out- extent of informality in countries where the quality of gov- weighed by the larger benefits of formality in countries ernance is sufficiently low--the threshold corresponding to where higher tax revenues help finance productivity- the levels of Colombia and Pakistan.11 enhancing public goods and a strong legal environment. Indeed, they find that most of the available indicators of Firm-level evidence on the determinants bad governance--including corruption, overregulation, of informality and weak legal environments--are positively and robustly Using survey-based, firm-level data for five Eastern European related to the size of the informal sector. Thus, high tax countries, Johnson et al. (2000) confirm some of the above rates can coexist with small unofficial economies, provided cross-country results. They find, for instance, that among that rules and regulations are not enforced in a discre- Russian and Ukrainian manufacturing firms, respectively, tionary way and that levels of corruption are kept under an average 41 and 29 percent of sales go unreported for tax control. In other words, as argued elsewhere in this report, purposes, compared to between 5 and 7 percent of sales 165 I N F O R M A L I T Y in Poland, Romania, and Slovakia. This is not surprising, Panama, Paraguay, and Peru. The relative importance of Johnson and coauthors argue, given that managers in other labor issues varies across countries, with regulations Russia and Ukraine face much higher taxes, report much on temporary work being mentioned more frequently in higher levels of bureaucratic corruption and Mafia extor- Argentina and Bolivia (after severance costs), as well as in tion, and exhibit a lower trust in their legal and court sys- Colombia and Panama, and retirement benefits being the tems. However, firm-level regressions on the determinants issue of second-most importance in Uruguay. of informality using the same data for three of the five In the case of Panama, we also find some evidence of a countries cited above suggest that only the prevalence of link between informality and the enforcement of tax regu- corruption--measured through firms' reporting of extrale- lations. Indeed, in this country we find that informality gal payments for services or government licenses--has a decreases with the percentage of firms in the corresponding significant relationship with the percentage of sales unre- city and sector that have been visited or inspected by tax ported for tax purposes, with no effects found for taxation officials during the previous year. For each percentage point or court efficiency. increase in the probability that a firm is visited by tax Enterprise survey data for Latin America and the inspectors, the fractions of unreported sales and workers are Caribbean also suggest that corruption is positively and reduced by between 1 and 2 percentage points. This result significantly related to informality. As seen in table 6.1, in is consistent with evidence on the higher prevalence of five of the seven countries for which data are available (the informal salaried workers among Brazilian firms located in exceptions being Panama and Peru), companies reporting areas where labor regulations are less tightly enforced. that bribing of government officials to "get things done" is Indeed, Almeida and Carneiro (2005) show that, for each a common practice in their line of business exhibit rates of 10 percent increase in the number of fines per 1,000 firms revenue and worker underreporting that are between 4 and issued due to irregularities associated with unregistered 8 percentage points larger than those of other firms. As workers, the fraction of informal employees falls by 1.2 per- suggested by Johnson et al., (2000), this result could be cent as a proportion of total employment. due to firms' underreporting some of their activities (sales Interestingly, Almeida and Carneiro also find that the and workers) in order to hide them from corrupt officials. lower level of informality resulting from a tighter enforce- Alternatively, if causality runs in the opposite direction, ment of labor regulations is associated with a decrease in bribes could be a condition for remaining partially infor- labor productivity--a reduction of 3.6 percent for each mal. Moreover, a complementary explanation is that firms 1 percentage point decrease in the proportion of informal that view the government as corrupt may also place a lower employment--and lower investments in capital and tech- value on the public goods that it provides, and thus have nology. This result is consistent with the findings of lower incentives for contributing to its financing. Scarpetta and Tressel (2004) that higher labor adjustment To evaluate the effect of labor regulations on informality, costs resulting from stricter employment protection legisla- we construct a dummy variable for firms stating that those tion may lead, at least in some industries and countries, regulations significantly affected their hiring and firing to lower levels of total factor productivity. In other words, decisions during the previous year. Both for the pooled it appears that, in some specific contexts, the added sample and for three individual countries--Argentina, flexibility resulting from informality--which is one of Colombia, and Mexico--we find that firms constrained by the above-mentioned private benefits from regulatory labor regulations evade a higher fraction of taxes and/or noncompliance--can facilitate the introduction of new social security contributions. In most cases, the cost of sev- technologies and enable firms to operate more efficiently. erance payments is the aspect of labor market regulations These benefits, however, may not necessarily prevail over firms most frequently report as the biggest obstacle to hir- the various above-mentioned costs of informality, which, as ing more workers (figure 6.8). The only exception, among argued below, may lead to a negative relationship between the eight countries for which data are available, is Mexico, informality and overall productivity. where severance costs are surpassed (as an obstacle) by the The results in table 6.1 also provide some support to the costs of health insurance contributions. Health costs are hypothesis that, where market support mechanisms and second in importance to severance payments in Colombia, links with large firms are better developed, informality 166 TABLE 6.1 Firm-level correlates of sales and employment underreporting Argentina Bolivia Colombia Mexico Underreported Underreported Underreported Underreported Underreported Underreported Underreported Underreported sales employment sales employment sales employment sales employment Variable (1) (2) (3) (4) (5) (6) (7) (8) Corruption dummy 6.436 5.655 4.462 6.603 3.119 4.784 7.222 4.850 (0.000)*** (0.000)*** (0.088)* (0.033)** (0.184) (0.043)** (0.001)*** (0.021)** Labor regulations 1.957 4.793 0.791 4.659 6.771 7.607 8.358 0.731 dummy (0.267) (0.004)*** (0.799) (0.204) (0.005)*** (0.002)*** (0.006)*** (0.803) 167 Tax inspections 0.014 0.054 0.089 0.370 0.209 0.094 0.040 0.056 (average n) (0.898) (0.615) (0.701) (0.176) (0.196) (0.564) (0.644) (0.502) Contract enforcement 0.086 0.116 0.170 0.100 0.019 0.128 0.288 0.227 (average %) (0.510) (0.352) (0.348) (0.640) (0.890) (0.356) (0.033)** (0.081)* Large clients dummy 5.775 2.954 1.271 1.378 3.927 3.719 2.360 0.125 (0.001)*** (0.087)* (0.673) (0.698) (0.053)* (0.069)* (0.364) (0.960) Bank loans (% of 7.292 10.611 34.587 1.148 1.679 5.503 17.277 5.986 firms) (0.461) (0.259) (0.232) (0.973) (0.916) (0.732) (0.156) (0.611) Observations (n) 744 744 432 432 833 833 1,009 1,009 Correlation of 0.50 (0.00) 0.43 (0.00) 0.64 (0.00) 0.67 (0.00) residuals (p-value of independence test) (Continued) TABLE 6.1 Firm-level correlates of sales and employment underreporting (Continued) Panama Peru Uruguay Pooled sample Underreported Underreported Underreported Underreported Underreported Underreported Underreported Underreported sales employment sales employment sales employment sales employment Variable (9) (10) (11) (12) (13) (14) (15) (16) Corruption dummy 2.246 0.469 2.281 2.948 7.300 3.085 4.832 4.835 (0.680) (0.932) (0.234) (0.171) (0.017)** (0.311) (0.000)*** (0.000)*** Labor regulations 4.139 2.513 2.015 1.141 1.209 2.385 1.937 3.196 dummy (0.438) (0.641) (0.288) (0.593) (0.624) (0.332) (0.062)* (0.002)*** Tax Inspections 1.840 1.270 0.024 0.022 0.439 0.337 0.050 0.022 (average n) (0.007)*** (0.065)* (0.800) (0.834) (0.132) (0.246) (0.237) (0.608) Contract enforcement 0.861 0.366 0.052 0.073 0.568 0.698 0.119 0.072 168 (average %) (0.191) (0.582) (0.612) (0.528) (0.195) (0.111) (0.009)*** (0.122) Large clients dummy 2.075 4.501 2.107 4.267 0.191 0.055 2.814 2.709 (0.635) (0.308) (0.280) (0.052)* (0.949) (0.985) (0.004)*** (0.006)*** Bank loans (% of 68.822 15.790 14.453 1.006 11.987 11.923 0.594 4.351 firms) (0.240) (0.789) (0.423) (0.960) (0.572) (0.573) (0.904) (0.383) Observations (n) 417 417 575 575 268 268 4,278 4,278 Correlation of 0.55 (0.00) 0.52 (0.00) 0.60 (0.00) 0.58 (0.00) residuals (p-value of independence test) Source: AuthorsÕ calculations, using the World Bank's Enterprise Survey Database. Note: Firm size, time in business, labor productivity, formality status at time of starting up, sector, and location are also controlled for. Estimation was performed for each country, using seemingly unrelated regressions. Probability values are in parentheses. * p < .1. ** p < .05. *** p < .01. I N F O R M A L I T Y, P R O D U C T I V I T Y, A N D T H E F I R M only significant difference is obtained for manufacturing, FIGURE 6.8 which shows higher social security evasion than all other Labor regulations described by firms as biggest obstacle to hiring workers sectors. At the country level, somewhat different patterns emerge in some cases: in Argentina, tax evasion is highest in manufacturing; in Bolivia, employment underreporting Argentina is highest in services; and in Mexico, tax evasion is lowest Bolivia in services. The above cross-sector differences are obtained after controlling for firm characteristics (for example, firm Colombia size, productivity, and location, among others), as well as for such sector-specific factors as the incidence of corrup- Mexico tion and the strength of regulatory enforcement (see regres- sors in table 6.1). Thus, there must be other unobserved Panama characteristics of firms operating in those sectors that make them more likely to evade government regulations. While Paraguay we can only speculate about such unobserved factors, they could include, for instance, a greater ability to avoid gov- Peru ernment enforcement (as in the case of firms engaged in local urban transport, short-term construction contracts, or Uruguay locally distributed manufacturing goods), and competitive- ness challenges faced by previously protected industries in 0 0.2 0.4 0.6 0.8 the context of trade liberalization (as in the case of manu- Retirement benefits Health insurance facturing firms using tax evasion as a way to compete with Work hours Temporary work low-cost foreign manufacturers). Severance cost The results discussed above also suggest that the factors Source: Authors' calculations, using the World Bank's Enterprise driving the underreporting of sales and workers are very Survey Database. similar. Moreover, the firms' decisions to evade taxes and social security appear to be quite interrelated, as revealed tends to be lower. In particular, firms that operate in sectors by the fact that the residuals of the corresponding regres- and regions where the use of third-party contract enforce- sions are highly correlated (table 6.1). These findings lend ment mechanisms is more frequent have lower tax and credence to the hypothesis that firms behave strategically social security evasion rates. On average, for the pool of the when evading taxes (income taxes, value-added taxes, and seven countries for which we have data, an increase of so forth) and hiring workers off the books, and that they 10 percentage points in the fraction of firms that use third understandably seek to be consistent in the information parties to solve commercial disputes--a change of about they report to the tax and social security authorities to pro- one standard deviation in that variable--is associated with tect themselves from possible audits. Moreover, firms prob- a reduction of about 1 percentage point in the fraction of ably decide simultaneously on their levels of tax and social unreported sales. Moreover, while we do not find evidence security evasion, taking into account both tax and labor for any impact of access to financial services on informality, market regulations and enforcement. the results in table 6.1 suggest that companies that sell A note of caution is in order, however, with regard to mainly to large firms (that is, companies with more than the regressions reported in table 6.1, as their explanatory 100 employees) exhibit rates of underreporting that are power is relatively small, with R-squared statistics ranging about 3 percentage points lower. from 5 to 20 percent. Thus, while those results illustrate As for differences in tax evasion rates across sectors, we the relevance of a number of factors underlying the large find in the pooled sample that the highest rates of sales size of the informal sector in several countries of the region, underreporting are in the construction and transport sector, we are unable to explain as much as 80 percent of the tax followed by manufacturing, and then by commerce and and social security evasion behavior of the corresponding services.12 In the case of employment underreporting, the firms. 169 I N F O R M A L I T Y Impact of simplified registration microenterprises that do not have salaried employees, and and tax systems for whom registration with IMSS is not mandatory. One While the existing statistical evidence suggests that reduc- possible, albeit somewhat surprising, interpretation of the ing the time and cost required for firm registration can con- conflicting conclusions reached by Kaplan, Piedra, and tribute to increases in the number of formally registered Seira and by Bruhn is that SARE could have had a sizable firms, the magnitude of the corresponding effects is still effect on the creation of new owner-only formal businesses, subject to some controversy. In particular, both Bruhn but a much smaller impact on the formalization of existing (2006) and Kaplan, Piedra, and Seira (2006) have analyzed informal microfirms. In any case, it appears that further the effect of a Mexican program that allows firm registra- research is needed to evaluate the impact on informality of tion procedures to be completed within three days--the simplified firm registration programs, such as SARE. so-called Rapid Business Opening System (Sistema de Complementary evidence concerning the impact on Apertura Rápida de Empresas [SARE]) program imple- informality of red tape reduction programs has been mented in about 30 Mexican cities. Kaplan, Piedra, and obtained from the analysis of the Brazilian Integrated Sys- Seira find that SARE has led to statistically significant, tem for Tax and Social Security Payments for Micro and albeit quantitatively small, effects in the flow of new regis- Small Firms (Sistema Integrado de Pagamento de Impostos tered firms, which would increase by between 4 and 8 per- e Contribucoes as Microempresas e Empresas de Pequeno cent as a result of SARE, implying about two to five new Porte [SIMPLES]) program. In a manner different from firms registered and 12 to 19 new formal jobs created per SARE, however, this program combines simplified firm municipality per month. They show that the effects of registration with lower taxes and social security contribu- SARE have been concentrated in the first 10 months after tions for micro- and small enterprises, allowing for an its implementation, which, they argue, suggests that the 8 percent reduction in the overall tax burden faced by eligible impact of the program is limited to the formalization of a firms. Both Fajnzylber, Maloney, and Rojas (2006a) and small fraction of the existing stock of informal firms. Monteiro and Assunção (2006) find that, at least during the Bruhn (2006), nevertheless, reaches somewhat different year following the implementation of SIMPLES, the pro- conclusions on the impact of SARE, possibly as a result of gram led to statistically significant increases in formal reg- the use of a different data source (employment surveys as istration rates of between 6 and 13 percentage points, opposed to official administrative records) and a different depending on the sample and methodology. While further estimation technique (one based on the different timings of research may be needed to establish whether the effects of implementation of the program across Mexican cities). SIMPLES were permanent or temporary, the above evi- Bruhn focuses on the effect of SARE on the fraction of dence suggests that there is a potential for increasing registered businesses, as captured by Mexico's national microfirm formalization by combining red tape reduction employment survey. She finds that the program had a much with tax relief measures for microfirms. larger effect than the one reported by Kaplan, Piedra, and Programs to simplify and reduce tax burdens for small Seira--namely, a 5.6 percent increase in the stock of regis- contributors, including individuals and small firms, have tered businesses. This implies that SARE can be credited been implemented in recent years in a number of Latin for about 1,000 new registered firms per county, on aver- American and Caribbean countries. These countries include age. Moreover, Bruhn shows that past informal business Argentina (the Monotributo program), Bolivia (the simpli- owners are not more likely to register their businesses after fied tax regime for small firms in selected activities [RTS]), SARE, but former wage earners with conditionally high Colombia (simplified value-added regime for small contrib- wages do become more likely to open a formal business. utors), Costa Rica (RTS), Chile (simplified income tax), The fact that Bruhn's estimates imply effects that are much Ecuador (simplified value-added tax), Mexico (simplified tax larger than those obtained by Kaplan, Piedra, and Seira can regime for small contributors), Nicaragua (single payment also be attributed to the different types of businesses that system for value-added and income taxes), Honduras (sim- are covered in the databases used in each of those reports. plified sales tax), Paraguay (single tax for owner-only firms), Indeed, while Kaplan, Piedra, and Seira look at firms regis- Peru (simplified unique tax and special income tax regimes), tered with the Mexican Social Security Institute (IMSS), and Uruguay (small-enterprise tax system).13 Given the evi- Bruhn's employment data also cover the large majority of dence from the Brazilian case, it appears that most countries 170 I N F O R M A L I T Y, P R O D U C T I V I T Y, A N D T H E F I R M in the region could benefit from a revision of their simpli- of overall productivity drive a large number of firms into fied tax systems for micro- and small enterprises in order to informality, consumer demand is also likely to be directed better suit them for the objective of increasing formality-- toward products and services whose production does not for instance, by combining them with simplified firm regis- exhibit increasing returns to scale, so one should not expect tration systems, such as Mexico's SARE. large static losses from informality driven by low firm size alone. Impact of informality on firm productivity Besides potential static inefficiencies associated with the and economic growth nonexploitation of economies of scale, a parallel concern is As argued above, many lower-tier microfirms may choose that unproductive firms are able to compete with their to operate informally as a result of their limited levels of lower-cost peers by means of avoiding taxes and regula- productivity and growth potential--levels that translate tions. Thus, informal firms may be able to stay in business into a small demand for market- and government-provided despite having higher operating costs--driven, for services that have formality as a precondition. Not surpris- instance, by lower levels of entrepreneurial ability. This ingly, when deciding to formalize, they are often more could slow down the creative destruction process by which motivated by sticks--avoiding fines and bribes--than by innovative, high-productivity firms expand to the detri- carrots like access to credit or formal contract enforcement. ment of less-productive ones. In other words, to the extent This, however, does not mean that increasing formality that increases in regulatory enforcement drive out of busi- does not in itself have the potential for increasing overall ness a large number of firms that self-selected into infor- productivity, through both static and dynamic channels--a mality because of having lower productivity than firms of possibility that we review next, both conceptually and from the same size operating in the same sectors, one could an empirical point of view. expect potentially large negative effects on aggregate pro- ductivity. One caveat to this argument, however, is that, as Static versus dynamic effects shown by Almeida and Carneiro (2005) for the case of labor To the extent that informality is associated with a prepon- regulations, informality may allow firms greater flexibility derance of small firms, there is a concern that it could lead in their employment and production decisions, which, in to considerable efficiency losses. This prediction, however, turn, could lead them to operate more efficiently. Whether depends on whether returns to scale are constant or increas- this effect dominates other factors that could lead to a neg- ing. As reviewed by Tybout (2000), the literature on the ative link between informality and firm productivity--for subject is divided between simulation studies, which often example, the self-selection of unproductive firms into infor- assume decreasing average costs, and survey-based esti- mality, the incentives to operate at a small scale to avoid mates, which generally suggest that the benefits from detection, and the inability to gain access to factor and increasing plant size are relatively small. Thus, while one- product markets--is a question that only empirical evi- person establishments are usually found to be less efficient dence can help resolve. than firms with at least some employees, returns to scale A parallel concern is that high levels of informality could among the latter firms are very close to unity and, at most, also have negative consequences on the incentives of formal mildly increasing. firms to innovate and adopt new technologies, which also This is not to say that increasing returns to scale are not the could reduce overall productivity growth. Several studies have norm in some specific industries, notably the most capital- emphasized these potentially negative dynamic implications intensive ones. However, because of their limited access to of informality. Thus, Capp, Elstrodt, and Jones (2005), capital and skilled labor, and to avoid cost disadvantages, Elstrodt, Lenero, and Urdapilleta (2002), Farrell (2004), micro- and small firms tend to locate in industries where effi- Kenyon and Kapaz (2005), and Palmade (2005) see informal- ciency losses associated with low scale production are limited. ity as one of the main causes for the gap in productivity levels This is facilitated by the fact that demand for such products is between developed and developing countries, inducing dis- negatively correlated with countries' per capita incomes, as tortions in investment decisions and limiting the growth Engel effects direct consumer demand toward simpler prod- potential of the corresponding economies. Studies performed ucts that can be efficiently produced with labor-intensive by the McKinsey Global Institute suggest that informality technologies. In other words, in countries where lower levels accounts for around 50 percent of the productivity gap 171 I N F O R M A L I T Y between countries like Turkey, Portugal, and the United effect of impeding the ability of the frontier quality pro- States; and for 30 percent of the productivity gap between ducer to set a price that would force all other producers--of Brazil and the United States (Farrell 2004). lower qualities--to leave the market. The impact on The main argument proposed by these studies is that research and development investments--thereby including firms that join the informal sector tend to become "trapped the expenditures involved in adapting foreign technologies in a self-reinforcing dynamic that confines them to subscale, to local conditions--is, however, ambiguous. Indeed, while inefficient, low-productivity work" (Farrell 2004, p. 30). informality decreases the market power and profit levels of Moreover, informality also has negative effects on the frontier producers, it also increases the life span of frontier investment decisions of formal firms, as it reduces their products by augmenting the quality improvements that are market share and profitability. The relative cost advantages needed to debunk current market leaders. As a result, invest- enjoyed by informal firms as a result of not paying taxes and ments in research and development--and growth--could not incurring the costs of regulatory compliance, however, increase or decrease, depending on which effect dominates. allow them to stay in business despite their low productiv- ity, which, as mentioned above, could distort competition Aggregate growth effects and limit the process of creative destruction. Moreover, as Despite the widespread belief that a large informal economy more productive firms also have fewer incentives to invest in hurts economic growth, cross-country comparisons do not innovation and technology adoption, the McKinsey studies find a robust association between informality and growth. (reviewed in Farrell [2004]) suggest that informality leads Figure 6.9 shows the estimated impact of informality on to an overall reduction in economic growth. growth, using two diverse informality indicators: self- One caveat to these arguments is that, from a theoretical employment and the Schneider (2005) estimates of the pro- point of view, technology adoption and innovation could portion of GDP produced by the underground economy in either decrease or increase as a result of unfair competition the period 1999­2000. In accordance with common wisdom, by informal firms. In particular, as argued by Cunha under both indicators informality appears to have a negative (2006), if technological change takes the form of the impact on growth: on average, the regressions suggest that discovery--or introduction into the country--of improved decreasing informal economic activity by 10 percent of GDP qualities for intermediary goods, informality may have the is associated with 0.6 percent higher economic growth. FIGURE 6.9 Estimated impact of informality on growth Informality - % of GDP Self-employment Without controls Controlling for initial education Controlling for regulation Controlling for financial depth Controlling for investment Controlling for corruption All controls 0.10 0.05 0 0.05 0.10 0.05 0 0.05 Estimated impact 95% confidence interval Sources: Authors' calculations from Barro and Lee 2000; Fraser Institute 2005; ILO 2005; Schneider 2005; and World Bank 2005. Note: The dependent variable is average growth of GDP per capita during the period 1990­2003. All regressions include a constant, regional dummies, and initial log of GDP per capita. Education represents the average years of education of the population over age 25. Regulation is an index of regulatory efficiency from the Fraser Institute. Financial depth represents the log of private domestic credit over GDP in 1990. 172 I N F O R M A L I T Y, P R O D U C T I V I T Y, A N D T H E F I R M Nonetheless, the estimated coefficients tend not to be surprisingly large in developing countries, such as Chile robust. In none of the presented regressions is the coefficient and Colombia. These results, however, do not necessarily associated with self-employment significantly different from constitute evidence that the creative destruction process, zero at the 5 percent level. To be sure, when we use the macro- initially described by Joseph Schumpeter, is alive and well economic estimates of informality, the basic regression with- in those countries, as the observed high turnover rates could out added explanatory variables shows a negative and be the result of high macroeconomic instability and of the statistically significant association between informality and preponderance of small firms--which have inherently growth. However, the relationship loses its significance when higher rates of failure. Thus, when corrected by volatility, we control for education, financial depth, or corruption.14 turnover rates are not larger in Latin America and the Thus, while previous studies have found a negative relation- Caribbean than in industrial countries. Moreover, while ship between informality and growth, they have relied on a there is evidence that entrants in Chile and Colombia are very narrow pool of observations (Loayza 1996) or have not slightly more efficient than incumbents, and that new low- controlledforrelevantcorrelatesofgrowth,suchasregulation, productivity firms tend to go out of business rapidly, the human capital, and initial GDP per capita (Schneider and impact of this process on overall productivity is found to be Klinglmair 2004). It is worth noting, however, that the fact relatively small. that we do not observe large effects of informality on growth after controlling for other standard growth determinants may Firm-level effects of informality just reflect our inability to empirically distinguish between As evidenced in figure 6.10, firms that report having thedirecteffectsofthosevariablesandindirecteffectsthrough started operations without formally registering--at least informality.Inotherwords,oneofthechannelsthroughwhich initially--exhibit, on average, much lower levels of output some of those standard drivers of growth operate could be per worker, even after controlling for firm size, time in increasing informality--for example, low human capital business, sector, and region. In other words, those that start reducing the opportunity cost of self-employment, or up informally are clearly at the bottom end of the produc- corruption diminishing the incentives to comply with tivity distribution of the corresponding industries and regulations--but this would not be apparent in the results regions. The difference in labor productivity between those reported in figure 6.9. firms and the ones that have always operated formally is 29 percent, on average, for the seven Latin American and Empirical evidence on creative destruction Caribbean countries analyzed here. The effects are largest in Latin America and the Caribbean region The arguments on the negative dynamic effects of infor- mality offered above should lead us to expect a less vibrant FIGURE 6.10 process of industrial evolution in developing countries than Estimated impact of informality on labor productivity (%) in industrial countries--a process that should be reflected in higher productivity dispersion and lower firm turnover. Argentina The evidence, however, suggests that the average distance Bolivia to "frontier" production technologies is similar in studies of Mexico developing and OECD countries, with average technical effi- Panama ciency levels equivalent to about 60­70 percent of the corre- Peru sponding best practices (Tybout 2000). These measures, it Pooled sample must be noted, are usually based on data that exclude microenterprises and low-productivity, owner-only firms, so 50 40 30 20 10 0 Labor productivity they probably underestimate productivity dispersion. As for the evidence on the rates of firm and job creation Source: Authors' calculations, using the World Bank's Enterprise Survey Database. and destruction, the evidence on whether it is lower or Note: The figure reports coefficients on formality indicators in regressions of the log of output per worker on a dummy for not similar in least-developed countries is ambiguous. Roberts having formally registered at the time of starting up, controlling and Tybout (1997) show that rates of firm turnover and for employment size, time in business, and sector and region dummies. market shares of recent entrants into the formal sector are 173 I N F O R M A L I T Y in Peru (50 percent), and, while lower in the remaining FIGURE 6.11 countries, they are statistically significant in four out of Effects of a 10 percent increase in tax and social security evasion seven countries, the exceptions being Uruguay, Panama, at the industry region level on individual firm productivity and Colombia (not shown).15 (pooled sample) Registered firms that report having started informally are only 6.6 percent of all formal firms (in the pooled sam- Labor ple), and respond for just 4.3 percent of total sales and productivity 3.6 percent of employment. Thus, the increase in aggregate productivity derived from hypothetically excluding them Total factor from their sectors--for example, by a stricter enforcement productivity of entry regulations--would be relatively small: about 0.8 percent, on average.16 However, to the extent that labor 12 10 8 6 4 2 0 productivity is higher in registered firms that started up Tax evasion Social security evasion informally but eventually registered, than in similar firms that started informally but never registered, the potential Source: Authors' calculations, using the World Bank's Enterprise impact on productivity of fully enforcing entry regulations Survey Database. Note: The figure reports coefficients on formality indicators in could be much larger. Thus, for instance, in a hypothet- regressions that control for employment size, time in business, fixed capital stocks (in TFP regressions), and sector and region dummies, ical scenario where all nonregistered informal Mexican using a pooled sample of seven countries. microfirms with no more than five employees (which repre- sent about 21 percent of total employment in that country) were to go out of business due to stricter enforcement of our informality measure is aggregated at the industry and entry regulations, and assuming that employment would region levels so it could hardly be affected by individual shift to firms with a 35.5 percent productivity advantage firm productivity. The results suggest that there are signifi- (based on the estimate for Mexico reported in figure 6.10, cant potential productivity gains to be obtained from which we consider a lower bound for the productivity dif- increased enforcement of tax and social security regulations. ferentials between formal and informal firms), the resulting Further evidence on the presence of a causal link impact on aggregate productivity could be as large as between informality and firm performance has been 6 percent.17 Off course, these are very imprecise, back-of- obtained using microenterprise survey data. In the case of the-envelope calculations that are aimed only at illustrat- Mexico, Fajnzylber, Maloney, and Rojas (2006c) show that ing a much more general point--namely, the possibility microfirms that report paying taxes exhibit higher levels of that informality is associated with lower levels of aggregate profit, even after controlling for employment size and capi- productivity, which should be considered in any analysis of tal stocks. This result is robust to the use of estimation its social costs and benefits. techniques that control for the impact of unobserved per- Informality, however, also takes the form of tax and social sonal and firm characteristics--for example, managerial security evasion among registered firms. In this respect, it ability--that could affect both the decision to formalize would be useful to know what would be the impact of mar- the firm and its performance. Quantitatively, we find that ginal reductions in tax evasion on firm and aggregate pro- firms that pay taxes exhibit between 15 and 60 percent ductivity. With that purpose, we estimate the effect of higher productivity levels, depending on the estimation average tax and social security evasion in a given sector and method and the performance variable used--either firm region on the level of productivity of individual firms oper- profits as reported in detailed microfirm surveys or self- ating in the corresponding areas. The results are presented employment income as reported in the Mexican Employ- in figure 6.11 for our pooled sample of seven countries. We ment Survey (figure 6.12). In addition, there is evidence that find that each 10 percent increase in average evasion rates is owners of formal firms are less likely to go out of business. associated with reductions in labor and total factor produc- One of the channels through which formality could tivity of 7 and 10 percent, respectively. These effects are not increase firm performance is by facilitating access to factor subject to the criticism of a possible reverse causality from and product markets. Cull, McKenzie, and Woodruff (2007) low firm productivity to higher firm informality. Indeed, show that access to credit among small Mexican retail firms 174 I N F O R M A L I T Y, P R O D U C T I V I T Y, A N D T H E F I R M FIGURE 6.12 FIGURE 6.13 Productivity difference between tax-paying Mexican microfirms Productivity and paid employment effects of exogenous increase in and informal firms with similar characteristics formality driven by Brazil's SIMPLES program Control Labor function productivity Ordinary Paid least squares employment Propensity Total factor score matching productivity 0 0.2 0.4 0.6 0 0.1 0.2 0.3 0.4 0.5 Firm profits Self-employment income OLS Matching Source: Fajnzylber, Maloney, and Rojas 2006c. Source: Fajnzylber, Maloney, and Rojas 2006a. is much higher among formally registered firms. While this estimated to be associated with a 5 percent increase in paid correlation could reflect the presence of unobserved personal employment, a 15 percent boost in total factor productivity, and firm traits linked to both formality and access to and a 35 percent increase in labor productivity (figure 6.13). finance, it is also consistent with informality reducing At least in the Brazilian context, however, only a small frac- microfirm access to formal loans. Thus, for instance, if tion of the revenue-increasing effects associated with for- microlenders make formal registration a requisite for grant- mality can be attributed to increased access to credit ing loans, formalization could allow credit-constrained markets and government-provided technical assistance. In microfirms to exploit the sizable returns to investment that contrast, the greater willingness of formal firms to operate have been estimated by Cull and coauthors--between 20 out of a fixed locale is responsible for as much as 50 percent and 33 percent per month for investments of about $140, of the increase in revenues among formal microfirms and for using data from a randomized experiment. a third of the corresponding total factor productivity Evidence consistent with some of the findings for Mexico increase. has also been obtained for Brazil on the basis of exogenous increases in formality rates associated with the introduction Conclusions of a simplified tax and registration system for micro- and The empirical evidence on aggregate negative growth small firms (the above-mentioned SIMPLES). Indeed, effects of informality is not conclusive, as informality tends econometric estimates that take advantage of changes in the to lose significance when other standard growth determi- incentives to formalize, introduced by this program, show nants are controlled for. This, however, could be due to the that it significantly increased access to credit among eligible fact that many of the standard drivers of growth are also firms, and altered the amount and composition of invest- likely to affect informality--for example, low levels of ment toward larger and longer-term projects (Monteiro and human capital or institutional quality leading to both Assunção 2006). Moreover, Fajnzylber, Maloney, and Rojas lower growth and higher informality--and it is difficult to (2006a) show that increases in formality driven by SIMPLES separate their direct growth effects from those that operate are associated with a higher use of paid labor, higher levels through larger informal sectors. of capital intensity, and increased labor and total factor pro- The microeconomic empirical evidence, on the other ductivity. In particular, increases in the rates of microfirm hand, is still quite limited due to the econometric registration that can be attributed to SIMPLES are difficulties associated with distinguishing the effects of low 175 I N F O R M A L I T Y productivity on informality (see chapter 5) from the reverse costs associated with operating legitimate businesses. effects operating from informality to productivity. How- Internet-based technologies and one-stop shops can be ever, the available evidence suggests that considerable effective tools to implement such programs, although their efficiency gains could be derived from the transfer of pro- effectiveness can be greatly increased if they are used in duction from low-productivity informal firms to their more conjunction with comprehensive reviews and revisions of productive formal peers. Similarly, the evidence indicates existing administrative processes. that the concerns associated with possible negative exter- We argue, however, that although eliminating unneces- nalities generated by high levels of tax and social security sary regulations and reducing excessive red tape could con- evasion could be well justified, as firms operating in indus- tribute to reducing the size of the informal sector, those tries and regions characterized by high levels of sales and actions should not be the exclusive focus of policy makers employment underreporting exhibit lower levels of labor engaged in attaining that objective. Indeed, the costs of and total factor productivity. Moreover, there is evidence regulatory compliance are only one among other factors indicating that exogenous increases in formality are associ- that may affect formality decisions. In particular, to attract ated with better firm performance, which should, in princi- more businesses into the formal economy, it is crucial to ple, translate into higher rates of economic growth. increase the potential benefits of regulatory compliance. In this context, the interest of policy makers and develop- This implies facilitating the ability of micro- and small ment practitioners in designing policies and programs to enterprises to tap into formal credit markets and improv- facilitate the formalization of small businesses and increase ing the provision of business development and training regulatory compliance by larger firms appears well justi- services available to formal firms. Moreover, it is important fied. In particular, the fact that efforts to decrease regula- to facilitate access to product markets through public tory burdens have recently become very popular does not procurement opportunities and supplier development pro- come as a surprise. Burdensome regulations and costly grams aimed at increasing links with larger private firms. bureaucratic requirements are indeed an important deter- Other ways of increasing the benefits of formality include minant of informality that may create barriers for increas- improving the quality of legal services available to small ing formal entrepreneurial activity. One challenge that businesses and creating mechanisms to provide information governments face in this respect is that of assessing their to entrepreneurs wishing to formalize their businesses, existing and new regulations to determine the extent to thereby encompassing advisory services on taxes and regu- which they are justified by public interests, associated, for lations, as well as information about financial and nonfinan- instance, with the protection of public safety or the envi- cial services available to them. ronment. Dealing with this challenge may require compre- Overall, a wider and integrated approach appears to be hensive regulatory assessments aimed at distinguishing necessary to switch the incentives of a large fraction of relevant from anachronistic regulations, as well as at identi- informal firms in the direction of formality. Such an fying those regulations that reflect private rather than pub- approach would likely have to combine both carrots (for lic interests, and that could represent important barriers to example, lower costs of formalization, better and more effi- formalization. Examples of such initiatives include national cient government services, and higher access to market- regulatory reviews, such as those implemented in leading and government-provided services for formal firms) with transition economies.18 sticks (such as increasing government enforcement of regu- Similarly, many developing countries are now engaged lations and the expected cost of being caught). Moreover, as in reducing the time and cost needed by businesses to argued later in this volume, it is crucial that both the obtain various government-issued permits and licenses.19 enactment and the enforcement of regulations are perceived Indeed, even well-designed and legitimate regulations may to be fair, as this is vital for maintaining "tax morale" and create barriers to formalization if they are badly enforced or increasing regulatory compliance. administered, creating excessive costs and uncertainty for The correct mix of policies, however, is likely to vary private businesses. Thus, a complementary approach to across countries and over time, depending on the relative reduce firm informality--to be pursued in parallel to regu- importance of the various determinants of informality. latory reforms--is the implementation of administrative Moreover, other aspects of public policy should be taken into simplification programs aimed at reducing the transaction account, including those related to the social consequences 176 I N F O R M A L I T Y, P R O D U C T I V I T Y, A N D T H E F I R M of drastically reducing the size of the informal sector, which, Risk Guide's index of law and order; and the ratio of government at least in Latin America, is currently responsible for a large expenditures to GDP. 10. The index of product market regulations is a composite of fraction of employment and income-generating opportuni- indexes of regulations in the areas of firm entry, trade barriers, finan- ties for poor households. In other words, policies aimed at cial markets, contract enforcement, and bankruptcy. reducing firm informality should be considered in conjunc- 11. A related point made by Schneider and Enste (2000) is that tion with the labor market and social protection issues asso- complex tax systems can make legal tax avoidance in the official econ- ciated with the possibility that large contingents of omy more profitable, and thus create disincentives for informality. As previously informal workers would have to shift to other a result, fiscal reforms that combine lower tax rates and simpler tax systems could not necessarily lead to smaller informal sectors; the segments of the labor market. Austrian 1989 reform is mentioned as an example. 12. Within manufacturing, the food and beverage industry Notes exhibits the highest levels of evasion, even after controlling for firm 1. As in the study by Djankov et al. (2003), we leave outside the and sector characteristics. scope of this chapter those "underground enterprises" that are 13. See González (2006) for additional details. devoted to criminal activities. 14. The relationship remains weak when changing the reference 2. Since firms are understandably reluctant to reveal information period from 1990­2003 to 1999­2003 or when using five-year aver- regarding tax and social security evasion, as well as informal pay- ages for the period 1980­2004. As the macroeconomic estimates of ments to corrupt officials, the corresponding survey questions are informality are cross-sectional, regressions with five-year averages can phrased in terms of the practices of "typical firms in this establish- only be performed with self-employment. ment's line of business." This is a standard approach taken to measure 15. It is worth noting that the estimated effects for most countries the prevalence of corruption used, for instance, by Johnson et al. are based on a sizable number of initially informal firms: between 5 (2000). Note also that figure 6.1 reports simple averages across firms. and 14 percent of all firms, or 355 in the pooled sample, 72 in 3. The high levels of tax evasion for Brazil are, to some extent, Argentina, 72 in Bolivia, 66 in Mexico, and 104 in Colombia. Our puzzling as tax revenues in that country have increased considerably estimates are arguably weaker in Panama, Peru, and Uruguay, where in recent years, reaching 34 percent of gross domestic product (GDP) there are, respectively, 9, 20, and 12 of those firms (between 2 and in 2002, driven partly by improvements in tax administration effi- 3 percent of all firms). ciency. However, respectively, 40 and 25 percent of federal tax rev- 16. This is assuming that the employment share of the initially enues come from indirect cascading taxes and from social security informal firms is taken over by competitors, which, per our estima- contributions and other payroll taxes that firms, as suggested by the tions, have a 29 percent productivity advantage (figure 6.9). survey data, appear to be quite successful in evading. 17. This assumption is subject to the criticism that, as argued in 4. The controls are those described in table 6.1. the beginning of this section, informal microfirms tend to locate in 5. De Soto (1989) is the seminal reference on the links between industries where efficiency losses associated with low scale produc- government regulations and informality. In their survey on the topic, tion are minimized. Moreover, note that we are assuming that all Schneider and Enste (2000) list taxes, social security contributions, those working in informal microfirms would find employment in for- and the intensity of regulations (including those of labor markets) mal firms. We base the estimated employment share of informal among the top causes for growth in the "shadow economy." Loayza microfirms on figures on registration rates from the Encuesta Nacional (1996) offers a theoretical model illustrating those effects and pro- de Micronegocios--27.5 percent for owner-only firms and 62.7 percent vides supporting empirical evidence for Latin America. Additional for firms with two to five workers--and on Encuesta Nacional de cross-country evidence on the links between tax and regulatory Empleo Urbano­based estimates of the employment shares of firms in burdens and the size of the informal economy is provided by Botero those size ranges--respectively, 15.3 and 27.5 percent. et al. (2003), Djankov et al. (2002), and Loayza, Oviedo, and Servén 18. See Djankov et al. (2003) for a description of the 1995­98 (2005). Hungarian regulatory review. 6. The surveys were done in the context of municipal adminis- 19. See the World Bank's Doing Business reports for global trative simplification projects supported by the International Finance reviews of country-level reforms and benchmarking exercises. Corporation. 7. In the theoretical model proposed by Sarte (2000), for instance, changes in the fixed cost of entering the formal sector do References not affect the level of informality--their effect is nonbinding--when Almeida, R., and P. Carneiro. 2005. "Enforcement of Labor Regula- the cost of operating informally is relatively low. tion, Informal Labor, and Firm Performance." Policy Research 8. These databases have been expanded and updated annually Working Paper 3756, World Bank, Washington, DC. through the World Bank's Doing Business project. Barro, R., and J.-W. Lee. 2000. "International Data on Educational 9. Loayza and Rigolini (2006) use the following empirical mea- Attainment: Updates and Implications." Working Paper 42, sures for the above-mentioned variables: the Fraser Institute's index Center for International Development, Harvard University, of credit, labor, and regulatory flexibility; the International Country Cambridge, MA. 177 I N F O R M A L I T Y Botero, J., S. Djankov, R. La Porta, F. Lopez-de-Silanes, and Johnson, S., D. Kaufmann, and A. Shleifer. 1997. "The Unofficial Econ- A. Shleifer. 2003. "The Regulation of Labor." Quarterly Journal of omy in Transition." Brookings Papers on Economic Activity 2: 159­239. Economics 119 (4): 1339­82. Kaplan, D., E. Piedra, and E. Seira. 2006. "Are Burdensome Regis- Bruhn, M. 2006. "License to Sell: The Effect of Business Registration tration Procedures an Important Barrier on Firm Creation? Evi- Reform on Entrepreneurial Activity in Mexico." Photocopy. dence from Mexico." Photocopy. Instituto Tecnológico Autónomo Massachusetts Institute of Technology, Cambridge, MA. de México, Mexico City. Capp, J., H. Elstrodt, and W. Jones, Jr. 2005. "Reining In Brazil's Kenyon, T., and E. Kapaz. 2005. "Rising Informality: Reversing the Informal Economy." McKinsey Quarterly. Available at http://www. Tide." Viewpoint Series Note 301, Private Sector Development, mckinseyquarterly.com. World Bank, Washington, DC. CIEN (Centro de Investigaciones Económicas Nacionales). 2006. Levenson, A., and W. Maloney. 1998. "The Informal Sector, Firm Economía Informal: Superando las barreras de un Estado excluyente. Dynamics and Institutional Participation." Policy Research Guatemala City, Guatemala. Working Paper 1988, World Bank, Washington, DC. Cull, R., D. McKenzie, and C. Woodruff. 2007. "Experimental Evi- Loayza, N. 1996. "The Economics of the Informal Sector: A Simple dence on Returns to Capital and Access to Finance in Mexico." Model and Some Evidence from Latin America." Carnegie-Rochester Photocopy. World Bank, Washington, DC. Conference Series on Public Policy 45: 129­62. Cunha, B. 2006. "Informality, Productivity and Growth." Photo- Loayza, N., A. M. Oviedo, and L. Servén 2005. "The Impact of Regu- copy. World Bank, Washington, DC. lation on Growth and Informality--Cross-country Evidence." Pol- de Soto, H. 1989. The Other Path: The Invisible Revolution in the Third icy Research Working Paper 3263, World Bank, Washington, DC. World. New York: Basic Books. Loayza, N., and J. Rigolini. 2006. "Informality Trends and Cycles." Djankov, S., R. La Porta, F. Lopez-de-Silanes, and A. Shleifer. 2002. Policy Research Working Paper 4078, World Bank, Washington, "The Regulation of Entry." Quarterly Journal of Economics 117 (1): DC. 1­37. McKenzie, D. J., and C. Woodruff. 2006. "Do Entry Costs Provide Djankov, S., I. Lieberman, J. Mukherjee, and T. Nenova. 2003. an Empirical Basis for Poverty Traps?" Economic Development and "Going Informal: Benefits and Costs." In The Informal Economy in Cultural Change 55 (1): 3­42. the EU Accession Countries, ed. B. Belev. Sofia, Bulgaria: Center for Monteiro, J. C. M., and J. J. Assunção. 2006. "Outgoing the Shadows: the Study of Democracy. Estimating the Impact of Bureaucracy Simplification and Tax Cut Elstrodt, H., P. Lenero, and E. Urdapilleta. 2002. "Micro Lessons on Formality and Investment." Photocopy. Department of Eco- from Argentina." McKinsey Quarterly 2: 1­8. nomics, Pontifícia Universidade Católica do Rio de Janeiro, Brazil. Fajnzylber, P., W. F. Maloney, and G. V. Montes Rojas. 2006a. "Does Palmade, V. 2005. "Rising Informality." Viewpoint Series Note 298, Formality Improve Microfirm Performance? Quasi-experimental Private Sector Development, World Bank, Washington, DC. Evidence from the Brazilian SIMPLES Program." Photocopy. Rauch, J. E. 1991. "Modelling the Informal Sector Formally." Journal World Bank, Washington, DC. of Development Economics 35: 33­47. ------. 2006b. "Microenterprise Dynamics in Developing Countries: Roberts, M. J., and J. R. Tybout, eds. 1997. Industrial Evolution in How Similar Are They to Those in the Industrialized World? Evi- Developing Countries: Micro Patterns of Turnover, Productivity, and dence from Mexico." World Bank Economic Review 20 (3): 389­419. Market Structure. Washington, DC: World Bank. ------. 2006c. "Releasing Constraints to Growth or Pushing on a Sarte, P.-D. G. 2000. "Informality and Rent-seeking Bureaucracies in String? The Impact of Credit, Training, Business Associations a Model of Long-run Growth." Journal of Monetary Economics 46 (1): and Taxes on the Performance of Mexican Microfirms." Policy 173­97. Research Working Paper 3807, World Bank, Washington, DC. Scarpetta, S., and T. Tressel. 2004. "Boosting Productivity via Inno- Farrell, D. 2004. "The Hidden Dangers of Informal Economy." vation and Adoptions of New Technologies: Any Role for Labor McKinsey Quarterly 3: 27­37. Market Institutions?" Policy Research Working Paper 3273, Fraser Institute. 2005. "Economic Freedom of the World: Annual World Bank, Washington, DC. Report." Available at http://www.fraserinstitute.ca. Schneider, F. 2005. "Shadow Economies around the World: What Do Friedman, E., S. Johnson, D. Kaufmann, and P. Zoido-Lobatón. We Really Know?" European Journal of Political Economy 21 (3): 2000. "Dodging the Grabbing Hand: The Determinants of 598­642. Unofficial Activity in 69 Countries." Journal of Public Economics Schneider, F., and D. H. Enste. 2000. "Shadow Economies: Size, Causes, 76: 459­93. and Consequences." Journal of Economic Literature 38 (1): 77­114. González, D. 2006. "Regimenes especiales de tributación para Schneider, F. G., and R. Klinglmair. 2004. "Shadow Economies pequeños contribuyentes en América Latina." Photocopy. Inter- around the World: What Do We Really Know?" Working Paper American Development Bank, Washington, DC. 1167, Center for Economic Studies and the Ifo Institute, Munich. ILO (International Labour Organization). 2005. Labor Statistics. Tybout, J. 2000. "Manufacturing Firms in Developing Countries: Available at http://laborsta.ilo.org/. How Well Do They Do, and Why?" Journal of Economic Literature Johnson, S., D. Kaufmann, J. McMillan, and C. Woodruff. 2000. 38 (1): 11­44. "Why Do Firms Hide? Bribes and Unofficial Activity after Com- World Bank. 2005. World Development Indicators. Available at http:// munism." Journal of Public Economics 76: 495­520. www.worldbank.org. 178 CHAPTER 7 Informality, Social Protection, and Antipoverty Policies SUMMARY: While formal sector workers often have access to generous social security packages, informal sector workers generally have more limited access to risk management instruments. In response to concerns about large gaps in coverage, several countries have launched or expanded noncontributory assistance programs to help breach the coverage gap, especially among the poor. A clear rationale for public intervention exists; however, weaknesses in the design and imple- mentation of many programs actually exacerbate people's lack of access to protection and create further incentives for infor- mality. Providing effective social protection for the citizens of the region will thus require a reengineering of many countries' existing programs. In the long run, this should include providing universal "essential cover" in health, de- linked from the labor contract and financed by general taxation; it should also include "poverty prevention" pensions aimed at the elderly poor combined, in an incentive compatible way, with individual savings as the mainstay of life-cycle consumption smoothing. Successful implementation of this agenda will, in most countries, require a number of short-to- medium-term measures that increasingly improve the efficiency and design of countries' social protection systems. T HE PREVIOUS CHAPTERS HAVE FOCUSED to ensure that programs are incentive-compatible with pro- on characterizing the informal sector of ductive employment, as well as with workers' and their the labor market in Latin America and the families' need for protection against health shocks, poverty Caribbean, as well as those people who work in old age, and other debilitating risks. in it. In characterizing informality, one core definition relates to the absence of worker coverage by tradi- Informality and social protection: Why policy tional social security programs--most notably health insur- makers should care ance and pensions, but often also other benefits available to Social protection has been defined in a number of different workers by virtue of their labor contract. This chapter ways in the literature on social policy and poverty reduction. focuses on the "flip side" of the informality coin from the Most commonly, however, social protection is defined as a perspective of the region's workers, that is, access to social range of measures adopted by governments to help people protection. Specifically, the chapter examines the state of manage risk more effectively--whether in the form of pro- workers' access to social protection in the region; the role of moting basic income security, protecting people from unan- public social protection--or social risk management-- ticipated shocks (such as family health shocks or economic mechanisms; and how deficiencies in the design of social downturns), developing and protecting the human capital of protection programs contribute to the persistence of infor- society's poorest members to strengthen their ability to mal employment, as well to as a larger failure to protect the prosper in the labor market, and/or ensuring basic service most vulnerable members of society. It then discusses how access to those outside the reach of traditional government the region's social protection systems may be reengineered (or private sector) programs. Social protection programs 179 I N F O R M A L I T Y typically focus on helping people prevent, mitigate, or cope actors (public and private sectors, employers and employees) with a range of risks. In doing so, programs that are well have contributed to social security programs and have, in designed and successfully implemented can contribute not return, been covered by relatively generous, multidimen- only to household welfare, but also to long-term economic sional benefit packages--often including health insurance, productivity, growth, and development. Social protection is old-age pensions, disability and workers' insurance, and, in typically delivered via a range of social insurance and/or some cases, housing, child care, and sports and recreation social assistance programs (box 7.1). benefits. At the same time, those outside the formal sector, Nearly all Latin American countries are characterized by both in urban and rural areas, have had much more limited what have been called "truncated welfare systems." These are access to formal risk management instruments or other gov- social protection systems in which, historically, formal sector ernment benefits. BOX 7.1 Social protection--strengthening people's abilities to manage risk and promoting long-term productivity, growth, and development People in developing countries face a range of risks. Some and families deal with temporary or chronic poverty at risks, such as economic recessions, harvest losses, natural different stages of their life cycle, and/or strengthen their disasters, and wars, affect whole societies or large groups. capacities to achieve higher standards of living. Others, such as the illness of family members, loss of Well-designed public social protection programs can household breadwinners' jobs, and crime, may affect only compensate for missing insurance markets or other pri- individual households. Social protection is defined as a range vate risk-mitigation instruments, and thus they create of public interventions that support society's poorest and opportunities among the poor for more productive most vulnerable members and help individuals, families, investments and higher incomes. Moreover, certain types and communities better manage risks by helping them of safety net programs--such as "conditional transfers" in prevent, mitigate, or cope better with adverse events. which payments are made contingent upon family invest- Public social protection can comprise a range of mecha- ments in children's health or schooling--both provide nisms, including regulation, government financing, short-term income support and strengthen longer-term direct provision of services, or provision of conditional or investments in children's education, health, well-being, unconditional transfer payments. Intended to augment, and productivity. not replace, family, community, and market-based risk Other policies and institutions also play important management mechanisms, such interventions comple- roles in social risk management. Labor market policies ment national economic policies and support strategies for and institutions play a critical role--by influencing the poverty reduction and human development. nature and extent of the risks workers face; by providing Public social protection measures in Latin America the framework in which certain programs, such as pen- and the Caribbean often are categorized into two main sions, health insurance, or workfare are accessed; and by groups: social insurance and social assistance. Social insur- providing opportunities for skills development or techni- ance includes a range of contributory programs, includ- cal training so that people can find more remunerative ing old-age pensions, health insurance, disability, and employment. Public health systems in developing coun- professional risk insurance, that are intended to help tries often perform a safety net function, providing subsi- cushion the impact of shocks affecting income/earnings, dized health services to the poor and others who lack health, and employment, and thereby prevent families health insurance. And where financial institutions func- from falling into poverty. These programs generally fall tion well, access to financial services can also contribute in under the heading of social security. Social assistance important ways to households' ability to manage risk, as a includes a variety of noncontributory safety net pro- complement to public social protection programs. grams, such as workfare, assistance to the disabled and indigent, and cash transfers, all of which help individuals Sources: de Ferranti et al. 2000; World Bank 2001, 2003b. 180 I N F O R M A L I T Y, S O C I A L P R O T E C T I O N , A N D A N T I P O V E RT Y P O L I C I E S While some workers, on having done an implicit or holds and firms), and even at the level of aggregate explicit cost­benefit analysis, voluntarily opt out of the sys- output (see, for example, Ravallion 2003). tem of social security, others may not have that choice. For · Failure to cover losses due to health or income shocks example, by virtue of residing in less favorable geographic can impose external costs on society. Indeed, the exis- locations (such as poor or remote rural areas), or lacking tence of externalities, along with other market fail- opportunities to work in the formal sector (their employers ures (for example, information failures), provides a may not offer benefits), many do not have access to formal powerful rationale for public intervention to improve risk management instruments. Still others--those who societal outcomes. through the course of their working lives frequently move between formal and informal employment--may find they Finally, from the perspective of the social protection system, are not eligible for some benefits during key stages in their the truncated welfare systems of Latin America and the life cycles due to program design features or eligibility Caribbean have tended to result in small, inefficient risk requirements. Historically, there has been a strong correla- pools and forgone savings, due to the absence of scale tion between income and social security coverage and, con- economies. comitantly, between poverty status and lack of access to Partly in response to concerns about the truncated wel- formal risk management instruments. fare state and partly due to political pressures associated There is ample evidence that individuals and families with democratization, several countries in the region have employ a number of strategies to manage risk--with vary- launched or expanded noncontributory assistance and/or ing degrees of success. If so, is lack of access to formal social targeted poverty reduction programs over the last decade to protection programs really a problem? The evidence, which help breach the coverage gap. These programs have often will be elaborated upon in this chapter, indicates "yes," on contributed in important ways to the welfare of the poor. At at least three fronts. First, from the perspective of household the same time, some of these programs have inadvertently welfare: brought with them their own set of challenges from the per- spective of the labor market. As this chapter will discuss in · While even poor, informal sector workers engage in further detail, problems in the design of social protection private risk management strategies, the evidence programs--and of the social protection system as a whole-- indicates that households are only partially successful can serve to generate adverse incentives in the labor market, in protecting themselves against the impoverishing stimulating informality and leaving people without ade- effects of shocks, whether caused by illness, quate protection from key risks. They can also have adverse disability, unemployment, or loss of income in old effects on economic productivity and growth.1 age. Thus, the design of social protection--both the tradi- · Poor and near-poor households generally have fewer tional "Bismarckian" social security system linked to the assets and/or risk management instruments at their labor contract and subsequent noncontributory assistance disposal, making them particularly vulnerable to programs--appears to have generated a number of critical shocks. Indeed, inadequate access to risk manage- challenges to ensuring access of country populations to ade- ment instruments can lead to families engaging in quate risk management instruments, particularly among harmful "coping" activities--for example, removing the poor. In this context, this chapter their children from school--that can make it more difficult for them to escape poverty, and serving to · reviews the state and recent evolution of social pro- perpetuate poverty across generations. tection in Latin America and the Caribbean; · examines the role and limitations of households' pri- Second, from the societal perspective: vate risk management strategies and outlines the rationale for public social protection in the region; · A growing body of evidence indicates that, in addi- · analyzes the key challenges policy makers face in tion to adverse welfare effects, too much uninsured making adequate risk management instruments avail- risk can have negative productivity and income able to the regions' citizens working in both the effects at the microeconomic level (that is, house- formal and informal sectors; and 181 I N F O R M A L I T Y · outlines key policy directions for ensuring that the social risk management concept organizes people's responses populations of Latin America and the Caribbean are to shocks into three broad categories: prevention (ex ante), adequately protected from key risks. mitigation, and coping (ex post) (see Holzmann and Jorgensen [2000] and World Bank [2001]). The comprehen- In developing directions for policy, the chapter builds sive insurance framework3 offers a tool for determining which upon the lessons from recent World Bank studies on social mitigation instruments and preventive measures will be protection policy, based on the concept of social risk man- most effective, given the expected size, frequency, and agement and on the economics of insurance.2 Individuals extent of externality of a range of possible financial losses; and societies can respond in a variety of ways to the the framework can also be used to identify when coping is prospect of economic losses associated with such shocks as an efficient, ineffective, or even damaging course of action illness, disability, or loss of income due to old age, and the (box 7.2). BOX 7.2 The comprehensive insurance framework--providing guidance on social protection policy making In classical theory, individuals facing the likelihood of a range of possible financial losses; the framework can also financial loss from an adverse event can either insure be used to identify when coping is an efficient, ineffective, against such a loss or take steps to lower the likelihood or even damaging course of action. From the perspective of that the loss will occur. The challenge that people face, financial protection, when prospective losses are small and therefore, is to choose the optimal mix of market insur- infrequent, it is more efficient for individuals to cope with ance, self-insurance, and self-protection. Both market and the loss after the fact than to insure. In that sense, full self-insurance transfer income from the "good" states to insurance is not efficient. But as prospective losses become the "bad" states of the world. Market insurance pools more frequent, it becomes relatively more efficient to risks across individuals, compensating for differing risks engage in prevention--to lower the probability of the loss among them. Where market insurance is available, it can occurring--and saving--to cover the costs of the loss. be purchased at a price--the insurance premium--which When a prospective loss becomes less frequent but reflects the size of the prospective loss and the probability increases in size, it becomes more efficient to engage in risk of the bad state coming about. Self-insurance--essentially pooling. For rare but large prospective losses, households individual or family saving--does not involve risk pool- also have incentives to engage in prevention, again to lower ing or compensation for risk differentials. Although it has the likelihood that the adverse event will occur. For losses no explicit price, its cost can be imputed from the expense that are both frequently occurring and catastrophic in size, people incur to save, for instance, in forgone consumption. there is little that individuals, households, or even markets Self-protection refers to measures that individuals or can do on their own; specific measures to create a larger households take, ex ante, to prevent an adverse event from risk pool are required. occurring. Self-protection reduces the probability that When markets fail or are missing, policy intervention losses will occur, and may reduce the size of a loss, should can be justified to strengthen people's ability to manage one occur. For simplicity, the literature often refers to risk. While insurance markets may fail due to the nature market insurance as risk pooling, to self-insurance as saving, of risk (prospective losses are both frequent and large), and to self-protection as prevention. When individuals and they also commonly fail due to information problems. households do not insure through risk pooling or saving, On one hand, individuals may lack critical information or engage in prevention, they are often forced to cope with on the nature and extent of the risks that they face, lead- the costs of losses in the wake of a shock. ing to a lack of demand for appropriate risk management The comprehensive insurance framework provides instruments; on the other hand, problems of "adverse guidance on which mitigation instruments--risk pooling selection" and "moral hazard," well-known in the insur- or saving--and preventive measures will be most effective, ance literature, can raise the price of risk mitigation given the expected size, frequency, and extent of externality of instruments, pricing the poor out of the market, or cause 182 I N F O R M A L I T Y, S O C I A L P R O T E C T I O N , A N D A N T I P O V E RT Y P O L I C I E S markets not to form at all. Finally, the failure of individ- · Government should help households build and protect uals or households to manage risks appropriately can their human capital. Investing in human capital-- often impose negative externalities on society. For exam- education, hygiene, and primary and preventive ple, in countries where a significant number of people health care--can be an effective and powerful means fail to insure, governments face a dilemma in which of prevention and can strengthen families' abilities politicians cannot credibly refuse to provide aid when a to cope effectively. However, where credit is con- large number of people suffer losses, and the burden of strained, individuals may under-invest in human these losses is transferred onto current and future tax- capital. To ensure appropriate human capital devel- payers. Similarly, in the case of health, communicable opment, governments can subsidize investments in diseases that go untreated due to lack of insurance can education and health for the poor. Due to the nature have important negative spillover effects on large swaths of health shocks and the presence of externalities, of society. Indeed, the degree of externality posed by government can support work to broaden and health risks and the public-good nature of many health strengthen risk-pooling instruments. treatments and interventions present a powerful addi- · It is less costly to help households mitigate losses than to tional justification for policy intervention. cope with them. The instruments for individuals and The comprehensive insurance framework provides families to pool risks and save are not always avail- guidance for public intervention regardless of the able. The resources to take preventive measures are prospective loss under consideration (for example, due to often scarce. Where individuals and families are illness, disability, unemployment, or income loss in old constrained, adverse coping can result. Some pre- age): vention, risk pooling, and saving are always desir- able. Effective policy should thus place priority on · Government should provide (or help provide) the instru- enabling individuals to insure against losses ments that the market cannot (or will not) provide. Risk through risk pooling and saving and to lower the pooling to cover certain losses (for example, lost probability of losses through prevention, rather earnings from becoming unemployed, the risk of than merely coping after a shock. poverty, disasters, and certain frequently occurring or preexisting health events with catastrophic costs) While market failures and externalities justify state does not exist in many contexts due to information intervention, the comprehensive insurance framework problems. Government can step in to correct market also highlights the risk of "government failure." This can failures by providing risk-pooling instruments. occur when governments "overreach," that is, where the · Government should provide (or help markets provide) state tries to do too much by providing coverage that superior instruments where only inferior instruments are households and markets can--or should--provide them- available. For risks best covered with individual sav- selves. Examples of this include when governments sub- ings, private agents may turn to "bad" saving sidize pensions for upper- or middle-class groups or instruments (for example, property or other nonliq- subsidize insurance for small and frequent health events uid assets for precautionary saving) because "good" (phenomena that have been common in Latin America instruments (such as diversified financial assets; safe, and the Caribbean). Overreach can lead to unsustainable reliable, and competitively priced forms of liquid fiscal burdens, even when coverage rates are low. Govern- savings; or credit) are not available. Moreover, ment interventions can also be "misaligned," that is, poorer households simply may not have much mar- focused on the wrong mix of instruments given the gin to save. Government can intervene to foster the expected size and frequency of loss, or excluding clear development of more efficient instruments for sav- "insurables" (such as low probability, high expected loss ing through prudential regulation of capital, credit, events) in mandatory risk-pooling packages. and insurance markets, as well as provide direct subsidies for households that are too poor to hold Source: Adapted from Baeza and Packard 2006; Gill and Ilahi 2000; Gill, savings or debt. Packard, and Yermo 2004; and Packard 2006. 183 I N F O R M A L I T Y The state of social protection in Latin America non-wage costs account for over 47 percent of payroll (see and the Caribbean Mason et al. forthcoming). The truncated welfare state in Latin America and the Carib- In response to fiscal pressures, as well as concerns about bean has stemmed largely from what is often called the low coverage and persistently large informal economies, a "Bismarck" model of social security, which has been devel- number of reform measure changes have been undertaken oped in most countries in the region since the early part of in Latin American and Caribbean countries since the the 20th century. The core of the Bismarck model is that 1980s, particularly with respect to pensions. This has social security coverage is based on the form of the worker's improved the fiscal sustainability in some countries, but labor contract. In general, workers acquire rights to a pack- increases in coverage have remained elusive. As discussed age of benefits--health insurance, disability, pensions, and briefly in chapter 1, social security coverage in most Latin so on--via employment in the formal sector of the econ- American countries remains relatively low, with limited omy rather than by virtue of working per se (or by virtue of progress recorded over the last decade (figure 7.1). Indeed, a more inclusive concept, such as "citizenship"). Such ben- in examining changes in pension coverage between the efits are typically financed through a combination of pay- mid-1990s and the early to mid 2000s, Rofman and roll taxes on the part of the worker, employer contributions Lucchetti (2006) find that in 9 out of 15 countries (for on the part of the firm, and, in many cases, subsidies from which comparable household survey data exist) coverage the government. In some instances, non-wage costs of rates have actually declined over time, and even in the few employment (payroll taxes plus other benefits) can be very countries where progress has been made (for example, high, affecting both employers' and workers' informal- Peru), advances have been small. formal sector decisions. In Colombia, for example, total As discussed in chapter 4, a combination of factors-- non-wage costs have been estimated at 53 percent of pay- macroeconomic, structural, and demographic--appears to roll, the highest in the region, followed by Mexico whose have contributed to observed patterns since the 1990s. This FIGURE 7.1 Pension coverage rates in Latin America and the Caribbean % of economically active population contributing to pension system 80 1990s 2000s 70 60 50 40 30 20 10 0 u or Per ador . de y ina Rica Bolivia R.B Mexico Brazil Chile Paraguay Nicaragua Colombia Ecu Salvad Guatemala El Argent Urugua Costa Venezuela, Source: Rofman and Lucchetti 2006. Note: The years for which household survey data are available across countries in the region are not identical. As such, Rofman and Lucchetti (2006) present figures based on available data in each country that comes closest to the 1995­2004 period. The years used are as follows: Argentina 1995­2004; Bolivia 1999­2002; Brazil 1995­2002; Chile 1996­2003; Colombia 1996­99; Costa Rica 1995­2004; Ecuador 2000­04; Guatemala 1998­2000; Mexico 1998­2002; Nicaragua 1998­2001; Paraguay 1999­2004; Peru 1999­2003; El Salvador 1995­2003; Uruguay 1995­2004; Republica Bolivariana de Venezuela 1995­2004. 184 I N F O R M A L I T Y, S O C I A L P R O T E C T I O N , A N D A N T I P O V E RT Y P O L I C I E S FIGURE 7.2 Social security coverage and GDP per capita a. Social security coverage in LAC by per capita GDP b. Changes in social security coverage in LAC versus (latest year) changes in GDP per capita, 1990s­2000s % of economically active population Average annual change in social security contributing to social security coverage (%) 70 10 y 0.0043x 4.1392 y 1.2298x 0.0183 8 60 R2 0.5553 R2 0.2124 6 50 4 2 40 0 30 2 20 4 6 10 8 0 10 0 2,000 4,000 6,000 8,000 10,000 12,000 14,000 3 2 1 0 1 2 3 4 GDP per capita (PPP, real 2000) Average annual GDP growth per capita (%) Sources: Author's calculations. Data on social security coverage from Rofman and Lucchetti (2006); data on GDP per capita purchasing power parity from World Development Indicators, various years. includes increases in labor force participation rates (particu- tends to decline (social security coverage tends to rise) as larly among females) during the 1990s, growth in nontrad- national income rises. Indeed, the same relationship able sectors driven by currency appreciations in the early between gross domestic product (GDP) per capita and social 1990s, and increased labor market rigidities, at least in security coverage that holds globally is observed when only Colombia and Brazil.4 In addition, social security reforms Latin American and Caribbean countries are analyzed. This themselves may have lowered the perceived benefits of the is true whether one looks either at GDP levels or GDP affiliation relative to its costs. In the case of pensions, for growth (figure 7.2a and b). As can be seen in panel b of the example, reforms that were designed to increase fiscal sus- figure, however, several countries that experienced positive tainability often raised contribution rates, minimum con- economic growth did not increase social security coverage tribution periods, and retirement ages, while lowering between the mid-1990s and the early to mid 2000s; in fact, replacement rates. These changes in the relative benefits and coverage actually declined in several positive-growth coun- costs of pension schemes may have more than offset the tries. This suggests that the factors leading to increased salutary effects associated with closer links between contri- informality--whether macroeconomic, structural, demo- butions and payments. As will be discussed further in this graphic, or related to social protection design--frequently chapter, a number of factors related to the design of social offset the positive effects of growth. security programs and to the incentives created by the Consistent with the evidence presented in the earlier broader constellation of social protection programs--both chapters, in nearly all countries in the region, coverage rates social insurance and social assistance--appear to affect cov- are significantly lower among low-income than high-income erage levels. workers (figure 7.3). In most of the region's countries, the Data indicate that social security coverage has failed to poorest are practically excluded from the system. The situa- increase (informality has failed to decline) in the region tion is similar when considering the employed population, despite net economic growth over the period. As was high- showing that pension systems suffer from major inequities lighted in chapter 1, global data show that informality in terms of access, even if differential unemployment is 185 I N F O R M A L I T Y FIGURE 7.3 Coverage rates in Latin America and the Caribbean, mid-1990s to early mid-2000s, by quintile of per capita income % of economically active population contributing to pension system 80 60 40 20 0 Bolivia Paraguay Peru Nicaragua Guatemala Colombia Ecuador El Salvador 80 60 40 20 0 Venezuela, R.B. de Mexico Argentina Brazil Uruguay Chile Costa Rica Quintile­I (1990s) Quintile­II (1990s) Quintile­III (1990s) Quintile­IV (1990s) Quintile­V (1990s) Quintile­I (2000s) Quintile­II (2000s) Quintile­III (2000s) Quintile­IV (2000s) Quintile­V (2000s) Source: Rofman and Lucchetti 2006. ignored. Most countries have coverage rates that are above to 59 percent of employed workers), it declined among all 50 percent for the highest quintile, but none has a rate other income groups. Declines in coverage were largest higher than 50 percent among the poorest workers. Analy- among those in the lowest quintile (from 48 to just under sis of coverage rates by worker education levels shows simi- 10 percent [Rofman and Lucchetti 2006]). lar patterns (see Rofman and Lucchetti 2006). Beginning in the 1990s--and partly in response to con- In addition, inequalities in access to social security have cerns about limited access to formal social security pro- tended to increase over time (figure 7.3). In Argentina, grams by large swaths of the region's population--a Uruguay, Chile, and Costa Rica, increases in inequality number of countries in Latin America and the Caribbean occurred, at least in part, because of noticeable declines in have launched or expanded poverty reduction and/or social coverage among those in the poorest quintile. Perhaps the assistance programs to help provide support and coverage most dramatic instance of this occurred in Argentina to the poor or extreme poor (for example, Argentina, (figure 7.4). While coverage rates in Argentina actually Bolivia, Brazil, Chile, Mexico, and Peru, among others). increased slightly among those in the top quintile (from 54 These efforts have taken a number of forms--including 186 I N F O R M A L I T Y, S O C I A L P R O T E C T I O N , A N D A N T I P O V E RT Y P O L I C I E S FIGURE 7.4 Pension coverage rates in Argentina, by quintile of per capita income % of employed workers 80 60 40 20 0 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 Quintile 1 Quintile 2 Quintile 3 Quintile 4 Quintile 5 Source: Rofman and Lucchetti 2006. subsidized health care and insurance (Acceso Universal para Uruguay and Argentina have small noncontributory prestaciones integrales y Garantías Explícitas [AUGE] in Chile, benefits that cover some impoverished individuals, age the Regimen Subsidiado in Colombia, and Seguro Popular in 70 or older, who do not qualify for contributory Mexico), noncontributory pension programs for the poor retirement benefits. In Brazil, there is a large quasi- (Bono Solidario [BONOSOL] in Bolivia, the Rural Pension noncontributory system that covers rural workers.5 Other Scheme in Brazil, and Pension Asistencial [PASIS] in Chile), countries, such as Mexico, have some noncontributory workfare programs (Jefes de Hogar in Argentina), and condi- schemes at the subnational level, including a program tional cash transfer programs (Bolsa Escola in Brazil, Famil- that covers the elderly in Mexico City. Unfortunately, ias en Acción in Colombia, and Oportunidades in Mexico). data available from household surveys do not generally Figure 7.5 presents the coverage of the elderly in Latin distinguish between beneficiaries of contributory and America between the mid-1990s and the 2000s. As with noncontributory programs (so most bars in figure 7.5 participation of the working-age population in formal only show total coverage, not what proportion of the cov- social security schemes, formal pension coverage among the erage is due to the contributory, quasi-, or noncontribu- current elderly population is extremely low in many coun- tory programs). There are a few countries--including tries in the region; rates are at 60 percent or higher only in Bolivia, Ecuador, Chile, and Costa Rica--where data on Costa Rica, Argentina, Bolivia, Chile, Uruguay, and Brazil. coverage under contributory and noncontributory pro- While formal pension coverage has increased in several grams are collected. As can be seen, where coverage of countries since the 1990s--for example, in Colombia, noncontributory programs is measured, they can often República Bolivariana de Venezuela, Costa Rica, and have a substantial impact on coverage of the elderly. Most Panama--some countries have actually registered declines dramatic is the case of Bolivia's BONOSOL program, in coverage of the elderly. Coverage in Argentina shows a which extends pension coverage to 58 percent of the particularly large decline, falling from 77 percent in 1995 elderly who are not otherwise covered by formal pension to 65 percent in 2004. programs.6 In this context, several countries in the region operate Beyond pensions, several countries have also introduced noncontributory pension programs, some of which have programs to provide subsidized health insurance (for exam- had an important impact on coverage of the elderly. ple, AUGE, Chile; Regimen Subsidiado, Colombia; Seguro 187 I N F O R M A L I T Y FIGURE 7.5 Pension coverage among the elderly in Latin America and the Caribbean, contributory and noncontributory programs Percent 100 1990s, noncontributory programs 2000s, noncontributory programs 90 1990s, contributory programs 2000s, contributory programs 80 70 60 50 40 30 20 10 0 u de Per Rica Mexico R.B. guay entina Bolivia Chile Brazil Republic Salvador Guatemala Paraguay Colombia Ecuador Panama Uru El Costa Arg Dominican Venezuela, Source: Rofman and Lucchetti 2006. Popular, Mexico) and/or conditional cash transfer programs FIGURE 7.6 focused on strengthening the human capital of the poor (for Coverage of two poverty-targeted assistance programs in Mexico: example, Bolsa Escola, Brazil; Familias en Acción, Colombia; Oportunidades and Seguro Popular (by decile, 2004) Red Solidaria, El Salvador; Programa de Asignacion Familiar, Percent Honduras; Programme Advancement Through Health and 100 Education, Jamaica; Oportunidades, Mexico; Red de Protección 90 Social, Nicaragua). In some cases, these programs have had 80 considerable outreach to the poor. In Mexico, for example, 70 the Oportunidades program (formerly known as Progresa) has 60 reached roughly 5 million very poor families, about two- 50 thirds of whom live in rural Mexico. 40 As can be seen in figure 7.6, poverty-targeted programs 30 such as Oportunidades and Seguro Popular in Mexico have 20 done a much better job of reaching the poor than tradi- 10 tional (contributory) social security programs. While 0 analysis of household survey data (as shown in figure 7.3) I II III IV V VI VII VIII IX X indicates that the distribution of formal social security ben- Oportunidades Seguro Popular efits is very regressive, coverage of Oportunidades and Seguro Popular is highest among the poor. Indeed, around 55 per- Sources: Adapted from Mason et al. forthcoming, based on data from Mexico's 2004 national income and expenditure survey, cent of households in the poorest decile were covered by Encuesta Nacional de Ingresos y Gastos de los Hogares. Oportunidades in 2004, and coverage declines significantly Note: Some Seguro Popular beneficiaries also participate in Oportunidades. Seventy-two percent of Seguro Popular as household income rises (figure 7.6). Coverage of Seguro beneficiaries participate in Oportunidades in the first decile, but the overlap decreases very rapidly with income. Popular is similarly progressive, albeit at lower levels of coverage in 2004; nonetheless, coverage of the program has grown considerably since that time.7 188 I N F O R M A L I T Y, S O C I A L P R O T E C T I O N , A N D A N T I P O V E RT Y P O L I C I E S FIGURE 7.7 Public social protection spending in Latin America and the Caribbean % of GDP 20 Mid-1990s Early 2000s 18 16 14 12 10 8 6 4 2 0 r de Peru Rica aguay Chile Brazil Jamaica R.B. Mexico Bolivia Ecuado Panama Uruguay Nicaragua Republic HondurasGuatemalaColombia Salvador Par Costa Argentina El inican Venezuela, Dom Source: World Bank staff calculations. Note: No mid-1990s data were available for Brazil, Ecuador, and Guatemala. A recent study on income redistribution and public American and Caribbean countries. For the countries in transfers in Latin America and the Caribbean shows that, in which disaggregated data exist, it can be seen that spend- general, social assistance programs have done better at ing on social insurance (that is, formal social security) reaching the poor than have traditional social security pro- generally make up the lion's share of spending; with a few grams, such as health insurance or pensions (Lindert, exceptions (such as Nicaragua and Honduras), spending on Skoufias, and Shapiro 2006). It is important to note, how- social insurance accounts for more than two-thirds of total ever, that the quality of targeting varies across types of pro- social protection spending (figure 7.8). At the same time, grams and across countries, and that not all assistance existing data suggest that the share of spending going to programs are targeted in a progressive manner. Scholar- social insurance has declined slightly in several countries ships, for example, as well as many food-based programs in since the 1990s, as the prevalence of poverty-oriented assis- the region, often lack effective, pro-poor targeting and are tance programs has increased. therefore more likely to benefit nonpoor families and indi- This growth of social assistance programs over the last viduals than the (intended) poor. 10 to 15 years has had at least two important effects on Regional efforts to extend social protection programs social protection in the region. First, an increasing number more effectively toward the poor can be seen, in part, in the of poor people have access to state benefits that play an figures on public spending on social protection between important social protection or poverty reduction function-- the 1990s and the 2000s. Average levels of public spending whether to help the poor build their human capital, protect on social protection in Latin America and the Caribbean them against basic health risks, or provide basic income increased over the period, from around 4.0 to 4.8 percent of security in old age. These benefits can often have an impor- GDP (figure 7.7). This is compared to 14.4 percent in the tant positive impact on the recipients, even though the Organisation for Economic Co-operation and Development unit transfers tend to be considerably smaller on a per- (OECD), 8.3 percent in the United States, and 16.3 per- person (or per-household) basis than formal social security cent in continental Europe.8 Clearly, the data show signifi- benefits (Lindert, Skoufias, and Shapiro 2006). Indeed, cant variations in the levels of spending across Latin there is a growing empirical body of literature on the 189 I N F O R M A L I T Y without risk management instruments. In some cases, FIGURE 7.8 workers may have access to health insurance or other formal Public spending on social insurance and social assistance in Latin America and the Caribbean, early 2000s risk management instruments by virtue of the employment status of a spouse, a parent, or other relative.10 Moreover, % of GDP individuals and households in Latin America and the 14 Social insurance Caribbean typically engage in a range of private risk man- 12 Social assistance agement strategies to prevent, mitigate, or cope with a 10 variety of risks and shocks. While, in principle, private risk 8 management strategies could include the purchase of pri- 6 vate insurance (for example, health, disability, and so on), 4 in practice, private insurance markets remain relatively thin in the region, and access to private insurance instru- 2 ments is generally confined to the wealthier segments of 0 societies. In this context, a number of private risk manage- de ico Peru Chile ntina R.B. Brazil Mex ment strategies are identified in the literature for Latin NicaraguaHondurasColombia Paraguay Arge America, including income diversification, adjustments of Venezuela, household labor supply, drawing down of household sav- Source: World Bank staff calculations. ings, sale of assets, and adjustments in household spending and/or consumption patterns. Such strategies are used both before the fact to reduce the likelihood of an adverse shock (or to mitigate the likely impact of a shock, should one positive impacts of conditional cash transfer programs, occur), and after the fact to help cope with or deal with the such as Oportunidades in Mexico or Familias en Acción in effects of an adverse event. Colombia, on human capital outcomes of the poor (see Evidence from Mexico, for example, indicates that Rawlings and Rubio [2005] for a recent summary of households send additional members into the labor force impact evaluation findings). Similarly, Brazil's rural in response to real or expected employment shocks pension program has contributed to observable declines in (Cunningham 2001). Evidence from Guatemala, Nicaragua, old-age poverty in that country (World Bank forthcoming). El Salvador, and Honduras shows not only that family At the same time, the growth of social assistance pro- members increase their hours worked to mitigate the grams has led to an increasing fragmentation in social pro- impact of an adverse event, but that they draw down finan- tection systems in the region, in which the poor (and some cial savings or other assets, if necessary, to protect their nonpoor) receive similar--albeit more modest--benefits income and consumption levels (World Bank 2003a). for free as formal sector workers receive by virtue of their Migration and remittances also make up a key element of payroll contributions. As will be discussed later in more household risk management in a number of Latin Ameri- detail, such well-intended efforts to make social benefits can and Caribbean countries--both ex ante, as preventive available to those outside the formal sector may--due to lack measures, and ex post, to soften the impact of a shock (see of incentive compatibility with contributory programs--be Arias [2004] and Beneke de Sanfeliú and Shi [2004] for creating disincentives to the formalization of the work- evidence from El Salvador). In Argentina, during the 2002 force. To the extent that such disincentives exacerbate dis- economic crisis, families altered their consumption pat- tortions in the labor market, they may also have adverse terns, buying fewer luxury goods and spending less on productivity effects.9 necessities, educational materials, and children's health vis- its (World Bank 2003a). When households' abilities to Private risk management and rationale for public cope with shocks are stretched, as they were during the social protection 2002 crisis in Argentina or after hurricane Mitch in While informal employment means that workers are Nicaragua, families often rely on wider social networks, not directly affiliated with formal social security programs, including memberships in community, religious, or neigh- this is not to say that informal workers are left completely borhood organizations, that can provide an alternative 190 I N F O R M A L I T Y, S O C I A L P R O T E C T I O N , A N D A N T I P O V E RT Y P O L I C I E S source of resources--as loans or gifts--in the event of an dynamics in rural El Salvador indicates, for example, that adverse shock (World Bank 2003a; Klugman, Kruger, and poor households take considerably longer to recover from an Withers 2003). income shock than nonpoor households (Rodriguez-Meza It is important to note that while some household risk and Gonzalez-Vega 2004). Evidence from rural China also management strategies, such as developing diversified indicates that income shocks have deeper impacts on poor income-earning portfolios, increasing adult labor supply, or households (Jalan and Ravallion 1999). While an income drawing down financial savings, may be seen as appropriate shock of 10 percent had only a 1 percent impact on the con- and rational responses to the reality of risk and shocks, sumption of the wealthiest households, it caused a 4 per- other strategies, such as engaging in distress sales of pro- cent decline in consumption among the poorest households. ductive assets like land, withdrawing children from school, Indeed, despite the best efforts of workers and their fam- or deferring utilization of preventive or curative health ser- ilies to protect themselves from risk, households that lack vices, may generate their own risks to long-term family access to formal risk management instruments remain vul- productivity and welfare. During the 2002 crisis in nerable to the impoverishing effects of shocks. The evi- Argentina, spending less on education and on children's dence suggests that families are particularly vulnerable to medical care raised critical concerns about the long-term the effects of health shocks (Baeza and Packard 2006; effects on children's human capital and long-term family Skoufias 2004). In addition to treatment costs, households welfare (World Bank 2003a). Moreover, a recent study of bear the cost of productive time lost from work, as well as Mexico indicates that children who are removed from the opportunity costs due to days spent taking care of fam- school in response to a shock are one-third less likely to ily members who are ill. Costs associated with illness can continue school than those who find ways to continue their drive families into poverty; for the already poor, the costs of education in the face of a shock (Sadoulet et al. 2004). a health shock can make escape from poverty more difficult. Thus, in the absence of adequate insurance or other formal Evidence from Argentina, Chile, Ecuador, and Honduras risk management mechanisms, families can risk long-term indicates that the impoverishing effects of health shocks losses in their human capital, adversely affecting their eco- can be significant. In Argentina, 5 percent of all nonpoor nomic productivity and increasing the likelihood of inter- households fell into poverty for at least three months in generational transmission of poverty. 1997 due to out-of-pocket spending on health (figure 7.9). A number of recent empirical studies--within and out- side Latin America--have tried to measure how effectively households smooth their consumption in the face of adverse income shocks. Specifically, these studies measure FIGURE 7.9 Nonpoor population who fell below the national poverty line due to the extent to which a shock to household per capita income out-of-pocket health expenditures translates into a shock to household per capita consump- tion. While the specific findings differ from country to Percent 12 country, these studies find that households are partially-- but not fully--effective at mitigating the impacts of shocks 10 to their incomes. Indeed, recent evidence from several Latin 8 American countries, including Mexico, Nicaragua, and 6 Peru, suggests that households in these countries are able to protect (or insure) only between 60 and 75 percent of their 4 per capita consumption in the face of an income shock 2 (Glewwe and Hall 1998; Klugman, Kruger, and Withers 0 2003; Skoufias 2004). In other words, a 10 percent shock to ina as a household's per capita income translates into a 2.5­4 per- Chile Ecuador cent decline in its per capita consumption on average. Argent Hondur The evidence also indicates that poor households are Sources: Baeza and Packard 2006, citing Maceira (2004) for often less able to smooth their consumption in the face of Argentina; Bitran, Giedion, and Muñoz (2004) for Chile; Montenegro (2004) for Ecuador; and Fiedler (2004) for Honduras. shocks than nonpoor households. A recent study of income 191 I N F O R M A L I T Y In Ecuador, 11 percent of nonpoor households fell below many families adequate protection from impoverishing the national poverty line for at least three months in 2000 risks and shocks. In this context, what is the role for public due to the direct costs of health care. In Honduras, in 2000, social protection policies and programs? From an economic 4 percent of nonpoor households fell into poverty, at least perspective, if private markets for insurance (or other risk- temporarily, due to health-related spending. pooling or savings instruments) existed and functioned Cross-country analysis of public and private spending on well, if actors in the market for risk management instru- health indicates that out-of-pocket spending on health care ments had adequate information, and if there were no exter- is relatively high in Latin America and the Caribbean, as nalities associated with inadequate risk management, then compared to other regions (Baeza and Packard 2006).11 In there may not be a compelling reason for the state to inter- Latin America and the Caribbean, 85 percent of all private vene. But private insurance markets are often missing--or spending on health care represents out-of-pocket spending are extremely thin--in Latin America and the Caribbean.13 by households. This compares with 72 percent, on average, Information problems abound, as do negative externalities in Europe and the OECD (which also have lower levels of associated with insufficient insurance. private health spending, as a share of total health spending, In terms of information, workers and their families com- than does Latin America). Moreover, the poor generally pay monly lack sufficient knowledge of the nature and extent of a higher share of health costs out of pocket than the nonpoor. the health risks that they face, as well as the likely direct And while participation in well-designed and well-func- (and indirect) costs associated with illness of the family tioning risk-pooling schemes reduces the likelihood of breadwinner or another family member. They also have less- falling into poverty in the wake of a health shock, relatively than-complete information on what their future needs will few of the poor (or near-poor) in the region participate in be--for example, at retirement. This latter issue causes effective risk pooling (Baeza and Packard 2006). what is commonly referred to as "myopia" in the pensions Another result of low social security coverage in the literature. Information problems--such as adverse selection region is that poverty among the elderly in Latin America and moral hazard--well-known in the economics of insur- and the Caribbean is often very high. A recent 19-country ance literature, similarly plague those who would supply study on old-age security found that the incidence of insurance or other risk-pooling instruments. Such problems poverty among the elderly in the region, measured in terms can work to raise the price of risk mitigation mechanisms of household per capita income, is often significantly higher beyond the reach of low-income groups; information prob- than among the population as a whole (Bourguignon et al. lems are also notorious for causing the private market for 2004). Indeed, this was found to be the case in 11 out of 19 risk-pooling to fail--and even preventing these markets countries studied. Elderly who fall outside the formal social from forming in the first place (Packard 2006). Informa- security system find themselves at particular risk. Simula- tion-related problems are often aggravated in low-income, tion analysis carried out by the authors suggests, however, high-poverty environments.14 that programs that provide minimum income support to the Insufficient risk management instruments can also have elderly poor could have an observable, positive impact on important negative externalities that argue for public inter- old-age poverty. This is consistent with real-life experiences vention. An individual's failure to manage risk can impose in Brazil and Bolivia. In Brazil, as noted above, the rural costs on others in society in one of several ways. First, in pension program is credited with contributing to observable countries where a significant number of people fail to declines in old-age poverty in recent years.12 A recent study insure, governments face a "Samaritan's dilemma"; politi- on Bolivia indicates that BONOSOL payments have had a cians cannot credibly refuse to come to the aid of a large positive and significant impact on food consumption among number of people who suffer a loss, and the burden of these recipient households, particularly in rural areas (Martinez losses is transferred onto current and future taxpayers 2005). (Packard 2006). But the external costs of individuals who are not adequately insured can extend well beyond the tax The role of the state burden; in the case of health, in particular, communicable Together, the evidence suggests that although households diseases that go untreated due to lack of insurance (or finan- rely on a number of private mechanisms for risk manage- cial protection) can have health, and therefore, basic welfare ment, this reliance on informal mechanisms does not afford impacts on large swaths of society.15 192 I N F O R M A L I T Y, S O C I A L P R O T E C T I O N , A N D A N T I P O V E RT Y P O L I C I E S In sum, missing or failed markets for insurance or other tics and endowments. For these workers, part of the explicit forms of risk pooling, information problems, and externali- or implicit calculations may involve their perception of the ties all serve to establish a role for the state in social relative costs and benefits of contributing to and affiliating protection from an economic perspective. Moreover, well- with formal social security. Other individuals, such as designed public intervention in social protection-- many informal salaried workers, find the choice to affiliate discussed in greater detail later in the chapter--can itself to social security made for them by employers and firms help to strengthen the efficiency of risk-pooling mecha- that have chosen (for whatever reason) to opt out of the sys- nisms, through widening and deepening of the risk pool tem. Chapters 5 and 6 have addressed the issue of firms' (which is often too small and fragmented, especially in relationships to informality. This chapter focuses in more developing-country contexts). But there are also very valid detail on attributes of social protection programs and, more reasons for public social protection that go beyond an eco- broadly, social protection systems (the agglomeration of nomics rationale. Among the most important include valid social protection programs in a country) that affect how distributional concerns in Latin America and the workers value the prospect of working in the formal versus Caribbean's highly unequal societies (see de Ferranti et al. the informal sector. Understanding these factors can help 2004), and concerns for poverty reduction. Such justifica- policy makers design a smarter, more effective system of tions, while not purely economic in nature, speak to the risk management for their citizens. types of societies that Latin American policy makers would like to foster (or as will be discussed in chapter 8, how to Costs, benefits, and the design of social redefine the nature of the social contract in the region). security programs A number of factors affect the locus of costs and benefits that Challenges for social protection in the face workers see with respect to participation or nonparticipa- of informality tion in formal social security programs, including physical Given the high human costs associated with lack of access accessibility (proximity) of benefits and services for affili- to appropriate risk management instruments, such as ated workers, the quality of services offered within the health insurance and old-age security, and the clear ratio- scheme, workers' valuations of the sometimes-complex nale for public intervention to ensure basic access, Latin bundle of goods and services that they are compelled to American and Caribbean policy makers face an important "purchase" via payroll taxes, program rules that make it challenge; to ensure that individuals and families have difficult for workers to qualify for certain benefits, and access to suitable risk management instruments in the face design peculiarities that "force" affiliated families to pay of significant levels of informality. At the same time, it is twice for the same services. These factors are now examined important to note that ill-designed interventions may serve in turn. to make things worse, not better. Indeed, as will be shown, there is ample evidence of "government failure" in Latin Accessibility of benefits America and the Caribbean that needs to be addressed as For many workers, particularly those in remote rural loca- part of any actions to strengthen risk management among tions, the lack of program-authorized health facilities the region's citizens. Therefore, an important, related chal- makes the choice of opting into the system untenable, even lenge is to ensure that the design and implementation of if they would otherwise be inclined to affiliate with a social risk management instruments (and related programs) are security system. The unavailability of the basic amenities consistent with improved risk management for the people associated with social security makes the costs of contribut- of the region, as well as with increased productivity and ing well above the benefits. Levy (2006a, 2006b) high- sustained economic growth. lights that this is a critical issue facing millions of workers As has been highlighted in earlier chapters, the informal in rural Mexico and in less-developed regions of that sector is highly heterogeneous. Some individuals (for exam- country. ple, many self-employed workers) are in that sector as a matter of choice, the result of an explicit or implicit assess- Program quality ment of the costs and benefits of formal versus informal Similarly, workers will also be inclined to favor informal employment, and/or in light of their personal characteris- employment if they perceive that the value of services 193 I N F O R M A L I T Y provided is low relative to its cost (that is, the cost of their minimum wage, if they do not have other sources of payroll tax contributions). In the case of health, for exam- income. Thus, the value of a pension benefit for a worker ple, workers may not value health insurance if the service at who earns close to the minimum wage would be practically authorized health facilities is unpredictable, involves long the same, whether or not he contributes to social security-- waiting times, and is of low quality. To the extent that low although the worker can access the benefit earlier if he or service quality drives some workers--particularly low-risk, she is a formal sector worker (Fernandes, Gremaud, and high-income workers--to affiliation with health insurance, Narita 2006). In Mexico, while all workers (and employers) this not only reduces participation in the system but serves are required to contribute to the housing subsidy, they are to "fragment" the risk pool, hampering the effectiveness of unable in practice to access those funds, due to the nature the insurance model. of the Mexican housing market (Levy 2006b; Mason et al. forthcoming).16 In this context, again, workers will view Perceived value of the "bundle of benefits" social security contributions--or at least the relevant com- In Mexico, the benefit package offered by the Mexican ponent--as a pure tax on their earnings. Social Security Institute (IMSS) has eight mandatory com- ponents: health insurance, retirement pensions, disability Paying twice for the same benefits insurance, professional risk insurance, life insurance, day Another common feature of formal social security in the care centers for workers, sports and cultural facilities, and region is that members of the same family--for example, housing credits. In Colombia, formal sector workers make husband and wife--are required to contribute individually mandatory payroll tax contributions toward a number of for health insurance benefits, even though nonworking benefits, including health insurance, pensions, and profes- wives (or husbands) would be covered by their working sional risk insurance, as well as toward worker training via spouses' insurance. In other words, members of the same the Colombian Training Institute (SENA), to the Colom- family are essentially required to pay twice for the same set bian Institute for Family Welfare (ICBF), and to the Cajas of benefits when both work in the formal sector. In this de Compensación Familiar, which provide services ranging case, where one formal sector earner is already contributing from cash subsidies to supermarkets to recreation for work- for health insurance, the contribution of the second earner ers. While some workers may value all elements of this may be viewed as a pure tax on his or her earnings. Indeed, package, others may value only some of them. For example, in Colombia, 44 percent of informal salaried workers and while single mothers may value child care benefits very 54 percent of informal independent workers state that the highly (indeed, they may be willing to pay more than they main reason they do not contribute to health insurance is are asked to contribute via payroll costs), other workers-- that they are already covered by a relative's insurance plan such as those without children--may not value it at all. In (see chapter 2, table 2.13). This particular design issue cases where workers do not value one or more components has become increasingly important over the last few of the benefits package, those components may be consid- decades as female labor force participation rates have risen ered by the worker as a "tax" on their earnings. in the region. In the case of Argentina, Galiani and In some cases, workers may face situations in which they Weinschelbaum (2006) argue that significant numbers of perceive contributions for part or all of the package of ben- female entrants to the workforce chose to be informal efits provided under social security as a "pure tax." In some because their husbands were already eligible for benefits, cases, this may be an issue of government credibility. For and that this phenomenon led to a 13 percent increase in example, if workers lack confidence in their government's informality between 1974­76 and 1999. commitment or ability to actually provide promised benefits--for instance, pension benefits to be paid 20 years Labor movements in and out of the formal sector in the future--then workers may view making pension Latin American and Caribbean social security systems are contributions as a pure tax. Moreover, certain aspects of also designed under the implicit assumption that workers social security design may cause workers to perceive social will spend their entire working lives in the formal sector. security contributions as a pure tax. In Brazil, for example, This approach is consistent with a long-standing view across the Social Security Law (Lei Orgânica da Previdência Social) Latin America that labor markets are highly segmented and guarantees all Brazilians over the age of 67 a benefit of one that flows of workers between the formal and informal 194 I N F O R M A L I T Y, S O C I A L P R O T E C T I O N , A N D A N T I P O V E RT Y P O L I C I E S sector jobs--and particularly from formal to informal sector FIGURE 7.10 jobs--are small. As was discussed in earlier chapters, there Distribution of Mexican workers' years insured by IMSS, appears to be more mobility and a greater flow of labor in 1997­2005, for low- and high-wage workers in IMSS in 1997 and out of the formal sector than was envisioned in the % of workers design of either health insurance or pensions. 45 Data suggest, for example, that there is considerable 40 movement of individual workers in and out of the formal 35 and informal sectors, even over relatively short periods of 30 time. Analysis of Mexico's 2005 National Urban Employ- 25 ment Survey indicates that about 11 percent of high-wage 20 15 workers in the formal sector during the first quarter of 10 2005 (those earning more than three minimum wages) had 5 moved to the informal sector by the end of the year (Levy 0 2006a). About 16 percent of low-wage workers in the for- 1 2 3 4 5 6 7 8 9 mal sector (those earning less than three minimum wages) Years had moved into the informal sector from the beginning to Workers with 3 MW Workers with 3 MW the end of the year. Flows moved in the other direction as well; roughly 11 percent of both low- and high-wage work- Source: Adapted from Levy (2006a), based on Razu (2006). Note: Of the almost 9 million workers analyzed, nearly 6 million ers who were informal in the first quarter had moved into were "low-wage" workers, defined as earning less than three times minimum wage, while roughly 3 million were "high-wage" the formal sector by the end of 2005. workers, defined as earning more than three times minimum wage. Overall, movement of Mexican workers is greater among low-wage workers than among higher-wage work- ers. This can be seen from data on the duration of roughly 9 million IMSS workers in the formal sector between 1997 system, this means that, over any given time period, IMSS- and 2005 (figure 7.10); only 11.6 percent of low-wage affiliated workers--especially low-wage workers--spend a IMSS workers spent the entire nine-year period in the IMSS considerable amount of time without access to IMSS health system. Moreover, on average, low-wage workers spent just insurance. This suggests that under Mexico's current social less than half of the period (4.3 years) in the system. In con- protection system, some workers may spend some time trast, over 42 percent of the higher-wage workers spent the periods covered by IMSS, and other periods covered by full nine-year period in the system. These workers also ministry of health programs (for example, Seguro Popular) spent a longer amount of time in the IMSS system over the with access to separate and mutually exclusive health facil- period--6.5 years, on average. It is worth noting, however, ities. This represents significant inefficiency in the system. that while higher-wage workers spent more than two years Similar conclusions can be drawn from data on years of longer in the system than low-wage workers during the worker registration with the main social security institu- period, they still averaged roughly 2.5 years out of the sys- tion in Uruguay, Banco de Previsión Social or BPS (Bucheli tem over the period.17 et al. 2006). First, these data also suggest that there is con- This mobility of labor has potentially important impli- siderable movement in and out of Uruguay's formal sector cations for workers' access to benefits and, thus, the design during a person's normal working life. This can be seen of social protection. For example, under current regula- in figure 7.11. If the data had a bimodal distribution--in tions, IMSS workers are required to accrue 25 years of work which one group of workers never or almost never con- experience "in the system" in order to qualify for the Mini- tributed to BPS and another group always or almost mum Pension Guarantee (MPG). But if the years of IMSS always contributed over their working lives--then one affiliation implied in figure 7.10 are representative, then it could say that the data show evidence of strong labor mar- would take a low-wage worker roughly 50 years of work to ket segmentation. But, as can be seen from the figure, the qualify for the MPG. For all practical purposes, therefore, data show a continuum of cases, with peaks around 17, 23, the MPG would be unattainable. Moreover, to the extent 25, 30, 35, and 38 years of service, suggesting more that the data reflect worker movement in and out of the mobility than segmentation.18 195 I N F O R M A L I T Y FIGURE 7.11 FIGURE 7.12 Density function for years of formal sector work among Cumulative distribution of years of formal sector service among 60-year-olds in Uruguay (accumulated ages 18­60) 60-year-olds in Uruguay (accumulated ages 18­60) 0.009 Percent 0.008 100 90 0.007 87% 80 0.006 70 0.005 60 62% 0.004 50 40 0.003 30 0.002 20 0.001 10 0 0 0 5 10 15 20 25 30 35 40 0 5 10 15 20 25 30 35 40 45 Years of service Years of service Source: Bucheli et al. 2006. Source: Bucheli et al. 2006. Second, the data suggest that there is a high risk that a For those workers--particularly low-income workers-- very significant share of the workers registered in the work who do not envision achieving the required vesting period, history of the BPS will not be able to comply with the con- such rules may provide a strong disincentive to entering dition of contributing 35 years to access the pension when the formal sector in the first place.21 they reach the usual ages for retirement (figure 7.12) Just how important are these factors? A large body of (Bucheli et al. 2006). In contrast to Mexico, 35 years of empirical literature from Latin America and beyond shows contributions are required not only to get a minimum pen- that high rates of payroll tax have a negative impact on for- sion, but to attain the right to any pension at all (if the mal sector employment. For example, Packard (2002) finds worker is below the age of 70).19 As can be seen from the that the higher payroll taxes for social security are associ- figure, Bucheli et al. estimate that by age 60, roughly ated with lower numbers of contributors in the workforce. 87 percent of all workers will not have accumulated the Fiorito and Padrini (2001) arrive at similar results in an 35 years of service necessary under the law to qualify for analysis of labor taxes in developed economies. pensions (figure 7.12). The authors estimate similarly that Several authors have estimated semi-elasticities of self- 72 percent of 65-year-olds will not have accumulated the employment with respect to a change in relative formal sec- necessary number of years. tor earnings (Krebs and Maloney 1999; Loayza and As in Mexico, the Uruguayan problem tends to be more Rigolini 2006; Maloney 2001). These authors find a range serious among low-income workers than among high- of estimates, depending on methodology used,22 and this income workers (and among private sector workers as has implications for the possible size of the impact. If, for opposed to those in the public sector). Low-income workers example, one assumes that 10 percent of the formal sector in Uruguay tend to be at the bottom of the distribution of earnings package is absorbed in unvalued benefits and per- years of formal sector work, while high-income workers haps that another 10 percent of the value of earnings is tied tend to be at the upper end of the distribution (Bucheli to inaccessible benefits (leading to a 20 percent decline in et al. 2006).20 Bucheli et al. estimate that just over 50 per- the relative attractiveness of formal labor), then the cent of 60-year-olds from the highest earnings quintile will authors' estimates suggest the size of the informal self- achieve 35 years of formal service (and registration in BPS), employed sector would increase by between 0.6 and but that almost no one from the lowest earnings quintile 6.0 percent, as a share of the workforce. In the case of will. As such, low-income workers in Uruguay--as in Brazil, Fernandes, Gremaud, and Narita (2006) simulate Mexico--face a particularly serious risk of not acquiring the impact of eliminating payroll taxes for unskilled work- the years of service necessary to qualify for pension benefits. ers (those earning up to one minimum wage). If combined 196 I N F O R M A L I T Y, S O C I A L P R O T E C T I O N , A N D A N T I P O V E RT Y P O L I C I E S with deductions of value-added tax (VAT) paid on capital In the last few years, however, several new social assis- investment, these authors find that eliminating payroll tance programs have been launched or proposed for poor taxes on unskilled workers would lead to a decline in infor- workers working in the informal sector. These include a mality of 1.5 percent.23 subsidized package of financial protection in health, called Seguro Popular, with an insurance premium that is 100 Adverse incentives within the broader social percent subsidized for extremely poor families; a pension protection system program, MAROP (Mecanismo de Ahorro para el Retiro Opor- The fact that there is greater mobility between the formal tunidades), that includes noncontributory transfers for the and informal sectors than has generally been thought cre- elderly and savings incentives for those in Oportunidades ates challenges not only for the design of specific social who work in the informal sector; and Fondo Nacional de security programs, but for social protection systems as a Habitaciones Populares (FONAHPO), a program of housing whole. As mentioned earlier in the chapter, the recent subsidies for poor informal sector workers.24 More recent growth of social assistance programs has also led to increas- proposals are on the drawing board for universal health ing fragmentation in social protection systems in the Latin insurance for all pregnant mothers and children under five American and Caribbean region. A key benefit of this years of age, and subsidized child care for working mothers approach has been that an increasing number of poor peo- in the informal sector.25 ple have access to programs that play an important social Several features are noteworthy about these recent or protection or poverty reduction function, providing free emerging social assistance instruments--health insurance, but relatively modest benefits to eligible families--those pension, housing, child care--relative to the IMSS pack- whose livelihoods exist largely outside the formal sector. age. First, although the benefit packages are not as gener- The main challenge relates to the fact that expanded social ous as those offered by IMSS, they are less costly than the assistance programs provided freely (and sometimes pro- IMSS package and, in some cases, essentially free. Second, vided conditionally on working in the informal rather than eligible beneficiaries can choose which of the programs the formal sector) may themselves be creating disincentives they would like to participate in; there is no mandatory to the formalization of the workforce who are taxed via bundle. And, third, beneficiaries must work outside the payroll contributions to gain the right to social security IMSS (formal sector) system to be eligible for these pro- benefits. grams. So, key questions that potential beneficiaries face An example of this can be seen in the case of Mexico. As are: Do I search for a job in the formal sector where I can noted, the benefit package offered by the IMSS has eight receive a bundle of benefits that I will pay for via my pay- mandatory components, including health insurance, roll contributions? or, Do I stay in the informal sector and retirement pensions, disability insurance, professional risk get similar--although somewhat less generous--benefits insurance, life insurance, day care centers for workers, that are essentially free? The answers lie, in part, in how sports/recreational and cultural facilities, and housing cred- highly workers value the net benefits of social security ver- its, paid for in part by workers' payroll taxes. In 1995, the sus social assistance benefits. While this emerging system government of Mexico launched a conditional cash transfer of assistance programs is too young to have enabled mea- program, called Progresa (subsequently renamed Oportu- surement of its effects on formal versus informal employ- nidades), which provided cash transfers to poor families on ment, the creation of a dual system--social insurance the condition that families make specified human capital versus social assistance--may be creating incentives for investments. The design of the program is oriented toward greater informal employment (or slower formal employ- strengthening the human capital of the poor through ment growth) in Mexico. incentives and investments in school-aged children. Fur- Levy (2006b) models the labor market decisions of util- thermore, the design of the program is such that support ity maximizing workers between the formal and informal will end prior to the time that a student-beneficiary gradu- sectors, given a predetermined and mandatory (take-it-or- ates from school and prior to the time that he or she makes leave-it) bundle of social security benefits available to for- a labor market decision (for example, agriculture versus mal sector workers, on one hand, and an unbundled industry versus services, rural versus urban, informal versus collection of social assistance benefits available to informal formal). sector workers on the other. Consistent with the situation 197 I N F O R M A L I T Y in Mexico and elsewhere in Latin America and the between 2002 and 2005, during a period of strong Caribbean, workers must pay for social security benefits via economic growth and formal employment growth in payroll taxes, while social assistance benefits are available Argentina. The authors investigate whether male Jefes par- for free to workers outside the formal sector. Workers may ticipants were less likely to accept formal jobs in this differ in their valuation of social security benefits as a result booming economy than their nonparticipant counterparts, of inherent differences in their personal preferences or due taking advantage of a short panel of data from the to differences in the quality of program benefits and ser- Argentine household survey. vices. When workers' valuations of the relative benefits dif- Gasparini, Haimovich, and Olivieri (2006) find some fer, they self-select into formal and informal sector jobs, evidence of an informality-incentive effect due to Jefes. affecting the composition of the labor market. A general Specifically, the authors find that the share of Jefes partici- conclusion of Levy's model is that when social (protection) pants who found a formal job was significantly lower than policy consists of many programs and policies with distinct the corresponding share of nonparticipants with similar rules of access (formal versus informal job status) and forms observable characteristics. In some cases the difference is of financing (payroll taxes versus general revenues), it can not only statistically significant but economically large. generate unanticipated labor outcomes; these resulting out- For instance, when carrying out the propensity score comes can have adverse effects on productivity and long- matching with the radius method, during the 2003­04 term growth (Levy 2006b). period, the difference between male Jefes participants and While the Mexico case may be particularly illustrative, Jefes nonparticipants in the share of workers moving to a Mexico is by no means alone on this issue. Colombia has a formal job is 5 percentage points. The authors also find subsidized health insurance regime (Regimen Subsidiado) that the informality-incentive effect is sensitive to the rela- that is available to poor informal sector workers; Argentina tive wage offers in Jefes and in the formal sector. Specifi- has an emergency employment program (Jefes de Hogar) in cally, the effect of Jefes on informality vanishes during the which participation is conditional on not being employed 2004­05 period when the gap between the Jefes transfer elsewhere. In fact, a recent study of the informality impacts (fixed in nominal terms) and wages in the formal sector of the Jefes de Hogar program in Argentina provides some greatly widened.26 preliminary evidence of how important these incentive In sum, evidence from the Jefes program in Argentina effects might be at the specific program level. In the midst suggests that the design of social assistance programs and of one of the most serious economic crises of its history, the incentives they create do indeed influence workers' Argentina implemented Jefes de Hogar (Jefes), a large decisions to seek formal versus informal employment. poverty alleviation program. This program combines the Additional analysis in other country and programmatic features of a workfare and a conditional cash transfer pro- contexts would be useful in deepening policy makers' gram. Jefes is aimed at providing cash transfers to those understanding of just how strong are these incentive unemployed household heads with children at school. The effects, both in absolute terms and relative to other factors. belief that poverty is closely related to unemployment led In the meantime, several factors may be thought to affect Argentina to include the unemployment requirement as a the relative strength of the incentive effects in a given con- targeting device. text. These factors include the following: In principle, conditioning on unemployment implies a full taxation on outside incomes for the program partici- · The more generous noncontributory program bene- pants; that is, getting a job means losing their program fits are relative to contributory program benefits, benefits. Under certain circumstances, therefore, Jefes's the stronger the incentive problem. This could unemployment requirement may create a disincentive for include relative wages, as in the case of Jefes, or the beneficiaries to search for a formal job--although in prac- relative size of non-wage benefits, all other factors tice, monitoring of Jefes's program requirements is not per- (including service quality) being equal. fect, so the incentive effects may not bear out in reality. · Program incentive effects are likely to operate The effect of Jefes on informality is, thus, an empirical mat- most strongly among workers at the margin of ter. In this context, Gasparini, Haimovich, and Olivieri formal and informal employment decisions. In that (2006) assess the impact of Jefes on labor informality context, social assistance programs (including 198 I N F O R M A L I T Y, S O C I A L P R O T E C T I O N , A N D A N T I P O V E RT Y P O L I C I E S noncontributory health and/or pension benefits) are market failure, and government failure for the design of the likely to have a smaller effect on labor market out- region's social protection systems. comes. When they are well targeted toward the very poor, and when the social security coverage gap is rel- Providing essential cover in health atively large, formal sector employment is concen- For guidance on how governments should intervene to pro- trated among those in the upper deciles. The issue of tect their citizens, it is useful to return to the comprehen- incentives between social security and social assis- sive insurance framework discussed earlier in the chapter tance is likely to be more important in the relatively (box 7.2). Baeza and Packard (2006) illustrate the applica- more developed countries in the region, with tion of this framework in the context of health. Illness moderate-to-high levels of both social security and comes in a variety of forms. Most frequent illnesses are not social assistance coverage. serious, nor do they imply large costs or financial losses. In · Certain types of social assistance programs may not fact, for 80­90 percent of health events that households have strong incentive effects on the workers' deci- will experience in their lifetimes, people rarely go to the sions to seek formal versus informal employment. doctor. Most symptoms--headaches, the common cold, Programs such as Oportunidades in Mexico (or similar adult diarrhea, and even minor fever--last fewer than three conditional cash transfer programs) that limit their or four days and can be easily treated with rest and nonpre- focus to strengthening children's human capital--for scription pharmaceuticals. To mitigate the financial losses example, education, health, and nutrition--but do from these relatively small, frequent symptoms, most not try to influence labor market choices, would not households are better off relying on prevention (for exam- be expected to create strong incentives for any spe- ple, good nutrition and good hygiene, preventive and cific forms of labor market participation. primary medicine, and exercise) and saving (that is, indi- · Conversely, those programs that tie participation to vidually assuming the cost of treatment and medication). a particular labor market state--for example, work- However, for less frequently occurring, more serious ill- ing outside the formal sector--would be expected nesses, the cost of treatment can increase rapidly. Indeed, to more strongly affect informal-formal sector for conditions such as complicated flu, pneumonia, bacte- decisions. rial bronchitis, or urinary infections, the cost of diagnosis, treatment, and resolution can be substantial. Medical con- Reengineering social protection to protect sultation becomes critical to identify more serious condi- all citizens tions. To cover the potential financial consequences of Poor access to basic risk management instruments, com- less-frequent sicknesses that are costly to treat, households bined with the existence of market failures (information are better off relying on risk-pooling arrangements. As problems, externalities) and pervasive government the cost of treating health events grows--and because the failure, highlights an urgent need to rethink--indeed, direct and indirect costs of an illness and its treatment can "reengineer"--social protection policy and systems in be impoverishing--it becomes critical for individuals and much of Latin America and the Caribbean. Specifically, households to find an effective mechanism to pool risks given the high and persistent levels of informality in the (Baeza and Packard 2006). region, it will be important to rethink the traditional Bis- Yet, as noted above, private markets for health insurance marckian model of social protection in which protection are thin or are missing in most Latin American and depends on the specific form of the labor contract. The col- Caribbean countries, due in part to information failures. lection of evidence suggests that a broader notion of who The small and fragmented risk pools created by existing has access to basic risk management instruments is health insurance schemes are relatively inefficient. More- needed--one based on assuring the basic protection and over, in the case of health, losses that go uncovered have the welfare of countries' citizens rather than of workers, as tradi- potential to impose significant external costs on others. As tionally and narrowly defined. Chapter 8 focuses in detail such, there is a strong case for public intervention to on the nature and evolution of the social contract between strengthen risk-pooling mechanisms and to expand the the Latin American State and its citizens. This section pool. Since public resources for health insurance tend to be focuses on potential implications of informality, risk, heavily constrained in the region, there is a case for extending 199 I N F O R M A L I T Y broad coverage via a package of minimum or "essential" direct package in Chile (box 7.3). It is important to underscore cover of catastrophic losses with high external costs (Packard that the idea of general-revenue financing of health insur- 2006). To ensure the greatest possible coverage, such a ance should apply solely to minimum essential cover; both package should be de-linked from the labor contract and the economics of insurance and a country's real-life fiscal financed through general taxation. Since even moderate constraints dictate that additional coverage, if desired, health costs can be catastrophic for the poorest households should be available on a contributory basis. In this context, in the region, the exact composition of the package should there is a critical role for public policy in fostering more reflect this fact. efficient health sectors, including the strengthening of both There are a number of arguments in favor of de-linking insurance and provision functions.27 the provision of essential cover in health from the labor contract and, instead, financing such provision through Strengthening old-age security general taxation. General taxation is potentially the most There is also a case for providing essential cover to the elderly in the efficient and also the most equitable financing mechanism form of a poverty prevention pension, focused on the poor, as part of a for risk pooling, depending on the progressivity of tax broader multipillar pension system. As a form of social insurance, collection instruments and subsequent patterns of public risk pooling would be central to the poverty prevention ele- spending (Mossialos et al. 2002; Savedoff 2004, cited in ment, and, because of the social costs associated with people Baeza and Packard 2006). Financing essential cover falling into poverty at an older age, there is a clear risk man- through general tax revenues also has the benefit of ensur- agement rationale for de-linking access to this poverty pre- ing that health risks are effectively pooled across the widest vention pension from the form of the labor contract, and possible risk pool. In the case of essential cover, Packard financing it through general revenues (Packard 2006). At (2006) argues that, given externalities, the social costs the same time, the comprehensive insurance framework of individuals failing to cover themselves and their highlights the importance of individual saving as a core dependents are high enough that "there is a clear risk man- element of a broader system of old-age security. Indeed, given agement rationale to take the `choice to cover' out of the the high probability of income loss in old age, saving should be the hands of employers and workers" (p. 25) by shifting financ- mainstay for earnings replacement during old age (Gill, Packard, ing away from payroll contributions and toward general and Yermo 2004). Lessons from recent experience make clear revenues. Moreover, in countries such as Mexico or Colombia that savings pillars should closely link benefits to contribu- where subsidized health regimes have been put in place to tions and do so in a similar way for most workers, regardless fill health coverage gaps left by the formal sector, general- of the status of the labor contract. In this context, the indi- revenue financing alleviates the problem of misaligned vidual capitalization schemes that have been introduced in labor market incentives associated with a system in Latin America over the last few decades are fully consistent which payroll tax­financed social security for the formal with the saving objective. sector "competes" with general revenue­financed (and In pursuing more effective protection against old-age government-subsidized) assistance programs for workers in poverty in the face of informality, Latin American and the informal sector (Levy 2006b). Caribbean countries face several important challenges While national health systems in Latin America and the regarding the design and operation of the poverty preven- Caribbean have long sought, in principle, to cover a wide tion and savings components, as well as the relationship range of health conditions via the public health services, between the two. Recent policy research on pension reform these systems have generally not conformed to sound insur- in Latin America and the Caribbean suggests, for example, ance principles, nor have they been able to deliver on their that from an institutional perspective, the poverty preven- promise with quality services. Overreach on the part of tion component is best financed and managed separately government has led not only to low-quality service in the from the savings component. Among other things, the insti- public health system, but unsustainable fiscal situations in tutional imperatives associated with managing a minimum some countries. In recent years, however, several countries poverty reduction payment are distinct from those associated have attempted to define more modest but implementable with the management and supervision of investments under minimum-benefit packages, such as the Regimen Subsidiado individual capitalization schemes (Gill, Packard, and Yermo in Colombia, Seguro Popular in Mexico, and the AUGE 2004). Moreover, weak design of one element can undermine 200 I N F O R M A L I T Y, S O C I A L P R O T E C T I O N , A N D A N T I P O V E RT Y P O L I C I E S BOX 7.3 Extending health insurance coverage by correctly aligning risk-pooling instruments The question of how best to protect households from the with the population covered by FONASA. The pilot financial consequences of illness and disability has been experience yielded positive results by increasing access to the subject of lengthy debate in Latin America. The core the services included in the package. Since January 1, of this discussion has focused on the trade-off between 2004, AUGE has been mandatory for all citizens and all the breadth of coverage (how many households can count public and private providers of health insurance. Both on some health protection) and the depth of coverage (the households covered by FONASA and those covered by package of care they can count on). The constitutions the private health insurers (called ISAPREs) are entitled of most Latin American countries guarantee the right of to the same minimum package of health services. citizens to good health and access to health services. In While it is too early to assess the effect of the Chilean theory, the public health system covers treatment of all reform on lowering vulnerability to poverty or easing the health conditions. However, in practice, health systems burden of health expenses for the poor, the "explicit enti- are financially constrained, and resort to rationing or tlements approach" embedded in the AUGE is already lowering the quality of care in order to comply with having both intended and unexpected effects in Chile and their budget constraints. At least until the mid-1990s, elsewhere in the region. For example, it has forced policy constitutional mandates existed alongside a large, unsat- makers to revisit the breadth and depth discussion of isfied demand for health care, and the promise of uni- health coverage. Before the reforms, all health services versal health care was not accompanied by effective were theoretically available to all citizens, but, in prac- instruments to achieve that. The casualties of this failure tice, nothing was actually guaranteed. The result was a tended to be the poor. Beginning in the mid-1990s, limited and poor-quality package of services, particularly however, several countries have introduced legislation for the poor. and sector reforms to transform health benefit packages Under the new explicit entitlement approach, a basic into well-defined and explicit entitlements guaranteed package of health coverage becomes legally binding for to all citizens. These reforms typically include legal mech- governments, throwing open the debate of whether to anisms for households to demand their entitlements guarantee a limited package to all or an extensive pack- from the state. age (often focused on the most costly health care) to a few. Since 2002, Chile has been gradually introducing a Moreover, these reforms have led to a much closer dia- new health insurance package (Plan of Universal Access logue between ministries of finance and health because with Explicit Guarantees, or AUGE). Rather than con- the legally binding nature of the package reduces the tinuing to offer unrealistic promises to cover all health space for fiscal adjustment in the health sector. Errors in needs, AUGE establishes a guaranteed minimum pack- defining the coverage of the package can thus have signif- age of basic health cover for all Chileans, along with icant fiscal consequences, putting a premium on effi- guarantees of attention within specified time periods, set ciency and management in the health sector. Indeed, ceilings on co-payments, and full subsidies for the poor- although largely unintended, most related reforms have est households. The new system is designed to eliminate provided governments with a powerful instrument to rationing and improve service. AUGE levels out premi- focus on broader efficiency-enhancing reforms in the ums and co-payments and reduces "cream skimming" by health sector. Guaranteeing a package requires clarity insurers. It represents an enforceable "patient's bill of regarding the quality of delivery, which, in turn, requires rights" for affiliates of FONASA (Fondo Nacional de complex monitoring systems, provider payment systems, Salud), the National Health Service, and is set up to and contracts or quasi-contracts between the public reduce waiting lines and other inefficiencies that can financing agency and health service providers. increase out-of-pocket costs, particularly for poorer households. AUGE was first tried out on a pilot basis Source: Baeza and Packard 2006. 201 I N F O R M A L I T Y the impact of another. For example, excessively generous or old-age security for all, including the poor and those badly designed poverty prevention components serve to who work outside the formal sector. In doing so, it reduce incentives for personal retirement savings and/or would move Chile from having two different exacerbate incentives for informal sector employment. systems--a defined contribution pillar system and a So what does this imply for the design of a poverty targeted social assistance pension--to a single inte- reduction component of a pension system? Given institu- grated system. The reform seeks to establish an tional and fiscal constraints that exist in the region--and appropriate mix of pooling and saving instruments, that vary significantly across Latin American and Caribbean an effective institutional structure for supervision countries--it is useful to draw on the discussion of incen- and management of the new system, and, impor- tives above, and to distinguish between those countries tantly, incentive compatibility of the poverty preven- with large coverage gaps, in which incentive issues may not tion and saving components of the system (box 7.4). be so important in the short run, and those with moderate- to-high levels of social security and social assistance cover- What about the design of the savings components in the age, where the incentive effects are likely to have greater face of pervasive informality? A key issue here is whether importance, even in the present. More specifically, these (or how) to mandate old-age savings among those who, by incentives could include the following: definition, operate outside of countries' administrative and enforcement mechanisms. Finding the appropriate balance · For countries with large coverage gaps, it makes between the mandatory savings pillar (so-called pillar 2) sense to focus first on the implementation of social and the voluntary savings pillar (so-called pillar 3) has been assistance pensions that are well targeted to the poor the focus of recent analysis of multipillar pension schemes (or extreme poor) and scaled to be fiscally sustain- in Latin America (Gill, Packard, and Yermo 2004). Part able.28 In this context, an important design challenge of the argument for mandating savings revolves around will be to set the level of the pension benefit large concerns about information failures and "myopia" that enough to provide protection against poverty, but might lead workers and households to under-save for retire- not so large that it creates significant adverse labor ment. Yet, working to ensure old-age savings opportunities market or saving incentives in the future when the for those in the informal sector would seem to argue for coverage gap between the social assistance pension strengthening voluntary savings instruments as part of a and social security closes.29 An important message multipillar system. It also argues for providing greater flex- here is that even in designing a well-targeted social ibility within the system to help raise the perceived bene- assistance pension, policy makers should consider not fits of savings relative to its costs. In this context, there will only short-term concerns about coverage, but also be gains from ensuring greater portability of savings and longer-term concerns about incentive compatibility benefits as workers move in and out of the formal sector. as the gap between contributory and noncontributory Providing greater flexibility in savings and investment programs closes over time. options, perhaps based on people's stage of the life cycle-- · For countries with moderate-to-high levels of social for example, and allowing lower levels of mandatory contri- security coverage, where incentive issues are likely to butions for younger workers, allowing younger workers to be more important in the short term, the need to select investments with higher risk-return profiles, while focus on the issue of incentive compatibility is more supporting lower-risk portfolios for older workers--may immediate. In such cases, in addition to having a also serve to raise people's demand for retirement savings. well-targeted (and well-means-tested) poverty pre- vention pension, some progressive reduction in the Unbundling complex, multidimensional benefit level with income would be advisable to help benefit packages maintain appropriate incentives throughout the As previously noted, a number of countries' social security broader system. In this regard, the pension reform systems require contributions to complex, multidimen- recently proposed by the government of Chile may sional benefit packages. As has also been noted, to the hold some valuable lessons for other countries in the extent that workers do not value one or more components region. The Chilean proposal focuses on providing of the package, those components represent a tax on 202 I N F O R M A L I T Y, S O C I A L P R O T E C T I O N , A N D A N T I P O V E RT Y P O L I C I E S BOX 7.4 Old-age protection in the new millennium: Chile's proposed pension reform In December 2006, the Chilean government sent to Con- Benefit structure under Chile's proposed pension reform (workers' total pension as a function of their self-financed pension) gress a draft law to reform the pension system. The pro- posed reform contains a broad set of measures, including Total pension the replacement of the existing targeted social assistance pension (PASIS) with a new solidarity pillar, extension of coverage to self-employed workers, norms to promote gender equity, and several norms aimed at improving the 200 performance and supervision of the individual accounts pillar. It is an ambitious and comprehensive proposal that builds on the existing system. To prevent old-age poverty, the current Chilean social 75 security system includes a minimum pension as part of its contributory pension scheme, as well as a social assis- tance pension. The minimum pension is available only to 0 0 200 Self-financed pension individuals who have contributed for at least 20 years to individual accounts, and there is some evidence that Source: Forteza 2007, based on Government of Chile 2006a, 2006b. many workers will not achieve the required contribution benefits to the self-employed as to salaried workers, history before retiring. At the same time, the social assis- including the solidarity benefits. Over time, participa- tance pension is targeted to the poorest. Thus, there is a tion of self-employed workers would become mandatory, concern that a significant share of the population might unless the person explicitly opts out. It is worth noting end up with very low--or no--pensions. that even though Chile currently has low coverage of self- The proposed reform would create a new solidarity employed workers, as in other Latin American countries, pillar that provides benefits to individuals in the lowest it does not have as high levels of informality among the 60 percent of income distribution who either have not self-employed (since most of them are registered and pay contributed to the system or have low contribution lev- other taxes). As such, the enforcement challenges associ- els. Those who have not contributed would receive a pen- ated with expanding coverage among the self-employed sion of Ch$75,000 (approximately US$142) per month. in Chile look relatively less onerous than in other coun- Those who have low contribution levels would receive a tries in the region. supplementary payment, such that the sum of their self- The reform effort also seeks to promote reductions in financed pension and the supplement would be no less the commissions charged by the pension fund adminis- than Ch$75,000 and no more than Ch$200,000 (approx- trators and to increase the return of the investments. imately US$380) per month. The size of the supplemen- Current commissions are considered high and are attrib- tary payment declines with the size of the pension that is uted to the lack of sufficient competition. The Superin- self-financed so that workers who can self-finance a pen- tendent of Pensions will organize annual auctions of sion of Ch$200,000 or more receive no supplement (see new affiliates based on the level of the commission; the box figure). Until that point, however, the total pension administrator who offers the lowest commission will get increases as the self-financed pension does, so that work- the new affiliates. The administrator will be required to ers receiving the supplement still have positive incen- keep the commission offered to new affiliates for at least tives to continue contributions. 18 months and extend this rate to all its affiliates. The The participation of self-employed workers in the cur- reform also seeks to raise the return of the funds by rent pension system is voluntary in Chile, and only about providing more flexibility to choose the composition of 5 percent of the self-employed have chosen to participate. The proposed reform would extend the same program (Continued) 203 I N F O R M A L I T Y BOX 7.4 Old-age protection in the new millennium: Chile's proposed pension reform (Continued) the portfolio. With more flexible supervision, administra- The pension system envisioned in the Chilean reform tors will have more responsibility for returns on the represents a movement away from what is predominantly funds. a single savings pillar to a more balanced multipillar, or The reform would also modify the institutional multi-tier, system. The proposal strengthens and better structure for managing and supervising the new system. integrates the assistance and redistributive components Specifically, a new Social Security Institute (Instituto de with the individual savings component of the system. It Prevision Social) would be in charge of the system of soli- improves the capacity of the pension program to protect darity pensions while a Superintendent of Pensions the less fortunate, without neglecting incentives and (Superintendencia de Pensiones) would supervise the whole avoiding strong fiscal impacts. The proposal also repre- system, including the Social Security Institute. In under- sents a valuable contribution in terms of the policy- taking the reform, the government of Chile envisions a making process. For example, there were extensive five-year transition period, with a progressive roll-out of consultations with stakeholders conducted by the com- the solidarity pillar. The fiscal implications of the reform, mission appointed by the government to develop the pre- while not trivial, appear manageable. The presidential liminary set of proposals. In addition, the government advisory commission overseeing reform of the social secu- chose to build its reforms on the existing system. The rity system estimates that the new solidarity pillar will result is a balanced project that improves the existing cost roughly 1 percent of GDP in 2025. The total fiscal system without disregarding the achievements made cost of pensions in that year would thus grow from a pro- under previous reforms. jected 1.6 percent of GDP in the absence of the reform to 2.5 percent of GDP with reform. Source: Forteza 2007, based on Government of Chile 2006a, 2006b. workers' earnings, generating disincentives to formal sector If governments see the need to provide a housing benefit, at employment. Such taxes adversely affect incentives for for- a minimum that benefit should be made voluntary or "elec- mal employment. tive." However, the rationale for using payroll taxes as an From the perspective of public social protection, some of instrument to finance such a benefit is unclear. If the objec- the components included in these complex packages have tive is to ensure that the poor have access to adequate hous- no clear risk-pooling or risk management rationale. Nor is ing, then a more efficient and appropriate way to provide there clear economic justification as part of a social security that benefit would be through direct subsidies targeted to package on the basis of providing public goods or positive the poor and financed by general revenues. externalities. Similarly, there is no clear justification for Some countries, such as Colombia, also provide subsidies payroll-tax financing. Take, for example, the cases of the to early childhood development financed through payroll sports and recreation or housing-related benefits in Mexico. taxes. Although investments in early childhood development Both of these benefits represent private goods; neither do have a strong "public goods" dimension, the rationale for involves risk pooling or social risk management. financing them through payroll taxes remains unclear. They, In such cases, social security packages would benefit too, are better financed through general taxation. from "unbundling" of what might be called "nonessential" elements. Unbundling would involve shedding--or Potential costs of social protection reform: making voluntary--those elements of the current social Financing essential cover security packages with no risk-pooling or risk management Reengineering Latin America's social protection systems rationale, or that finance private as opposed to public and, in particular, ensuring minimum essential cover in goods. The objectives of unbundling are to increase the health and pensions via de-linked social protection programs benefits relative to the costs of social security packages and will have important fiscal implications for most countries in focus social security on its core social protection functions. the region. First, de-linking implies eliminating payroll 204 I N F O R M A L I T Y, S O C I A L P R O T E C T I O N , A N D A N T I P O V E RT Y P O L I C I E S taxes (at least in these areas) and replacing them with rev- health, the expected fiscal impact of essential old-age cover enue sources that are less distortionary, such as the VAT, or depends critically on several factors, including desired cover- income or property taxes. Second, in most countries in the age levels, expected demographic changes, the proposed size region, providing essential cover will imply the need for of the benefit, whether benefits are indexed to prices or greater public resources to health and pension coverage. wages, and whether savings incentives are offered. For the The precise amounts of additional resources required and purpose of this exercise, estimates were made using a pro- the appropriate revenue source in each country will depend posed benefit of half a minimum wage (roughly equivalent to on a number of factors, including but not limited to the size the extreme poverty line in Mexico), indexed to real wages. of the coverage gap, the size of the benefit package being Under this scenario, and accounting for the aging of the offered by the government, and the country's fiscal space. Mexican population, it is estimated that the incremental fis- Calculating the fiscal impacts of providing essential cal costs of implementing a targeted social assistance pen- cover is a difficult exercise, as estimates are sensitive to sion would be 0.3 percent of GDP between now and 2010, changes in assumptions. In the case of health, for example, rising to 0.7 percent of GDP in 2050. Data from other coun- estimating the cost of providing essential cover is tricky, as tries in the region suggest that the fiscal impact of extend- the results are extremely sensitive to the definition of the ing pension coverage will vary depending on the program's size and content of the benefit package. Moreover, in a design and the country's institutional characteristics-- number of countries, there are potentially important fiscal although the estimates are generally consistent with the savings associated with micro-efficiency reforms in the findings for Mexico. The government of Chile, for example, health sector that would be important to facilitating essen- estimates that the proposed solidarity pillar will translate tial cover in health. Measuring the extent of those savings into a cost of around 1 percent of GDP by 2025 (box 7.4), and how they offset the additional costs of expanding cov- while a forthcoming study on pensions in Brazil estimates erage is also a challenge. Nonetheless, attempts to estimate that spending on the government's extensive rural and social the potential costs are important, and can be useful in iden- assistance pension schemes is equivalent to about 2 percent tifying general orders of magnitude of impact. of GDP (World Bank forthcoming). Recent World Bank estimates in the case of Mexico What might such reforms imply from the perspective of are illustrative. In terms of providing universal essential the tax system and, in particular, if such programs were to be cover in health, it is assumed that the unit cost of such pro- financed via general taxes rather than through payroll taxes? vision is equal to the unit cost of Seguro Popular, Mexico's Analysis was undertaken for Mexico to estimate the potential subsidized health insurance program for the poor. Launched effects of financing these reforms via the VAT. Specifically, it in late 2004, Seguro Popular covers a packet of basic health was assumed that all households would be taxed under an care, including preventive, outpatient, inpatient, emer- increased VAT. It should be noted that the current VAT in gency, and surgical interventions.30 This is a much more Mexico includes a number of exemptions, such as on food modest package of coverage than is afforded by the health and medicines. Estimates of the impact on VAT rates were insurance packages in IMSS and ISSSTE (the Social Security undertaken--with the assumption that these exemptions Institute for Mexican civil servants at the federal level). In were kept in place as rates were raised. this context, it is assumed that wealthier households would One concern sometimes voiced about raising the VAT is be free to purchase additional (supplementary) health insur- its costs to the poor. To address potentially adverse impacts ance in the private market, according to their preferences on the poor of an increased VAT, the estimates assume that and willingness to pay. Under this scenario, it is estimated the poorest Mexicans are directly compensated for the addi- that the additional fiscal costs of providing essential cover in tional tax impact.31 While direct compensation for higher health (that is, costs above and beyond what the govern- value-added taxes would not be provided to the nonpoor, it ment of Mexico currently spends on health care provision) should be noted that those working in the formal sector would be approximately 1.3 percent of GDP. would simultaneously benefit from the fact that they would The potential fiscal costs of providing a poverty preven- no longer have to contribute to the payroll tax. Moreover, tion pension were also estimated for Mexico--specifically, given that part of the payroll tax is paid by employers, cur- the costs associated with providing a targeted social assis- rent formal employees may be expected to receive part of tance pension to all elderly among the extreme poor. As with the current employer tax burden back in the form of higher 205 I N F O R M A L I T Y salaries.32 While these latter benefits to formal workers exemptions of the VAT on capital goods, then the adverse are noted, they were not included explicitly in the VAT effect on productivity and growth would disappear. calculations. In considering a shift from payroll tax to general Estimates of the cost of VAT financing of essential cover revenue­financed social protection, it is important to rec- for all Mexicans were undertaken, including provision for ognize that policy makers in Latin America are not starting the poorest 20 percent of Mexican households to be com- with a blank slate, but operate in a well-established policy, pensated for their increased tax burden. This is equivalent institutional, and political environment. That said, most to the roughly 5 million households currently covered OECD and many developing countries--all with their own under the Oportunidades conditional cash transfer program, history of policies and institutions--have already started to and the compensation could in principle be targeted using make the transition. Even countries with long traditions of the Oportunidades beneficiary registry. These estimates sug- Bismarck-type social insurance systems, such as Spain, are gest that to cover additional public spending equivalent to moving toward general taxation to finance minimum levels 1.3 percent of GDP--the amount necessary to provide of risk-pooling coverage (see box 7.5). In fact, countries minimum essential cover in health--an increase in the VAT that have opted for general-revenue financing typically rate of about 6.4 percentage points would be necessary, in started with fragmented voluntary and then mandatory the absence of a payroll tax and under the current VAT social insurance systems similar to the types found in Latin structure, including exemptions. That would imply an America and the Caribbean. Even in Germany, birthplace increase from the current VAT rate of 15.0 percent to one of of the Bismarckian social insurance model, recent proposals 21.4 percent, a change that is not trivial. to reform segments of the welfare system aim at moving In the case of providing a targeted social assistance pen- health finance toward general taxation. sion, resource needs increase over time before leveling out, The legacy of past social protection policies creates and this is reflected in the necessary VAT increases. Specifi- important challenges for policy makers in switching from cally, financing a well-targeted social assistance pension today's model based on payroll taxes to a new model based would require an increase of the VAT rate by approximately on general-revenue financing. However, lessons from recent 1.3 percentage points at present, but would require a reform countries, such as Spain, can provide some guid- cumulative increase in the VAT rate of roughly 3.4 percent- ance; some experience with this type of shift also exists in age points by 2050 (as the costs associated with providing a the Latin American and Caribbean region. Chile's pension social assistance pension, under current assumptions, rise). reform of 1981 included shifting some elements of payroll It should be noted, that, under both the health and pension tax­financed social insurance to general revenues, and was scenarios, generalizing the VAT to include taxes on food accompanied by a substantial decrease of payroll taxes and medicines--were such a move politically feasible-- (Gruber 1995). Thus, technical solutions exist for how to would have a significant mitigating impact on the need to implement the shift to a social protection model that can raise rates, even if compensation of the poor for broader cover all citizens.33 VAT taxation were included. Relying on the VAT to finance a de-linked social protec- Managing the transition from here to there tion system may not be the optimal solution in all coun- Latin America and the Caribbean countries exhibit a high tries, however; indeed, other instruments, such as income or degree of heterogeneity in levels of social protection cover- property taxes, may be more appropriate, given the specific age (the "flip side" of informality). They also demonstrate context. Recent analysis suggests that in Brazil, for exam- great heterogeneity in social protection spending and the ple, switching from a payroll tax to a VAT in its current capacity to dedicate additional resources to social protection form could have a negative impact on economic growth in in the short-to-medium term. Social protection institutions the long run (Fernandes, Gremaud, and Narita 2006). This and institutional capacity also differ from place to place. So is because the VAT in Brazil does not completely exempt while some countries may be in a position--and have the investment goods from the taxation, and this additional political will--to effect significant changes in the structure taxation on capital goods has a long-term negative impact and financing of social protection in the near term, others on growth. Nonetheless, Fernandes, Gremaud, and Narita will need to focus on more incremental challenges, as they also find that, in Brazil, if measures were taken to permit move toward longer-term goals of ensuring universal access 206 I N F O R M A L I T Y, S O C I A L P R O T E C T I O N , A N D A N T I P O V E RT Y P O L I C I E S BOX 7.5 De-linking health coverage from employment status: Spain's shift to general-revenue financing of essential social insurance The health-financing system in Spain underwent radical Consistent with the reforms introduced by the Gen- changes in the 1980s and 1990s, when that country eral Health Care Act, the sources of funding for the shifted from a social insurance system financed by payroll health care system were drastically modified in 1989. taxes to a national health service financed by general tax- Beginning in that year, new budgets were financed ation. Today, almost 100 percent of public expenditures 70 percent from general taxation and only 30 percent in health are financed from general taxation. from payroll tax contributions. Spain's transition to democracy and the constitution of In the mid-1990s, consistent with agreements 1978 gave new impetus to health care reform. A separate signed by political parties and trade unions (known as organization was established within the social security the Pactos de Toledo), 100 percent of financing would system for the administration of health care services. Most come from general taxation and individual contribu- health care programs and organizations were consolidated tions were to be progressively phased out by 2000. In under the umbrella of the Ministry of Health, which was 1999, one year ahead of schedule, the entire health care established as an independent entity in 1981. budget in Spain was financed from general taxation. In 1986, the General Health Care Act was approved Today, Spanish regions receive funding for health care following almost four years of public and parliamentary as part of the general funding from the central govern- discussion. The act provided a unified legal framework for ment. Funding is proportional to population adjusted many of the previous piecemeal reforms and called for a for such factors as age distribution, number of tempo- tax-based financing system. Publicly managed health ser- rary residents, and services provided to the national sys- vices were consolidated in a single national organization tem or to neighboring regions. Supplemental funding and in a small number of regional organizations (Cataluña is also supplied from fiscal revenues raised in the and the Basque Country), within the framework of the region. newly decentralized state. The Spanish National Health Currently, only workers' compensation for work- System was subsequently devolved to the 17 Autonomous related injuries and diseases is financed from individual Communities that, since 2001, have fully managed their contributions of employers and employees. Regional Health Services and together make up the Spanish National Health System. Source: Fernandez 2004, in Baeza and Packard 2006. to basic risk management mechanisms in health and old- To this end, governments in the region would benefit age security. from taking several sets of incremental reforms in the short- Indeed, in pursuing a long-term vision and strategy for to-medium term that would contribute in important minimum essential cover in social protection, most coun- ways to the long-term goals of extending social protection tries in the region will likely need to pursue a series of dis- in health and old-age security to all its citizens. These crete measures, rather than "big bang" types of reforms. reforms are beneficial in their own right. At the same time, This may be particularly pragmatic given that the needed they can help pave the way toward a system of essential cover reforms require the support of a diverse set of institutional that facilitates greater mobility in the labor market, not and political actors. In this context, it will be important to based on the informal-formal distinction but on sensible ensure that any short-term measures countries take are con- design of social protection and on flexible, incentive- sistent--or at least not inconsistent--with the governments' compatible labor market institutions. Key priorities include long-term visions and agendas. This will be especially the following: important if governments choose to pursue more inte- grated social protection systems that align social objectives · Unbundling of complex, multidimensional benefit packages. of better risk management with economic objectives of The objective of unbundling would be to increase the higher productivity and growth. benefits relative to the costs and improve the public 207 I N F O R M A L I T Y goods content of social security packages. While · Strengthening program design to account for and enable health and pension benefits would remain at the core of greater worker mobility. This should include: the benefit package, unbundling would involve shed- revising overly burdensome vesting periods, where they ding--or making voluntary--those elements of the exist, in the region's pension programs to facilitate present social security packages that currently worker access and mobility. In this context, Latin represent pure or partial taxes on workers or finance American and Caribbean countries should con- "private" as opposed to "public" goods. sider adopting the types of points programs used · Improving program/benefit quality in health and pensions. in some European countries that enable workers to Again, the objective here would be to increase access pension benefits in proportion to their time the relative benefits associated with participation in in service. the social security system. Key areas for attention enabling spouses and/or secondary family workers to opt out include: of specific benefits, suchashealthinsurance,soastoavoid micro-efficiency reforms in the health sector of most coun- making families pay twice for the same coverage. tries in the region. Recent studies of the health sec- fostering portability of health benefits and retirement tor in the region (for example, Baeza and Packard savings across jobs, sectors, and occupations. 2006; Mason et al. forthcoming) highlight several · Establishing consistency and incentive compatibility of pro- key areas that are critical to strengthening social gram structures and benefits across different parts of (and financial) protection in health, including: countries' social protection systems (both within * separation of the purchaser/insurer and health social security and across social security and social care provider functions, assistance) to minimize adverse incentives and pro- * moving from historical budget processes to ductivity effects. This includes efforts to "harmonize" production-based budgeting and strategic rules, eligibility requirements, and benefits levels purchasing, across programs and institutions--both in health and * developing minimum benefit packages (essen- in old-age security. tial cover) that are consistent with the funda- While the actions listed here constitute an important mental principles of insurance, and short-to-medium-term reform agenda for most countries * establishing--or strengthening--consumer choice in the region, these actions are particularly important in among service providers. countries in which there is a high degree of integration and efficiency reforms in the region's pension systems. Among movement between the formal and informal sectors, as well the priorities are: as in countries where the gaps in coverage between formal * lowering costs and administrative fees to affili- social security and emergent social assistance programs are ates and improving risk management (Gill, relatively small. Packard, and Yermo 2004); further reductions in commissions, along with efforts to raise net Conclusion rates of returns, would go far to improve the Nearly all countries in Latin America and the Caribbean attractiveness of the funded savings pillars, remain characterized by "truncated welfare systems," in * strengthening voluntary savings instruments for which those in the formal sector have access to an often gen- old age, and erous multidimensional package of social security, while * creating greater flexibility in the savings pillars, those outside the formal sector--whether in urban or rural including greater portability and flexibility in areas--have much more limited access to government ben- savings and investment options to reflect differ- efits and/or formal risk management instruments. Recent ent stages of workers' life cycles (for example, progress in extending social security to uncovered portions lower mandatory contributions for younger of the population has been disappointing, at best. Overall, workers and/or greater flexibility in the invest- social security coverage has failed to increase (informality ment-risk profile across the life cycle); such flex- has failed to decline) despite economic growth over the ibility would help raise the implicit benefit­cost period; and coverage has actually declined in a number of ratio among potential affiliates. countries over the last decade. Moreover, in nearly all 208 I N F O R M A L I T Y, S O C I A L P R O T E C T I O N , A N D A N T I P O V E RT Y P O L I C I E S countries in the region, coverage rates are significantly example, information problems and externalities) lower among low-income than high-income workers. In abound. As such, there is a strong case for public many countries, the poorest workers and families are practi- intervention to strengthen households' abilities to cally excluded from the system. manage risk. Partly in response to concerns about the truncated wel- · Third, alongside market failures that justify public fare state, several countries in the region have launched or action, there is evidence of widespread government expanded noncontributory assistance and/or poverty reduc- failure in the design and implementation of social tion programs over the last decade to help breach the cover- protection. These problems serve to exacerbate peo- age gap and provide support and coverage to the poor or ple's lack of access to appropriate risk management extreme poor. A number of these efforts (for example, con- mechanisms. For example, ditional cash transfer programs) have contributed in impor- at the level of specific programs, design problems tant ways to the short-term welfare and long-term raise the costs of participating in social security opportunities of the poor. These efforts are welcomed and relative to its benefits, causing some workers to are to be encouraged. At the same time, the creation opt out of the system; design issues also serve to and expansion of certain types of assistance programs, espe- hinder some workers' eligibility for benefits; cially those that link eligibility to work in the informal at the level of social protection systems--the sector (or to being unemployed), may be creating incentives constellation of contributory social security and that constrain the growth of formal sector employment as noncontributory social assistance programs-- well as long-term economic growth in the region. These programs often compete, creating adverse labor incentive effects are particularly serious in countries where market incentives and outcomes. workers perceive that the benefits of affiliating with social · Finally, the combination of inaccessibility to basic security are low relative to the costs. In sum, the region risk management instruments, market failure, and faces both new and long-standing challenges in providing government failure creates an urgent need to rethink appropriate risk management instruments to its people. and, in fact, "reengineer" social protection in the Against this background, this chapter has reviewed the region--looking beyond the traditional Bismarckian state and recent evolution of social protection in Latin model in which protection is based on one's labor America and the Caribbean; examined the role and limita- contract to one that ensures protection to people on tions of households' private risk management strategies the basis of citizenship. and outlined the rationale for public social protection in the region; analyzed the key challenges policy makers face So what does this imply for the future of social protection in making adequate risk management instruments avail- in the region? Drawing from the economics of insurance, able to the regions' citizens, both those working in the for- the chapter has outlined a long-term agenda for social pro- mal sector and those in the informal sector; and proposed tection reform in the region. In the case of health, because several key directions for policy to ensure that all citizens of shocks that go "uncovered" can impose significant external Latin American and Caribbean countries are adequately costs on society, there is a case for providing a package of protected from key risks, especially those associated with minimum essential direct cover, de-linked from the labor health shocks and poverty in old age. Several key messages contract and financed through general taxation. In the case emerge from the analysis: of old-age security, there is also a case to provide essential cover in the form of a poverty prevention pension focused · First, despite ample evidence that individuals and on the poor, and as part of a broader multipillar pension their families engage in a number of private risk system that includes provisions for individual retirement management strategies, many people in the region-- savings. Whether countries should focus first on a well- particularly those in low-income households--remain targeted social assistance pension for the poor or on devel- vulnerable to the impoverishing effects of health opment of a more integrated system of old-age security shocks and to poverty in old age. depends on a number of factors, including the size of the · Second, private insurance markets are generally thin current coverage gap, the country's fiscal space, and local or missing in the region and market failures (for institutional capacity. In either case, it will be important 209 I N F O R M A L I T Y for policy makers to ensure incentive compatibility 3. The framework is based on Ehrlich and Becker (1972) and between the poverty prevention pension and the savings Gill and Ilahi (2000). 4. For a more detailed discussion of the factors that have driven component, which should be the mainstay of earnings increases in informality in the region, see chapter 4. replacement during old age. 5. While the rural pension system in Brazil is formally contribu- For a variety of reasons, including those related to fiscal tory, contributions are not linked to salaries or individual income, and institutional capacity, movement to minimum essen- but to rural production. Benefits are not linked in any way to past tial cover in health and old-age security--de-linked from contributions, but are defined in relation to the minimum wage, and the labor market and financed by general taxes--represents the system is heavily subsidized. For a detailed discussion, see Schwarzer and Querino 2002. a long-term agenda for many countries in the region. In 6. Indeed, the BONOSOL program covers nearly 75 percent of this context, it will be important for countries to orient the population and in a very egalitarian manner. As can be seen in their short-to-medium-term policy agendas in ways that figure 7.5, in the absence of this scheme (that is, relying only on con- are consistent--or at least not inconsistent--with their tributory pensions), Bolivia would be among the lowest-coverage longer-term vision. This will be critical if the region's countries in the region (Rofman and Lucchetti 2006). governments are to align more effectively the objectives of 7. Absolute levels of coverage by Seguro Popular, which was launched in 2004, have increased since the data shown in figure 7.6 better social risk management with those of higher produc- were collected. Newer data suggest that Seguro Popular covered as tivity and growth. many as 11.5 million individuals by the first quarter of 2006 (Gaki- To this end, governments in the region will benefit from dou et al. 2006). According to the October 2005 census, 21 percent making incremental reforms that improve the efficiency of of Seguro Popular affiliates belong to the second income (wealth) decile existing systems, as well as establishing greater consistency and 19 percent belong to the poorest decile. The fraction declined and incentive compatibility across program benefit struc- sharply as income increases. 8. See Lindert, Skoufias, and Shapiro (2006), who present OECD tures. Several sets of actions will contribute to short-term data from 2001. Using a smaller sample of Latin American countries, improvements in social protection while moving countries they find that average public social protection spending in the region in the direction of essential cover in the long term. This was approximately 5.7 percent of GDP around the late 1990s and includes efforts to improve the benefit­cost ratios of pro- early 2000s. grams via unbundling of complex, multidimensional bene- 9. To the extent that financing of expanded assistance programs fit packages; undertaking microefficiency reforms in comes at the expense of infrastructure or other investments that increase a country's competitiveness, such efforts may also have unin- countries' health and pension systems; and harmonizing tended negative effects on long-term growth (see Levy 2006b; World rules, eligibility requirements, and benefits levels across Bank forthcoming). programs and institutions. Such measures will enable 10. As chapter 2 shows, obtaining access to health insurance greater worker mobility and provide the foundation for through a relative is an important reason for not affiliating in more effective social protection for all citizens. Colombia (table 2.13). 11. Notable exceptions to this are seen in Colombia and Uruguay (Baeza and Packard 2006). Notes 12. This appears to be the case, notwithstanding important ineffi- 1. Depending on how programs are financed, legitimate efforts ciencies in Brazil's pension system. For more detailed analysis of the to cover increasingly large segments of the population could also have strengths and weaknesses of the Brazilian pension system, see World unintended negative effects on a country's long-term growth. For Brazilian Bank (forthcoming). example, Levy (2006b) argues that increases in social protection 13. In Mexico, for example, only 2 percent of households have pri- spending in Mexico have come at the expense of investments in infra- vate insurance (Mason et al. forthcoming), while in Colombia, less structure that are necessary to keep Mexico's economy competitive in than 5 percent of formal independent workers and less than 1 percent the global economy and, if continued, will have adverse effects on the of informal workers (either independent or salaried) have private country's growth performance. Similar concerns have been raised health insurance (see ch. 2, table 2.12). about the costs of high (and inefficient) levels of social protection 14. Indeed, in Bolivia, 42 percent of informal salaried workers and spending on infrastructure investment and, thus, future growth and 55 percent of informal independent workers reported that the main competitiveness in Brazil (see World Bank forthcoming). reason for not contributing to social security was lack of knowledge 2. This conceptual framework has been developed and extended about how the system works (see ch. 2, table 2.12). progressively in a series of studies on social protection in Latin 15. High rates of immunization play a prevention role not only America and the Caribbean, including Baeza and Packard (2006), de for those directly immunized, but also for society in general. As such, Ferranti et al. (2000), and Gill, Packard, and Yermo (2004). they have a strong public goods element to them. So, too, do goods 210 I N F O R M A L I T Y, S O C I A L P R O T E C T I O N , A N D A N T I P O V E RT Y P O L I C I E S that help prevent the transmission of sexually transmitted diseases 25. Universal health insurance for pregnant mothers and children (for example, HIV/AIDS), such as condoms, which not only serve to under five years of age and subsidized child care for those working in protect the individual users, but also help to slow the spread of such the informal sector were campaign proposals during the presidential diseases more generally (Packard 2006). campaign of Enrique Calderón and, at the time of this writing, were 16. Data collected by the IMSS between 1997 and 2005 indicate under development by the new Calderón administration. that roughly two-thirds of the almost 9 million IMSS-affiliated workers 26. While Gasparini, Haimovich, and Olivieri's (2006) results are these are "low-income workers," earning less than three minimum illustrative, they need to be interpreted with some care. Their find- wages (Levy 2006a). It is possible that most if not all of these workers ings are not robust to all econometric specifications tested. Moreover, are unable to access the IMSS housing subsidy (or their related they note that the control group may well differ from the treatment contributions). group in some dimensions that are not observed and cannot be con- 17. These figures almost certainly understate mobility across the trolled for in the data. If such factors affect the probability of finding formal and informal sectors in the Mexican economy, as the data do a formal job, then their econometric results could be biased. not capture rural workers. It is not uncommon for rural workers to 27. See the section titled "Managing the transition from here to migrate for work--and move in and out of formality--on a seasonal there" later in this chapter. For more detail on leverage points for basis (Levy 2006a, 2006b). improving health sector efficiency in Latin America and the 18. At first glance, these patterns appear to be at odds with other Caribbean, see Baeza and Packard (2006). recent findings for Argentina and Uruguay, based on similar types of 28. Estimates of the likely fiscal costs of a well targeted social administrative data. For example, Farall et al. (2003) and Bertranou assistance pension are presented later in the case of Mexico. and Sánchez (2003) for Argentina and Lagomarsino and Lanzilotta 29. How social assistance pension benefits are adjusted for infla- (2004) for Uruguay present bimodal densities of pension contribu- tion over time can also affect the incentive structure across programs. tions (although with significant numbers of individuals in intermedi- If benefits are adjusted by the Consumer Price Index rather than by ate cases, so there still appears to be a continuum of cases in their real wages, they will keep their value in terms of protection against data). It is important to note, however, that these other analyses do poverty, but become an increasingly less attractive alternative to not control for the age of the workers. This is important, as younger participation in the savings pillars as the economy and people's real workers (aged 18­26) tend to be overrepresented among those with a incomes grow. low density of contributions, while older workers (aged 40­48) tend 30. The unit cost of Seguro Popular was equal to Mex$8,500 in the to be overrepresented among those with a higher density of contribu- fall of 2004. This included a cost of Mex$7,500 for the basic package, tions. Once the age is controlled for, one finds the type of multimodal but an additional Mex$1,000 allocated for catastrophic health care. It distribution found in Bucheli et al. (2006) and shown in figure 7.11. should be noted that although Seguro Popular is called "insurance," at The authors thank Alvaro Forteza for clarifying this issue. present there are very few premiums charged to affiliates under the pro- 19. Only when the worker reaches age 70 is he or she entitled to a gram. In principle, premiums will be charged on a graduated scale, special (small) pension--although even that requires a minimum of based on household income, with the poorest Mexicans receiving a full 15 years of contributions to qualify. (100 percent) subsidy, but with progressively wealthier households 20. As might be expected, Bucheli et al. (2006) also find that paying progressively higher premiums (see Mason et al. forthcoming). achieving the required years of formal sector service is more difficult 31. A study by Levy and Dávila (2003) examines the option of among private sector than among public sector workers. raising value-added taxes as a revenue-enhancing measure and then 21. Another related issue concerns the general lack of portability providing compensating transfers to the poor. They similarly find in the region's pension systems, whether in and out of the formal sec- this approach a viable and efficient option. tor, across the private and public sectors, or across specialized occupa- 32. Estimates of the potential costs of promoting health insurance tional schemes that exist in some countries. Although there is little portability in Chile (Baeza and Copetta 1999) suggest that moving or no direct evidence to date on the effects of the lack of portability of from payroll tax to general revenue-financing of health insurance pensions on informality, it stands to reason that by curbing job could have a neutral fiscal effect or even lead to tax savings--at least mobility, lack of portability also impedes entry (and exit) from the in the Chilean context--if greater formal sector participation and formal sector. earnings due to the removal of payroll taxes translate into higher 22. Maloney (2001) and Loayza and Rigolini (2006) use cross- earnings, and, thus, more income tax revenues. sectional estimation techniques on cross-country data, while Krebs and 33. For a more detailed discussion of informality, taxation, and the Maloney (1999) do simulations based on labor transitions in Mexico. social contract, see chapter 8. 23. In the Brazilian case, if capital is not exempted, the authors find a decline in physical accumulation that decreases the demand for References formal workers in the long run and leaves informality relatively unchanged (Fernandes, Gremaud, and Narita 2006). Arias, O. 2004. "Poverty in El Salvador during the 1990s: Evolution 24. It should be noted that the MAROP program was launched in and Characteristics." Photocopy. Poverty Reduction and Eco- 2006 by the Fox administration, but as of the writing of this chapter, nomic Management Unit, Latin America and the Caribbean it has not been implemented. Region, World Bank, Washington, DC. 211 I N F O R M A L I T Y Baeza, C., and C. Copetta. 1999. "Análisis conceptual de la necesidad paper for the regional study, Beyond Survival: Protecting Households y factibilidad de introducir mecanismos de ajuste de riesgo en el from Health Shocks in Latin America. World Bank, Washington, DC. contexto de portabilidad de los subsidios públicos en el sistema de Fiorito, R., and F. Padrini. 2001. "Distortionary Taxation and Labor seguros de salud en Chile." CLAISS, Santiago, Chile. Market Performance." Oxford Bulletin of Economics and Statistics 63 Baeza, C., and T. Packard. 2006. Beyond Survival: Protecting Households (2): 173­96. from Health Shocks in Latin America. Washington, DC: World Bank. Forteza, A. 2007. "Old-age Protection in the New Millennium: Beneke de Sanfeliú, M., and M. Shi. 2004. "Dinámica del ingreso Chile's Proposed Pension Reform." Background note for the rural en El Salvador." Photocopy. FUSADES, San Salvador, El regional study, Beyond Survival: Protecting Households from Health Salvador. Shocks in Latin America. World Bank, Washington, DC. Bertranou, F., and A. Sánchez. 2003. "Características y determinantes Gakidou, E., R. Lozano, E. González-Pier, J. Abbott-Klafter, J. T. de la densidad de aportes a la Seguridad Social en la Argentina Barofsky, C. Bryson-Cahn, D. M. Feehan, D. K. Lee, H. 1994­2001." International Labour Office, Geneva. Hernández-Llamas, and C. J. L. Murray. 2006. "Assessing the Bitran, R., U. Giedion, and R. Muñoz. 2004. "Fondos de riesgo, Effect of the 2001­06 Mexican Health Reform: An Interim ahorro y prevención: estudio regional de políticas para la protec- Report Card." Lancet 368: 1920­35. ción de los más pobres de los efectos de los choques de salud: Galiani, S., and F. Weinschelbaum. 2006. "Modeling Informality Estudio de caso de Chile." Background paper for the regional Formally: Households and Firms." Photocopy. World Bank, study, Beyond Survival: Protecting Households from Health Shocks in Washington, DC. Latin America. World Bank, Washington, DC. Gasparini, L., F. Haimovich, and S. Olivieri. 2006. "Labor Informal- Bourguignon, F., M. Cicowiez, J. J. Dethier, L. Gasparini, and P. ity Effects of the Programa Jefes de Hogar." Photocopy. CEDLAS Pestieau. 2004. "What Impact Would a Minimum Pension Have (Centre for Distribution, Labor and Social Studies, National Uni- on Old Age Poverty? Evidence from Latin America." Paper pre- versity of La Plata), Buenos Aires, Argentina. sented at the Keeping the Promise of Old-age Security Confer- Gill, I., and N. Ilahi. 2000. "Economic Insecurity, Individual ence, Bogotá, Colombia, June 22­23. Behavior, and Social Policy." Office of the Chief Economist, Latin Bucheli, M., N. Ferreira-Coimbra, Á. Corteza, and I. Rossi. 2006. America and the Caribbean Region, World Bank, Washington, "El acceso a la jubilación o pensión en Uruguay: ¿cuántos y DC. quiénes lo lograrían?" Economic Commission for Latin America Gill, I., T. Packard, and J. Yermo. 2004. Keeping the Promise of Social and the Caribbean, United Nations, Montevideo, Uruguay. Security in Latin America. Washington, DC: World Bank. Cunningham, W. 2001. "Breadwinner Versus Caregiver: Labor Force Glewwe, P., and G. Hall. 1998. "Are Some Groups More Vulnerable Participation and Sectoral Choice over the Mexican Business to Macroeconomic Shocks Than Others? Hypothesis Tests Based Cycle." In The Economics of Gender in Mexico: World, Family, State, on Panel Data from Peru." Journal of Development Economics 56: and the Market, ed. E. G. Katz and M. C. Correia, 85­132. 181­206. Washington, DC: World Bank. Government of Chile. 2006a. "Informe del Consejo Asesor Presiden- de Ferranti, D., G. E. Perry, F. H. G. Ferreira, and M. Walton. 2004. cial para la Reforma Previsional." Santiago, Chile. Inequality in Latin America: Breaking with History? Washington, ------. 2006b. "Reforma Previsional. Protección para la Vejez en el DC: World Bank. Nuevo Milenio." Santiago, Chile. de Ferranti, D., G. Perry, I. Gill, and L. Servén. 2000. Securing Our Gruber, J. 1995. "The Incidence of Payroll Taxation: Evidence from Future in the Global Economy. Washington, DC: World Bank. Chile." Working Paper 5053, National Bureau of Economic Ehrlich, I., and G. Becker. 1972. "Market Insurance, Self-Insurance Research, Cambridge, MA. and Self-Protection." Journal of Political Economy 80: 623­48. Holzmann, R., and S. Jorgensen. 2000. "Social Protection as Social Farall, R., D. Aisenberg, M. Bourquin, E. Caviglia, F. Dabusti, Risk Management: Conceptual Underpinnings for the Social Pro- S. Mychaszula, A. Sanmiguel, and D. Zagaglia. 2003. "Estudio sobre tection Sector Strategy Paper." Journal of International Development la frecuencia de los aportes en el SIJP. Una primera aproximación 11: 1005­27. hacia las historias laborales." In Historias Laborales en la Seguridad Jalan, J., and M. Ravallion. 1999. "Are Poor Less Insured? Evidence Social, 17­36. Social Security Secretariat Publication Series, Min- on Vulnerability to Income Risk in Rural China." Journal of Devel- istry of Labor, Employment, and Social Security, Argentina. opment Economics 58: 61­81. Fernandes, R., A. P. Gremaud, and R. D. T. Narita. 2006. "Tax Klugman, J., D. Kruger, and K. Withers. 2003. "Consumption Risk Reform and Formalization of the Economy: Simulating Different and Smoothing during Disasters: The Case of Hurricane Mitch in Alternatives for Brazil." Photocopy. Department of Economics, Nicaragua." Photocopy. World Bank, Washington, DC. University of São Paulo, Brazil. Krebs, T., and W. Maloney. 1999. "Quitting and Labor Turnover: Fernandez, J. M. 2004. "Background Note on Spain's Health Reform Microeconomic Evidence and Macroeconomic Consequences." for the Regional Study, Beyond Survival: Protecting Households from Policy Research Working Paper 2068, World Bank, Washington, Health Shocks in Latin America." World Bank, Washington, DC. DC. Fiedler, J. 2004. "Out-of-pocket Health Expenditures, Risk Pooling, Lagomarsino, G., and B. Lanzilotta. 2004. "Densidad de aportes a Savings, and Prevention: A Honduras Case Study." Background la Seguridad Social en Uruguay: Análisis de su evolución y 212 I N F O R M A L I T Y, S O C I A L P R O T E C T I O N , A N D A N T I P O V E RT Y P O L I C I E S determinantes a partir de los datos registrales de historia laboral Rawlings, L., and G. Rubio. 2005. "Evaluating the Impact of Condi- (1997­2003)." Document prepared for the team for workers' rep- tional Cash Transfer Programs." World Bank Research Observer 1: resentation at the Banco de Previsón Social, Montevideo, 29­55. Uruguay. Razú, D. 2006. "Competencia Desleal entre Políticas Públicas en Levy, S. 2006a. "Productividad, Crecimiento y Pobreza en México: México: El Modelo de Seguridad Social vs. Programas Asisten- ¿Qué sigue después de Progresa-Oportunidades?" Photocopy. ciales." Photocopy. Mexico City. World Bank, Washington, DC. Rodriguez-Meza, J., and C. Gonzalez-Vega. 2004. "Household ------. 2006b. "Social Policy, Productivity and Growth." Back- Income Dynamics and Poverty Traps in El Salvador." Photocopy. ground paper for the regional study, Beyond Survival: Protecting Ohio State University, Columbus, OH. Households from Health Shocks in Latin America. World Bank, Rofman, R. 2006. "El Sistema de Pensiones." Bolivia Policy Notes, Washington, DC. World Bank, Washington, DC. Levy, S., and E. Dávila. 2003. "Taxing for Equity: A Proposal to Rofman, R., and L. Lucchetti. 2006. "Pension Systems in Latin Reform Mexico's Value-added Tax." In Latin American Macroeco- America: Concepts and Measurements of Coverage." Social Pro- nomic Reforms: The Second Stage, ed. J. A. Gonzalez, V. Corbo, A. O. tection Discussion Paper 0616, World Bank, Washington, DC. Krueger, and A. Tornell. Chicago: University of Chicago Press. Sadoulet, E., F. Finan, A. de Janvry, and R. Vakis. 2004. "Can Condi- Lindert, K., E. Skoufias, and J. Shapiro. 2006. "Redistributing tional Cash Transfer Programs Improve Social Risk Management? Income to the Poor and the Rich: Public Transfers in Latin Lessons for Education and Child Labor Outcomes." Social Protec- America and the Caribbean." Social Protection Discussion Paper tion Discussion Paper 0420, World Bank, Washington, DC. 0605, World Bank, Washington, DC. Savedoff, W. 2004. "Tax-Based Financing for Health Systems: Loayza, N., and J. Rigolini. 2006. "Informality Trends and Cycles." Options and Experience." Health Financing Issues Paper EIP/ Photocopy. World Bank, Washington, DC. FER/HFP/PIP.04.4, Department of Health Systems Financing, Maceira, D. 2004. "Mecanismos de protección social en salud e Expenditure, and Resource Allocation, World Health Orga- impacto de shocks financieros: El caso de Argentina." Background nization, Geneva. paper for the regional study, Beyond Survival: Protecting Households Schwarzer, H., and A. C. Querino. 2002. "Beneficios sociales y los from Health Shocks in Latin America. World Bank, Washington, DC. pobres en Brasil: Programas de Pensiones no Convencionales." In Maloney, W. 2001. "Self-employment and Labor Turnover in Devel- Pensiones No Contributivas y Asistenciales, ed. F. Bertranou, C. oping Countries: Cross-country Evidence." In World Bank Econo- Solorio, and W. van Ginneken, 63­124. Santiago, Chile: mists' Forum. Washington, DC: World Bank. International Labour Organization. Martinez, S. 2005. "Pensions, Poverty and Household Investments in Skoufias, E. 2004. "Household Vulnerability and Its Determinants in Bolivia." Photocopy. Department of Economics, University of Rural and Urban Mexico." Photocopy. World Bank, Washington, California at Berkeley. DC. Mason, A. D., J. Saavedra, M. L. Escobar, G. Lopez-Acevedo, and World Bank. 2001. Social Protection Sector Strategy: From Safety Net to M. Rubio-Sanchez. Forthcoming. "Strengthening Social Protec- Springboard. Washington, DC. tion in Mexico: Recent Progress, Future Challenges." Mexico ------. 2003a. "Argentina Crisis and Poverty 2003: A Poverty Transition Policy Notes, World Bank, Washington, DC. Assessment." Report 26127-AR, Poverty Reduction and Eco- Montenegro, F. 2004. "Household Health Expenditures, Financial nomic Management, Latin America and the Caribbean Region, Protection, and Poverty in Ecuador." Background paper for the Washington, DC. regional study, Beyond Survival: Protecting Households from Health ------. 2003b. "Volatility, Risk, and Innovation: Social Protection Shocks in Latin America. World Bank, Washington, DC. in Latin America and the Caribbean." SPectrum, Washington, DC. Mossialos, E, A. Dixon, J. Figueras, and J. Kutzin. 2002. Funding Health ------. 2005. "Shocks and Social Protection: Lessons from the Cen- Care: Options for Europe. Berkshire, UK: Open University Press. tral American Coffee Crisis." Report 31-8570-CA, Poverty Reduc- Packard, T. 2002. "Are There Positive Incentives from Privatising tion and Economic Management and Human Development Social Security? A Panel Analysis of Pension Reforms in Latin Sector Management Units, Latin America and the Caribbean America." Journal of Pension Economics and Finance 1 (2): 89­109. Region, Washington, DC. ------. 2006. "On Social Protection Policy and the Informal ------. Forthcoming. "Brazil: Towards a Sustainable and Fair Pen- Economy: An Essay." Background note for the regional study, sion System." Photocopy. Latin America and the Caribbean Beyond Survival: Protecting Households from Health Shocks in Latin Region, Washington, DC. America. World Bank, Washington, DC. ------. Various years. World Development Indicators. Washington, Ravallion, M. 2003. "Targeted Transfers in Poor Countries: DC. Revisiting the Trade-offs and Policy Options." Social Protection Discussion Paper 0314, World Bank, Washington, DC. 213 CHAPTER 8 The Informal Sector and the State: Institutions, Inequality, and Social Norms SUMMARY: This chapter argues that the state­society interactions underlying informality, whether leading to exclusion or to exit, ultimately constitute an indictment of the overall effectiveness and legitimacy of the state. As noted before, bur- densome business and labor market regulations, poorly designed social protection systems, and weak enforcement capabili- ties bias the cost­benefit assessments of firms and workers in favor of informality. Furthermore, a collective perception of ineffectiveness, unfairness, and illegitimacy of the state's actions, in terms of who it represents and serves, can give rise to a social norm of noncompliance with taxes and regulations (a "culture of informality"), which further undermines the state's capacity to enforce the law and to provide effective public services. This is partly related to the inability of the state to redress the long-standing high inequality in the region, resolve social tensions, and uphold the rule of law; and it crys- tallizes in a dysfunctional underlying "social contract." In Hirschman's terms, it is a lack of voice and loyalty. Thus, addressing informality in Latin America requires bolstering the competency and legitimacy of the state by delivering the correct set of "carrots" in more equitable public policies and programs to foster a sense of greater inclusion and responsive- ness, while carefully considering the incentives created to become formal or informal, and by wielding adequate "sticks" through political resolve and evenhanded enforcement of laws and regulations. T HIS REPORT HAS VIEWED INFORMALITY other regulations. For instance, the high tax compliance through the lens of the relationship between rates observed in developed countries cannot be explained economic agents and the state. Several chap- only by the deterrent effect of the chance of being caught ters have discussed how specific policies in evading and the ensuing penalties. the areas of labor legislation, business regu- In Latin American economies, with large informal sec- lation, social protection, and taxation can have a critical tors, the individual cost­benefit analysis leading to exit impact on microlevel decisions that foster or preclude from taxation and other regulations, and from participating agents' participation in the formal economy. These policies in the formal circuit of taxes and transfers, in the Hirsch- define a comprehensive set of incentives to which individu- manian sense, may be influenced by a collective perception als and firms respond by weighing the costs and benefits that the state's action is ineffective, inefficient, and unfair. of participating in different markets and the enforcement In a social exchange view, willingness to comply with regu- efforts and capabilities of the state. This chapter argues that lations may be affected by the individual perception of these economic responses are also influenced by how the effectiveness of the government in providing services. agents--individually and collectively--perceive and define Moreover, a collective perception--the "sentiment" within a relationship with the state. In some situations, there are a group--about the performance of the state influences the "social norms" that are influenced by the perception of the social norm regarding compliance. For instance, if more effectiveness of the state and the collective projects it repre- people operate in the informal sector, it might be easier for sents, and that foster willingness to comply with taxes and each citizen to do so; for a given level of state enforcement 215 I N F O R M A L I T Y effort, one person will be less likely to be sanctioned if lots inequitable and of low quality. Firms, workers, and citizens of people also evade taxes or do not comply with other reg- in general make decisions based on the state's enforcement ulations; the psychological/ethical costs of evading will be capacity and on their perceptions about the effectiveness of lower if most people in one's peer group do so, and hence the state and of prevalent social norms. These decisions, in one's tax morale and disposition to comply with regula- turn, affect the capacity of the state to enforce regulations tions will be lower. This interdependent behavior might and provide high-quality public goods and services for all. generate a social multiplier that might make a specific Alternatively, an economy may be in a "good equilib- social norm more prevalent. This type of mechanism might rium" where a large mass of the economy operates formally be behind what has been dubbed a "culture of informality" and public goods and services are provided effectively and that prevails in many Latin American countries. where the government is able to enforce the agreed regula- Informality is also a reflection of mechanisms that tions and taxes--which, in turn, is facilitated by a social exclude large segments of the citizenry from education, norm that induces people to comply. Even if economies are, health care, and judiciary services; and from economic in general, not "stuck" in any of these equilibria, in many opportunities through a segmented labor market and cases they have features that resemble these feedback loops imperfection in other factor markets. This exclusionary that might impede the reduction of informality. process is related to the extremely high and persistent lev- Throughout this report we have described many features els of inequality, which are rooted in differences in power, that relate to different measures or dimensions of informal- voice, and influence; and which, as reported here, seem ity and that, indeed, are reflections of a systemic failure. empirically highly correlated with informality. Among those features, we have Latin America's low level of trust in the state, its culture of informality, and often the design of its regulations and · low levels of participation in the social security system policies reflect what we might call a "dysfunctional social · low coverage of many social insurance schemes, espe- contract" under which the state is not complying with its cially among poor people designated roles and individuals therefore see little point in · a large number of small firms (and larger ones) that playing by its rules. We use the expression social contract to partially or completely evade tax, labor, and business refer to some degree of societal consensus over basic aspects regulations of the operation and role of the state relative to the private · low-quality regulation that increases red tape sector and among citizens. In this usage, social contract refers · exclusion in the access to property rights, judiciary to key aspects of a social equilibrium, including beliefs and services, and other public services actions of citizens, organized groups, and state actors. · low-quality public provision of many social services Among the aspects that enable us to characterize this con- (such as health care or education) tract are the structure of taxation and social expenditures · low levels of trust in the state and in the fairness of (Lledo, Schneider, and Moore 2004), the performance of the dominant arrangements state in using citizen's taxes in delivering public goods, and · low and uneven enforcement the structure and effectiveness of the social protection sys- · with exceptions, low levels of tax collection, related tems (Birdsall and Menezes 2005; ECLAC 2006). to low compliance and low tax bases. Heuristically, this situation may be characterized by a "bad equilibrium" where certain norms are being upheld, Each of those features is, in itself, a reflection of a dys- implicitly or explicitly. The state might be ineffective, able functional individual and collective interaction with the neither to enforce well its regulations--the sticks--nor to state, intimately linked with the state's inability to perform lure firms and individuals to formality by providing quality effectively and equitably its main roles--roles of remedy- public goods and services for all--the carrots. ing market failures, coordinating the provision of public Many Latin American countries seem to be in a situation goods, and maintaining a level and equitable playing field. where the share of informality is high; trust in the state is Seen from a less state-centered perspective, the features low; tax morale and regulatory compliance are low and, mentioned above may also be a reflection of dysfunctional hence, enforcement of regulations and tax collection are social equilibrium in the "horizontal" relationships among generally low; and public provision of public services is citizens--how they interact with and the degree to which 216 T H E I N F O R M A L S E C T O R A N D T H E S TAT E : I N S T I T U T I O N S , I N E Q U A L I T Y, A N D S O C I A L N O R M S they trust each other. In other words, several of the phe- simultaneously in several realms. As an illustration, con- nomena that may be related to the notion of informality sider this nontaxonomic list: are, in the end, a reflection of the way individuals interact both with the state and with each other. 1. obtaining the protection of labor laws, paying contri- In this chapter, we explore several characteristics of the butions, and getting social security benefits for the social contracts that prevail in the different countries of the worker and his or her family region. This chapter also will present suggestive evidence 2. avoiding the costs of being caught not complying of the kind of feedback loops mentioned above. We dwell with regulations, given state enforcement technology on the different manifestations of informality that concern 3. contributing to the provision of public goods and the relationship between the individual and the state, services to society as a whole, and trusting the state emphasizing how social norms and social interactions also in doing it might affect the decision maker. We discuss and present 4. avoiding the peer-pressure cost of being singled out some evidence related to perception and performance of the as a cheater. state and how they correlate with informality. The chapter then focuses on the roles of taxes and how tax compliance is To assess costs and benefits, the worker will take into related to the real and perceived role and effectiveness of consideration a number of factors that have the state as the the state. We also discuss the structure of taxes and trans- main actor. Among others, the informality decision might fers and its relation to inequality and informality, as well as be affected by the following: other elements that illustrate the exit or opting-out mecha- nisms that are observed in the region. · The direct costs and benefits of formality: The assessment of point (1), above, depends on the costs of labor market Social norms, the state, and informality regulations and a comparison of those costs with the Latin America's high level of informality is a manifestation valuation of the benefits provided to the individual of disconnects between the state and citizens, in part as a (for example, comparing payroll tax payments with result of failures of the state in its various roles. Here we health benefits received, as discussed in chapter 7). present some evidence of firms' and individuals' perceptions Note, however, that what the government can pro- of the performance of the state, and how these perceptions vide depends on its fiscal capabilities, which, in turn, correlate with indicators of informality. To frame the analy- depend on the decisions of agents whether to be for- sis of these correlates, we provide first a framework to dis- mal (and pay taxes and contributions). Massive opt- cuss different dimensions of the decision to participate in ing out may generate a negative feedback loop that the formal economy. might move the country to a bad equilibrium. · State enforcement capacity: The assessment of point (2) Dealing--or not--with the state will depend on the perceived capacity of the state to Throughout the report, we have stressed the cost­benefit enforce labor, tax, and other regulations. This is a analysis that individuals undertake in deciding what sector function of the enforcement technology used and the to work in, whether to register their firm, and whether to perceived probability of being caught. This is influ- pay their taxes or risk detection and punishment. In this enced by individual risk aversion and by the collective section, we further explore how individual perceptions perception of this probability. In addition, however, about the performance of the state and how social norms the formation of these collective perceptions may be and social interactions might also impinge on these deci- influenced by social multipliers. Another negative sions to opt in or out.1 Consider the case of a worker or feedback loop is possible here because low enforce- a microentrepreneur who has two employment/business ment capacity implies fewer resources for the state. opportunities that are somewhat comparable in some other · Individual perceptions about government effectiveness: The dimensions (such as net earnings), but that differ in the fact assessment of point (3), above, might depend on how that one is formal and requires complying with all regula- effective and fair the individual perceives government tions, and the other is informal. From the individual point institutions to be in fulfilling its role. The individual of view, the occupational choice entails having a position might not be amoral and might decide to comply 217 I N F O R M A L I T Y with regulations not only as a result of the cost­benefit from paying directly through taxes, in part because it is too analysis of a utility maximizer, but might also be costly for the state to go after them. At the same time, factoring in the degree of citizen/taxpayer satisfaction however, they are excluded from the benefits of being pro- with, confidence in, and trust in government. A per- tected by the systems regulated by the state or are excluded ception of state incompetence, unfairness, and corrup- from receiving public services. Among many others, an tion may affect willingness to comply. The literature example of this exclusion is the inadequate access that the finds that, in collective action settings, individuals poor have to the judiciary. If, for instance, individuals do adopt not a purely materialistic, calculating posture, not perceive that property rights will be enforced, commu- but a more complex, emotional, and reciprocal stand.2 nities will maintain traditional mechanisms to enforce · Social norms and the social multiplier: The assessment of property rights.3 Another example of exclusion is the exis- point (4) depends on what others believe about the tence of large segments of the population that are not cov- role and performance of the state, that is, the social ered by any social protection mechanism, the so-called norm--the pattern of behavior that constitutes a cus- truncated welfare state discussed in chapter 7. And, in a tomary rule that coordinates actions among people most extreme case of informality generated by an exclu- (Young 2006) and is sustained by social approval-- sionary process, large segments of the country, often in that influences collective behavior toward the state. rural areas, have never been reached by the state. A serious Deterrence is not enough to explain the observed lev- manifestation of this is the problem of unregistered births els of tax and regulatory compliance, and at least parts and undocumented citizens which is still significant in of those levels are explained by social interactions. several countries, as the evidence available for Bolivia, Moreover, if one's own behavior changes not only Brazil, the Dominican Republic, Nicaragua, and Peru because of the influence of an exogenous determinant demonstrates (see box 8.1). but also because of the change in the behavior of a A perception that there is an incompetent state due to cor- reference group, the result is a social multiplier. ruption or ineffective governance generates low trust in the state. In addition to directly reducing the benefits of The revision of these dimensions of the formality deci- engaging in formal interactions, this might reduce the sion illustrates that informality connects to issues pertain- willingness to comply voluntarily with regulations. Invest- ing to the reality of and perception about the state. The ment climate survey data concerning the efficiency of gov- answer to each of the questions above depends on some ernments to deliver services show that Latin America characteristics and capabilities of the state and the services appears to fare particularly poorly,4 featuring the lowest it provides, which affect the assessment of the private proportions of firms being satisfied with the efficiency of benefit of engaging with the state. But it also depends on government services (panel [a] of figure 8.1). Across coun- a system of individual and collective beliefs about the tries, there is a negative relationship between a proxy of effectiveness and fairness of the existing arrangements. informality and governments' effectiveness (even after con- trolling for the level of development). Panel (b) of figure 8.1 Informality and performance and perceptions shows significant negative correlation between a proxy for about the state government effectiveness5 and a proxy for informality. Many Latin American states share the characteristics of what Informality also tends to be higher in countries that are can be called an exclusionary state, in itself another manifes- perceived as more corrupt (figure 8.2). As discussed in tation of an imperfect social contract. It is a contract as long chapter 6, data for five Latin American countries suggest as it is an implicit arrangement through which the society that the perception of corruption is positively and signifi- has given the state some of the roles mentioned above. cantly related to informality.6 Companies reporting that However, in Latin America this contract is failing to bribing government officials to "get things done" is a com- define social and economic arrangements that are inclusive mon practice in their line of business exhibit higher rates of and that provide fair rights and responsibilities to all. This revenue and worker underreporting (see table 6.1). This generates a perception that the state is not complying fully could be interpreted as firms hiding some of their activities with its role in the social contract. Manifestation of this is a from corrupt officials. But a complementary explanation is political equilibrium where certain groups are exempted that firms that view the government as corrupt are less 218 T H E I N F O R M A L S E C T O R A N D T H E S TAT E : I N S T I T U T I O N S , I N E Q U A L I T Y, A N D S O C I A L N O R M S BOX 8.1 The extreme of informality and exclusion: Being undocumented A significant share of the population in several Latin an indication that a key reason for the high rates is the American countries lacks official identity cards, and in this prohibitive cost of traveling to an urban center where sense is largely invisible to the state. This is an extreme of there is a government office. Other reasons that are often the continuum of informality by which groups of individ- suggested are a lack of knowledge of the importance of uals are absent from the formal circuits of the economy birth certificates and the legal quagmires surrounding and any interaction with the state. Lack of documentation registration (for example, if parents are undocumented, it can have serious consequences, such as making an individ- may be very difficult or impossible to register a child). ual vulnerable to exploitation by employers because of the The effects of not being registered are serious. Using inability to seek legal redress, preventing access to basic household surveys, Duryea, Olgiati, and Stone (2006) public services and transfer programs, and curtailing the found that for the nearly 300,000 children between 7 and accumulation of human capital through public education. 9 years old who did not matriculate in Brazil, the main rea- Due to the nature of the phenomenon, regional estimates son cited for not enrolling was the lack of documentation. of the magnitude of this population are hard to come by. Econometric analysis for the Dominican Republic found However, some representative statistics are available. The that lack of documentation was one of the strongest Inter-American Development Bank (IDB 2006) reports predictors of school enrollment in higher grades. The that demographic and health surveys conducted in International Program on the Elimination of Child Nicaragua in 2001 showed that 17 percent of people over Labor­International Labour Organization found that 15 years of age did not have a national identity (ID) card. 50 percent of children living in Velleda Morales, a largely In the Dominican Republic, a 2004 living conditions sur- indigenous community in Honduras, did not attend school vey estimated that about one-fourth of the poorest seg- because they lacked a birth certificate. The inability to ment of the population lack birth certificates and identity increase one's human capital through education has long- cards. A survey for a planned project in Argentina indi- term consequences on earnings potential. In contrast, the cates that 14 percent of potential beneficiaries lacked Inter-American Development Bank (IDB 2006) reports national ID cards in one municipality and 17 percent in that access to health care services for nondocumented another municipality. The UK Department for Interna- people is much easier in most Latin American countries tional Development found that, in 2002, between than access to education--with the exception of Colombia 750,000 and 2 million Bolivian citizens were "function- where health services require two forms of identification. ally undocumented." Furthermore, it was found that, in Countries in the region have responded to the regis- some areas of Bolivia, 90 percent of the population lacked tration crisis by combining social targeting programs and a valid form of identification, and the majority was not registration initiatives. The Inter-American Develop- included in the civil registry at all. ment Bank (IDB 2006) presents a number of examples, Undocumented adults were, in most cases, once undoc- such as Chile where registration has become a part of the umented children who have continued living in a state of nationwide Chile Solidario program that offers, among official nonexistence. Statistics for birth registrations are other things, health care services, counseling, and educa- more readily available in the region, Duryea, Olgiati, and tion services to poor families. Similarly, the Brazilian Stone (2006) find that the lack of birth registration varies government has introduced special boats that offer a mul- from 8.4 percent in Peru to 25.8 percent in the Domini- titude of services to isolated populations living deep in can Republic. They also find that birth underregistration the Amazon region, including medical services, birth rates are much higher in rural areas than urban areas. In registration, civil marriage, voter registration, and mili- Peru and Bolivia, rural rates are between 25 and 40 per- tary conscription. Similar initiatives are being developed cent higher than in urban areas; however, in Nicaragua, in the Dominican Republic. Despite these initiatives, rural rates were 200 percent higher than urban rates. additional government efforts and innovations are needed Furthermore, the highest birth underregistration rates in to include these invisible segments of the population in Nicaragua are found in areas suffering extreme poverty, the mainstream of the economy and the social contract. 219 I N F O R M A L I T Y FIGURE 8.1 Informality and government effectiveness a. Firms with favorable b. Correlation of informality perception of government's effectiveness with government effectiveness Percent Share of self-employed 70 coef .074, t 4.65 0.2 60 50 0.1 VEN BOL COL 40 PER HON GUA JAM CHI 0 ECUARG BRA COS 30 MEX PAN SAL 20 0.1 10 0 0.2 East Asia Middle South Asia Europe and Sub- Latin America 1.5 1.0 0.5 0 0.5 1.0 1.5 and East and Central Asia Saharan and the Pacific North Africa Africa the Caribbean Goverment effectiveness Source: Author's estimates, based on World Bank World Development Indicators and Worldwide Governance Indicators, (2005); and investment climate surveys. Note: Figure b shows partial correlations controlling for gross domestic product (GDP) per capita at purchasing power parity (PPP). Government Effectiveness Index measures the quality of public service provision, the bureaucracy, the competence of civil servants, the independence of the civil service from political pressures, and the credibility of the government's commitment to policies. Higher values correspond to a more effective government. Firms' perception of government effectiveness is defined as the proportion of firms that report that their government is efficient in delivering services. FIGURE 8.2 FIGURE 8.3 Informality and corruption Share of firms confident that the judiciary will enforce contractual and property rights Share of self-employed coef .07, t 4.56 Percent 0.4 90 80 0.2 70 60 VEN BOL 50 PER COL 0 GUA HON 40 ARG ECU CHI BRAPAN JAM COS MEX SAL 30 20 0.2 10 1.0 0.5 0 0.5 1.0 1.5 0 Control of corruption index and ca Asia Africa Africa and Source: Author's estimates, based on World Bank World East OECD c Asia Asiacifi Development Indicators and Worldwide Governance AmeriCaribbeanSouth Pa Europe East Indicators, (2005). MiddleNorth Central the and the Latin Sub-Saharan and Note: Figure shows partial correlations controlling for GDP per capita at PPP. The control of corruption index is a measure of Source: Investment climate surveys, (2005), as reported in perceptions, where corruption is defined as the exercise of public World Bank (2007). power for private gain, with higher values corresponding to less perception of corruption. willing to pay taxes and finance the state. As will be dis- generates very low levels of trust in the judiciary. On aver- cussed later, this unwillingness relates to the discussion of age, firms in Latin America are less confident that their the so-called tax morale. judicial system will enforce contractual and property rights Many judicial systems in Latin America are perceived as disputes than are firms in other regions of the world incompetent, inefficient, and unfair; and this perception (figure 8.3). Within the region, there are large differences, 220 T H E I N F O R M A L S E C T O R A N D T H E S TAT E : I N S T I T U T I O N S , I N E Q U A L I T Y, A N D S O C I A L N O R M S with more than 70 percent of firms being confident in the Latin American countries, is clearly negatively correlated judicial system in Chile and Costa Rica compared with less with informality, even after controlling for GDP. More gen- than 30 percent in Ecuador and Guatemala. Evidence pre- erally, an indicator of the rule of law from the WGI data- sented by Biebesheimer and Payne, (2001) shows that base, which reflects perceptions of enforcement capacity, of although 65 percent of citizens were reasonably confident the efficiency of the police and judicial systems, as well as in their judicial systems in Europe, only 35 percent the popular observance of the law, is negatively correlated expressed such confidence in Latin America. In questions with informality. Broadly, Latin American countries follow asked in recent investment climate surveys, which included closely the regression line in these graphs. a larger sample of small firms, the percentage of firms that perceived courts as fair, impartial, and uncorrupt was less FIGURE 8.4 than 25 percent in Panama, Mexico, Argentina, Bolivia, Share of firms that consider the court system fair, and Peru (figure 8.4). impartial, and uncorrupt As discussed by Ronconi (2006), there is ample evidence Percent that people do not trust the judiciary, in part because 70 judges are perceived as corrupt and as benefiting the 60 wealthier side in a lawsuit (Eyzaguirre 1996). Although 50 systematic evidence is lacking, he argues that it is plausible that low levels of trust in the system translate into fewer 40 employees whose labor rights might have been violated 30 bringing their cases to courts because those employees have 20 low expectations of success or because they perceive that 10 they will lack the financial resources to pay for their 0 defense--a perception that often is justified. This percep- ama ina tion reduces the probability that employers will be fined Peru Uruguay Pan Mexico Bolivia Colombia Argent and, hence, induces a social norm of noncompliance. As observed in figure 8.5, impartiality of courts, which Source: Author's estimates, based on investment climate surveys (2005). the previous figures show is perceived as low in a sample of FIGURE 8.5 Informality and state competence indicators Share of self-employed Share of self-employed coef .03, t 4.3 coef .073, t 4.32 0.2 0.2 0.1 0.1 VEN VEN PER BOL COL COL BOL PER GUA HON JAM HON ECU GUA JAM BRA CHI CHI 0 ARG ECU PAN MEX COS 0 ARG MEX BRA PAN COS SAL SAL 0.1 0.1 0.2 0.2 4 2 0 2 4 1.0 0.5 0 0.5 1.0 1.5 Impartiality of courts Rule of law index Source: Author's estimates, based on World Development Indicators (2005); Fraser Institute (2004); and Worldwide Governance Indicators (2005). Note: Figure shows partial correlations controlling for GDP per capita at PPP. Impartiality of courts is defined as the degree to which a trusted legal framework exists for private business to challenge the legality of government actions or regulation. The rule of law index measures, in broad terms, the respect of citizens and the state for the institutions that govern their interactions. 221 I N F O R M A L I T Y A related view of informality posits that it is the conse- (table 8.1). The first two variables, business flexibility and quence of weak Latin American states that assign them- law and order, that represent the opportunity cost of infor- selves an unbearable and unmanageable regulatory load, mality, and government expenditures (which the authors use with a scope that goes beyond their enforcement capacity-- as a proxy for the capacity to enforce) that represent the direct what Centeno and Portes (2006) call a frustrated state.7 cost of informality all have the expected signs and are signifi- Chapters 1 and 2 reviewed the evidence supporting the cant when they replace GDP in the specification. When the belief that regulatory barriers to entry are positively related four variables are included, significance is lost because of the to informality. Djankov et al. (2002) and Loayza, Oviedo, strong correlation with GDP per capita. and Servén (2005), for example, find that regulatory barri- Within the same theoretical framework, a simple empir- ers to entry are positively related to informality. ical exercise was conducted to analyze alternative institu- Loayza and Rigolini (2006) put several of the elements of tional indicators and alternative variables to approximate this literature in a theoretical framework and hypothesize informality, namely, the share of the informal economy, the that the size of the informal sector increases with the regula- ratio of self-employment to total employment, and the tory burden but decreases with the efficiency in the provision ratio of workers without contribution to pensions or social of services and with compliance enforcement. This long-run security to the total number of workers.9 Even controlling relationship is tested empirically through a regression in for GDP per capita, in most cases the institutional variables which informality, proxied by the ratio of self-employment have a significant and negative sign (table 8.2). The to total employment, depends on the flexibility of the busi- dummy for Latin America is significant in specifications ness environment, the quality of public services (such as the for two of the proxies of informality.10 prevalence of the rule of law, the efficiency of police and judi- A final strand of the literature emphasizes the possible cial systems), and an indicator of the government's ability to existence of a captured state, where a political equilibrium is monitor taxes and enforce regulations, proxied by govern- preserved such that elites (business, public sector, or labor) ment expenditures as a percentage of GDP.8 As generally interact explicitly or implicitly with the state to maintain found in the cross-country literature, GDP explains approxi- rents, even if that implies the exclusion of certain segments mately 80 percent of the variation of this proxy of informality of the population. The typical situation is that of a populist TABLE 8.1 Long-run informality relationships OLS Self-employment rate (as ratio to total workers) (I) (II) (III) GDP per capita (in logs, annual) 0.0759*** 0.0516*** 0.0043 0.0060 Business flexibility (index from Fraser Institute, 0.0293** 0.0167* range: 0­10, country average) 0.0111 0.0092 Law and order (as % of GDP, country average) 0.0457*** 0.0191*** 0.0072 0.0050 Government expenditure (as % of GDP, country average) 0.0050** 0.0015 0.0022 0.0015 Constant 0.9065*** 0.6954*** 0.9030*** 0.0388 0.0666 0.0424 Observations (n) 525; 42 525; 42 525; 42 R2 0.80 0.72 0.85 Source: Loayza and Rigolini 2006. Note: OLS = ordinary least squares. Robust, country clustered estimations, standard errors. * significant at 10 percent. ** significant at 5 percent. *** significant at 1 percent. 222 T H E I N F O R M A L S E C T O R A N D T H E S TAT E : I N S T I T U T I O N S , I N E Q U A L I T Y, A N D S O C I A L N O R M S TABLE 8.2 Indicators of informality and institutional indicators Definition 3: Definition 1: Definition 2: Share of persons without access Share of informal Economy Share of self-employed to pension Indicators (I) (II) (III) (I) (II) (III) (I) (II) (III) GDP (in logs) 5.23 3.12 3.83 7.88 7.60 6.68 22.70 21.67 23.53 (3.72)*** (2.20)** (2.49)** (5.55)*** (5.05)*** (4.41)*** (10.41)*** (9.11)*** (9.28)*** Regulatory quality index 3.42 3.44 5.59 (1.96)* (2.00)** (2.07)** Rule of law 6.19 3.60 6.36 (3.69)*** (2.05)** (2.33)** Government effectiveness 5.03 4.82 3.79 (2.78)*** (2.75)*** (1.28) Latin America 8.89 6.34 7.52 2.45 0.71 0.74 14.06 11.97 13.61 (3.27)*** (2.33)** (2.74)*** (1.07) (0.28) (0.31) (3.69)*** (2.99)*** (3.38)*** Observations (n) 110 110 110 60 60 60 98 98 98 R2 0.47 0.52 0.49 0.74 0.74 0.75 0.82 0.82 0.81 Source: Author's estimates, based on World Development Indicators and Worldwide Governance Indicators, 2005. Note: Absolute value of t-statistic is shown in parentheses. Regulatory Quality Index: higher values correspond to a more effective regulatory policy. * significant at 10 percent. ** significant at 5 percent. *** significant at 1 percent. government, with a political base in the middle and low- Some stylized facts of the tax structure middle class, that is unwilling to eliminate subsidies to Tax revenues in Latin America remain below the interna- public services, even if that crowds out expanding the tional norm. Figure 8.6 shows that almost all Latin services to poorer segments with less political clout. American countries lie below the regression line relating tax collection to per capita GDP. Lledo, Schneider, and The tax side of the social contract Moore (2004) show that the average tax take has been per- in Latin America sistently smaller in Latin America than in OECD coun- Tax compliance is one of the main components of the deci- tries, while it has been fairly similar to the Asian average sion to engage in the formal economy. In fact, as mentioned (see table 8.3). The median country in the region collects in chapter 6, some of the private "benefits" of informality 4 percentage points of GDP less than would be expected, are related to cost savings derived directly from eluding the given its level of development (Perry et al. 2006). The payment of taxes and avoiding the often complex adminis- average figure for the regions stands at 14.2 percent of trative procedures associated with tax and regulatory GDP in 2000­03 (15.8 percent of GDP including compliance. The large informal sector both reflects ineffi- social security contributions). Only Brazil, Nicaragua, ciencies and inequities of the tax system and defines the Trinidad and Tobago, and Jamaica have tax burdens-- magnitude of the challenge to improve tax collection in the without social security contributions--above 20 percent. region. As long as tax compliance is a social transaction Guatemala, Paraguay, and the República Bolivariana de between citizens and the state, it is a key element of the Venezuela barely reach 10 percent. social contract. We review here the main features of the tax Indirect taxes, in the form of taxes on domestic and system in Latin America and discuss tax compliance behav- internationally traded goods and services, represent the ior and selected policy issues. bulk of Latin American tax revenues. They make up about 223 I N F O R M A L I T Y observe and monitor incomes; and it is consistent with a FIGURE 8.6 concentration of taxes on businesses, presumably medium- Central government tax revenue and GDP per capita size and large businesses that are easier to monitor. The per- Total tax revenue (% of GDP) centage of corporate income tax revenues triples the 30 percentage of personal income tax revenues. Further, when benchmarked by GDP per capita, Latin American collec- 25 JAM tions of personal income taxes appear unduly lower than 20 HON CHIARG those in comparable countries. It is interesting to note that URU DOMBRA 15 BOL NIC VEN in all regions, except Latin America and Africa, direct taxes PERCOL COS ECUGUA PAN MEX 10 have increased their share during the last decades. Tax collection has been rising modestly in most of Latin 5 6 7 8 9 10 11 America since the early 1990s (Lora 2006b). Most of the Log GDP per capita increase in tax collection since the late 1980s is explained Source: Author's estimates, based on World Development Indicators by value-added tax (VAT) collections, which increased from (2006) and Tanzi (2000). 2 percent of GDP in the mid-1980s to an average of 5 per- cent in 2003,11 becoming one of the most important 60 percent of total tax revenues--almost twice as impor- sources of revenue for governments in the region. The tant as taxes on income, profits, and capital gains. By income tax collection, however, remained flat at around 4 per- comparison, in developed countries tax revenues from cent of GDP during the same period, while revenues from international trade and from domestic goods and services trade, excise, and other taxes fell sharply (figure 8.7). make up 40 percent of total tax revenues (see table 8.4). Part of the trend observed in VAT collection is related to a Undertaxation of income, wealth, and property is a persis- small increase in the VAT rate since the mid-1980s, reaching a tent characteristic of Latin American tax systems. In Latin median rate of 15 percent by 2003 (figure 8.8). There is still America, direct taxes are about a third of tax collections; great rate disparity in the region, with rates ranging from in Europe, half; and in North America, that share reaches 5 percentinPanamato25percentinBrazil.Tarifftaxratesand 80 percent. All the difference is explained by individual dispersion fell sharply as a consequence of trade liberalization. income taxes. These are a third of total tax collections in Corporate and personal income tax rates also experi- Europe and more than 60 percent in North America. In enced important changes. Corporate rates decreased from Latin America, the share is in the single digits--lower than 41 percent in 1985 to 29 percent in 2003. To encourage any other region of the world. The share of corporate direct investment in an environment with higher capital mobil- taxes, however, is not small, and it is only below that in ity, these rates were set below the maximum rate in devel- Asia. This situation is a reflection of a very low capacity to oped economies. The personal income tax shows quite TABLE 8.3 Comparative perspective of tax burdens and structures (percent of GDP) Tax revenue Direct Domestic indirect International trade Economic region 1975­85 1986­97 1975­85 1986­97 1975­85 1986­97 1975­85 1986­97 OECD 28.3 34.2 9.9 12.4 8.6 10.5 1.0 0.5 Asia (South and Southeast) 14.0 15.8 5.3 5.6 4.2 5.8 4.0 3.4 Africa 17.6 18.8 5.7 6.3 4.3 5.7 6.4 5.6 Latin America 14.9 15.2 3.8 3.4 5.0 5.5 3.1 2.2 Caribbean 23.3 22.4 6.4 0.0 6.7 6.2 6.3 5.4 Source: Lledo, Schneider, and Moore 2004. Note: Total tax revenue includes social security contributions. 224 T H E I N F O R M A L S E C T O R A N D T H E S TAT E : I N S T I T U T I O N S , I N E Q U A L I T Y, A N D S O C I A L N O R M S TABLE 8.4 Tax structure by region, selected years, 1975­2002 (percentage of total tax revenue) Income taxes Domestic goods and services General International Region Total Individual Corporate Total consumption Excises trade North America 1975­80 78.4 56.9 20.5 15 7.7 6.5 6.6 1986­92 78.8 63.5 14.4 17 9.8 6.3 4.3 1996­2002 83.3 66.3 15.8 14.8 8.8 5.1 1.8 Latin America 1975­80 32.7 11.1 17.6 40.4 17.1 19.3 26.8 1986­92 31.1 8.5 17.6 47.3 20.9 21 21.5 1996­2002 30.4 6.2 18.5 56.3 34 16.1 13.3 Western Europe 1975­80 42.7 33.3 8.5 50.6 28.6 16.5 6.7 1986­92 43.4 32.9 9.3 53.4 33.4 14.9 3.2 1996­2002 47.2 32.8 13 52.4 31.8 15 0.3 Asia 1975­80 38.8 22.9 20.5 37.2 14.3 18.3 24.1 1986­92 39.3 20.8 19.2 39.5 17.4 16.7 21.2 1996­2002 46.9 24.2 21.4 40.2 19.6 15.3 12.9 Africa 1975­80 32.1 14.6 16.1 29.7 18.4 13.5 38.2 1986­92 27.4 14.6 11.4 31.9 18.3 11.9 40.7 1996­2002 30.7 17.7 11.6 36.2 21.8 11.3 33.2 Source: Bird and Zolt 2005. varied approaches in the region. For example, Bolivia has a rose from 24 percent in 1985 to 34 percent in 2000, while flat rate tax of 13 percent, whereas Colombia has a number income tax productivity also increased but at a lower pace of progressive rates (132 in all), ranging from 0.26 percent (figure 8.9). These average figures show still very low levels to 35 percent; Chile similarly has a very progressive sched- in the region, although they hide a great deal of hetero- ule with eight brackets and with minimum and maximum geneity. There is great disparity in the region in the overall rates of 5 percent and 40 percent, respectively. Personal effectiveness in using the existing tax structure. For exam- income maximum rates also decreased sharply in the ple, as shown in figure 8.10, VAT productivity ranges from period. On the other hand, personal exemption levels a high of 0.64 in Chile to a low of 0.17 in Guatemala. increased from an average of 60 percent of GDP per capita Differences in productivity across countries are probably in 1985 to 230 percent in 2003 (an unusually high level by linked to different VAT structures and to tax expenditures international standards), and the income levels taxed at (that is, to multiple tax exemptions and to deductions maximum rates (the cutoff for the upper-income bracket) granted to specific sectors, particular geographic regions, were lowered sharply. It is important to note that, in some or, in some cases, to specific taxpayers (see table 8.5), as countries, the personal income tax exemption allows not well as government's ability and willingness to administer only for the poor but also for many middle-income and rich efficiently the existing taxes. Exemptions are particularly workers to be exempted from income taxes. pervasive in the region and are a manifestation of state Overall tax productivity (the ratio between real and capture by specific groups. In addition to direct loss of rev- potential tax collection, given the basic or maximum rate) enues, these loopholes and distortions usually create oppor- has increased during the last 15 years. VAT productivity tunities for tax elusion. There are also important differences 225 I N F O R M A L I T Y FIGURE 8.7 FIGURE 8.8 Tax revenue of Latin America and Caribbean countries Average tax rate of Latin America and Caribbean countries Total tax revenue % of GDP Percent 16 60 Mean 50 15 40 30 14 Median 20 10 13 0 5 12 1985 1995 2003 1985 1995 2003 1985 1995 2003 1985 199 2003 7 8 198519861987198819891990199119921993199419951996199 199 199920002001200220032004 Tariff Corporate income tax rate VAT Individual income tax rate Total tax revenue by tax type % of GDP 16 Source: Lora 2006b. 14 12 The cross-country literature also supports the view that 10 the higher the tax burden, the larger the informality. In 8 cross-country studies, Johnson, Kaufmann, and Zoido- 6 Lobatón (1998), using indicators of perceptions of the tax 4 burden--including both how high tax rates are and the 2 discretionary power of authorities in administering and 0 operating the system--find a large positive effect on infor- 7 8 mality. They also show that what matters to explain 198519861987198819891990199119921993199419951996199 199 199920002001200220032004 informality are the administration and operation of the tax Other taxes Excise taxes Trade taxes system rather than the established rates as a key correlate. Income taxes VAT Friedman et al. (2000) find that higher tax rates are not correlated with a larger informal sector. In fact, the oppo- Source: Based on Lora 2006b. site might be true: countries with high tax rates may be those with high benefits of formality. in the administration capacity. Chilean tax administration practices would appear to be over three times more effective Tax compliance behavior than those in Guatemala in raising VAT revenues (Alm and From a utilitarian viewpoint, individuals regard tax obli- Martinez-Vazquez 2007). gations casually and display no particular moral aversion to From a different perspective, the productivity of VAT is evading if they feel they can safely do so. The taxpayer is an undercut by informality and difficulty in bringing small isolated expected utility maximizer who makes rational businesses and individuals supplying certain services into portfolio decisions under uncertainty, given an informa- the system. As observed, there is a clear inverse correlation tional set. In this view, people pay taxes exclusively between VAT productivity and informality (figure 8.10). depending on their perception of being detected and sanc- So the relatively low level of tax collection in Latin America tioned (Alm et al. 1995; Cowell 1990). They do not relate is a result of tax systems characterized by poor capacity of tax payment to perceptions of the quality or fairness of the administration, excessive exemptions, and narrow tax bases. public services received, to perceptions of a moral obliga- As will be discussed in the following section, however, the rich tion, or to any social norm. If enforcement is weak and the in Latin America pay a much larger share of taxes than the possibility of a sanction is low, predicted tax evasion would share paid by the rich in developed countries. be high. The policy implication of this approach is that 226 T H E I N F O R M A L S E C T O R A N D T H E S TAT E : I N S T I T U T I O N S , I N E Q U A L I T Y, A N D S O C I A L N O R M S FIGURE 8.9 Average VAT and income tax productivity in Latin American countries VAT productivity Income tax productivity Percent Percent 38 16 36 15 34 14 32 13 30 12 28 11 26 Income tax revenue GDP 24 VAT revenue (GDP imp exp) 10 personal income tax rate 22 VAT rate 9 corporate income tax rate 20 8 2 3 4 5 6 7 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 1985 1986 1987 1988 1989 1990 1991 199 199 199 199 199 199 1998 1999 2000 2001 2002 Source: Lora, Cardenas, and Mercer-Blackman 2005. FIGURE 8.10 VAT productivity and informality indicators VAT productivity as percentage of GDP VAT productivity as percentage of GDP CHI CHI 60 60 ECU ECU 50 50 PAR VEN HON VEN HON BRA URU URU BRA 40 40 COS COS ARG PER ARG PER 30 PAN 30 PAN COL COL MEX MEX BOL 20 BOL 20 NIC DOM NIC GUA DOM GUA 20 30 40 50 60 70 40 50 60 70 80 90 Share of informal economy Share of people without access to pension Source: Author's estimates, based on Alm and Martinez-Vazquez (2007), and World Development Indicators, 2005. TABLE 8.5 compliance with taxes--and presumably with most Tax exemptions in Latin America (percent of GDP) regulations--will depend on the economic consequences of detection and punishment; therefore, greater compliance Country Total will depend on deterrence mechanisms. Argentina 2.4 But tax compliance seems influenced by factors beyond Brazila,b 1.4 economic ones. For instance, in the United States, the Chile 4.2 Colombia 9.2 percentage of individual income tax that is audited is quite Ecuador 4.9 small, often less than 1 percent; typically, it is even lower in Guatemala 7.3 Mexicoa Latin America (Alm and Martinez-Vazquez 2007). And, in 6.3 Peru 2.5 any case, very few noncompliance complaints are punished Uruguay 5.3 effectively (Tanzi 2000). Given the level of fines and audit rates in most countries, and the available estimates of risk Source: Based on Gómez-Sabaini 2005. aversion, deterrence models are not able to predict the Note: GDP = gross domestic product. observed levels of compliance (Alm and Torgler 2006; Feld a. Corresponds to federal or central government. b. Contributions from profits and social security are included. and Frey 2007), and differences in public attitudes toward 227 I N F O R M A L I T Y tax laws seem to play a role. In some nations, individuals between states and citizens and, as such, views this tend to view paying their taxes as an important civic oblig- exchange as the "bedrock of the social contract" (p. 290). ation and, for that reason, are motivated to pay. These social norms of tax compliance are measured in An individual's compliance is related to her or his belief the empirical literature by what is termed tax morale.12 For that compliance is the social norm. Perception of fairness, example, Torgler (2005) performs a multivariate analysis of trust, and legitimacy in the system might influence a social tax morale for Latin America using data from the Latino- norm that leads to higher voluntary compliance. Citizens barometro. The 1998 data come from approximately 15,000 perceive their taxpayer relationship with the state as one of individuals in 10 countries in the region. He finds that exchange or as a contract. They will avoid taxes if they per- Mexico stands out as a country with low tax morale, while ceive they are not getting quality government services for South American countries generally have lower tax morale the taxes levied on them. Cowell (1990) posits also that the than do Central American and Caribbean countries. Also, degree of taxpayers' satisfaction with the government he shows that a large majority of individuals perceive tax affects evasion decisions. If they perceive that this relation- collection as largely arbitrary and unfair (only 23 percent of ship is not in equilibrium, moral costs of evading fall and those surveyed by Latinobarometro in 2003 thought tax col- tax morale is crowded out (Torgler 2005). Using survey and lection was "impartial"). Spicer and Becker (1980) provide tax return data from a sample of 800 well-off taxpayers in evidence of a "fairness effect" whereby those who believe New York, Scholz and Lubell (1998) found that trust in they are not being treated fairly by the tax system are more government and in fellow citizens' keeping their side of the likely to evade. social contract significantly influences tax compliance Cross-country data provide suggestive evidence of the "even after controlling for the influence of any internalized relationship between willingness to comply with tax regu- sense of duty and self-interested fear of being caught" lations and perceptions of government's performance. Fig- (p. 412). Bergman (2002) discusses the contrast between ure 8.11 shows a clear negative correlation between tax northern Europe, where tax evasion historically has been morale and the perception that the government is run lower than in Italy, Greece, and Portugal. He also discusses according to the interests of a few, a measure of state cap- how, on the Iberian peninsula, democratization and the ture. There is also a positive correlation between tax morale expansion of the welfare state led to improvements in com- and the perception that the government spends taxpayers' pliance. This strand of the literature views tax compliance money wisely. In both cases, correlations are significant as influenced by a "social exchange," a social transaction even after controlling for GDP. FIGURE 8.11 Tax morale, state capture, and perception that the government spends taxpayers' money wisely Tax morale Tax morale 0.2 COS coef .88, t 2.8 0.2 COS coef .66, t 1.85 SAL VEN GUA GUA SAL 0.1 BRA 0.1 PAR PAR BRA NIC NIC HON COL ARG ARG HON COL 0 0 CHI CHI MEX URU PAN MEX 0.1 ECU URU 0.1 PER PAN ECU PER BOL BOL 0.2 0.2 0.15 0.10 0.05 0 0.05 0.10 0.15 0.10 0.05 0 0.05 0.10 0.15 State capture Government spend money wisely Source: Author's estimates, based on data from World Development Indicators (2005), and Latinobarometro (2004). Note: Figure shows partial correlations controlling for GDP per capita at PPP. State capture is proxied by an indicator of the perception about the economy being run according to the interests of a few. To construct the indicator we ask: "In general terms, would you consider that the country is governed according to the interests of a few or is governed for the benefit of the country?" 228 T H E I N F O R M A L S E C T O R A N D T H E S TAT E : I N S T I T U T I O N S , I N E Q U A L I T Y, A N D S O C I A L N O R M S But perceptions of individuals regarding the effective- transition to a democratic system after 1975 and after join- ness and fairness of state actions may change through time. ing the European Union. The country has adopted major tax For example, Martinez-Vazquez and Torgler (2005) studied policy and tax administration reforms, an extensive redirec- changes in tax morale in Spain, using data from the World tion in public expenditures with the development of a social Values Survey and the European Values Survey, to analyze welfare system, and a significant push for decentralized gov- the evolution at four benchmark years: 1981, 1990, 1995, ernance. Martinez-Vazquez and Torgler argue that Spain and 1999/2000. Spain has undergone fundamental changes succeeded in designing general institutional reforms, in the role and effectiveness of the public sector since its including tax policy and tax administration reforms, as well BOX 8.2 Local taxation and social norms Sokoloff and Zolt (2005) maintain that, in the 19th cen- In the 20th century in North America, the federal tury, North American and Latin American economies government's share of taxation started to increase, and, at raised national/federal-level taxes in a very similar way, both the national and the local levels, the structure of with high reliance on trade and excise taxes. Where the two taxation shifted to one that relied on property, income, areas differed was in taxation at the local level. Local gov- and sales taxes. In Latin America, however, local govern- ernments were far more prominent in North America, and ments and local taxation never grew significantly. Most of they relied for their revenue primarily on property taxes-- the increase in taxation there occurred at the national a fact that suggests a rather progressive taxation structure. level and was much more timid. As shown in table 8B.1, Local governments grew on the basis of these resources and there was basically no significant increase in the tax take were given by local communities the task of providing as a percent of gross domestic product during the first six public education and investing in roads and other infra- decades of the 20th century, and only between the 1970s structure. This dynamic suggests closer local control, and the 1990s was there some increase. This increase in higher accountability, and the generation of a social norm taxation involved greater reliance on consumption taxes that induced people to pay taxes. In Latin America, local in the period, reaching basically half of the tax collection governments were much smaller, and both local and in the 1990s. Property taxes made a very small contribu- provincial or national governments did not rely on prop- tion to revenues collected. erty taxes; rather, they relied on mechanisms that placed a smaller relative burden on the elites (see table 8B.1). TABLE 8B.1 National-level government tax revenue in Latin America, 1900­2000 (percent of GDP) Country 1900 1910 1920 1930 1940 1950 1960 1970 1980 1990 2000 Argentina 10 7 5 7 8 10 10 8 13 10 14 Bolivia -- -- -- -- -- -- 5 10 5 14 18 Brazil 10 11 9 8 10 7 7 10 10 24 23 Chile -- -- -- -- 9 11 17 16 32 21 24 Colombia -- -- -- -- 4 7 8 10 12 13 14 Costa Rica -- -- -- -- -- 10 12 14 18 23 21 Mexico 5 4 6 7 9 8 9 16 16 15 Peru -- -- -- -- -- 11 16 16 17 13 16 Uruguay -- -- -- -- -- -- -- -- 22 24 28 Venezuela, R. B. de -- -- 8 9 12 18 27 19 26 24 20 Source: Sokoloff and Zolt 2005. Note: -- = not available. 229 I N F O R M A L I T Y as in changing the extent to which citizens identify them- differences in taste or temperament. Inconvenient though selves with the state and the national institutions of the it might appear for neat, individualistic models of eco- country itself--processes that had a strongly positive effect nomic behavior, people do seem to take into account the on tax morale and practically doubled the tax effort for gen- `climate' within the group or groups to which they eral revenues in the country (from 22 percent of GDP in belong" (p. 102). This relates to the literature of strong rec- 1976 to 40 percent of GDP in 2002). Concomitantly, a iprocity, which is the behavior of so-called emotional and major increase in formal employment was observed during moral reciprocators who condition their contributions to the 1990s, and tax revenues collected from small firms collective goods on the contributions of others, even in increased dramatically (Farrell 2006). fleeting transactions with multiple actors whose behavior The cross-country correlations presented above suggest they cannot keep track of and whose identities they can that willingness to pay taxes is related to perceptions about never discern (see also Falk and Fischbacher 2005). the performance of the state, among other things. Bergman Schelling (1978) initially posited the idea that interdepen- (2002) goes a step further and tries to test explicitly tax dent behavior can generate multiple equilibria through a compliance at the country level using micro data surveys in "social multiplier effect" in which one person's behavior Chile and Argentina. He finds that, in Chile, citizens com- influences his or her neighbor's behavior (Glaeser, Sacerdote, ply more and are more willing to abide by the rules. Confi- and Scheinkman 2003).13 An individual's perception of dence in public institutions generates trust in their ability the extent of evasion is a powerful predictor of compliance to use public resources to fund social policies and fight behavior: the higher an individual believes the rate of tax poverty. In Argentina, on the other hand, there is lower cheating will be, the more likely he or she is to cheat as satisfaction with less trust in public institutions, and this is well. Individuals prefer to contribute if they believe others behind the observed lower levels of solidarity (see box 8.3). are inclined to contribute, but free-ride if they believe that In a related report, Bird, Martinez-Vazquez, and Torgler others will do so (see, for instance, Alm, Sánchez, and De (2006) explore the role of governance and, more generally, Juan 1995; Andreoni, Erard, and Feinstein 1998; Kahan the quality of the responsiveness of governments to citi- 2005; and Sheffrin and Triest 1992). These behaviors that zens' demands over tax effort. Their basic hypothesis is condition tax compliance and the incentives to informality that although traditional "supply factors" (such as trade on "what others do" tend to generate strategic comple- openness, sectoral composition of output, GDP per capita, mentarities:14 if more people operate in the informal sec- and structure of tax bases) clearly matter in explaining tax tor, it might be easier for anybody to do so; for a given effort, there is also a need to account for citizen attitudes in level of state enforcement, an individual will be less likely response to government performance as shaped by societal to be sanctioned if other people also evade taxes. Con- institutions. To account for such "demand factors," or soci- versely, given specific changes in tax policy, each person's etal institutions, they study the explanatory power of tax compliance might increase not only because of the pol- quality of governance indicators (including government icy change, but also because of the change in peers' behav- effectiveness, rule of law, control of corruption, regulatory ior. Levi (1998) provides a double contingency approach: quality, and voice and accountability). Using a cross- citizens will pay taxes according to a social norm of reci- section of developed and developing countries, and control- procity between taxpayers and the state; at the same time, ling for the above-mentioned supply factors, they show they will comply based on the perceived fairness of a that these "demand" factors matter and that the quality of collective behavior.15 Using data from Latinobarometro, institutions have a strong positive effect on tax effort. Torgler (2005) finds that knowing about other individuals In addition to this "social exchange" mechanism--that avoiding taxes has a significant negative effect. He finds is, I comply because I trust in the state, and the govern- that if people trust that others will obey the law and ment is responsive, and so I have a moral obligation--tax if people trust the president of the country, tax morale compliance decisions are also affected by social interac- is higher. tions (Andreoni, Erard, and Feinstein 1998). In a multi- The interaction of trust among individuals, a perception country study of tax compliance, Cowell (1990) reports, of fairness in existing institutional arrangements, and a "These systematic differences among countries and among perception that the state fulfills its part of the social con- groups within one country cannot be dismissed as innate tract do not seem to hold in Latin America, and the state 230 T H E I N F O R M A L S E C T O R A N D T H E S TAT E : I N S T I T U T I O N S , I N E Q U A L I T Y, A N D S O C I A L N O R M S BOX 8.3 Tax compliance, social norms, and trust in the state: The contrasting cases of Chile and Argentina in the late 1990s Bergman (2002) contributes to the literature on compli- TABLE 8B.3.1 ance by providing us with two interesting case studies. He Determinants of reported willingness to increase tax burden, Chile uses a multivariate ordinary least squares regression to determine the relationship between trust and perception Independent variable Coefficient variables taken from surveys conducted in Chile and Collective perception of honesty 0.079** Argentina and the "reported willingness of taxpayers to It is justified to cheat 0.130*** comply with taxes in order to fund social policies" (p. 294). Trust in and approval of public services 0.163*** Perception of detection capability 0.036 The surveys Bergman used were conducted in R2 0.12 Argentina in 1997 on a sample of 549 individuals, and in Chile in 1998 on a 1,200-person sample. In both Source: Bergman 2002. countries, 70 percent of the active population paid taxes. ** significant at 5 percent. *** significant at 1 percent. In the 1990s, Chile's level of value-added tax (VAT) eva- sion averaged 22 percent, compared with a rate of 55 per- TABLE 8B.3.2 cent in Argentina during the same period. Seventy OLS--effects of attitudes and tax experience of the willingness to percent of Chilean taxpayers report VAT evasion as very comply, (controlled for age and gender), Argentina hard or somewhat difficult. This is in contrast to 67 per- cent of Argentines who "agree or strongly agree with the Independent variable Coefficient statement, `People think that in this country evading Satisfaction with public services 0.173*** taxes is easy'" (p. 269). In addition, both countries show Tolerance of cheating 0.211*** a different prioritization of social welfare. In Chile, 76 Feels guilt at evasion 0.289** Social sanction by peers if taxes are not paid 0.144** percent of respondents somewhat agreed or totally agreed R2 0.26 with the statement "I am ready to pay more taxes if they are channeled to benefit the poor." Argentina's survey Source: Bergman 2002. shows wide dissatisfaction with social services. Bergman ** significant at 5 percent. *** significant at 1 percent. (2002) reports that Argentines "do not challenge the legitimacy of the tax system" (p. 269), but are dissatisfied negative coefficient; and, finally, the detection variable with the quality of public services. again seems to have no effect. Social sanctions and Using that information Bergman finds that, in Chile, guilt--variables not available for Chile--show, surpris- both perceptions of others' honesty and disapproval of ingly, that there is a lack of social norm and individual cheating are correlated with citizens' willingness to increase guilt to pay taxes (see table 8B.3.2). their tax burden, and are statistically significant (see table Bergman concludes that there is a strong link between 8B.3.1). Trust in public agencies (police, public health, citizens' satisfaction with public services and their will- courts, customs, and treasury) also has a positive effect on ingness to comply with tax responsibilities or to increase willingness to pay more taxes. The model does not find their tax share in Argentina and Chile. any effect on the responses caused by the individual's sub- Chile has a higher level of "social solidarity" (that is, jective estimate of the tax authority's detection capacity. willingness to pay higher taxes) as a result of the popu- In the case of Argentina, the dependent variable is lace's trust in public institutions' commitment to allevi- actually less stringent because it inquires only about the ating poverty. In contrast, Bergman concludes from the willingness to comply with taxes--not to increase one's tax Argentine empirical analysis that the lack of satisfaction burden--to fund social programs. Satisfaction with the and trust in public sector institutions is behind the low performance of public services (health, education, secu- levels of solidarity. Moreover, legal enforcement there is rity, and infrastructure) has a positive influence on will- perceived as weak, as is social enforcement of norms. ingness to comply; tolerance of cheating has a significant Source: Based on Bergman 2002. 231 I N F O R M A L I T Y might lack the legitimacy for citizens to feel obliged to matter of political will because the technology is available comply with these regulations. Even if systematic micro and administrative capacity increases are definitely within evidence is still scarce in the region, the cases of Chile and the range of options of the much-modernized Latin Ameri- Argentina suggest a large contrast in the social norm that can tax systems. As an example, in 1947, the United States prevails in those countries. In the same way, attitudes data had a GDP per capita that was lower than the one in from Latinobarometro suggest that there is high variance of Argentina and somewhat above the ones in Chile, Costa tax morale in the region. Low tax morale leading to tax eva- Rica, and Mexico in 2000. However, in that year the United sion is related to the Hirschmanian exit option: if the state States already raised 56.6 percent of its revenues from per- does not give people public goods that they value, and sonal and corporate income taxes (Schmitt 2005). In Latin there is no societal pressure for contributing to or partici- America, that figure today is 30 percent. In table 8.6, survey pating in the circuit of taxes and transfers established by data reveal some of the problems faced in Latin America. the state--and, consequently, there is a social norm of non- Levels of bribery and corruption are, in general, higher than compliance--then people will have a strong incentive to go in other regions. Underreporting and evasion seem also to and remain underground. This is illustrated in figure 8.12, be higher, as sales amounts reported to authorities by a typ- which suggests that tax morale is, as a consequence, ical firm in Latin America are about three-fourths of actual robustly negatively related to informality in Latin America, sales, with East Asia and the Pacific the only region where using two different indicators of informality. This suggests that figure clearly is lower. that countries with high informality tend to be those where There is large scope to increase collection through the social norm is not conducive to complying with tax improvements in tax administration and tax policy. Bird, regulations. In the case of Latin America, in addition to the Martinez-Vazquez, and Torgler (2006), Schmitt (2005), and problems of perception that the state is weak and does not Tanzi and Zee (2000) discuss several areas where the tech- respond to the interests of the majority, the social norm nology is available, and the challenge is more in the politi- may be affected by the operation of tax administrative sys- cal realm. Among others, Latin American tax authorities tems. If there is a widespread perception that the govern- have to move aggressively in professionalizing tax adminis- ment is not willing to fully detect and penalize evaders, tration management and in putting firewalls between tax then noncompliance is legitimized and the prevailing administration decisions and tax policy decisions. Signifi- social norm is reinforced. cant progress has already been made in areas like computer- Latin American countries currently lack the capacity to ization, control of large taxpayers, and use of withholding enforce tax collection as in OECD countries. In a way, it is a systems, together with third-party information. In addition, FIGURE 8.12 Tax morale and informality indicators Tax morale data from Latinobarometro Tax morale data from World Values Survey Tax morale Tax morale 0.2 VEN coef .004, t 1.94 0.2 coef .01, t 2.99 COS GUA 0.1 0.1 NIC HON BRA ARG MEX COL CHI ARG VEN 0 0 PER CHI MEX URU 0.1 BOL ECU 0.1 PER PAN 0.2 0.2 20 10 0 10 20 20 10 0 10 20 Share of informal economy Share of informal economy Source: Author's estimates, based on data from World Development Indicators, 2005, World Values Survey, and Latinobarometro, 2004. Note: Figure shows partial correlations controlling for GDP per capita at PPP. 232 T H E I N F O R M A L S E C T O R A N D T H E S TAT E : I N S T I T U T I O N S , I N E Q U A L I T Y, A N D S O C I A L N O R M S TABLE 8.6 Comparative efficiency and corruption in tax administration: Survey evidence for 2005­06 Middle East East Asia Europe and Latin America and North Sub-Saharan Indicator and Pacific Central Asia and Caribbean Africa OECD South Asia Africa Taxes Average time firms spent in meetings 4.91 2.78 2.89 3.52 1.65 3.37 5.08 with tax officials (days) Time to prepare and pay taxes (hours) 270.06 437.92 548.80 281.36 197.19 331.71 394.00 Corruption Unofficial payments for typical firm 1.81 0.76 1.48 2.72 0.13 1.28 1.64 to get things done (% of sales) Firms expected to give gifts in meetings 33.59 42.84 30.40 40.09 28.26 44.27 18.86 with tax inspectors (% of total) Value of gift expected to secure 1.82 1.36 4.08 1.30 0.55 2.04 3.52 government contract (% of contract) Informality Sales amount reported by a typical firm 69.30 89.35 76.51 73.55 93.55 93.7 78.39 for tax purposes (% of total) Source: Alm and Martinez-Vazquez 2007; data from the World Bank Enterprise Surveys. Note: OECD = Organisation for Economic Co-operation and Development. in most countries in the region, significant efficiency gains staffed, and compounded with a legal system too weak to can be obtained by eliminating or lowering exemptions that criminalize evasion effectively. Many of the steps that can create loopholes in value-added, income, and property taxes. be taken to improve capacities on this front are technically But incentives to register with the tax authorities are feasible in almost all countries. Although it is difficult to necessary to increase the tax base--both carrots and sticks. disentangle their deterrent effects empirically, there is con- Examples of carrots are programs like the earned income sensus that auditing, fines, and business closures are essen- tax credit that provides tax credits to the labor income of tial tax enforcement tools. In one of the few studies in Latin families whose annual earnings remained below a certain America, Bergman (2003) examines the cases of Argentina threshold, often gradually phased out (see box 8.4). These and Chile, two countries that, as noted before, at the begin- types of programs have had tremendous success in reducing ning of the 1990s, found themselves with very similar tax poverty in OECD countries, and have the virtue that trans- policies and macroeconomic conditions but that have since fers are implicitly conditional on the family having some- diverged in their tax compliance; today, Argentina has one in regular employment. approximately double the VAT evasion of Chile, which is Another example of policies oriented to lure the taxpayer 22 percent. Bergman posits that the key reason for this has is changing the tax administrator's approach to a "service been the inability of Argentina's tax system to create a per- paradigm" so as to enhance its role as a facilitator and a manent, credible threat to the noncompliant. Argentina has provider of services to taxpayers. This might comprise granted a total of 24 major amnesties since 1974, and sur- actions promoting taxpayer education, developing services veys show that the population believes bribing auditors is to assist taxpayers in filing returns, broadcasting advertise- common and tax audits have low detection rates. Further- ments that link taxes with government services, lowering more, audits in Argentina have been performed almost compliance costs, simplifying taxes and their payment, exclusively on large taxpayers, and smaller firms are aware of and promoting a taxpayer--and a tax administrator--"code that fact. In contrast, Chile has had no major amnesties, and of ethics." the population perceives it is very hard to bribe auditors. On the side of "sticks," audits and penalties are core Although it also focuses more on high-level taxpayers, instruments of tax enforcement policy. The audit function Chile's audit selection system includes in its design a larger in most of the region is weak, often underfunded, under- number of smaller firms. The efficacy of audits has been very 233 BOX 8.4 Earned income tax credits: Transfers that encourage formal employment Fighting poverty and creating incentives to work are also differ among countries (see figure 8B.4.2); indeed, the challenges faced by many governments. One experience Dutch have chosen not to phase out at all when the maxi- that potentially can offer insights for Latin American mum credit level is reached. Their motivation has been to countries is the earned income tax credit (EITC). In respond to the key theoretical criticism of the labor tax 1975, the United States introduced an innovative social credit, which is that it might provide disincentives policy tool that used employment incentives as opposed to working or earning more in the phase-out range to public assistance to aid low-income people by provid- (Hoffman and Seidman 2003; Hotz and Scholz 2003). ing tax credits to the labor income of families whose FIGURE 8B4.1 annual earnings remained below a certain threshold. Value of the EITC, by income, unmarried filers,* 2005 More than 30 years later, the EITC has become the Credit value largest antipoverty program in the United States: in 4,500 $4,400 2005, 22 million poor working families benefited from 4,000 the $34 billion in credit. Other countries, such as Belgium, 3,500 Canada, France, the Netherlands, New Zealand, and the 3,000 United Kingdom have since introduced their own ver- $2,662 sions of earned income tax credits. Using such credits in 2,500 countries with large informal sectors could induce large 2,000 segments of the population to register, increasing the 1,500 ability of the state to observe income for tax or transfer 1,000 purposes. 500 $399 The basic design of the credit is this: for each addi- 0 tional earned dollar, workers receive a tax credit, up to an 0 5,000 10,000 15,000 20,000 25,000 30,000 35,000 Income (US$) income level considered by the government as the national poverty threshold; after this, the credit plateaus No children One child Two or more children and steadily phases out until an earnings ceiling is reached Source: Holt 2006. where eligibility stops. Different thresholds apply to dif- Note: Data from Internal Revenue Service. (*) Married couples filing jointly are eligible for slightly higher credit amounts in the "phase- ferent numbers of children (see figure 8B.4.1). Schedules out" range of the EITC. FIGURE 8B4.2 Labor tax credits in selected countries, 2001 Tax credit (in US$) 4,500 4,000 WFTC EITC 3,500 3,000 2,500 2,000 1,500 1,000 LTC 500 PPE 0 0 1,800 3,600 5,400 7,200 9,000 10,800 12,600 14,400 16,200 18,000 19,800 21,600 23,400 25,200 27,000 28,800 30,600 Income (US$) Source: Detragiache 2001. Note: WFTC working families tax credit, UK; EITC earned income tax credit, USA; LTC labor tax credit, the Netherlands; PPE prime pour l'emploi, France. 234 T H E I N F O R M A L S E C T O R A N D T H E S TAT E : I N S T I T U T I O N S , I N E Q U A L I T Y, A N D S O C I A L N O R M S Other internationally differentiating characteristics are been split into the child tax credit and the working tax whether the annual credit is refundable or if it applies credit, was that the credit encouraged more hours of work only to cover one's tax burden, whether families without by entitling only individuals working more than 16 hours children are eligible, and whether there is a minimum a week to the credit and providing small additional credit number of hours one must work to be eligible. for those working more than 30 hours a week. The U.S. EITC has been studied extensively, and the The earned income tax credit model's ability to be tai- foremost objective of the credit--to reduce poverty--has lored to fit specific country environments is a characteris- been a widely documented success. Between 1995 and tic that makes this formula attractive. In the Latin 1999, the EITC was responsible for an overall reduction American and Caribbean context, over the medium term, in the U.S. poverty rate of 1.5 percentage points tailored credits could be not only a potent tool for (Hoffman and Seidman 2003). Estimates suggest that, poverty reduction but also an instrument to combat the without the EITC, the poverty rate among children in the region's high level of informality. People would have United States would be 25 percent higher (Greenstein greater incentives to register with tax authorities to 2005). Indeed, the credit has been so successful that, by receive the credit, providing corresponding incentives for 2006, 18 individual states in the United States had insti- individuals working in informality to exert pressure on tuted similar credits for state taxes. The program has also their employers to become formal. In addition, disbursal positively affected participation rates among one of the of the credit using official financial intermediaries could most vulnerable populations in the United States, single be used as a tool to encourage participation in the finan- mothers. Meyer (2001) finds that 60 percent of increases cial system. In terms of innovative regional design and in single-mother labor participation between 1984 and targeting, we can envision, for example, that poor fami- 1990 are due to the EITC and other tax. lies involved in the care of the elderly would receive an The EITC is not the only tax credit model that seeks adjusted credit in the same way that families with chil- "to make work pay." In 1999, the United Kingdom intro- dren currently do. A primary issue would be the develop- duced the working families tax credit (WFTC). Brewer ment of good targeting systems in the absence of good et al. (2005) find that the WFTC was responsible for a income indicators to ensure that the poor are the ones 5.11 percent increase in the labor supply of single mothers receiving the working credit. Targeting models currently in the United Kingdom between 1999 and 2003. The used in conditional cash transfer systems, for example, direct poverty effects of the WFTC have been more could provide guidance as to how, in the Latin American ambiguous. A key feature of the WFTC, which has now context, a working credit initiative could be realized. different, with Chile focusing on enhancing compliance by effects on tax evasion in a country where tax authorities do matching computerized and third-party information. not have the ability to produce a credible threat of detec- Argentina relied more on penal sanctions that end up tion and evenhanded enforcement of sanctions. being less effective, given the low expected probability of being detected. Inequality, taxes, and transfers Bergman (2003) provides complementary evidence of One of the elements that influence the social norms related the specific impact of audits on tax evasion in Argentina. to compliance with regulations is the perception of how He finds that, in both Argentina and Chile, audited and resources given to the state will be used. On one hand, sanctioned taxpayers decreased their future compliance in deficient public services encourage people to opt out of pub- subsequent years, presumably in an effort to recoup the lic service systems. On the other hand, a state that is per- losses prompted by the audit fines, but the decrease was ceived as unfair will lack legitimacy. One element that feeds more moderate in Chile. The author attributes this to the the perception of unfairness is the real and perceived fact that, in Chile, the threat posed by the tax authorities is structure of incidence of taxes and transfers along the considered credible, as discussed above. The conclusion of income scale. Are taxes excessively concentrated and this analysis is that tools of deterrence may have unforeseen deemed unfair by some? Are transfers--public expenditures, 235 I N F O R M A L I T Y in general--excessively concentrated? Are there segment of services, such as to the judiciary and to police protection. the population that are excluded from the taxes and transfer Very much to the point of this report, informal workers are mechanism? Are public resources used to equalize opportu- one of the key excluded groups. nities in a way that is consistent with the implicit social In an exercise that aggregates social spending, Breceda, consensus? Do they generate an exclusionary mechanism Rigolini, and Saavedra (2007) compare patterns of progres- that leaves part of the population out of the social contract siveness for a sample of Latin American countries with the and consequently renders the state less legitimate? United States and the United Kingdom. They find that, in We discussed above some stylized facts of the tax side of addition to significant differences in average levels, there is the equation. As we saw, tax collection in the region is below a strong contrast between Latin America and the United its expected value, given the level of development. However, Kingdom in terms of the progressiveness of social spend- social spending does not show a clear pattern. Overall, the ing. In Latin America, social spending is slightly biased in ratio of social spending to GDP in Latin America seems to be favor of the rich: on average, social spending to the poorest in line with the region's level of development. There is, how- quintile corresponds to 92 percent of social spending to the ever, a large variance within the region (Lindert, Skoufias, richest quintile, against 233 percent for the United and Shapiro 2006). Argentina, Brazil, and Uruguay spend Kingdom.16 Latin America seems, therefore, to be closer to around 20 percent of GDP in social areas, whereas, at the U.S. model, which has a ratio of 107 percent. Although the other extreme, the Dominican Republic, Ecuador, variation in the Latin American sample is quite large (in Guatemala, and El Salvador show social expenditures below Honduras--the least progressive state in the sample--the 7 percent of GDP. Mexico stands out as a richer country with poorest quintile receives 57 percent of what the richest relatively low social expenditures as a percentage of GDP. quintile receives, whereas in Colombia the ratio climbs to Similar to the small increase in taxes observed since 1985, 108 percent), all countries remain far away from the pro- there is clear evidence of a slightly larger increase in social gressiveness of the European welfare states, and all are less expenditures. Without exception, social expenditures have progressive than the United States. increased in all the countries of the region, particularly in The key element of this discussion for the issue at hand countries that started at a low basis in the early 1990s. For is that there is a large segment of the population for which the region as a whole, the increase was from 12.8 percent of the state is not providing basic services, for which the state GDP in 1990/91 to 15.1 percent in 2002/03. is not a guarantor of minimum opportunities, and for Despite this increase in social spending, patterns of which the provision of public goods is insufficient and, regressivity remain and within-country inequities continue equally important, inequitable. to be very large. In education, spending is mostly progres- sive, except for tertiary education, and has increased in all Inequality and the system taxes and transfers countries. But, despite the improvements, quality indica- It is well known that Latin America is the most unequal tors perform poorly and within-country variance is region in the world, as measured by income distribution extremely large. In the case of health care, Latin America (de Ferranti et al. 2004; Perry et al. 2006). Recent research has witnessed undeniable progress in the provision of basic work at the World Bank (2006) has shown that inequality services, such as basic health and nutrition services, and of opportunities is one of the key factors. Differences across indicators like maternal mortality and immunization rates individuals in access to assets, public goods, services, and have improved dramatically. But, again, differences in the voice (for which individuals could hardly be responsible quality of access by income groups are still extremely large because the differences arise at the moment of birth) (see de Ferranti et al. 2004; World Bank 2006). Spending determine differences in their ability to contribute to on social security, which has expanded the most, is fairly development and to build their own future. Accumulated regressive. The pension system is the worst offender, and differences in opportunities translate eventually, among one of the biggest problems, as explained in chapter 7, is the other ways, to differences in income. low coverage. Some social assistance programs are fairly pro- A significant part of the observed inequality is due to the gressive, but, in general, they represent a small fraction of effects of state intervention (or lack thereof) more than to spending. In addition, in Latin America there is abundant pure market outcomes. As discussed in Perry et al. (2006), if evidence of deficiencies and inequities in access to other income inequality in Latin America is compared with that 236 T H E I N F O R M A L S E C T O R A N D T H E S TAT E : I N S T I T U T I O N S , I N E Q U A L I T Y, A N D S O C I A L N O R M S of Europe, there is evidence that a significant part of the dif- many European countries (as a proportion of their income), ference between the levels of disposable income inequality given the high inequality, their contribution to total tax in the two regions is due to the different impact of taxes and revenues is much larger (despite the elements of state cap- transfers: they reduce market income inequality consider- ture mentioned above, including the high level of tax ably in Europe, and very little in Latin America.17 Interest- exemptions that, in many cases, favor the rich dispropor- ingly, the same can be said even when comparing the Latin tionately). This may place a strain on the social contract American countries with the United States, a country that because the richest quintiles have to pay for most of the has the reputation of not being too distributive (Alesina and state provision of public services without seeing sizable Glaeser 2004). The inability or unwillingness of Latin benefits in return. These findings give further support to American political and economic systems to improve the the argument that an increase in the efficiency of the sys- distribution of income is not a new event, nor has it passed tem and an increase in the tax base are the key avenues to unnoticed by previous observers. For instance, Kuznets increase tax collection in the region. (1955) argued about the "the failure of the political and Other studies that look only at taxes find small progres- social systems of underdeveloped countries to initiate the sivity and, hence, a small redistributive effect of taxes in governmental or political practices that effectively bolster Latin America (Chu, Davoodi, and Gupta 2000). More- the weak positions of the lower-income classes" (p. 24; cited over, Engel, Galetovic, and Raddatz (1999) find that, in in Beramendi and Díaz-Cayeros 2006). Chile, which has the highest tax productivity of the Breceda, Rigolini, and Saavedra (2007) present an inci- region, taxes are slightly regressive. They argue that the dence analysis of both social spending and taxation for more unequal the pretax distribution--as is the case in seven Latin American countries, and make the comparison Chile and most of the region--the less the redistributive to the United Kingdom and the United States. Consistent effect of progressive taxation. A lot more is achieved with previous studies, they find that, in Latin America, through better taxation, fewer loopholes, and higher levels absolute levels of social spending are fairly flat across and quality of spending. income quintiles--and in some countries they are even An unfortunate characterization of Latin America is one regressive (see figure 8.13). Nonetheless, they find taxation of low-quality, ineffective provision of public services that to be quite progressive, particularly income and value- reflects an unresolved problem of high inequality of oppor- added taxes. On average, the richest income quintile in tunities and is correlated with an extremely high level of Latin America pays 22 times more taxes than the poorest inequality of current incomes. This is within a context of a quintile. This remains close to the difference in the United taxes and transfers framework that does not redistribute States (19 times) and much higher than in the United effectively. Kingdom (6 times). Moreover, in Latin America, the share In addition to how inequality interacts with the taxes of taxation paid by the richest income quintile averages and transfers structure, there are several channels through 61 percent. This remains significantly higher than the which inequality might be linked explicitly to informality, share paid by the richest quintile in the United Kingdom although the intuitive relationship is not well studied yet. (43 percent), and similar to what the richest quintile Ahmed, Rosser, and Rosser (2004) find that the informal contributes in the United States (58 percent). sector share is a significant determinant of income inequal- Both features make the structure of social spending and ity in a sample of 52 countries (as cited in Davis [2007]). taxation in Latin America closer to that of the United Chong and Gradstein (2007), explore the opposite chan- States than to that of the United Kingdom (where social nel, which is of inequality as a mechanism that generates spending is more progressive and taxation is less so). The more informality. They model and test that relationship comparison, therefore, strongly indicates that Latin Ameri- empirically and find a significant positive relationship. Fur- can countries resemble a minimalist welfare state similar to ther, they argue that the effect of high inequality may be the one in the United States, more than a Europe-like one. exacerbated in the context of low institutional quality. The The extremely high inequality levels observed in Latin reason they postulate is that, given market imperfections, America make the transition difficult toward a more pro- when institutional quality is low, protection of property gressive welfare state. In particular, although the rich in rights in the formal sector is weak and resources largely are Latin America are taxed equal to or less than the rich in up for grabs. "Poor individuals whose endowments are 237 I N F O R M A L I T Y FIGURE 8.13 Social spending and taxation by income quintiles Argentina Honduras Amount per capita as % of GDP per capita Amount per capita as % of GDP per capita 25 25 15 15 5 5 0 0 5 5 15 15 25 25 35 35 45 45 55 55 65 65 Quintile 1 Quintile 2 Quintile 3 Quintile 4 Quintile 5 Quintile 1 Quintile 2 Quintile 3 Quintile 4 Quintile 5 Mexico Peru Amount per capita as % of GDP per capita Amount per capita as % of GDP per capita 25 25 15 15 5 5 0 0 5 5 15 15 25 25 35 35 45 45 55 55 65 65 Quintile 1 Quintile 2 Quintile 3 Quintile 4 Quintile 5 Quintile 1 Quintile 2 Quintile 3 Quintile 4 Quintile 5 United States United Kingdom Amount per capita as % of GDP per capita Amount per capita as % of GDP per capita 25 25 15 15 5 5 0 0 5 5 15 15 25 25 35 35 45 45 55 55 65 65 Quintile 1 Quintile 2 Quintile 3 Quintile 4 Quintile 5 Quintile 1 Quintile 2 Quintile 3 Quintile 4 Quintile 5 Other social spending Cash transfers Health Education Other taxes VAT sales tax Income tax Source: Breceda, Rigolini, and Saavedra 2007. 238 T H E I N F O R M A L S E C T O R A N D T H E S TAT E : I N S T I T U T I O N S , I N E Q U A L I T Y, A N D S O C I A L N O R M S relatively limited are at a disadvantage in extracting a larger FIGURE 8.14 share of the resources, hence, find it beneficial to move into Informality and inequality the informal sector, where although less productive, they are able to fully retain their production output. High inequal- Share of persons without pension 40 ity, exacerbated by low institutional quality, magnifies this coef .72, t 3.54 effect, implying a positive relationship between inequality PER ARG PAR 20 ECU MEX BOL and the size of the informal sector" (p. 160). NIC DOM GUA VEN HON This result is very relevant in the context of Latin Amer- 0 URU SAL BRA ica. Income inequality reflects itself in differences in voice, COL JAM CHI PAN power, and influence. The lack of influence usually leads to COS 20 capture and to a perception that the state is run according to the interests of an elite. Among other effects, this leads 40 to lower tax morale and higher informality. Precisely, better institutions might be hampering the possibility that 60 inequality of income may lead to inequality of power.18 80 Bird and Zolt (2005) show evidence of a negative effect of 20 10 0 10 20 inequality on tax effort. They argue that highly unequal Gini coefficient distributions of income, typical of Latin America, can lead Source: Author's estimations, based on World Development to low levels of solidarity by the elites toward poorer Indicators (2005). Note: Figure shows partial correlations controlling for GDP per groups. For example, low tax effort in Central America is capita of PPP. often interpreted as the unwillingness of a small elite to finance publicly provided goods because they can opt out and finance their own services. selves, can explain a high share of the variation in informality. But an argument that is valid for the elites who might It is then difficult to disentangle the independent effect of influence policies in their favor is not valid for the relatively inequality on informality from the effect of institutional vari- rich or the middle class. When there is extreme inequality, it ables. The evidence presented by Chong and Gradstein is more difficult to collect revenues in a fair and efficient way. (2007) is suggestive, but more analysis is needed to under- In Latin American countries, despite paying proportionally stand better how the institutional setting may affect the less than their peers in richer countries, the relatively rich channel through which inequality affects informality. (that is, the small middle class that exists in Latin America) pay a disproportionate share of all revenues. And, given that Informality: A reflection of a broken services provided by the state are of low quality, what they social contract? get back from the state is not aligned with what they pay-- Is Latin America in a "bad" equilibrium? although they may be getting more service from the state Economies can land in different equilibria if social norms than the poor are getting. This misalignment implies that and social interactions that lead to specific collective behav- the opportunity cost of tax compliance is even higher. iors are strong.19 You make the queue or skip the line. You We performed several cross-country estimations to test stop at the red light or you keep moving. As discussed the robustness of the result of a positive relationship between above, tax and regulatory compliance, one of the consum- inequality and informality, given the level of development mate collective action problems of public policy and a key and other institutional characteristics. As shown in fig- factor behind informality, may be affected by social interac- ure 8.14, the two variables are positively related after control- tions. Individuals will be more inclined to pay taxes if they ling for GDP, and the results are robust to using different believe the government is complying with its fair share of informality indicators. We wanted to further analyze whether the social contract and is using public resources effectively. the effect of inequality is conditioned by institutional charac- They will be more willing to comply if they believe others teristics. However, the econometric analysis (not reported) do so, prompting still others to comply, and so forth and so shows that both inequality and GDP are highly correlated on until a highly cooperative state of affairs takes root. with institutional and structural variables that, by them- Conversely, people will be inclined to evade tax obligations 239 I N F O R M A L I T Y if they believe others are inclined to do so. Such interde- fair deal--through different mechanisms--by the current pendencies tend to generate patterns of collective behavior arrangements, and that they are justified in avoiding mak- with specific reinforcing mechanisms. ing a contribution to the system whenever possible. Latin America seems to be in an equilibrium in which From both ends of the socioeconomic scale we tend to trust in the state is low; tax collection and compliance with observe a large amount of exit, in Hirschman's terminol- regulations are low in most countries; and public provision ogy. We can use a cost­benefit decision scheme, like the of public services is inequitable and of low quality, both one we presented in the second section of this chapter, now directly because of lack of resources and because of low gov- from the point of view of a citizen from the highest income ernment effectiveness, even given the available resources. quintile who is considering his or her relationship with This unfortunate characterization reflects an unresolved the state and with society at large when it comes to decid- problem of high inequality of opportunities correlated with ing whether to evade taxes. In many Latin American coun- an extremely high level of inequality of current incomes. tries, such a "rich" citizen is likely to opt out of public This happens in the context of a structure of taxes and services and into the higher-quality private provision of transfers, that does not redistribute effectively. old-age insurance,20 security services, education, and Low taxes are mostly related to low tax productivity. health care. A citizen is likely to feel that such state And, as discussed above, this is mainly related to low tax services are of very little value to him or her. When assess- compliance, narrow tax bases, and excessive exemptions. ing how valuable such state services are for the population The empirical analysis presented above suggests that tax at large, he or she is likely to share the perception that compliance is affected by deterrence and by a complex of these things that are of little worth to him or her are not factual beliefs and emotional dispositions, whereby social useful, effective, of high quality, or fair for citizens in other norms are not necessarily conducive to complying with strata. This negative view of the worth of state services is taxes and other regulations--a fact that has been proxied by likely to be compounded by a generalized perception of low tax morale. The evidence in Latin America also points patronage and corruption in the government generally and to a negative relationship between tax morale and informal- in social assistance particularly; and by the social norm ity. These actually are different sides of the same coin. Low externalities that provide an implicit validation of wide- willingness to pay taxes and incentives to operate infor- spread tax evasion.21 In some Latin American societies, mally are shown to be empirically correlated with percep- such as Argentina, we see an increase in socioeconomic tions of incompetence of the state (proxied by indicators of segregation, which might bring further difficulties for government effectiveness, rule of law, impartiality and fair- weaving together a social contract down the road. Poor ness of courts, for example). Evidence also suggests the and rich citizens tend to live in worlds further apart in importance of fairness in the use of public resources. In this terms of the schools their children go to, where health context, informality is fueled by a structure of regulations services are received, and where their homes are located, that, in a cost­benefit analysis, may provide a rationale for with gated communities and shanty towns as clusters, each firms and workers to operate underground, as discussed in time placed farther apart. chapter 5, fostered by a social norm not conducive to com- However, many of the rich, usually the richest of the plying with regulations and by weak social and administra- rich, may also be capturing the state and using its relatively tive sanctions. In this context, there are multiple examples larger influence to maintain tax privileges or lobby to of ways in which both the poor and the rich might feel maintain oligopolistic structures. Corporatist groups and excluded and find it convenient to opt out, usually partially, certain unions may use political pressure to maintain privi- from the system. leges that perpetuate unequal structures of power and of opportunities. In any of these cases, we are confronting Opting out . . . but not completely weak and incompetent states--not necessarily small Telling the story in terms of the beliefs and actions of agents states--on one side, incapable of constraining the influence at different levels of the socioeconomic scale, we might say of powerful groups (Guerrero, López-Calva, and Walton that, in many Latin American countries, the rich, the mid- 2006) and, on the other side and related to this, incapable dle class, and the poor might feel they are not getting a of providing services and public goods in a fair manner. 240 T H E I N F O R M A L S E C T O R A N D T H E S TAT E : I N S T I T U T I O N S , I N E Q U A L I T Y, A N D S O C I A L N O R M S Poor people feel even more disengaged. The Hirschman- incorporation and looked like the construction of a social ian notion of exit applies also to the lower end of the socioe- contract that allowed for the inclusion of increasing seg- conomic distribution: the poor do not pay many taxes, but ments of the population. But that welfare system reached they also do not get much from the state (see box 8.5). full coverage, was strongly stratified, and turned out to They feel an adverse differential access to public goods, to be financially unsustainable. Many segments of society, property rights, to protection under the law, and to judi- including rural citizens and urban marginal areas, never ciary services. This fosters the use of informal mechanisms participated. and reduces the incentive to participate in the formal cir- As a result of an explicit agenda to improve fiscal sus- cuit. The poor will organize themselves for self-protection tainability and to insert some "market principles," that sys- as well, will invade public property, and, in some unfortu- tem and the political institutions behind it were reformed. nate cases, will take justice into their own hands. Poor The social security system was reorganized along market citizens are more likely than their wealthier neighbors to be principles. It is not clear yet that the new system will informal, to participate in clientelistic networks, and to imply a fast incorporation of segments that traditionally have a negative view of the state and the extant social have been left out. Moreover, the evidence shown in this arrangement. A difference with respect to the wealthy is report and elsewhere points to increasing informality in the that, in many cases, the poor have never been part of the 1990s. formal system anyway. More than "exiting" the system, the But different countries seem to be going in different poor have never entered the system. This situation is part of directions, and societal consensus regarding the potential a culture of informality in which the state is basically future path varies. In Argentina, for example, there is evi- absent and a social contract is basically broken. dence of increasing inequities in access to social protection It is easy to see that precisely these attitudes have and less clarity about the effectiveness of the current important reinforcement effects. The low willingness to arrangements. One possible reading is that now we have a contribute is part of the explanation of the fiscal limitations patchwork of elements of the previous covenant, mixed that impinge on the low quality and coverage of the ser- with the recent add-ons, in the form of social assistance vices provided by the state and that, in turn, feeds back programs that were effective during the time of the crises into low trust in the state and low tax morale, as discussed and efficient in some dimensions, but that have not yet above. There is no clear evidence that the process of opting added up to anything coherent, let alone an integrated set out is increasing, but there also is no evidence at all to the of social policies that has garnered societal consensus. Some contrary. At least it is safe to say that some people opt out interactions between increasing informality and a transfor- (exit) from a more inclusive social contract, while others mation of politics seem to be feeding a negative loop. continue to be excluded from it. It happens not only in the Countries such as Argentina that, several decades ago, "flat" horizontal sense, but also in a "vertical" sense in seemed to define crucial political decisions in a centralized which whole groups "collectively" exit through what one bargaining arena defining national policies now are moving might call "local" reciprocity dynamics. to an increasingly territorialized model of construction of The specific experience regarding all the different political power in which focused social assistance programs, aspects of exit and exclusion that are behind the large informality, clientelism, and new forms of political partici- informal sector in Latin America varies notoriously. Below, pation (such as piquetes) seem to reinforce each other. we describe briefly three particular cases to illustrate this Chile is a very different case. It is the only economy in heterogeneity. the region that has been able to attain very low levels of extreme poverty, with a Chile Solidario program that Heterogeneity in Latin America has more beneficiaries in 2003 than what the Encuesta de Most Latin American countries attempted at some point in Caracterización Socioecónomica Nacional survey reported as the 20th century to form a social contract that centered on extreme poor. One of the key--and first--elements of the the labor market, with some strong institutional actors, Chilean process that started with the return to democracy such as unions. That model, linked to the development in 1990 was its ability to negotiate a tax reform as one of model in vogue at the time, provided some progressive its first steps. It was able to convince the elites (businesses 241 I N F O R M A L I T Y BOX 8.5 Expansion of private security services in Managua Hirschman (1970) posits that when discontent with the has seen an annual growth in the rate of crime of approxi- performance of an organization to which one belongs, a mately 10 percent since 1990. Crimes against people rep- person has two options: "exit" the organization and find resent the greatest increase--362 percent over 13 years. another whose performance is superior, or "voice" one's Rodgers reports that ". . . regionally, the Nicaraguan dissatisfaction in the hope of prompting the organization police force has the lowest number of police personnel to improve. In that framework, the phenomenon of per capita and per crime, the lowest budget per crime, the wealthier Latin American citizens seeking alternative lowest budget per police officer, and the lowest average health care, education, and security providers is seen as salaries in Central America. As crime has exploded in their exercise of the exit option. In a flight to quality, they the capital so has the private security market" (p. 117). abandon the state and seek private providers. Such aban- He states that, in 1990, there was one registered private donment can have far-reaching effects on the state. Citi- security firm in the capital. This number climbed to 14 in zens may be less inclined to pay taxes for services they 1996 and was 56 in 2003. In that same year, 9,017 people don't use or value, so the state's capacity to perform its were registered as private security guards. The increasing duties will be weakened. The provision of public security trend has not abated, despite the government's efforts to to a nation is one of the oldest and most firmly established increase resources for the police force and a 33 percent responsibilities of the state. Centeno and Portes (2006) increase in the number of officers between 2000 and 2005 aptly describe the state of public security in the region at (Gómez 2005). Recent statistics show that, in 2005, there the dawn of the 21st century in these terms: "In Latin were 67 private security companies covering 4,153 loca- America, the incapacity of the state to protect the citi- tions, with 9,329 guards and 6,805 weapons. Goméz zenry has led to the massive growth of private security ser- delves further into the phenomenon of private security in vices, the withdrawal of the wealthy into fortress-like Nicaragua to reveal a crossover between public and pri- gated communities. . . . In the absence of credible enforce- vate security provision. He observes that some of the ment of rules, people take things into their own hands . . ." highest-ranking former police officers have become active (p. 14). in providing private security. Rodgers (2004) explores issues of security, criminality, The message seems to be that when the state doesn't and spatial segregation in Managua, Nicaragua. The city deliver, people take things into their own hands. and high-income earners who would pay about two-thirds extremely high and, symptomatically, very high in the of the new tax burden) that, in the midst of uncertainty, it policy agenda. was a small price to pay for the return to socioeconomic The opposite of Chile might be Guatemala, where trust peace. The combination of tax reform and explicit social in institutions is low and the presence of the state in several policy objectives made this a step toward building a social socioeconomic realms is very limited. Tax collections are contract. The reform was approved by Congress six weeks among the lowest in the region, and informality is around after the new Concertación government started (box 8.6). 80 percent. The peace accord signed in 1996, after 30 years Chile reformed its pension system and has recognized that of civil war, included a political pact for the reconstruction the increase in coverage is still too small. It is therefore of the country; and it established the need for increasing proposing a package of reforms to the system aimed at tax collection by at least 50 percent before 2000 (that is, to achieving universal coverage in a way that would be incen- 12 percent of GDP). The increase in tax collection was tive compatible with keeping individual savings as the earmarked for education, health care, housing, and justice. mainstay of the system and that is fiscally sustainable (see In 1998, the government rescheduled the agreement. In chapter 7). Trust in institutions is high and tax productiv- 1999, a large commission prepared the draft of the new fis- ity is the highest in the region. Informality is less than 15 cal pact. During 2000, the negotiations among all political percent. Inequality of income in Chile, however, is still parties, the business elites, and other organizations failed, 242 T H E I N F O R M A L S E C T O R A N D T H E S TAT E : I N S T I T U T I O N S , I N E Q U A L I T Y, A N D S O C I A L N O R M S BOX 8.6 Negotiating tax reform and the start of the social contract, Chile, 1990 In a difficult transition environment, the center-left Con- In addition to this negotiation strategy, led by certación coalition, which won Chile's first democratic elec- Alejandro Foxley (then minister of finance), policy suc- tion in 17 years, successfully passed a tax reform bill within cess resulted from Concertación's willingness to be moder- the first months of the new party's term, earmarking greater ate, as evidenced in the final substance of the reform. tax revenue for increased social expenditures. Boylan Initially Concertación wanted to raise the percentage of (1996) identifies two main reasons for this unexpected pol- gross national product tax revenue to 3 percent rather icy success: a willingness to be moderate and an engaged than 2 percent. The RN negotiators pushed for the negotiation strategy. The final version of the tax package reduction of the corporate tax increase from 20 to consisted of four main components. It included an increase 15 percent. From the RN's perspective, the business in the corporate tax rate from 10 to 15 percent and a change elite's foresight that the "social debt" of the years of dic- in the progressive income tax categorization system that tatorship had somehow to be paid and its realization put more people in the highest bracket. Third, to control that, compared with other taxes in the region, the pro- tax evasion, the highest contributors in the agriculture, posed taxes were not high played a role in ensuring the transportation, and mining sectors were no longer taxed on successful negotiations and passage of the reform. estimated profits but on actual profits. Finally, the VAT Finally, Concertación accepted the RN's proposed increase increased from 16 to 18 percent. in the VAT, thus showing a willingness "to tax their Boylan (1996) argues that it was crucial that the own," not only the business elite. In terms of selling the details of the tax reform were negotiated in extraparlia- reform to the general public, the earmarking of the addi- mentary meetings with a targeted group from the major tional revenue to explicit social policy gave a sense that opposition party. The targeted politicians of Renovación the reform was a step toward building a social contract. Nacional (RN), a party whose agrarian base strongly iden- Furthermore, Boylan (1996) argues that Chilean tax tified with the outgoing regime, were known for their reform was more than a targeted policy success: the technical competence, consensual approach to politics, moderate and risk-averse strategy followed by the new and their desire to transform the right into a democratic government "played a crucial role in soldering the frag- force. Concertación was thus able to develop the tax reform ile rule-making environment at the delicate moment of with a small, less hostile faction of opposition, who then regime change" (p. 8). sold the reform to the rank and file of their party. This way, by the time the reform was introduced into parlia- ment, it was almost a fait accompli. Source: Based mainly on Boylan 1996. among other reasons, because of the lack of leadership in failed to increase coverage. Minimum pension programs the executive (ICEF 2005). Some changes implemented in have been launched in several states, and the flagship 2001 were harshly resisted by the business elites, which antipoverty program, Oportunidades, covers 80 percent of in that year alone presented 31 constitutional appeals the poor and is considered an international best practice. against tax increases. The courts lined up with business But further expansion of minimum pension programs or interests, and the tax take is still around 11 percent of any other initiative aimed at universalizing social insurance GDP. No need to continue the narrative. Point made. clashes with a hard budget constraint. The Fox administra- Mexico is an intermediate case. Filgueira (2005) classi- tion was never able to pass a tax reform, and Mexico is the fies it among those dual social states with a large social pro- OECD country with the lowest tax collection--well below tection system that covers a bit less than half of the what would be expected for its level of development. Trust population and with rigid labor legislation that hampers in institutions is weak, and the social norm regarding tax formal job creation. Mexico reformed its private pension compliance is probably behind the low VAT productivity system, which has made it financially healthier but has of 24 percent. Organized public sector labor is politically 243 I N F O R M A L I T Y very strong, and neither labor reform nor public pension expand the tax base, incorporate more citizens in the formal sector reform has been possible politically. Still, the coun- economy, and increase tax compliance. The rich in Latin try is looking to improve the quality of its public perfor- America contribute a much larger share of the tax collection mance, and tax reform is again on the agenda. than what is observed in richer regions, and further increases in tax rates might not be socially tolerable or economically Conclusions efficient. Furthermore, in attempting to increase tax collec- We have discussed here that to understand informality it is tion, the region faces the challenge of reducing exemptions, critical to understand several interaction mechanisms which often is a reflection of state capture. On the other between the state and the citizens. Individuals explicitly or hand, Latin American governments would have to improve implicitly decide whether and how to engage with the their performance--both the quality of their expenditures mandates and institutions of the state, weighing costs and and the mechanism for citizens to monitor them. Recent benefits and state enforcement capabilities. These decisions mainstream literature is less pessimistic about the effect of are conditioned by social norms shaped by how agents indi- taxes (and therefore the size of government) on growth, sug- vidually and collectively perceive and define a relationship gesting that Latin America may be on the side of the curve with the state. As documented here, cross-country evidence where taxes, as a whole, may be growth enhancing through based on opinion surveys and investment climate surveys, as the public-goods channel. This implies that increases in well as on a few comparative studies, suggests that percep- both taxes and expenditures in Latin America might be tions of government effectiveness and of the performance of growth enhancing, but only if government effectiveness services like the judiciary system in Latin America are below increases dramatically. This will facilitate establishing the those observed in other regions. And, as summarized here conditions for a gradual change in individual and collective and discussed extensively in the literature, informality is beliefs and attitudes regarding the real and perceived rela- negatively related to institutional quality indicators. This tionship between citizens and the state, which is essential to environment is consistent with a social norm that is not start reducing the high levels of informality in the region. conducive to complying with regulations. The strong interest in initiatives to improve government Taxation, which lies at the heart of a social contract, is effectiveness suggests that many countries are steering in one of the areas that has been most studied regarding the the right direction. role of social norms. As discussed here, tax morale--a social Reducing informality is a daunting task and a critical norm about a citizen's willingness to pay taxes--seems to development challenge for the region. It requires not only be correlated with several measures of state performance increasing overall productivity and growth in the economy and with informality, both in Latin America and globally. and improving regulations in labor and product markets, Low willingness to pay taxes and incentives to operate in- but also pursuing a long-term agenda that could move formally are related to perceptions--supported by reality-- countries faster to a better equilibrium. In other words, it is of incompetence of the state and lack of fairness in the use an agenda that includes building a better social contract of public resources. Moreover, the high and persistent lev- from which fewer people are excluded and in which there els of inequality, and the prevalence of a structure of taxes are fewer incentives to opt out from it. As has been and transfers that is not efficient in leveling the playing described throughout the chapters of this report and in the field and improving the equality of opportunities, fuel a ever-expanding body of literature on informality, the policy perception of state ineffectiveness. Many among both the agenda in the areas of labor, credit, business services, cost of rich and the poor, through different mechanisms, may find registration, taxes, business regulation, property rights, it convenient to opt out, which leads to higher informality. and access to judiciary service, among others, is critical to Tax collection is low, given the region's level of develop- making progress in fostering access to the formal economy. ment, which reveals that higher levels of taxation might be Policies in these areas, if designed in a consistent and needed to move toward the path of development. But this integrated fashion, can help reduce informality as part of begs the question, How and why should taxes be raised the concomitant process of increasing productivity and when much of the region perceives that the performance of incomes. But those policies should be part of a road map so the state is bleak? On one hand, even if the objective of rais- that partial steps might be taken over time, using political ing taxes is taken at face value, the main challenges are to windows of opportunity for reform. 244 T H E I N F O R M A L S E C T O R A N D T H E S TAT E : I N S T I T U T I O N S , I N E Q U A L I T Y, A N D S O C I A L N O R M S Notes cheat on taxes if you have the chance?" Tax morale variables used are 1. The following two sections draw heavily on Saavedra and expressed as the percent responding that it is never justified. Tommasi (2007). 13. Methodological difficulties in the ability to identify and 2. This discussion also relates to the fact that the willingness to calculate precisely social interaction parameters, in general, have contribute to finance welfare support is heavily influenced by percep- been discussed in detail by Manski (2000). Glaeser, Sacerdote, and tions about how deserving the recipients are, by how "close" the per- Scheinkman (2003) review the scarce recent empirical work that tries son feels to the recipients (ethnically, culturally, and so forth), and by to identify social multiplier effects. Their results suggest that social perceptions about the adequacy of state services and the like (Fong, interactions may be large. They posit that, if one person's proclivity Bowles, and Gintis 2005; Lindert 2004). to certain behavior influences his or her neighbor's behavior, policy 3. This is an example of opting out, which is discussed in more changes to address that behavior will have a direct effect and an indi- detail later in the chapter. rect one through social influence. "The presence of positive spillovers 4. The investment climate surveys used for these calculations or strategic complementarities creates a social multiplier where contain 11 countries from Latin America (Bolivia, Brazil, Chile, aggregate coefficients will be greater than individual coefficients" Costa Rica, Ecuador, El Salvador, Guatemala, Guyana, Honduras, (p. 2). Empirically, the estimated ratio of aggregate coefficients Nicaragua, and Peru). The sample contains 42 percent of small (estimated, say, at the city level) to individual coefficients is the social firms (20 employees or fewer), 38 percent of medium-size multiplier. The writers apply this approach to three settings. They firms (21­100 employees), and 20 percent of large firms (more than find that, among Dartmouth University students, one predetermined 100 employees). variable had a bigger impact on joining a fraternity at higher levels of 5. Data from the World Bank's Worldwide Governance Indica- aggregation. Using crime data, they find a very large social multi- tors (WGI) database are used extensively in this chapter. See plier; and, using data on wages and human capital, they find a social Kaufmann, Kraay, and Mastruzzi (2006). The Government Effective- multiplier at higher levels of aggregation. Another good overview is ness Index is a measure of the quality of public service provision, the provided in Gintis et al. (2005). quality of the bureaucracy, the competence of civil servants, the inde- 14. In game theory parlance, the strategies of players A and B are pendence of the civil service from political pressures, and the said to be strategic complements if when player A increases a component credibility of the government's commitment to policies. of his or her strategy, player B will want to do so also. In other words, 6. The Control of Corruption Index from WGI is a measure of per- the cross derivative of the payoff function of B with respect to the ceptions, where corruption is defined as the exercise of public power for action of A and his or her own action is positive. Strategic comple- private gain, with higher values corresponding to less corruption. mentarities, if strong enough, lead to multiplicity of equilibria. 7. They argue that the liberal reforms implemented in the mid- 15. The prospect of shame or potential stigma and guilt has a 1980s were an attempt to reduce the regulatory intent and, at the similar effect. The more likely an individual believes it is that he or same time, increase regulatory capacity. she will be condemned by others if caught, the more likely he or she 8. It should be noted that this last variable can be interpreted is to refrain from evading (Grasmick and Scott 1982; Kahan 2005). plainly as the size of the state, which may or may not be correlated 16. The authors make these calculations with and without pen- with enforcement capacity. sions. These figures exclude pensions because of the problems that 9. Share of informal economy (informality definition 1 through- exist in calculating how much people who receive the transfer had out the text) is the Schneider measure of the average size of the infor- paid during their working lives. In any case, the inclusion of mal economy. The share of self-employed (informality definition 2) is pensions--evenly if precisely calculated--will imply that social the ILO measure of self-employed workers in the workforce. The spending would be even more biased toward the rich. share of person without access to pension is computed based on 17. For further discussion, see Goñi, López, and Servén (2006). WDI and Centro de Estudios Distributivos Laborales y Sociales 18. High quality of institutions may explain the fact that, in for 14 Latin American countries (informality definition 3). These Chile, high income inequality is not translated into lower trust in the definitions are similar to those discussed in chapter 1. state, low tax morale, and, consequently, high informality. 10. This is probably capturing Latin American higher levels of 19. This section draws heavily on Saavedra and Tommasi (2007). inequality because the coefficient is not robust to the inclusion of that 20. In the specific case of pensions, privatization from pay-as-you- variable. go systems has not always implied an increase in trust in the system 11. This increase is after several years of declining tax revenues and uncertainty about future benefits, and political--and, in many caused by low growth, high inflation, and ineffective collection countries, economic--risk has not diminished (Kay 2003; Spiller and efforts. Tommasi forthcoming); so "private options" may imply alternatives, 12. Tax morale information is collected in the World Values Survey such as old-age savings accounts in foreign financial institutions, and the Latinobarometro, which are representative surveys collected in other assets, or reliance on family networks for the provision of 80 (worldwide) and 17 (Latin American) countries. At the national protection in old age. level, these surveys have a sample size of at least 1,000 observations 21. In many cases, like Argentina--and unlike Chile--(Bergman per country. Both the values survey and Latinobarometro ask the ques- 2002, 2003), there is a negative peer-pressure effect in which tax tion, "Could you tell me in a scale from 1 to 10 (where 1 is never jus- evasion has wide social acceptance, particularly among the better-off tified and 10 is totally justified) whether you think it is justified to populations. 245 I N F O R M A L I T Y References Chu, K., H. Davoodi, and S. Gupta. 2000. "Income Distribution and Ahmed, E., J. B. Rosser, and M. Rosser. 2004. "Income Inequality, Tax, and Government Social Spending Policies in Developing Corruption, and the Non-Observed Economy: A Global Perspec- Countries." Working Paper 214, World Institute for Development tive." Photocopy. Available at http://cob.jmu.edu/rosserjb/NOE. Economics Research, United Nations University, Helsinki. Inequality.Global.doc. Cowell, F. 1990. Cheating the Government: The Economics of Evasion. Alesina, A., and E. Glaeser. 2004. Fighting Poverty in the US and Europe: Cambridge, MA: Massachusetts Institute of Technology Press. A World of Difference. Oxford, UK: Oxford University Press. Davis, Lewis S. 2007. "Explaining the Evidence on Inequality Alm, J., and J. Martinez-Vazquez. 2007. "Tax Morale and Tax Eva- and Growth: Informality and Redistribution." The B. E. Journal sion in Latin America." Photocopy. Georgia State University, of Macroeconomics 7 (1) article 7. Available at: http://www. Atlanta. bepress.com/bejm/vol7/iss1/art7. Alm, J., I. Sánchez, and A. De Juan. 1995. "Economic and Noneco- de Ferranti, D., G. E. Perry, F. H. G. Ferreira, and M. Walton. 2004. nomic Factors in Tax Compliance." Kyklos 48 (3): 3­18. Inequality in Latin America and the Caribbean: Breaking with History? Alm, J., and B. Torgler. 2006. "Culture Difference and Tax Morale in Washington, DC: World Bank. Europe and the United States." Journal of Economic Psychology 27 (2): Detragiache, E. 2001. "The Labor Income Tax Credit in an Interna- 224­46. tional Perspective." Kingdom of the Netherlands Country Report Andreoni, J., B. Erard, and J. Feinstein. 1998. "Tax Compliance." 01/96, International Monetary Fund, Washington, DC. Journal of Economic Literature 36: 818­41. Djankov, S., R. La Porta, F. Lopez-de-Silanes, and A. Shleifer. 2002. "The Beramendi, P., and A. Díaz-Cayeros. 2006. "Distributive Tensions in Regulation of Entry." Quarterly Journal of Economics 117 (1): 1­37. Developing Federations." Paper presented at the 102nd Meeting Duryea, S., A. Olgiati, and L. Stone. 2006. "The Under-Registration of the American Political Science Association, Philadelphia, of Births in Latin America." Working Paper 551, Inter-American August 31­September 3. Development Bank, Washington, DC. Bergman, M. 2002. "Who Pays for Social Policy? A Study on Taxes ECLAC (Economic Commission for Latin America and the and Trust." Journal of Social Policy 31 (2): 289­305. Caribbean). 2006. Shaping the Future of Social Protection: Access, ------. 2003. "Tax Reform and Tax Compliance: The Divergent Financing, and Solidarity. New York: United Nations. Paths of Chile and Argentina." Journal of Latin American Studies Engel, E., A. Galetovic, and C. Raddatz. 1999. "Taxes and Income 35: 593­624. Distribution in Chile: Some Unpleasant Redistributive Biebesheimer, C., and M. Payne. 2001. "IDB Experience in Justice Arithmetic." Journal of Development Economics 59: 155­92. Reform." SDD Technical Paper 101, Inter-American Develop- Enste, D. H., and F. Schneider. 2000. "Shadow Economics: Size, Causes, ment Bank, Washington, DC. and Consequences." Journal of Economic Literature 38: 77­114. Bird R., J. Martinez-Vazquez, and B. Torgler. 2006. "Societal Insti- Eyzaguirre, H. 1996. "Institutions and Economic Development: Judi- tutions and Tax Effort in Developing Countries." In The Chal- cial Reform in Latin America." DPP Strategy and Policy Working lenges of Tax Reform in the Global Economy, ed. J. Alm, J. Paper 103, Inter-American Development Bank, Washington, DC. Martinez-Vazquez, and M. Rider, 283­338. New York: Springer. Falk, A., and U. Fischbacher. 2005. "Modelling Strong Reciprocity." In Bird, R., and E. Zolt. 2005. "Redistribution via Taxation: The Moral Sentiments and Material Interests: The Foundations of Cooperation Limited Role of the Personal Income Tax in Developing Coun- in Economic Life, ed. H. Gintis, S. Bowles, R. T. Boyd, and E. Feher, tries." Working Paper 0508, International Tax Program, Institute ch. 6. Cambridge, MA: Massachusetts Institute of Technology Press. for International Business, Joseph L. Rotman School of Manage- Farrell, D. 2006. "Tracking the Informal Economy." Business ment, University of Toronto. Week, May 8; available at http://www.mckinsey.com/aboutus/ Birdsall, N., and R. Menezes. 2005. "Beyond the Washington mckinseynews/tacklingeconomy.asp. Consensus: A New Job-based Social Contract in Latin America." Feld, L., and B. Frey. 2007. "Tax Compliance as the Result of a Psy- Foreign Affairs en Español 5 (3). chological Tax Contract: The Role of Incentives and Responsive Boylan, D. 1996. "Taxation and Transition: The Politics of the 1990 Regulation." Law and Policy 29 (1): 102­20. Chilean Tax Reform." Latin American Research Review 31 (1): 7­31. Filgueira, F. 2005. "Welfare and Democracy in Latin America: The Breceda, K., J. Rigolini, and J. Saavedra. 2007. "Taxes and Transfer Development, Crises and Aftermath of Universal, Dual, and in Latin America." Photocopy. World Bank, Washington, DC. Exclusionary Social States." Prepared for the United Nations Brewer, M., A. Duncan, A. Shepard, and M. Suárez. 2005. "Did Work- Research Institute for Social Development Project on Social Pol- ing Families' Tax Credit Work? The Final Evaluation of the Impact icy and Democratization, New York. of In-work Support on Parents' Labor Supply and Take-up Behav- Fong, C., S. Bowles, and H. Gintis. 2005. "Reciprocity and the ior in the UK." Photocopy. Institute for Fiscal Studies, London. Welfare State." In Moral Sentiments and Material Interests: The Centeno, M. A., and A. Portes. 2006. "The Informal Economy in the Foundations of Cooperation in Economic Life, ed. H. Gintis, S. Shadow of the State." In Out of the Shadows: Political Action and the Bowles, R. T. Boyd, and E. Feher, ch. 10. Cambridge, MA: Mass- Informal Economy in Latin America, ed. P. Fernández-Kelly and achusetts Institute of Technology Press. J. Shefner. State College, PA: Pennsylvania State University Press. Friedman, E., S. Johnson, D. Kaufmann, and P. Zoido-Lobatón. 2000. Chong, A., and M. Gradstein. 2007. "Inequality and Informality." "Dodging the Grabbing Hand: The Determinants of Unofficial Journal of Public Economics 9: 159­79. Activity in 69 Countries." Journal of Public Economics 76: 459­93. 246 T H E I N F O R M A L S E C T O R A N D T H E S TAT E : I N S T I T U T I O N S , I N E Q U A L I T Y, A N D S O C I A L N O R M S Gintis, H., S. Bowles, R. T. Boyd, and E. Feher, eds. 2005. Moral and E. Feher, ch. 12. Cambridge, MA: Massachusetts Institute of Sentiments and Material Interests: The Foundations of Cooperation in Technology Press. Economic Life. Cambridge, MA: Massachusetts Institute of Tech- Kaufmann, D., A. Kraay, and M. Mastruzzi. 2006. "Governance Mat- nology Press. ters V: Governance Indicators for 1996­2005." Available at Glaeser, E. L., B. I. Sacerdote, and J. A. Scheinkman. 2003. "The http://ssrn.com/abstract=929549. Social Multiplier." Journal of the European Economic Association 1 Kay, S. J. 2003. "Pension Reform and Political Risk." Prepared for (2­3): 345­53. the LASA XXIV International Congress, Dallas, March 27­29. Gómez, J. L. R. 2005. "The Political Economy of Nicaragua's Insti- Kuznets, S. 1955. "Toward a Theory of Economic Growth." In tutional and Organizational Framework for Dealing with Vio- National Policy for Economic Welfare at Home and Abroad, ed. R. lence." Photocopy. Crisis States Research Center, Universidad Lekachman, Garden City, NY: Doubleday. Centroamericana, Managua, Nicaragua. Levi, M. 1998. "A State of Trust." In Trust and Governance, ed. V. Gomez-Sabaini, J. 2005. "Evolucion y situacion Tributaria Actual en Braithwaite and M. Levi, 77­101. New York: Russell Sage America Latina: Una Serie de Temas para la Discusion." Presenta- Foundation. tion to CEPAL, Santiago de Chile, Chile, September. Lindert, K., E. Skoufias, and J. Shapiro. 2006. "Redistributing Goñi, E., J. H. López, and L. Servén. 2006. "Fiscal Reform for Social Income to the Poor and the Rich: Public Transfers in Latin Equity in Latin America." Photocopy. World Bank, Washington, America and the Caribbean." Social Protection Discussion Paper DC. 0605, World Bank, Washington, DC. Grasmick, H., and W. Scott. 1982. "Tax Evasion and Mechanisms of Lindert, P. 2004. Growing Public: Social Spending and Economic Growth Social Control: A Comparison with Grand and Petty Theft." Since the Eighteenth Century. Cambridge, UK: Cambridge Univer- Journal of Economic Psychology 2 (3): 213­30. sity Press. Greenstein, R. 2005. "The Earned Income Tax Credit: Boosting Lledo, V., A. Schneider, and M. Moore. 2004. "Governance, Taxes Employment, Aiding the Working Poor." Photocopy. Center on and Tax Reform in Latin America." IDS Working Paper 221, Budget and Policy Priorities, Washington, DC. Available at Institute of Development Studies, Brighton, UK. http://www.cbpp.org/7-19-05eic.pdf. Loayza, N. 1996. "The Economics of the Informal Sector: A Simple Guerrero, I., L. F. López-Calva, and M. Walton. 2006. "The Inequality Model and Some Evidence from Latin America." Carnegie- Trap and Its Links to Low Growth in Mexico." Working Paper 298, Rochester Conference Series on Public Policy 45: 129­62. Stanford Center for International Development, Stanford, CA. Loayza, N., A. M. Oviedo, and L. Servén. 2005. "The Impact of Regu- Hirschman, A. O. 1970. Exit, Voice, and Loyalty: Responses to Decline in lation on Growth and Informality--Cross-country Evidence." Pol- Firms, Organizations and States. Cambridge, MA: Harvard icy Research Working Paper 3263, World Bank, Washington, DC. University Press. Loayza, N., and J. Rigolini. 2006. "Informality Trends and Cycles." Hoffman, S. D., and L. S. Seidman. 2003. Helping Working Families: Policy Research Working Paper 4078, World Bank, Washington, The Earned Income Tax Credit. Kalamazoo, MI: Upjohn Institute. DC. Holt, S. 2006. The Earned Income Tax Credit at Age 30: What We Know. Lora, E. 2006a. "El Futuro de los Pactos Fiscales en América Latina." Brooking Institution Research Brief. Washington: Brookings Paper prepared for the forum Políticas Económicas para un Nuevo Institution Press. Pacto Social en América Latina, Fundación CIDOB, Barcelona, Holtz, J., and J. K. Scholz. 2003. "The Earned Income Tax Credit." October. In Means-Tested Transfer Programs in the United States, ed. R. A. ------. 2006b. "Trends and Outcomes of Tax Reform." In The State Moffitt. Chicago: University of Chicago Press. of State Reforms in Latin America, ed. E. Lora. Washington, DC: ICEF (Instituto Centroamericano de Estudios Fiscales). 2005. "Diez Claves World Bank. para un Pacto Fiscal." Boletín de Estudios Fiscales 1. Lora, E., M. Cardenas, and V. Mercer-Blackman. 2005. "The Policy IDB (Inter-American Development Bank). 2005. The Politics of Poli- Making Process of Tax Reform in Latin America." Prepared for the cies: Economic and Social Progress in Latin America. Washington, DC. Inter-American Development Bank Workshop on State Reform, ------. 2006. "Hidden in Plain Sight: Undocumented Citizens Public Policies, and Policymaking Processes, Washington, DC. in Latin America." Ideas for Development in the Americas 9; avail- Manski, C. 2000. "Economic Analysis of Social Interactions." Journal able at http://www.iadb.org/news/articledetail.cfm?Language= of Economic Perspectives 14 (3): 115­36. EN&artid=2697&artType=WS. Martinez-Vazquez, J., and B. Torgler. 2005. "The Evolution of Tax Johnson, S., D. Kaufmann, and A. Shleifer. 1997. "The Unofficial Morale in Modern Spain." Working Paper 2005­33, Center for Economy in Transition." Brookings Papers on Economic Activity 2: Research in Economic Management and the Arts, Basel, 159­239. Switzerland. Johnson, S., D. Kaufmann, and P. Zoido-Lobatón. 1998. "Regulatory Mendez, J. E., G. O'Donnell, and P. S. Pinheiro, eds. 1999. The (Un) Discretion and the Informal Economy." American Economic Review Rule of Law and the Underprivileged in Latin America. North Bend, 88 (2): 387­92. IN: University of Notre Dame Press. Kahan, D. 2005. "The Logic of Reciprocity: Trust, Collective Action, Meyer, B. 2001. "Taxes, Welfare, and Work by Single Mothers." and Law." In Moral Sentiments and Material Interests: The Foundations NBER Reporter Online, available at www.nber.org/reporter/fall01/ of Cooperation in Economic Life, ed. H. Gintis, S. Bowles, R. T. Boyd, meyer.html (accessed February 2007). 247 I N F O R M A L I T Y ------. 2002. "Labor Supply at the Extensive and Intensive Economic Research) Political Economy workshop. Available at Margins: The EITC, Welfare, and Hours Worked." American http://www.yale.net/polisci/info/conferences/Political_Economy_ Economic Review 92 (2): 373­79. Post-Industrial_Societies/Sokoloff.pdf. Perry, G., O. Arias, H. López, W. F. Maloney, and L. Servén. 2006. Spicer, M. W., and L. A. Becker. 1980. "Fiscal Inequity and Tax Evasion: "Poverty Reduction and Growth: Virtuous and Vicious Circles." An Experimental Approach." National Tax Journal 33: 171­75. World Bank, Washington, DC. Spiller, P., and M. Tommasi. Forthcoming. "Public Policy in Rodgers, D. 2004. "Disembedding the City: Crime, Insecurity, and Argentina." In. The Institutional Foundations of Public Policy in Spatial Organisation in Managua, Nicaragua." Environment and Argentina, ed. P. Spiller and M. Tommasi. Cambridge, UK: Urbanization 16 (2): 113­24. Cambridge University Press. Ronconi, L. 2006. "Notes of the Judiciary and Informality." Photocopy. Tanzi, V. 2000. "Taxation in Latin America in the Last Decade." Saavedra, J., and M. Tommasi. 2007. "Informality, the State and the Working Paper 76, Center for Research on Economic Develop- Social Contract." Photocopy. Universidad de San Andrés, Buenos ment and Policy Reform, Stanford University, Stanford, CA. Aires, Argentina. Tanzi V., and H. Zee. 2000. "Tax Policy for Emerging Markets: Schelling, R. 1978. Micromotives and Macrobehavior. New York: Norton. Developing Countries." Working Paper WF/00/35, International Schmitt, J. 2005. "Is It Time to Export the U.S. Tax Model to Latin Monetary Fund, Washington, DC. America?" Challenge 48 (3): 84­108. Torgler, B. 2005. "Tax Morale in Latin America." Public Choice 122: Scholz, J. T., and M. Lubell. 1998. "Trust and Taxpayers: Testing the 133­57. Heuristic Approach to Collective Action." American Journal of World Bank. 2005. World Development Indicators. Available at: Political Science 42: 398­417. www.worldbank.org. Sheffrin, S., and R. Triest. 1992. "Can Brute Deterrence Backfire? ------. 2006. World Development Report 2006: Equity and Develop- Perceptions and Attitudes in Taxpayer Compliance." In Why ment. Washington, DC. People Pay Taxes: Tax Compliance and Enforcement, ed. J. Slemrod. ------. 2007. "Job Creation in Latin America and the Caribbean: Ann Arbor, MI: University of Michigan Press. Recent Trends and the Policy Challenges." Photocopy. World Sokoloff, K., and E. Zolt. 2005. "Inequality and the Evolution of Bank, Washington, DC. Institutions of Taxation: Evidence from the Economic History Young, H. P. 2006. "Social Norms." Paper prepared for the New Palgrave of the Americas." Paper presented at NBER (National Bureau of Dictionary of Economics. Basingstroke, UK: Palgrave Macmillan. 248 ECO-AUDIT Environmental Benefits Statement The World Bank is committed to preserv- Saved: ing endangered forests and natural · 42 trees resources. The Office of the Publisher has · 3,508 lbs. of solid chosen to print Informality: Exit and waste Exclusion on recycled paper with 30 per- cent post-consumer waste, in accordance · 29,117 gallons of with the recommended standards for paper wastewater usage set by the Green Press Initiative, a · 8,493 lbs. of net nonprofit program supporting publishers greenhouse gases in using fiber that is not sourced from · 58 million BTUs of endangered forests. For more information, energy visit www.greenpressinitiative.org. nformality is a manifestation of a flawed relationship between citizens I and their state. While the term means different things to different peo- ple, it almost always has negative connotations--including inferior working conditions, low productivity businesses, unfair competition, and disrespect for the rule of law. It is not surprising then that the rise observed in informality in Latin American and Caribbean countries over recent decades has garnered attention as a potential brake on improvements in growth and social welfare, and as a force corrosive to the integrity of society. Informality: Exit and Exclusion analyzes informality in Latin America, exploring root causes and reasons for and implications of its growth. The authors use two distinct but complementary lenses: informality driven by "exclusion" from state benefits or the circuits of the modern economy, and driven by voluntary "exit" decisions resulting from private cost-benefit calculations that lead workers and firms to opt out of formal institutions. They find both lenses have considerable explanatory power to understand the causes and consequences of informality in the region. Informality: Exit and Exclusion concludes that reducing informality levels and overcoming the "culture of informality" will require actions to increase aggregate pro- ductivity in the economy, reform poorly designed regulations and social policies, and increase the legitimacy of the state by improving the quality and fairness of state institutions and policies. Although the study focuses on Latin America, its analysis, approach, and conclusions are relevant for all developing countries. Informality: Exit and Exclusion will be of value to professionals and academ- ics studying labor market, social protection, tax, microenterprise develop- ment, and urban public policies, and to those working in government, international organizations, research institutions, and universities. ISBN 0-8213-7092-8