Report No. 24408-BR Brazil Jobs Report (In Two Volumes) Volume II: Background Papers December 20, 2002 A Joint Report by The World Bank Instituto de Pesquisa With the Support of Brazil Country Management Unit Econ6mica Aplicada the Government of Brazil Latin America and the Caribbean Region Ministry of Labor and Employment Document of the World Bank Currency Unit - Real (R$) April 2002: R$2.36 The Metric System is used throughout the report. HS@AL YEAR January 1 to December 31 Vice President LCR: David de Ferranti Director LCC5C: Vinod Thomas Director LCSHD: Ana-Maria Aniagada Lead Economist: Joachim von Amsbarg Task Manager lndermn S. Gill - iii - TABLE OF CONTENTS **i3 AS 0yLo It]I,011 IN Ia f-il:pl.1o I ROADMAP ................................................ 1 I INTRODUCTION ........................................................... 5 11 CONCEPTUAL AND EMPIRICAL MOTIVATION ................................................ 7 A New Conceptual Framework ...........................................7 Some Empirical Evidence ............................................. 12 III THE BRAZILIAN CASE ........................................................... 15 Further Empirical Evidence .................... ........................ 15 IV POLICY IMPLICATIONS ........................................................... 23 Less Segmentation versus More Integration ........................... ................. 23 Labor Policy Orientations and the Macroeconomic Environment . . 25 V CONCLUSION ................................................. 30 I INTRODUCTION ............................................................................. 40 11 THE CORE OF THE INFORMAL SECTOR: THE SELF-EMPLOYED . . 4 1 1 [[ BUT DON'T INFORMAL WORKERS EARN LESS? ..................................9...... 46 IV ISN'T TI-IE INFORMAL SECTOR TIE SAFETY NET FOR DISPLACED FORMAL SECTOR WORKERS'? ................................ 47 V DOESN'T THE L.ARGE SIZE OF THE INFORMAL SECTOR IMPLY LARCE LABOR MLARKET DISTORTIONS 9 ................................... 50 VI ISN'T INFORMAL WORK PRECARIOtS? .........9.......................... 5 1 VII DOESN'T THE INFORMAL SFCTOR SHOW LOWER OVERALL PRODUCTIVITY GROWTH? ................................ 53 VIII ISN'T FORMALITY LARGELY A WAY FOR LARGER FIRMS TO AVOID PROTECTING THEIR WORKERS? ................................ 53 IX DOESN'T THE LARGE NUMBER OF WOMEN IN INFORMAL SELF-EMPLOYMENTr SUGGEST DISCRIMINATION IN THE FORMAL SECTOR? ......................................... 54 X AREN'T THE INFORMAL SALARIED THE MOST DISADVANTAGED? ....................... 55 XI CONCLUSION ......................................................... 56 I INTRODUCTION ......................................................... 61 11 l,ABOR L.AWS, UNION LEGISLATION, AND COLLECTIVE BARGAINING PROCESSES62 Labor Laws and Labor Courts .................................................... 62 Union Organization and Collective Bargaining .................................................... 63 IlI CHARACTERISTICS OF UNIONIZED WORKERS AND UNION DENSITY ..................... 66 Characteristics of Unionized Workers ............................ ........................ 67 Union Density ..................................................... 70 IV UNIONS AND WAGES ......................................................... 72 Union Wage Premium .................................................... 72 Unions and wage dispersion .............. ....................................... 73 Unions and Collective Bargaining .................................................... 77 V DIsCUSSION ......................................................... 79 V• RECOMMENDATIONS ......................................................... 82 - iv - llMoMM 1W (OF M M MD COLLECOQE SAUN . LE,` I BACKGROUNI)ANDSCOPEOFTHESIrUDY ........................................................ 96 11 UNIONIZATION AND COLIECTIVE BARGAINING: SOME STATISTICS ................... 97 111 A THEORETICAI, REVIEW OF UNIONISM AND COLLECTIVE BARGAINING ... 99 What Unions Do .................................................. 99 Bargaining Coordination .................................................. 100 IV EMPaRuCAL EVIDENCE FROM MICRO DATA ...................................................... 102 Union Non-union Wage Mark-up .................................................. 103 Other Union Effects .................................................. 106 V EMPIRICAL EVTDENCE FROM MACRO DATA ..................................................... 108 Union Density and Bargaining Coverage .................................................. 108 Bargaining Coordination and Comparative Economic Performance ................... 111 VI CONCIUSIONS ........................................................ 112 I INTRODLUCTION ........................................................ 120 fl ORIGINAL FUTNCTIONS AND PRINCIPLES OF THF BRAZILIAN LABOR JUSTICE SYSTEM ........................................................ 120 III THE LOGISTICS OF OPERATIONS ............................ ............................ 121 The Structure of the Courts .................................................. 121 The functioning of the labor justice and negotiations practices .............. ............ 122 The cost of accessing the labor justice .................................................. 125 IV iOW LABOR RELATIONS ARE AFFECTED BY' HE FLNCTION ING OFTI IE COURiS 126 V A PROPOSAL FOR REFORM ................ ........................................ 127 VI CONCLUSION ........................................................ 128 I INTRODUCTION ...................................................... 13 1 11 HISTORICAL BACKGROUND OF THE MINIMUM WAGE ..................... ................. 132 H11 CURRENT MINIMUM WAGE POLICY ............................... ...................... 133 IV PROFILE OF WORKERS WHiO EARN MINIlMUM WAGE . ...................................... 134 V IM-PACTON THELABORMARKET .................... ................................. 134 Estimates of the Employment-Effect of the Minimum Wage ............................... 136 The Impact of the Minimum Wage on Other Wages ........................................... 142 VI IMPACT OF THE MINIMUM WAGE ON POVERTY . ............................................... 146 Vil THE FISCAI.IMPACTOFMINIMUMWAGEPOLICY ....................................... 149 Impact on Payroll ............................................... 149 Impact on the Social Security System Budget ............................................... 150 VIII TooLs To AID IN REFORMING MINIMLUM WAGE POLICY . . 151 National vs. Regional Minimum Wages ............................................... 151 An Occupational or Sectoral Minimum Wage ............................................... 152 Differentiated Coverage ............................................... 153 Criteria for Adjustment and Management of the Minimum Wage Policy ............. 153 IX CONCLUSIONS ................................................................. 154 I INTRODUCTION ..................................... 164 11 CONCEPTUJAiL FRAMEWORK ............. 165 Competitive Model .......................... 165 Two Sector Model ........................... 167 T1I THE CORRFCT MlNIMUM WAC,E l,EVEI ..............................9........................... 169 IV IMPACT OF THE MIINIMUM WAGES ON THE WAGE DISTRIBLITION ....................... 170 Is the Minimum Wage Binding? ........................................... 171 How Does the Minimum Wage Affect the Wage Distribution? ............................ 171 V IMPACTOFNSWONEMPLOYMENT ....................................I ..... 173 Single labor market ........................................... 173 Covered and Uncovered Sectors ............. .............................. 174 VI POVERTY TMPACTS ............................................... 174 Poverty without risk pooling ........................................... 176 Who Earns the Minimum Wage? ........................................... 176 Poverty and Inequality with risk pooling at the household level .......................... 178 Vni FISCAL IMPACTS OF THE MINIMUM WAGE ............................................... 179 VIII CONCLUSIONS AND IMPLICATIONS FOR POLICY ........................................... 181 I INTRODUCTION ............................................... 189 11 UNEMPLOYMENT INSURANCE SYSTEM AND THE FGTS ................................... 190 The Brazilian Unemployment Insurance System ........................................... 190 The FGTS ........................................... 191 11I DATA ISSLuES ............................................... 192 IV UNEMPLOYMENT RATES IN BRAZIL .................... ........................... 193 V THE PROBABILITY OF BECOMING UNEMPLOYED ........................................ ; 196 .Descriptive Statistics .................................... 196 Evolution of the Probability of Becoming Unemployed Conditional on Employment Status .............................. 197 Probability of Becoming Unemployed Conditional on Individual Characteristics 199 VI UNEMPLOYMENT DTRATION ....................................... 202 Data and Descriptive Statistics .................................... 202 Evolution of unemployment duration .................................... 204 Kaplan Meier Estimates .................................... 207 Conditional unemployment duration and exit states .................................... 210 Vll POLICY IMPLICATIONS AND CONCILUDING REMARKS . 213 Unemployment Insurance and Informality .214 FGTS, Ul and Unemployment .215 Human Capital and U .216 Special support for poor workers .216 I INTRODUCTION ..................... 219 Types of Policies .............. 219 11 COMPENSATORY POLICTES AS A FORM OF WORKER PROTECTION .220 -vi - Description of the Main Compensatory Programs for Worker Protection ........... 221 Analysis of Program Targeting .......................................... 222 Protection Programs, Behavior of Economic Agents and Labor Market Performance .................................. 225 III DISTRIBUTIVE POLICIES: THE MINIMUM WAGE . .............................. 231 The Level of the Minimum Wage in Brazil .................................. 231 Impact of the Minimum Wage on Poverty in Brazil .................................. 234 IV DIRECT STRlUCTURAL POLICIES ..................................... 238 Targeting of Structural Policies .................................. 238 The Nature of Unemployment .................................. 238 Reasons for Low Productivity .................................. 240 Types of Structural Policy and Their Rationale .................................. 240 The Role of the State .................................. 241 Evaluating impact .................................. 242 Operational issues .................................. 243 V SUMMARY AND CONCLUSIONS ..................................... 244 A 10$ ,W-W3EO f X9 I TNTRODUCTION ..................................... 261 11 ACTIVE LABOR MARKET PROGRAMS: AN OVERVIEW ..................................... 262 Expenditures on Active Programs in OECD Countries ....................................... 262 III THE IMPORTANCE OF GOOD EVALUATIONS . .................................. 264 Impact Evaluation Techniques ........................................ 265 Relative Strengths of Techniques ........................................ 268 The Importance of Costs ......................................... 269 IV AN INTERPRETATION OF EVALLATION RESULTS . .............................. 269 Public Works Programs/Public Service Employment ........................................ 270 Job Search Assistance/Employment Services ......................................... 270 Training and Retraining Programs ........................................ 272 Wage/Employment Subsidies (WES) ........................................ 273 Micro-Enterprise Development (Self-employment Schemes) ............................. 275 Summary of Evaluative Evidence ........................................ 277 V CONCLUSION ............................................ 278 - vii - List of Tables Table 3.1 Sources of Real Exchange Rate Variations and Labor Market Behavior ........................................................ 20 Table 3.2 Policy Priorities During Alternative Macroeconomic Scenarios .................................................................. 30 Table 4.1 Real Hourly Wage Differential - Formal Salaried to Informal Self Employed .................................................. 43 Table 4.2 Rationale for not Preferring to Move to a Formal Sector Job, Brazil (%) ........................................................ 43 Table 4.3 Motivations for becoming seH-employed (in %) .................................................................. 44 Table 4.4 Sector of Origin of the Unemployed .................................................................. 49 Table 5.1 Characteristics Of Unionized And Non-Union Workers .................................................................. 85 Table 5.2 Proportion Of Unionized Workers By Region And Occupation .................................................................. 86 Table 5.3 Probability of Union Membership .................................................................. 87 Table 5.4 Union Density .................................................................. 88 Table 5.5 Union Density By Group .................................................................. 89 Table 5.6 Differential Of Log. Of Real Hourly Wages .................................................................. 89 Table 5.7 Union Wage Mark-Up ................................................................... 90 Table 5.8 Wage Dispersion Among Unionized And Non-Union Workers .................................................................. 90 Table 5.9 Simulation Of Standard Deviation Of Unionized And Non-Union Workers' Wages ........................................ 90 Table 5.10 Retums To Human Capital Variables In Unionized And Non-Union Workers Wage equatons, 1999 .......... 91 Table 5.11 Schooling Coefficients Among Unionized And Non-Union Workers, 1999 ................................................... 91 Table 5.12 Standard deviation of the inter-industry wage differential .................................................................. 92 Table 6.1 Union Density And Bargaining Coverage .................................................................. 98 Table 6.2 Union-nonunion Wage Mark-up for Selected Countries .................................................................. 104 Table 6.3 Union Density and Economic Performance in the OECD Counbies ............................................................. 109 Table 6.4 Bargaining Coverage And Economic Performance: A Summary Of The Relevant Studies .......................... 110 Table 7.1 Rights Contested in Labor Courts 1995 .................................................................. 123 Table 7.2 Percentage of Resolved Disputes in each Hierarchical Level .................................................................. 123 Table 7.3 Acceptance of Claims by the Labor Justice .................................................................. 125 Table 7.4 Workers Going to Labor Courts by Wage Level .................................................................. 127 Table 8.1 Profile of Workers who Eam Minimum Wage in Brazil ................................................................... 159 Table 8.2 Profile of Workers who Earn Minimum Wage .................................................................. 160 Table 8.3 Impact of the Minimum Wage on the Payroll, in R$ Million .................................................................. 161 Table 8.4 Impact of the Minimum Wage on the Social Security Deficit .................................................................. 161 Table 8.5 Matrix of Options for Minimum Wage Policy in Brazil .................................................................. 162 Table 8.6 Summary of the Results of Studies on the Impact of the Minimum Wage in Brazil ...................................... 163 Table 9.1 Ratio of the Minimum Wage (Mw) to Comparison Wages .................................................................. 170 Table 9.2 Overview of the Impact of Minimum Wage on Poverty .................................................................. 175 Table 9.3 Wage Eamers at or Below The Minimum Wage .................................................................. 177 Table 9.4 Household Characteristics .................................................................. 178 Table 9.5 Summary of the Literature on the Employment Effects of Minimum Wages ................................................. 183 Table 10.1 Unemployment Rates by Education and Household Status .................................................................. 195 Table 10.2 Descriptive Statistics .................................................................. 197 Table 10.3 Probability of Becoming Unemployed .................................................................. 201 Table 10.4 Descriptive Statistics of Duration Data ................................................................... 204 Table 10.5 Cox Proportional Hazard Model .................................................................. 209 Table 10.6 Competing Risk: Basic Sample .................................................................. 211 Table 10.7 Competing Risk: Formal Sample .................................................................. 212 Table 12.1 Expenditures on Labor Market Programs (Selected OECD Countries) ...................................................... 264 Table 12.2 Effectiveness of Wage Subsidy Programs ................................................................... 275 Table 12.3 Failure Rates of MEDA Business .................................................................. 276 Table 12.4 Overview of Active Labor Programs ................................................................... 278 - viii - Ust of Figures Figure 2.1 Real Effective Exchange Rate, Relative Wage and Share of Informal Workers ............................................ 13 Figure 2.2 Some Evidence for Segmentation in Brazil ...................................................... 14 Figure 2.3 Wage and Employment Okun Coefficients: 198901 to 199904 ...................................................... 16 Figure 2.4 Real Effective Exchange Rate, Relative Wage and Share of Informality in Brazil ......................................... 19 Figure 2.5 Recursive Cointegration Tests Between Real GDP and Employment: ...................................................... 36 Figure 2.6 Recursive Cointegration Tests Between Real GDP and Real Wages ...................................................... 36 Figure 2.7 Self-Employed versus Informal ...................................................... 37 Figure 2.8 Self-Employed versus Informal ...................................................... 38 Figure 2.9 Self-Employed versus Formal ...................................................... 39 Figure 2.10 Formal versus Informal ...................................................... 39 Figure 3.1 Self-employment in the Working Population ...................................................... 41 Figure 3.2 Informal Salaried by Gender ...................................................... 42 Figure 3.3 Unpaid by Gender ...................................................... 42 Figure 3.4 Ratio of Self-employment Eamings to Formal Sector Wages, late 1990s ..................................................... 46 Figure 3.5 Ratio of Informal to Formal Sector Wages, late 1990s ...................................................... 46 Figure 3.6 Colombia, Relative Sector Size and Earnings ...................................................... 48 Figure 3.7 Mexico, Relative Sector Size and Earnings ...................................................... 48 Figure 3.8 Brazil, Relative Sector Size and Eamings ...................................................... 49 Figure 3.9 Self-Employment vs. Industrial Productivity ...................................................... 50 Figure 3.10 Women's Participation in the informal sector ...................................................... 55 Figure 4.1 Union Density vs. Formalization, 1999 ...................................................... 93 Figure 4.2 Union Density (0/6) ...................................................... 93 Figure 4.3 Schooling and union density - industry, 1999 ...................................................... 94 Figure 4.4 Distribution of Real Hourly Wages - Union Members, 1999 ...................................................... 94 Figure 4.5 Distribution of Real Hourly Wage - Non-union Members, 1999 ...................................................... 95 Figure 5.1 Macro-economic Constraints and Labor Market Structure ...................................................... 118 Figure 8.1 Minimum Wages and Unemployment ...................................................... 166 Figure 9.1 Different Definitions of Unemployment Rates ...................................................... 194 Figure 9.2 Unemployment Rates by Gender ...................................................... 195 Figure 9.3 Unemployment Entrance Probability Conditional on Employment Status .................................................... 199 Figure 9.4 Mean Unemployment Duration by Gender ...................................................... 205 Figure 9.5 Mean Duration by Working Class ...................................................... 206 Figure 9.6 Unemployment Duration by Wage Groups ...................................................... 206 Figure 9.7 Estimated Kaplan Meier Survivor Function by Gender and FGTS ...................................................... 207 Figure 11.1 OECD Average ALMP Expenditures ...................................................... 263 Figure 11.2 Labor Market Expenditures ...................................................... 263 - ix- ACKNOWVLEDGEMENTS This report is the result of collaboration between Brazil's Institute for Applied Economic Research (IPEA) and the World Bank between 2000 and 2002. The task was jointly managed by Indermit Gill (Lead Economist, World Bank) and Ricardo Paes de Barros (Director of Social Studies, IPEA-Rio de Janeiro), with the support of Andreas Blom (Economist, LCSHD) and Jose Marcio Camargo of the Pontifical Catholic University of Rio de Janeiro. The report was prepared under the overall guidance of Roberto Martins (President, IPEA), Gobind Nankani and Vinod Thomas (Country Directors during 2000-2002), and Xavier Coll and Ana-Maria Arriagada (Sector Directors during 2000-2002), and Suman Bery and Joachim von Amsberg (Lead Economists during 2000-2002). Volume I consists of the Policy Report, prepared by Indermit S. Gill, Andreas Blom, Ricardo Paes de Barros, Carlos Henrique Corseuil (Economist, IPEA) and Mirela Silva (Economist, IPEA). Volume II, prepared by Andreas Blom and Indermit Gill, contains the background papers commissioned for the report, that diagnose labor market developments in Brazil, present relevant intemational experience, and discuss the policy implications. The authors of the papers are Jose Marcio Camargo (PUC-Rio de Janeiro), Francisco Galr&o Carneiro (Universidade Catolica de Brasilia), Jorge Saba Arbache (Universidade de Brasilia), Ricardo Paes de Barros, William Maloney (Lead Economist, LCRCE), Wendy Cunningham (Economist, LCSPR), Norbert Fiess (Economist, LCRCE), Marco Fugazza (Consultant, CERAS, Ecole Nationale des Ponts et Chausses, Paris), Dorte Domeland-Narvaez (Consultant, Universidad Pompeu-Fabra, Barcelona), Amit Dar (Senior Economist, HDNSP) and Andreas Blom. Edward Amadeo (former Minister of Labor, Brazil), Marcelo Neri (Funda,co Getulio Vargas-Rio.de Janeiro), Rosane Mendonga, and Lauro Ramos (IPEA-Rio de Janeiro) provided valuable feedback and guidance in Brazil, and Marito Garcia (Education Sector Manager, LCSHD), Gillette Hall (Social Protection Economist, LCSHD), Robert Holzmann (Director, HDNSP), Truman Packard (Social Protection Economist, LCSHD), Mark Thomas (Senior Economist, LCC5C), Dorte Verner (Senior Economist, LCC5C) and Joachim von Amsberg (Lead Economist, LCC5C) at the World Bank. The participants of a two-day workshop in Washington on January 9-10, 2002 to discuss the main findings and policy messages influenced the form and contents of both volumes. The team expresses its appreciation to Gabriela Falconi (LCSHD) and Andreas Blom for organizing the workshop and facilitating IPEA's participation, and to Truman Packard for moderating the discussions effectively. Participants in two workshops organized by HDNSP also provided useful feedback; the critical commentary of Paul Lundall (Deputy Director, Development Policy Research Unit, University of Cape Town) is especially appreciated. The peer reviewer for this task is Gordon Betcherman (Senior Economist, HDNSP), who provided detailed comments on both the concept paper and an earlier draft, and to whom the team owes a debt of gratitude. Finally, we wish to acknowledge the continued collaboration of the Government of Brazil, especially the support of the Minister of Labor, Mr. Paulo Jobim Filho, Mr. Francisco Dornelles, the former Minister of Labor, and Mr. Roberto Martins, President, IPEA, for this task. - 1 - 1. ROADMAP TO VOLUME II: DETAILED REPORT 1.1. This is the second of two volumes of the World Bank's uBrazil Jobs Report", which studies the Brazilian Labor Market. Volume I summarizes the analysis and the policy recommendations building upon analysis from background papers and other relevant research. Volume il presents the ten background studies in their entirety. This introductory chapter provides a roadmap to the volume by recapitulating each chapter. 1.2. Chapter 2 by Norbert Fiess (World Bank), Marco Fugazza (CERAS, Ecole Nationale des Ponts et Chaussees) and William Maloney (World Bank) analyzes the functioning of the labor market for the success of macro-economic policies, like fiscal adjustment, stabilization and trade liberalization. The chapter contrasts the traditional "dualistic" view of labor markets in less developed countries with a new "integrationist' view.1 The authors argue that these views are not necessarily incompatible. The labor market functioning in three Latin American countries-Mexico, Colombia and Brazil- is examined. The study finds that in Brazil in the 1990s, the labor market oscillated between segmentation and integration of the formal and informal labor markets. They show that the degree of labor market segmentation is not uniquely determined by the existence of labor market rigidities, but varies with the country's macroeconomic environment. The policy implications are compelling: only if there exists evidence of segmentation, is a large informal sector a signal of distorting labor market regulations, in which case a reduction of labor market rigidities would be welfare improving. In the opposite case, where there is evidence of integration then the informal sector may prove to be an important source of efficient employment and reducing labor market rigidities may have a weak or even counter-productive effect. 1.3. In Chapter 3, Wendy Cunningham (World Bank) discusses the definition of the informal sector and the reasons for its existence drawing on cross-country evidence. Issues on which three decades of research have failed to reach a consensus. Broadly speaking, the small-scale, semi-legal, often low-productivity, frequently family-based and perhaps pre-capitalistic enterprises continue to employ between 30% and 70% of the urban work force in Latin America. A long tradftion views informal workers as comprising the less-advantaged sector of a dualistic or segmented labor market. This chapter challenges the traditional dualism theory and seeks a better understanding of the informal sector: why it exists, and the quality of its jobs. It goes through a series of assumption about the informal sector and presents recent evidence to test the validity of the assumption. The review shows that the informal sector is a dynamic sector that is an outgrowth of the entrepreneurial sector in any country in the developing or developed world. Its workers are rational people who largely select informal sector employment due to many non-wage 1 The traditional "dualistic" view sees the informal sector as the residual of a highly distorted formal sector labor market, where unions and/or govemment regulations keep wages above their equilibrium level and thereby ration workers into the informal sector where they are left unprotected by labor legislation. Participation in the unregulated sector of the economy is thus seen as involuntary. The outcome is a segmented labor market. The aintegrationist" view perceives the flow of workers between the informal and the formal sector as voluntary. -2 - and - in some instances- wage benefits. These findings suggest that policies that attempt to create a single market by eliminating labor market rigidities are perhaps misguided. Instead, the sector should be brought into the formal tax and benefit system, as in developing countries, by improving the public goods associated with the formal sector. 1.4. In Chapter 4, Jorge Saba Arbache (University of Brasilia) analy?es the effect of unions on the labor markef in Brazil. The study attempts to answer the following questions: How do labor laws and union legislation affect the collective bargaining process? Do unions influence wage formation and income distribution? Do unions increase labor market rigidity? He finds evidence that institutions are essential in explaining the behavior of unions, their effects on income distribution, macroeconomic stability and the current state of collective bargaining in Brazil. The paper concludes with recommendations for enhancing the role of unions in the collective bargaining process. 1.5. Chapter 5 by Amit Dar (World Bank) and Francisco Galrao Cameiro (Catholic University of Brasilia) focuses on the economic impact of the freedom of awociation of workers and employers within an interational context. The author considers two questions similar to those in the previous chapter: (a) what are the microeconomic effects of unions and employers' organizations (unionism); and (b) how do different institutional approaches to collective bargaining affect macroeconomic performance? This study answers the question by reviewing intemational evidence primarily from OECD countries. The review suggests that the economic consequences of unionism differ significantly depending upon the interaction between unionism and the economic, political and institutional environment in which collective bargaining takes place. 1.6. Chapter 6 by Jos6 Marcio Camargo (Pontiffcia Universidade Cat6lica do Rio de Janeiro) examines the role of the laborjustice in Brazil. This institution has been the subject of intense debate among Brazilian labor market economists. Opinions differ on its impact on the performance of the labor market. In general, analysis suggests that it plays an important role in the protection of less organized workers, but few analysts have focused on the effect of the institution on the functioning of the labor market, which is the aim of this paper. The author argues that labor justice currently gives incentives to workers and employers that distort their behavior towards a low quality work-relationship, and therefore reduces productivity and real wage gains, specially for the unskilled workers. These points are illustrated by data from a sample of labor justice cases in Minas Gerais. The study concludes by recommending that the labor courts concentrate on its core functions within the judiciary branch. 1.7. In Chapter 7, Francisco Galrao Carneiro (Catholic University of Brasilia) reviews the principal findings on the impact of the minimum wage on the ?razifian labor market. The effects of the minimum wage on employment, wage level, poverty and fiscal deficit are analyzed. The chapter opens with a brief history of the minimum wage policy in Brazil and then characterizes the typical minimum wage earner in Brazil. Special attention is paid to the various methodologies and databases used to analyze the impact of the minimum wage on employment, wage levels, and poverty. The overview is intended to aid discussions about the future of minimum wage policy in Brazil. The author finds a certain degree of consensus among the reviewed studies. The minimum wage has (a) a negative effect employment in the formal sector, (b) only a marginal and positive increase on wages above the minimum wage, (c) an insignificant effect on poverty and (d) a significant impact on the deficit in the social security system. -3- 1.8. Chapter 8 by Wendy Cunningham (World Bank) examines the effectiveness of minimum wage regulation as a poverty alleviation tool in developed and developing countries. The discussion puts into perspective the analysis in the previous chapter by examining how minimum wage policies have affected labor market outcomes, poverty and government's fiscal accounts intemationally. Although the evidence is scarce, particularly for developing countries, the minimum wage is not an effective poverty alleviation tool. Instead it serves to reshuffle households who are clustered near the poverty line. However, it is a useful tool for decreasing inequality, as it does pull up the wages of all workers, particularly in developing countries. Furthermore, its reach is wider than previously thought. As evidence from Latin America shows, the minimum wage is more binding in the informal sector than in the formal sector. Moreover the minimum wage is commonly a unit of measure for social programs. To improve the effectiveness of the minimum wage in the fight against poverty, policy makers need to (a) better understand the poverty impacts of the minimum wage in their country, since poverty evidence only exists for industrialized countries, (b) link the minimum wage to long-run poverty rather than short-run political gains, (c) identify other social programs necessary to compensate for the welfare reducing impacts of the minimum wage, and (d) de-link the minimum wage from social programs such that fiscal concems do not interfere with the minimum wage adjustment criteria. 1.9. Chapter 9 by Dorte Domeland (Universidad Pompeu Fabra, Spain) and Norbert Fiess (World Bank) discusses the Brazilian unemployment insurance system and the Fundo de Garantia do Tempo de Servigo (FGTS). After a description of the two systems, the study analyzes the access and impact of the system using data from the Monthly Employment Survey (PME). They find that the Brazilian unemployment insurance system is characterized by a low replacement ratio and a short benefit duration. The authors examine the determinants of the probability of becoming unemployed and of the unemployment duration. The estimations show, on the one hand, that informal sector employees who have neither access to unemployment insurance nor FGTS are most likely to become unemployed in Brazil. On the other hand, unemployment duration among formal sector workers is higher for those who received FGTS before becoming unemployed. The study argues that the system suffers from two shortcomings: (a) more than half of the population is excluded from participating in the programs, since it is restricted to formal sector workers only, (b) the two programs overlap, since the unemployment insurance is conditional on having access to the FGTS. 1.10. In Chapter 10, Ricardo Paes de Barros, Mauricio Blanco Cossfo and Jorge Luiz Teles (IPEA) describe and investigate the efficiency of Brazilian active labor market policies for poverty reduction. The paper is motivated by a recent rise in permanent unemployment, which increases the importance of labor market inventions for poverty alleviation. Three types of labor market policies are investigated; Compensatory, Distributive and Structural. Using the "Perquisa sobre Padroes de Vida (PPVJ' and the national household survey (PNAD), the authors show that (a) the income compensatory programs are badly targeted towards middle-income workers, (b) the minimum wage as a distributive policy only marginally increases the wage of low-income families, and (c) structural policies, like training and job-matching, are far from reaching their potential impact on poverty. -4 - 1.11. Chapter 11 by Amit Dar (World Bank) describes active labor market policies in an international perspective. The study summarizes the impact of active labor market policies on labor market outcomes drawing on more than 70 reports.2 The purpose of this paper is to review intemational experience with active labor market policies and to discuss their applicability in more developing countries, such as Brazil. Most of the evaluations come from industrialized countries, though some evidence presented also comes from developing economies. The impacts are found to be mixed, with many programs assessed to have little or no impact on the employability or eamings of participants. The chapter completes the review by assessing the key considerations involved in balancing active and passive labor market policies. 2 Active labor market policies (ALMP) consist of a wide range of activities, intended to increase the quality of labor supply and to increase labor demand; or to improve the matching of workers and jobs (e.g., through job search assistance). ALMPs are usually regarded as a valuable altemative to 'passive" support (e.g. unemployment insurance, safety net). - 5- 2. LABOR MARKETS AND ECONOMIC ADJUSTMENT: THE BRAZILIAN CASE Prepared by Norbert Fiess, Marco Fugazza and William Maloney 2.1. Over the last two decades, most Latin American countries went through the experience of fiscal adjustment, stabilization and trade liberalization. Recent, growing attention has been devoted to assessing the role of the labor markets in the transmission process of such economic policies to the population's standards of living, in particular the poor3. One of the main conclusions of these studies is that market imperfections prevent an efficient allocation of resources, and hence are likely to mitigate the success of such policies and a fair distribution of their potential benefits. 2.2. The predominant conceptual approach adopted in the existing labor market literature is based on the assumption that labor market institutions are always a source of distortion. It corresponds to the traditional udualistic" view of labor markets in less developed countries4. This view sees the informal sector as the residual of a highly distorted formal sector labor market, where unions and/or government regulation keep wages above their equilibrium level, rationing workers into the informal sector where they are then left completely unprotected by labor legislation. Participation in the non-regulated section of the economy is thus seen as non-voluntary, with workers regarding the regulated section always as the better alternative. The perceived reaction of a dualistic labor market to negative shocks is then also predetermined. As wages in the formal sector are downwardly rigid, wages in the informal sector will fall in response to the increase in informal labor supply. The outcome is a segmented labor market, where relative eamings and relative employment shares of the formal and informal sector move in opposite directions5. 2.3. However, there is empirical evidence that the labor market does not exclusively behave in a segmented fashion6. In particular, the observed increase in infornality over the last decade in Latin America reflected prolonged episodes where-despite potentially distorting labor market regulations7-participation rates in the formal and informal sector move up and down together, that is, formal and informal labor markets behave in an uintegrated" rather than a segmented or dualistic fashion. These studies attempt to explain the integrated behavior of labor markets and present the altemative view that as a first approximation, the informal sector should be treated as an unregulated/unprotected entrepreneurial sector that may be desirable to many workers. In other 3 See, e.g, Horton and Mazumdar (1994). 4 The Harris-Todaro (1970) model is perhaps the seminal statement of this view. 5 Agenor (1996) presents a more sophisticated approach, although it stands within the dualistic paradigm, based on a fully articulated theory of the channels through which wages in the formal and informal sectors interact. 6 See, for instance, Vijverberg (1986), Yamada (1996), Maloney (1998a), Aroca and Maloney (1999). 7 See Davila Capalleja (1994) and Marquez (1997). -6 - words, the choice to participate in the informal sector is a rational one. This rationality could be linked to elements like the misalignment of implicit and explicit labor taxes with perceived benefits or the desire of workers to retain a degree of independence in their work8. 2.4. In this study, following Fiess, Fugazza and Maloney (2001), we argue that for any given developing country labor market, the dualistic and integrated views are not necessarily incompatible with each other and could be seen as special cases of the same model. The informal sector may well be a prosperous non-traded goods sector, but one which also contains involuntarily informal workers.9 We believe that this is a more realistic characterization of labor markets in low and middle income countries. We show that in this more realistic setting, wage differentials across sectors are not exclusively due to rigidities and, hence, the removal of such rigidities would not lead to wage equalization even with perfect mobility of factors of production. The degree of labor market segmentation is not determined by the existence of labor market rigidities per se. From a policy point of view, this is both somewhat novel and important. 2.5. But this is not all. This approach also illustrates that the extent to which labor market rigidities are binding, viz., the degree of segmentation of labor markets, is not determined exclusively by labor legislation but varies depending on the macroeconomic stage that the economy is passing through. That is, we propose that the macroeconomic environment itself can affect the degree of segmentation in the labor market and consequently the impact of policy interventions. Viewed another way, the expected outcome of stabilization and adjustment policies varies whether the labor market behaves like an integrated or a segmented market. The approach thus postulates an interdependence between labor market composition and macroeconomic environment. This study attempts to provide a better understanding of these linkages and, eventually, to equip policy makers with a more sophisticated instrument for policy analysis. 2.6. The policy implications emerging from our study are quite compelling: only if there is clear evidence of segmentation, a large informal sector can be a signal of effective labor market distortions and policy could then consist of reducing labor market rigidities. However, if there is evidence of integration or only weak evidence of segmentation then the informal sector may prove to be an important source of efficient employment and reducing labor market rigidities may have weak or even counter-productive effects on labor composition. 2.7. We use this approach to briefly contrast labor market functioning in three Latin American countries-Mexico, Colombia and Brazil-in order to draw policy lessons in specific settings. The result of the exercise (the details of which are in Fiess, Fugazza and Maloney (2001)) is that Mexico appears to have had the most "flexible" or integrated labor markets until the mid-1 990s, Colombia exhibits traits that can be better classified as segmented or dualistic, and Brazil appears to oscillate between these two forms during the 1990s. The case of Brazil is examined further in some detail, and policy implications are drawn. The main conclusion is that labor policy priorities change 8 See Maloney (1 998b) for a comprehensive presentation of the arguments. 9 This characterization is consistent with a description proposed by Fields (1990) which distinguishes an 'easy-entry' informal sector (the involuntary tier) and an 'upper-tier" informal sector (the voluntary tier). -7 - depending upon macroeconomic conditions. This paper makes an attempt to establish a mapping between macroeconomic developments and relevant labor policies. 2.8. The remainder of the paper is organized as follows. Section 2 provides the conceptual and empirical background of the study, and also presents additional evidence for other Latin American countries. Section 3 presents detailed empirical findings for the Brazilian economy and relates the Brazilian experience to conceptual insights. General and Brazil-specific policy implications are developed in section 4. Finally, section 5 summarizes the main results of the study and concludes. 1- IM.INS I M MW 2.9. The main scope of this section is to present some empirical findings for Colombia, Mexico and Brazil that sustain the view that despite the presence of potentially distorting labor institutions, labor markets can behave like integrated markets and not only like segmented markets. The empirical analysis is based on a simple model that combines Lucas's (1978) idea of workers characterized by differing levels of entrepreneurial ability, with a small economy trade model of the style most recently elaborated by Obstfeld and Rogoff (1996) and capital adjustment costs, which relate to the Tobin's q model to capture the existence of imperfections in capital markets. This simple model also captures the argument that the macroeconomic environment plays a central role in defining the degree of segmentation of the labor market. A New Conceptual Framework 2.10. There are different definitions of informality in the developing economics literaturel 1, which rely on a variety of criteria such as establishment size, type of employment, technological or capital level, income level, legal status, type of goods produced, and type of transactions that are carried out.12 The last criterion refers to the fact that activities being encompassed in the informal sector are unregulated. As discussed by Charmes (1990), some of these criteria overlap in practice.13 The formal sector is then characterized by large firms producing essentially traded goods and complying with fiscal and labor legislations. Workers classified as belonging to the formal sector are those who benefit from labor unions influence in the wage formation process and labor protection. 10 The empirical results and the conceptual framework presented in that section are drawn from Fiess et al. (2001). 11 See Peattie (1987), Turnham and Er6cal (1990), James (1992), Tokman (1992), Portes and Schauffler (1993) for comprehensive overviews. 12 We usually distinguish three major sectors in the labor market in developing countries: the rural sector, the informal sector and the formal sector. Our study concentrates on urban areas where only the last two sectors are present. 13 In this study, the informal sector is characterized by self-employed individuals or small privately owned enterprises (wfth no more than six workers) producing essentially services and non-traded goods and, not (fully) complying with government imposed taxes and regulations (mainly labor related). The reason why informal wage workers are excluded relies on the fact that the informal wage sector takes on the form of a transitory sector essentially populated by young people closely linked to self-employment activities (see Maloney (1999) and Packard (2001)). This is reinforced by our empirical investigation for Brazil which shows posHtive co-movements in relative eamings and labor force size of informal wage workers and self-employed. Tests results are reported in Appendix 2. - 8 - 2.11. Just as there are more than one criteria for an empirical delineation of the formal from the informal sector, there are different interpretations of why informality arises in the first place. These can be classified into two categories. The traditional "dualistic" view sees the informal sector as the residual of a highly distorted labor market in the formal sector. Unions and/or govemment regulation push wages above their equilibrium level, rationing workers into the informal sector where they are not protected by labor legislation. Participation in the non-regulated section of the economy is non- voluntary, the regulated section always offering a better alternative. Thus, distorting labor market regulations are the cause for a formal-informal segmentation of the labor market. As part of this segmentation, relative wages and relative employment shares of the formal and informal sector move in opposite directions. Implying, that, e.g., wages of informal sector workers relative to those in formal employment fall at the same time as the employment share of the informal sector increases. 2.12. A more recent line of thought presents the view that as a first approximation, the informal sector could be treated as an unregulated/unprotected entrepreneurial sector that offers work and working conditions that are desirable to many workers. In other words, the choice to participate in the informal sector is a rational and unforced one. This view also implies that there is voluntary movement of workers both to and from the formal sector. This rationality could be related to elements like the desire of independence and/or the misalignment of implicit and explicit labor taxes with perceived benefits. This "integrated labor markets" view stands in sharp contrast with the dualistic view. 2.13. In this study we argue that the two views are not necessarily incompatible with each other and should instead be seen as special cases of the same model. The informal sector may very well be a dynamic and entrepreneurial sector, but one which also contains unfortunate workers holding poor jobs. The relative shares of voluntary and involuntary workers will reflect the degree of segmentation in the labor market which can vary depending on the shocks to the economy. And depending upon the general health of the economy, one of these phenomena may dominate the other. The main technical characteristics of the model are presented in Box 2.1 and Box 2.2. 2.14. Our conceptual approach offers a rich framework for policy analysis as it can nest both the dualistic and integrated views characteristics in a non-trivial way. Wages differential across sectors are not due to rigidities exclusively and the removal of such rigidities would not lead to wages equalization even in a context of perfect mobility of factors of production. These features reflect a more realistic view of the economy and enrich the discussion of possible policy implications. 2.15. To reiterate, the usefulness of our conceptual framework is twofold. First, it offers insights that can be used to qualify empirically a labor market: o the market is segmented if eamings in the informal sector and the size of the latter, relative to their respective counter-part in the formal sector, move in opposite directions: relative eamings fall while participation in the informal sector increases o the pattem of relative eamings and participation inherent to segmentation can be observed only if there are effective downward wage rigidities in the formal sector (the labor legislation is binding) o downward wage rigidities in the informal sector are of first order relevance in the event of a negative shock (technology and/or preferences) to the formal sector and of second order - 9- relevance in the event of a negative shock (technology and/or preferences) to the informal sector any evidence of positive co-movements in relative earnings and participation allows to conclude in favor of an integrated labor market 2.16. Second, accounting for the nature of the labor market (segmented versus integrated), and observing movements of macroeconomic indicators like the real exchange rate, enables the identification of the underlying macroeconomic environment. Put another way, our framework allows us to determine the likely impact of policy shocks, i.e., changes in the macroeconomic environment, oil employment and eamings according to the nature of the labor market. We delve deeper into the analysis in the next section -10- Bon 2.1 The Conceptual Framework The Core Model - Open economy model based on Obstfeld and Rogoff (1996) and Lucas (1978) - Agents in the economy: o households: consume goods available in the economy, hold foreign traded bonds o producers: firms of indeterminate size and self-employed - Sectors of production: o Formal sector: production of traded goods, wage work, compliance with regulations o Informal sector: production of non-traded goods, self-employment, non compliance regulations, non perfect access to physical capital - Production factors characteristics: o Capital: - mobile both intemationally and across sectors - not instantaneously: capital market is not perfect for informal producers (installation costs o Labor: - constant total labor force - workers are homogenous when salaried in the formal sector but differ in terms of their entrepreneurial capability when self-employed - mobile across sectors - not instantaneously because of capital market imperfections - Properties of the framework: - interactions between macroeconomic variables and the labor market - steady state analysis and transitional dynamics due to non instantaneous adjustment of both capital and labor The Govemment Govemment characteristics can be easily introduced in the core model and can include some or all of the following elements. The listing is obviously not exhaustive. - Expenditures: - Consumption of traded and/or non-traded goods - Production of services that can be used as intermediate inputs - Social security benefits and assistance - Debt interests payments - Revenues: - Taxes: payroll, corporation and individuals income taxes, indirec4 taxes, tariffs ( according to the above specification, none of these taxes are expected to be voluntary paid by individuals involved in informal activities) - Other sources of revenues can consist of domestic and foreign credit and transfers from the central bank including seignorage and/or interest receipts on net foreign assets The Labor Market - Segmentation implies that the formal section of the labor market is characterized by downward rigidities in wages which can be related to the presence of unions and/or indexation practices and/or legal minimum wages - It is further possible to look at the impact of upward/downward pressures on formal wages due to the introduction of specific labor market policies - 11 - Box 2.2 Fundamental Equilibrium Relations in the Core Model We present steady state responses of the variables of:interest to the various shocks that can possibly occur in our framework. The following equations are used to formulate Table 1. s ~~~~~~PN The relative price of non-traded goods in terms of traded goods is denoted by p that is, p =--, where PN denotes the price of non traded goods and e the price of traded goods (the nominal exchange rate). In other words p stands for the inverse of the real exchange rate. Then, a real appreciation would correspond to a rise in p. A Tsubscript refers to the traded goods sector and a N subscript to the non-traded goods sector. A L subscript refers to formal labor and a se subscript to self-employment., q refers to the share in the respective sector total earnings while So refers to the share in total revenues. Then, (PLT denotes the share of formal workers labor eamings in total revenue. Nse is the self-employed labor force, 0 is the intra-temporal elasticity of substitution and ?is the share of individuals' revenue spent on traded goods. A stands for a technology shock. Average eamings are denoted by w. E denotes the nominal exchange rate, and r the real interest rate. Finally, variables with hats refer to rates of change- Ax=-) x P 7LN=(AT +,O)-A +(1-.- (1) If only technological shocks are considered, we obtain the standard Balassa -Samuelson result (Balassa (1964). N,~~~~~~~~ ~~AT +e(..J'rps NX, =0S21 pANO(-70(-0 l)-kT ) 1 [-> P, -FP.- a, N O N (-S 9) -1 )1 (2) WT 1 [AT ( 717LT)r] (3) q1LT #7N + + (1 1j + T N (4) 17LT 7N where Q. = [( V-.,e)T +. LT1 ' = [ se and N*" represents the initial steady state level self-employ state level self-employment. - 12 - Some EmpHricaG EvNdence 2.17. We use a multivariate Johansen approach (Johansen 1988) to explore cointegration relationships in relative earnings, relative sector size and the real exchange rate with the aim of establishing the degree of dualism-as defined above-in the labor market in Mexico, Brazil and Colombia. This exercise can give us a preliminary answer to the question: whether and to what degree are the labor markets in these three countries 'flexible" in the broadest terms, viz., whether key eamings and employment indicators respond to macroeconomic shocks in a manner that is consistent with efficient functioning of markets. The empirical analysis is also used to extract some conclusions about the source of real appreciations in the three countries. These results are explored in more detail in the next section. 2.18. Figure 2.1 displays the time series of the three variables under consideration for the three countries respectively. We limit ourselves here to a qualitative presentation of the results and refer the reader to Fiess, Fugazza and Maloney (2001) for technical details. The period under consideration runs from 1985:01 to 1999:02 for Colombia, from 1987:Q1 to 1999:Q1 for Mexico and from 1992:01 to 2000:04 for Brazil. By looking first at the whole sample, we find that for Mexico the test of integrated markets can be strongly rejected while the test of segmentation cannot be rejected. For Brazil, the hypothesis of integration cannot be rejected and the hypothesis of segmentation is rejected at the 1% level. For Colombia, the hypothesis of segmentation cannot be rejected while the hypothesis of integration can be. 2.19. In all three cases, however, tests of the stability of the cointegration space suggest the possibility of different relationships in different sub-periods. This finding is consistent with the previous argument that rigidities in the formal sector may bind in some periods and not in others depending on the macroeconomic environment. o For Menico, the segmentation hypothesis is only accepted from 1995 onwards and the hypothesis of integration cannot be rejected prior to 1995. This suggests that the Mexican labor market behaves pretty much like a flexible market in the recovery period leading up to 1995. o For Colombia, statistical tests support the hypothesis of a segmented labor market post 1995 onwards and never support the view of integration for any time sub-sample. These findings are in accordance with the view that regulations in the Colombian labor market are binding strongly and that the dualistic view reflects relatively well its functioning. o In Brazil, despite full sample evidence of integration, there is evidence of segmentation during the pre-Real plan period and in the post Asian crisis period (see Figure 2). These findings indicate that the labor market nature in Brazil is likely to have changed over the decade. - 13- Figure 2.1 Real Effective Exchange Rate, Relative Wage and Share of Informal Workers Mexico 1.1 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~2.1 1 0 9 | wJ ~> 1 S E2 cm ~~~~~~~~~~~~~~~~~1.9wL 00.9 0C,) 0 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~1.8 -08 0 0 ~~~~~~~~~~~~~~~~~~~~~~~~~~~1.7 U. 07- 1.6 0.6 1.5 N 0) 0 0 _ N~ CX et Lt CD N CO 0 C D C O 6 ) 0 0 4 C O L O c o F l c o 0 ) 0) 0) OD a) 0 ) 0) 0) 0) 6) 0 0) 0) 0) ------- Real exchange rate wage F/SE - ForrrfaVSE Colom bia 4 1.3- 1.2 3.5 0.9 2.5 0.8 : 0 ~07 0.6 *. _ --_. . -_ 0.5- , .*-- W | | | | | | | I, | | | I, | . | * g " " I,, | | | | I, | g | I, g | | ., . - 1 0.4 .. .... . ..... LO D S m 0 o _n co I,- co CD N C co co ) co co CD CD 0) 6) 6) 6) a) 6) a) 0) 0) 0) 0) 0) a) a 6) 6) 0) 0) 6) 0) 0) | ------ Real exchange rate w age F/SE- FormnaSE Brazill 2.3 - 1.5 2.1 1.4 , 1.9 1.3 0 9w \;"-\ ' 1.2 ~ 0 16 113 - . 0.7 -....... 0.8 N U) CDO Ns co 6 0 0 C' CO 0 U) co N0 06 co co 0 0 0 OD co c 0 0) 0) a) 0) 0) 0) 0) 6) 6) 0) 6) 6) 6) 6) 6)~~0 6) ) 6) 6 F., 6) F, 6) 6 F --- Real exchange rate wage F/SE - ForrnaVSE Source: Fiess, Fugazza and Maloney (2001) - 14 - Figure 2.2 Some Evidence for Segmentation in Brzil A recursive test of segmentation In Bramil 1.6 - 1 = 10% level of signffiance 1 4 1 2 1 0 0.8 0.6 04- I...I 1990 1991 1992 1993 1994 1955 1996 1997 1998 1999 2000 Note: Figure 2.2 shows a recursive test of the hypothesis of segmentation in the Brazilian labor market. The hypothesis of segmentation cannot be rejected for test statistics below 1 and the shaded areas are thus indicative for segmentation. Starting with an original sample from 1982:01 to 1989:04, the estimation period is increase by one observation at a time and the corresponding test statistic is calculated. It is important to realize that the periods where the hypothesis is rejected do not indicate exact break points. Much rather, they reveal what a researcher would find Kf she had only access to data stretching from 1982:01 to the specific point in time where the test statistic is calculated. 2.20. Brazil represents an interesting case as periods of evidence in favor of segmentation and periods of evidence in favor of integration altemate. This again strongly supports the argument that the degree of segmentation in labor markets, even though strictly linked to the severity of regulations, is also a function of the specific macroeconomic conditions prevailing in an economy at a given moment in time. 2.21. The Brazilian experience of the 1990's gives us the opportunity to assess practically the foregoing argument. Indeed, the economy went through three macroeconomic policy interventions of major importance: trade liberalization starting at the beginning of the decade, disinflation with the Real plan launched in 1994 which was followed by a conspicuous fiscal adjustment in the late 1990's. - 15 - 2.22. This section is devoted to a closer examination of the Brazilian case. We focus mainly on the 1 990s as we do not find any significant variation in the labor market behavior in the 1 980s'4. We first present evidence of changes in the labor market behavior across the period based on Okun coefficients for total employment and average labor earnings. Then we briefly discuss the major macroeconomic events of the decade and provide possible and plausible explanations, based on the insights of our conceptual framework from the overall behavior of the economy. Further Empirical Evidence 2.23. While the cointegration approach in the last section allowed an identification of segmentation and integration based on negative and positive co-movement of relative wages and sector sizes, in this section we make use of a different technique to identified segmented and integrated labor market episodes based on the sensitivity of employment and real wages with respect to output based on Okun coefficients. As we would expect, quantities adjust more than prices in a segmented market and the reverse holds in an integrated market. 2.24. To investigate the sensitivity of real wages and employment with respect to output we follow Gonzalez Anaya (1999) and estimate rolling Okun coefficients for total employment and real wages for 1982 to 2001.15 The advantage of a rolling regression approach is that as the estimation procedure moves through the sample, new observations are picked up and changes in the Okun can be attributed to changed circumstances in the latest incorporated period. The resulting period-specific wage and employment Okun coefficient are graphed in Figure 3. 2.25. As already mentioned, we expect quantities to adjust more to macroeconomic shocks than prices in a segmented market. This appears to be the case for the years 1991 and 1992. Even though wage reaction is not monotonic, it roughly remains less pronounced than employment reaction during these two years. The Collor Plan failed to rein in inflation and the economy was on a recessive path with increasing fears of the effects of a rapid and incisive opening of the economy. The return of hyperinflation without perfect indexation of workers' wages allowed firms to adjust through prices rather than quantities in 1993 and in the first half of 1994. In that period the labor market behaved in an integrated manner, and the economy recovered from a period of negative growth. 14 As suggested in the previous section, the labor market behaved essentially as an integrated market over the full decade. The usual explanation relies on the capability of firms to adjust real wages through imperfect indexation. 15 As Okun's law was developed for total employment and overall GDP, we do not distinguish between the formal and informal sector. -16- Figure 2.3 Wage and Employment Okun Coefficients: 198901 to 1999Q4 0~~ A - 0.6 0 4- 0 3- O.2 - _ - - - - - -- OKUN-N 0 0- 1990 1991 1992 1993 1994 1995 1996 1997 1999 1999 Note: Shaded areas indicate segmentation, i.e. situations where employment Okun coefficients react stronger than wage Okun coefficients. 2.26. In July 1994, as the Real Plan was implemented, inflation fell sharply and indexation practices were stopped. As a consequence firms were less able to erode real wages and adjustment occurred increasingly through quantities. This corresponds to the second shaded area in Figure 3. In the last quarter of 1994 and the first quarter of 1995, the labor market behaved in a segmented manner until the economy adapted to the low-inflation environment, reverting then to an integrated market. With the Asian crisis, once again, adjustment through quantities picked up while wage reactions were more muted. This corresponds to the third shaded area covering the last quarter of 1997 and beginning of 1998. The labor market again behaved more like an integrated market in the rest of 1998. Generally speaking, therefore, the Real Plan period was characterized by a roughly integrated labor market, with the exception of the period that follows the Asian crisis. However, since the devaluation in January 1999, the labor market has been showing an increased tendency to adjust more through quantities than prices, a phenomenon that corresponds to increased segmentation. -17 - Box 2.3 Empirical Estimation of Okun Coefficients Qkun, coefficient ,reflect oth'e s'ensitivityof4 re6al Wages, and emploe6 t to changes in p We follow Gon'alez Anaya (1999), a4nd estimate M4olingr^'Okun coefficierits for 1982'to 2001. Rolling coefficients irnplies that these magnitudes ,are eestirnated for consecutive Isb-p6ri6d&. The a.dvantage' of a rolling regression approach is that as'the estimation procedure moves through'thesa mple, new observations are picked up and a change in the Okun coefficient can be attributed to an in'iovation in'the latest'period; We make use of"the OkUn{(,,962) initial me,thod which.relates first differences;o une, rntand output. ,Although Okun's initial appioach has' givenwayto4more sophisticated methods, itstill represents a simple ,,nd efficient tool. - We use quarterly data for real wage, total employment,and seasonallye adjusted GDP. The wage data is from PME and the output data from IBGE. .' Sinice we deal with' quarterly data, weJintroduce a lag'structure to reflect the fac0 <0 <0 AN <<0 <0 >0 - _ <0 >0 >0 <0 >'0 >0 18 Indeed, a joint test of long-run exclusion and weak-exogeneity of the real interest rate could not be rejected (LR-test: X2(2)=2.91, p-value=0.23). -21 - Box 2.4 The Impact of Macroeconomic Policies in the Core Framework V _ ' ' . ., ' ' _ , .,, ,- -' 'V Trade Liberalization' Production. If the traded goods sector is competing, directly. with imports,then liberalizing trade ,corresponds to, a negative shock to the formal-,sector in 'the short (medium) run. However, -the sign bf the' shock is expected to be reverted in the longer'run,beca6sse, of imported technological improvements. lhputs costs are also expected to fall a ,fortiori if trade liberalization is accompanied by financial liberalization and/or. deepening. Then the cost of borrowing '.physical capital is expected to fall and.borrowing constraints to be eased, possibly also in the informal sector. ,C,onsumption. Consumption-is, expe6ted to boom because of the fall in the price of traded goods. The inherent positive income effect also leads to a rise in the.le'vel'of.consurnption of non-traded goods., IGovemment. If trade liberalization irmplies a conspicuous fall in - government revenues because of lower tariffs proceedings, then other.tax rates may,have to be increased and/or expenditures reduced. Any rise in taxation could generate a bias towards informality. In general, if the government adjusts immediately for revenue losses, then the fiegative impact on formal production is. likely to be more, pronounced. If the govemment adjusts its expenditure level -by reducing its consumption of non-traded goods then this would have the effect of a negative shock to the informal sector. The overall income effect is discussed in more detail in the fiscal adjustment section. Exchange-Rate-Based Disinflation' Production. Exchange-rate-based disinflation has the effect of a negative shock to the traded-goods sector as it corresponds to a fall in the price of the traded good. . , Consumption. A fall in the inflation` rate leads to two contrasting effects. First, -lower prices imply a higher consumption . based real interest rate,.which tends to.Jower, the,current level of consumpti6n. Second, lower price levels reduce the opportunity cost of holding money (or equivalently it generates a positive real balances wealth effect), which tends to increase the current level of consurmption. Then, the predominant effect has to be determined empirically. Govemment. Lower inflation implies'lower revenues from seignor,age., Then, govemment may be willing to reduce expenses and/or increase tax rates, to avoid fiscal crisis like in the case oftrade. liberalization. Fiscal Adjustment ' . ' - - 'Both, a trade liberalization and an exchange-rate-based disinflation, program, are expected, to be coupled with fiscal 'adjustment to offset losses in gove'rnment revenues. Fiscal adjustment could consist either in an' increase in taxation or 'a reduction in expenditures or both. ' ' - - , , - Higher taxation, besides having a negative impact on formal production, m'akes the informal sector more attractive as the latter is characterized by tax evasion. This is specially true if formal production (firms and/or workers eamings) is the target of higher tax rates. This would also be true, afthough to a lower extent, if indirect taxes are raised. As to a reduction in the level of expenditures, its impact depends essentially on the consequent changes in expenditure composition. If reduction in consumption of non-traded goods is the main target then the informal,sector is directly affected and this could increase the'degree' of segmientation in the labor market.- ' The purpose of fiscal'adjustment can also be the reduction of govemment.debt per se. In that,case interest rates could be expected to fall consequently. , v - 2.38. The Collor period is characterized by evidence of a segmented labor market. We also have that the price of non-traded relative to traded goods is falling dramatically. The real interest rate remains at very high levels. All these elements seem to indicate that through this period the formal sector was undergoing a negative shock or at least was behaving as if. The fact that the economy was opening progressively and in a credible manner makes this explanation quite plausible. The end of the Collor brought back some positive growth prospects and the labor market appeared to be less - 22 - segmented. At the same time the real exchange rate path was reverted. As real interests rates remained very high, essentially because of the return of hyper-inflation, a possible explanation of the phenomenon could be a shift in preferences towards the non-traded goods sector. 2.39. The Real Plan has been a path-breaking economic event. After a relatively short period of adjustment, characterized by segmentation in the labor market due to a substantial rise in labor costs brought about by a fall in inflation which led to depreciation, we observe a sharp increase in the relative price of non-traded goods essentially driven by a surge in consumption. The latter was due to the fall in the cost of holding money, the relaxation of liquidity constraints and increased access to credit brought about by financial liberalization. There is no clear evidence of any positive shock to the traded goods sector. On the contrary, productivity improvements appear to have been the consequence of lower worker participation in that sector. The expansion of the informal sector together perhaps with a possible positive shock could have explained the fall, even though not a dramatic one, of the relative price of non-traded goods in the period leading up to the Asian crisis, the Russian crisis and the subsequent recessionary period that includes the devaluation of January 1999. Again labor regulations appear to have been binding beginning in late 1997. Even though the economy as a whole has been hit by these external shocks, the formal sector which was more exposed to foreign investments may have suffered the most. 2.40. The year of 1998 is characterized by negative growth rates, a falling relative price of non- traded goods in terms of traded goods, falling self-employed eamings relative to formal wages and an increasing participation in the informal sector. These elements corroborate evidence of a predominant negative shock to the formal sector which has led to a higher degree of segmentation in the labor market. The first signs of overall economic recovery are perceived in late 1998 and are strengthened by the devaluation that occurred at the beginning of 1999. There is some evidence of an industry-led recovery which explains the reverted tendency in the real exchange rate. We may also interpret the real appreciation as the outcome of a negative shock to the non-traded goods sector as there is still some evidence of a segmented labor market to date. The contribution to gross domestic product of the industrial sector has risen from 34.6 % in 1998 to 35.5 % in 1999 while that of the services sector has fallen from 62.3 % in 1998 to 61.1 in 1999. These tendencies were maintained in 2000. GDP growth has jumped to more than 4 % and the relative size of the formal sector has increased in terms of the size of the informal sector. The degree of segmentation has fallen since 1999 and there is some evidence (Figure 2.4) of the return of job creating growth after a mitigated decade. 2.41. The decade of the 1990s has been characterized by a growing informal sector. This trend has not exclusively been the result of a increasing degree of segmentation in the labor market. Rather it has been the result of the conjunction of a series of economic events that hit negatively and predominantly the formal sector and of always threateningly binding labor legislation. In addition, a general shift of preferences in favor of non-traded goods also has contributed to this pattem in the composition of the economic activity. 2.42. Nevertheless, since the beginning of 2000 there are signs of a long-term oriented recovery in the formal sector. This may indicate an increase in productive efficiency and competitiveness in the formal sector which have been expected since the beginning of the 1990's. -23 - 2.43. Empirical evidence presented in previous sections supports the view that neither a purely "dualistic" nor a purely "non-dualistic" approach can be applied to labor markets in less developed economies without referring to any further qualification. We argue that an approach able to encompass both views is rather desirable, specially in terms of policy implications. We argued previously that in the presence of a labor market that behaves like an integrated labor market, then, reforming labor regulations should not be the core of policy orientation as there is no evidence that the latter are either strongly binding and/or distorting. Rather, efficiency improvements should be eased in both formal and informal sectors. On the contrary, if there is evidence of segmentation in the labor market, then reforming the labor regulation is likely to generate efficiency and eventually equity gains. In the case of a labor market that behaves both like a segmented and an integrated market, then policy orientation should be defined according to both set of characteristics. A rise in the degree of segmentation at some points in time underlines the necessity to reform labor market legislation in order to dampen the effects of activity downturns. Nevertheless, evidence of integration in other periods is the signal of a potentially efficient labor market and, policies aiming at reducing operational costs and at improving economic growth potential in all sectors of the economy are likely to be efficiency and equity enhancing. Again, labor-related policy orientations could not be disentangled from other macroeconomic objectives as the two are expected to interact. Less Segmentation versus More Integration 2.44. A higher degree of segmentation is the outcome of a negative shock to the economy in the presence of downward wage rigidities in the formal sector. Sources of downward rigidities include wage inertia due to indexation mechanisms (backward-looking contracts)'9, unions and government imposed wage floors. Then, we would expect the removal of these downward rigidities to generate strong efficiency gains as worker flows would be driven essentially by earnings differentials considerations20. 2.45. It has been argued that the implementation of macroeconomic reforms can take advantage of a dramatic economic environment21. However, as far as labor institution reforms are concerned, the validity of this argument can be questioned. 2.46. Recently, various authors presented the view that labor institution reforms have proved difficult to implement essentially because of political opposition22. Lora and Pages (1996) argue that this is effectively the case in most of Latin American countries. Political opposition is mounted by formally employed workers as they constitute the labor cohort who benefit directly from labor market institutions. Indeed, the latter generate rents that take the form of high relative wages and high relative employment security. Moreover, in contrast to informal and unemployed workers, formal 19 See Crowley (1997) for an extensive analysis. 20 Earnings would also include any kind of non-wage benefits as specified in Cuninngham and Maloney (2001). 21 See Rodrik (1996) for a general discussion of the argument. 22 See, for instance, Saint-Paul (2000) for a comprehensive theoretical analysis and further references. - 24 - workers are usually strongly represented in the political decisional process despite the fact that they do not represent the largest cohort in the labor force. As a consequence, the interest of workers with formal work contracts would prevail, even though the inherent economic outcome proves to be suboptimal and often unfair. 2.47. The Cardoso experience can be an illustration of the above arguments. The Real plan in Brazil was seen as a necessary adjustment program and eventually put the economy on a sustained path for economic development. However, the implementation of potentially beneficial labor reforms were refrained by political opposition. Indeed, as the economy, driven by a consumption boom, recovered rapidly from the initial negative impact of the Real plan, labor reforms have been perceived as an unnecessary additional sacrifice by the concemed individuals and their representatives. 2.48. The above process is likely to be reinforced by electoral concerns of politicians. Indeed, formal workers usually express their voting right more extensively than other workers. In other words, in a context where political interaction is taken into consideration, it is not surprising to not witness radical and effective changes in labor legislation within short spells of time. 2.49. In terms of policy implications, efforts put toward the implementation of radical labor reforms may not be highly effective unless they can be inserted into a broader set of policy interventions. 2.50. The previous sections suggest that in a market that behaves like an integrated market, the informal sector and specially the self-employed, are likely to present the features of a dynamic and potentially highly productive sector. In this context, the main challenge for policy makers is to make the informal sector to act like the formal sector with respect to compliance with social security contributions and taxation. Indeed, when there is evidence of labor market integration, tax and social security contribution evasion should be a major concem for policy makers. 2.51. In the same way that rigidities in the formal sector represent rents to formal workers, tax evasion can be seen as a rent seeking activity for informal individuals that is detrimental for the economy. 2.52. However, it is important to stress that tax evasion and non compliance with regulations are only a characteristic of the informal sector, not its very and only nature. Then, it could be counter- effective and harmful to impose on the informal sector compliance with taxes and regulations as long as other possible production impediments are not removed from that sector. Indeed, despite signals of a quasi well-functioning of the labor market, inefficiencies may still prevail in the informal sector. These inefficiencies result from the existence of costs not related to the very process of production. As such labor regulations for formal firms, weak judicial protection and high cost of capital can represent a bulk of "excessive" costs of production for informal producers23. 2.53. Weak judicial protection, as well as weak political representation, are essentially due to the fact that activities in the informal sector are not fully legal. Therefore, producers can not exercise full property rights over their capital and product and contracts can not be enforced through the judicial 23 De Soto (1989) presents evidence of the phenomena for Peru. - 25 - system. This translates into abnormally higher transaction and monitoring costs for the.informal sector compared to the formal sector. 2.54. Capital accumulation is also severely affected by non fully enforceable contracts. Indeed, effective borrowing rates paid by informal producers are higher than those paid by their formal counterpart. 2.55. A promotion of the efficiency of productive activities in the formal sector would contributes to lowering the cost of compliance with taxation and regulations to informal workers. This cost would be further reduced through a social security reforms that strives to align the costs with the benefits of contributors. 2.56. In the light of the above discussion, we argue that pursuing in the first place higher integration rather than lower segmentation, leads eventually to lower segmentation. Higher labor market integration implies better production conditions and a higher ability to take advantage of govemment-provided services in the informal sector. Subsequently the informal sector becomes increasingly attractive to all workers, implying that opposition to labor institution reforms could be weakened because of lower rents attached to formal activity. Moreover, the political weight of informal workers can be expected to increase, which could further contrast formal workers opposition. This tendency would be made even stronger by social security reforms that would ensure compliance of the full labor force. 2.57. In a country like Brazil, where segmentation and integration periods alternate, working for higher integration of the labor market is likely to produce almost immediate results. 2.58. In generic terms, the medium-to-long run policy objective is to fully align interests of participants in the formal sector with the interests of the participants in the informal sector, that is, to make sector-participation arbitrage rely exclusively on the skills and intrinsic ability of the individual. 2.59. The policy intervention orientation in the short-to-medium-run depends on the one hand on the behavior of the labor market and on the other hand on decision-making efficiency of the political process. Labor Policy Orientations and the Macroeconomic Environment 2.60. In this section, we present "anti-segmentation" and "pro-integration" policies suitable for macroeconomic environments inherent to the three type of macroeconomic policies analyzed previously. Trade Liberalization 2.61. Economic opening and integration are expected to affect negatively the traded goods sector as competition from foreign producers increases. Thus, trade liberalization is likely to affect labor eamings negatively. However, an offsetting positive income effect may appear due to a fall in the general price index as tariffs on imported goods are lowered. When wages are relatively flexible in the formal sector, those who flow to the informal sector are expecting to pursue higher income. Then, no segmentation should be identified. Moreover, the real exchange rate could either appreciate or depreciate depending on the amplitude of the offsetting income effect. - 26 - 2.62. In the presence of downward wage rigidities in the formal sector, workers are expected to be rationed out from the formal sector and the fall in labor eamings is expected to be essentially concentrated in the informal sector. In that case the degree of segmentation is expected to increase, relative eamings and participation in the informal sector move counter cyclically and, the real exchange rate depreciates. 2.63. The negative initial impact of trade liberalization would be accentuated if the govemment had to rise new taxes to offset the fall in tariffs revenues. 2.64. In the traditional view, a reduction in trade barriers is expected to enhance efficiency in the formal sector and to foster an adjustment in relative prices that leads to a reallocation of production resources towards that sector. However, the previous section indicates that in the presence of formal sector rigidities, the latter outcome may not be obtained in the short-to-medium run. Then, it is the role of policy intervention to smooth transitional negative effects. 2.65. In the formal sector, policy interventions must aim at facilitating efficiency improvements, specially for those firms that are directly affected by trade liberalization. Measures could consist of lowering the costs of borrowing physical capital by accompanying a trade liberalization with financial market liberalization and deepening. Nevertheless, firms may be induced to adopt more capital intensive technologies if regulations in the labor market remain unchanged. As we pointed out previously, reforms in the labor market may not be implemented rapidly. As a consequence, an alternative policy direction, which would complement efficiency improvements and contain substitution of labor, could be based on the implementation of specific training schemes. These training programs should provide workers with the necessary skills to catch up with expected technological improvements. All workers should be entitled to these schemes. Obviously, the cost of large scale training schemes may prove to be prohibitive, specially in the context of a macroeconomic adjustment program that requires budgetary restrictions. Then, there may be some scope for subsidizing temporarily and decreasingly formal producers. Subsidies could be made conditional on non-decreasing employment levels within a firm. However, the govemment would have to ensure that financial support is terminated within a relative short period of time as to avoid the perpetuation of rents previously attached to trade barriers, which are usually hold responsible for the lack of efficiency enhancing investments. 2.66. Even though the informal sector is not directly affected by trade liberalization, labor rationing in the formal sector due to rigidities is expected to lead to lower eamings and thus to worsen the plight of the lowest income group. Providing better access (full property rights) to cheaper capital to informal producers would contain segmentation and the progression of poverty. In the very short-run financial resources should also be devoted to a direct transfer to the poorest. It may be the case that part of these transfers would be used to build up some physical capital stock. Thus temporary income support programs may have long run poverty reducing effects. 2.67. When wages are relatively flexible, we would expect the impact of trade liberalization to be less harmful than in the presence of downward rigidities. However, labor regulations that do not lead essentially to wage rigidities but that are responsible for high labor costs (i.e. firing restrictions and indemnities, social security contributions ) may induce formal firms to replace labor with capital, putting the sector on a path of job destruction, despite efficiency improvements. -27 - Exchange-Rate Based Disinflation 2.68. A fall in the rate of devaluation is also equivalent to a negative shock to the formal sector. However, as inflation is curbed, we can expect aggregate consumption to react favorably to the disinflation program due to a positive wealth effect. The associated rise in consumption will also fall on non-traded goods and the exchange rate may appreciate as a consequence. 2.69. In that context, downward wage rigidities in the formal sector may not translate into a dramatic rise in the degree of segmentation as long as consumption reacts strongly and rapidly. Even though the formal sector is harmed by the program, workers are not necessarily rationed out in the presence of a strong rise in demand, which leads to a rise in the price of non-traded goods relative to traded goods. Indeed, the latter effect pushes earnings in the informal sector up and those workers initially rationed out may benefit from the rise in the price of non-traded goods. The labor market could then behaves as an integrated market. 2.70. Evidence of segmentation implies that either the negative shock to the formal sector is substantial and/or consumption reacts timidly and/or slowly. The Brazilian experience at the beginning of the Real Plan seems to fit the "and" scenario. There was initially a large negative shock to the formal sector: Then, consumption started to react upward and the real exchange rate started to appreciate after a period of pronounced depreciation. 2.71. Exchange-rate-based disinflation and trade liberalization have a similar impact on the economy. Both programs correspond to a negative shock to the formal sector and both attempt to generate efficiency improvements in sectors producing exported goods or in sectors which are competing directly with imported goods. However, while in the case of trade liberalization downward wage rigidity could be considered as an exogenous feature, within a disinflation program downward rigidity can be expected to be increased (generated) by the latter. Indeed, inflation usually represents a good means for firms to contain real wages progression. 2.72. The policy orientations exposed in the foregoing analysis hold in the context of a price stabilization program. Efficiency improvements should be eased controlling for the impact on employment. Nevertheless, there is stronger scope for a re-framing of the wage bargaining process and/or for acting on wage floors in order to offset the rigidities induced by the fall in inflation rates. 2.73. Theoretically, highly centralized or decentralized structures are the most desirable in terms of employment outcomes24. Nonetheless, highly centralized structures may give more space to an authoritarian government and could then generate non-optimal labor markets outcome. Ideally, bargaining should occur at the individual level. Only in that situation, all fringe benefits and costs incurred by both parts of the bargain can be on the table25. However, information is neither perfect nor complete and unions may play an essential role in the access to legal information and thus to guarantee a fair bargain. 24 See for instance Mc Donald and Solow (1992). 25 See Fugazza (2001) for a theoretical illustration. - 28 - 2.74. Another way to contain real labor costs is to erode the minimum wage if the latter influences substantially the wage distribution in the formnal sector. In the case of Brazil, the impact of the minimum wage seems to be far stronger on the informal sector than the formal. This corresponds to the so-called 'lighthouse effect"26, where, though not enforced by law, the minimum wage appears to be a benchmark for fair remuneration27. Then, by containing the real improvement of the minimum wage the govemment is likely to harm the poorest individuals concentrated in the informal sector without restraining job losses in the formal sector. In that context, the govemment has an additional instrument, besides direct financial support, to limit an eventual negative impact of disinflation on poverty. Note, that disinflation perse can be seen as a poverty reducing measure as poor individuals have usually few inflation hedges with which to protect themselves against price increases and are the ones who most suffer from the inflation tax28. Fiscal Adjustment 2.75. Fiscal adjustment could represent a policy goal per se. However, in most developing countries, fiscal adjustment has been a companion to attempts of macroeconomic adjustment. Brazil is no exception as the fiscal adjustment program set in the second half of the nineties was essentially the consequence of fiscal indiscipline in the aftermath of the Real plan. Public expenditures increased despite the dramatic drop in inflation tax revenues. Fiscal alarm was sounded in large part by regional public administrations because of a lack of cooperation with and, guidance from the central administration2g. 2.76. In a simplified understanding, fiscal adjustment can take the form of cuts in public expenditure3O and/or a rise in tax revenues31. Both interventions are directed toward the containment and possibly the reduction of soaring fiscal deficits. However, their respective impact could differ substantially despite identical accounting properties. 2.77. Lowering govemment spending eventually leads to a fall in absorption and can be biased either toward the forrnal or toward the informal sector. However, aggregate demand could in turn be stimulated by lower interest rates due to lower govemment borrowing. It is also argued that in such a context productive resources would be redirected toward the private sector which is usually thought to be more productive than the public sector. 2.78. When the govemment decides to reduce public expenditures by lowering its consumption of non-traded goods then, a real depreciation is likely to occur and eamings in the informal sector are 26 See Ned and al. (2000) and Amadeo and al. (2000) for a detailed identification and analysis of this effect. 27 Maloney and Nufiez (2001) present some evidence of 'lighthouse effect' also for Mexico, Argentina and Uruguay. 28 See Agenor (1999) for a comprehensive discussion. 29 See IMF (1998). 30 Cuts in public expenditure can take either a direct or indirect form. The latter can correspond to changes in the governments hiring and wage-setting policies, and/or the imposition of constraints on the growth in (or reduction in the level of) public sector employment. 31 An indirect way to rise tax revenues is to increase the relative price of public services. -29 - expected to fall relative to labor eamings in the formal sector. The degree of segmentation increases if worker migration toward the formal sector is refrained by downward wage rigidities. When spending cuts are achieved through lower public employment and wages are downward rigid in the formal sector, then again we expect the degree of segmentation to increase. 2.79. A rise in taxation, either direct (income taxation and payroll taxes) or indirect (VAT), represents a negative shock to the formal sector. The shock is likely to be stronger in the case of a rise in direct taxation and is amplified by the presence of downward rigid wages. The latter implies that even parts of the tax increase can not be passed onto wages. Whether formal wages are downward rigid or not, individual migration to the informal sector may be also motivated by tax evasion considerations32. Hence, the outcome of higher taxation may lead to a fall in the tax base and possibly to a fall in tax proceedings. The effects of a rise in indirect taxation are likely to be more diluted as part of it can fall on consumers through price adjustments. However, as indirect taxation is a regressive tax, the outcome for the poor can be worsened. In order to circumvent such a non desirable outcome it usually recommended to appropriately target the basket of goods on which to impose higher tax rates33. 2.80. Another alternative to raising tax revenues consists of an enlargement of the tax base. To do so, a government can decide either to force those who are evading to contribute and eventually penalize them, or to induce them to participate in fiscal schemes on a voluntary basis. While the former approach would not necessitate any particular measure but fiercer and more efficient tax audit agencies, the latter would be effective only if individuals find an incentive to refloat their activity. Tax evasion is usually the consequence of the perception by the evader of the existence of pure taxes. Then, voluntary contribution would be facilitated by aligning costs and benefits of fiscal participation. This is particularly true for social benefits contributions. The necessity of reforming the welfare system, essentially the pensions scheme, in Brazil and in general in Latin American countries have been underlined in the past years, the rationale being essentially a reduction of huge and unsustainable fiscal deficits. This rationale is reinforced by the possibility of attracting informal workers, specially the self-employed. In that context, efforts directed towards increasing the degree of integration as mentioned in previous sections, would contribute to an acceleration of this process. Moreover, a by-product of a reform of social benefits schemes could be a fall in employers payroll taxes and thus eventually a rise in formal employment. 2.81. Within a fiscal adjustment scheme based in part on the reform of the welfare system, there is also scope for a reform of severance payments aiming at avoiding the 'pure cost" feature of existing schemes. Indeed, severance payments should not represent a firing restriction but rather a financial cushion for the dismissed worker. Financial support during unemployment spells is motivated by the existence of imperfections in the financial market that limit individuals capability to insure themselves against income drops and thus to smooth consumption. However, if severance payments are given this characteristic, then they would make unemployment benefits redundant if the latter are paid only to workers which are formerly employed. Unemployment insurance schemes, where they exist, are characterized by very low levels of coverage in developing countries including Latin American 32 See Loyaza (1995). 33 World Bank Poverty Report -30- countries34. In Brazil for example, only 11.8 % of unemployed workers benefit from the scheme. This may reflect either very low replacement ratios or inefficient administrative management of the scheme or both. In other words, unemployment insurance schemes do not appear to play fully their role of financial cushion specially for the poorest individuals. Moreover, if the payment of unemployment benefits is conditional on former formal employment, then all informal workers, who include the poorest individuals, are not eligible for such a scheme. In order to make policy intervention more transparent for firms and workers, it would be preferable to guarantee financial support to the poorest through income support programs and give severance payments the role of unemployment insurance. More precisely, the worker should be able to cash in her contributions in any situation. Usually a large part of the contributions to severance payment schemes are paid by the employers. Unconditional severance payment would allow for downward pressure on wages, independently of the degree of centralization of the bargaining process. Indeed, workers (unions) intemalize the future financial transfer in case of the interruption of the job relation and thus are ready to give up some present income35. 2.82. Table 2.2 summarizes previous results. Table 2.2 Policy Priorities During Alternative iRscroeconomic Scenarios ._ _ YTradeUibweraIization ' lIDisin4flation - | FiscalAdjustment | i Pro-Integration [Lower cost of borrowing Lower cost of Social security reform physical capital borrowing physical Strengthening property capital rights Strengthening property rights Anti-Segmentation Training schemes Collective bargaining Severance payments Lower cost of borrowing reform reform physical capital Lower cost of Payroll tax reform Temporary and borrowing physical conditional subsidies capital Temporary and l ___________________ _____________________ conditional subsidies 2.83. Since nearly 60 percent of the workforce in developing countries are employed in the informal sector, an analysis of labor markets in developing countries cannot ignore the importance of the labor market composition, i.e. the relative size of the formal and informal sector. In this paper, we argue that the labor market composition is not determined only by the existence of labor market 34 See De Ferranti and al. (2000). 35 See for instance, Cahuc and Zilberberg (2000). -31 - rigidities per se, but is also a function of the macroeconomic environment, which determines if labor market rigidities become binding or not. The paper argues that labor market composition and macroeconomic environment are intimately linked and need to be analyzed jointly, as: * the macroeconomic environment itself can affect the degree of segmentation in the labor market and consequently the impact of policy interventions * the expected outcome of macroeconomic policies such as stabilization and adjustment policies varies whether the labor market behaves like an integrated or a segmented market 2.84. We empirically test the hypotheses of segmentation and integration using recursive cointegration techniques and rolling Okun coefficients. Within the first approach, evidence of integration or segmentation is nested within the hypothesis of a positive or negative co-movement of relative formaVinformal real wages and participation rates. Regarding the latter approach, the hypothesis of segmentation is accepted if quantities react stronger than prices to changes in output, i.e. employment Okun coefficient are more sensitive than wage Okun coefficients. 2.85. Both approaches identify altemating periods of labor market integration and segmentation during the 1990. As the 1990 decade in Brazil has been characterized by a growing informal sector, this trend is not the exclusive result of a increasing degree of segmentation in the labor market. Rather it appears to be the result of the conjunction of a series of economic events and of always threateningly binding labor legislation. In addition, a general shift of preferences in favor of the non- traded goods also contribute to the pattem of the economic activity composition. 2.86. In terms of policy implications, the analysis undertaken in the paper allows us to underline the following major points: * When a labor market behaves like an integrated market at some points in time, then trying to increase integration rather that to decrease segmentation proves to be a better policy orientation * In a political economy context, increasing integration may lead to lower segmentation . Higher integration contributes to ease compliance with social security contributions and taxation among informal individuals * Full integration can not be observed as long as pure taxes features prevail in the economy - 32 - References: Amadeo E., I. S. Gill and M. C. Neri, (2000), 'Do Labor Laws Matter ? The 'Pressure Points' in Brazil's Labor Legislation", in 1. S. Gill and C. E. Montenegro, "Readdressing Latin America's 'Forgotten Reform', Quantifying Labor Policy Challenges in Argentina, Brazil and Chile", manuscript, IBRD. Aroca, P and W. F. Maloney (1999), "Logit Analysis in a Rotating Panel Context and an Application to Self-Employment Decisions," IBRD Working Paper 2069. Latin America and Carribean Region, Poverty Reduction and Economic Management Unit, World Bank, Washington D.C. Processed. Agenor, R., (1996), 'The Labor Market and Economic Adjustment," IMF Staff Papers, 43, 261-335. , (1999), "Stabilization Policies, Poverty and the Labor Market", unpublished manuscript. Balassa B., (1964), 'Trade prospects for developing countries", Yale University. Economic Growth Center. Publications 382, New Haven. Brock P., (1988), 'Investment, the Current Account, and the Relative Price of Non-Traded Goods in a Small Open Economy", Journal of Intemational Economics, 24, pp. 235-253. Cahuc P. and A. Zylberberg, (2000), "Redundancy Payments, Incomplete Labor Contracts, Unemployment and Welfare", CEPREMAP Working Paper 00/06. Cheung, Y.-W.; and K.S. Lai, (1993), "Finite sample sizes of Johansen's likelihood ratio tests for cointegration", Oxford Bulletin of Economics and Statistics, 55, 3, 313-328. Conley, John P and William F. Maloney. 'Optimal sequencing of credible reforms with uncertain outcomes", Joumal of Development Economics, 48:151-66, October 1995. Crowley J., (1997), "The Effects of Forward- Versus Backward-Looking Wage Indexation on Price Stabilization Programs", IMF WP/97/38. Davila Capalleja, E., (1994), "Regulaciones Laborales y Mercado de Trabajo en Mexico", in Gustavo marquez ed., Regulacion del Mercado de Trabajo en America Latina, Centro Intemacional para el Desarollo Economico, San Francisco. De Ferranti D., Perry G., I. S. Gill and L. Serven, (2000), "Securing our Future in a Global Economy", World Bank Latin American and Caribbean Studies, Washington D.C. De Soto H., (1989), The Other Path, Harper and Row, New York. Edwards, Sebastian and Alejandra Cox-Edwards, (2000), "Economic reforms and labor markets: policy issues and lessons from Chile", National Bureau of Economic Research. Working Paper Series (U.S.); No. 7646:1-66, Boston. Evans, D.C. and B. Jovanovic (1989), " And Estimated model of Entrepreneurial Choice under Liquidity Constraints, Joumal of Political Economy 97:4 pp 808-826. Fajnzylber P. and W. Maloney, (2000), "Labor Demand and Trade Reform in Latin America", unpublished manuscript. - 33 - Fields, G. S., (1990), 'Labor Market Modelling and the Urban Informal Sector: Theory and Evidence", in Tumham D., B. Salom6 and A. Schwartz (eds.), The Informal Sector Revisited, OECD, Paris. Fiess N., M. Fugazza and W. Maloney, (2001), "Exchange Rate Appreciations, Labor Market Rigidities, and Informality", World Bank forthcoming WP. Fugazza M., (2001), "Search Bonuses and Hiring Subsidies: A Theoretical Comparative Analysis", unpublished manuscript. Gil-Diaz, F. and A. Carstens (1996) "One Year of Solitude: Some Pilgrim Tales about Mexico's 1994- 1995 Crisis," American Economic Review, 86,2: 164-169. Gonzalez Anaya, (1999) Harris J. and M. P. Todaro, 'Migration, Unemployment and Development: A Two-Sector Analysis", American Economic Review, 60, pp 126-43. Heckman J. and, C. Pages, (2000), "The Cost of Job Security Regulation: Evidence from Latin American labor Markets", IDB Working Paper 430. Horton S., R. Kanbur and D. Mazumdar, 1994, Labor Markets in an Era of Adjustment, The World Bank, Washington D.C. IMF, (2001), "Brazil: Selected Issues and Statistical Appendix', IMF Country Report 01/10. IMF, (1998), "Brazil: Recent Economic Developments", IMF Country Report 98/24. Johansen, Soren, (1988), "Statistical Analysis of Cointegrating Vectors", Joumal of Economic Dynamics and Control, 12:231-54 June-September 1988. Johansen, Soren and Katarina Juselius, (1992), uTesting structural hypothesis in a multivariate cointegration analysis of PPP and UIP for UK", Joumal of Econometrics, 53,169-209. Johansen, Soren and Katarina Juselius, (1994), "Identification of the long-run and short-run structure. An application of the ISLM model", Journal of Econometrics, 63, 7-36. Krebs, T. and W.F. Maloney (1999), "Quitting and Labor Turnover: Micro-economic Evidence and Macro-economic Consequences." IBRD Working Paper 2068, Latin America and the Caribbean Region, Poverty Reduction and Economic Management Unit, World Bank, Washington, D.C. Processed. Levenson, A. and W.F. Maloney (1988), "The Informal Sector, Firm Dynamics and Institutional Participation." IBRD Working Paper 1941, Latin America and the Caribbean Region, Poverty Reduction and Economic Management Unit, World Bank, Washington, D.C. Processed. Lora, E. and G. Marquez, (1998), "The Employment Problem in Latin America: Perceptions and Stylized Facts", IDB Working Paper 371. and, M Olivera, (1998), "Macro Policy and Employment Problems in Latin America", IDB Working Paper 372. and, C: Pages, (1996), "La legislacion Laboral en el Proceso de Reformas Estructurales de America Latina e el Caribe", unpublished manuscript. - 34 - Loayaza N., (1996), "The Economics of the Informal Sector: A Simple Model and some Empirical Evidence from Latin America", Camergie-Rochester Conference Series on Public Policy, 45, PP. Mac Donald I. and Solow R., (1992), "Wage Bargaining and Employment", Economic Models of Trade Unions, pp 87-104. Magnac T., and J-M. Robin, (1996), "Occupational Choice and Liquidity Constraints", Ricerche Economiche, 50,105-133. (1998a), "The Structure of Labor Markets in Developing Countries: Time Series Evidence on Competing Views," IBRD Working Paper 1940, Latin America and the Caribbean Region, Poverty Reduction and Economic Management Unit, World Bank, Washington, D.C. Processed. (1998b), "Are Labor Markets in Developing Countries Dualistic ? ", World Bank Policy Research WP 1941. (1999), "Self-Employment and Labor Tumover: Cross-Country Evidence. The Structure of Labor Markets in Developing Countries: Time Series Evidence on Competing Views," IBRD Working Paper 2102, Latin America and the Caribbean Region, Poverty Reduction and Economic Management Unit, World Bank, Washington, D.C. Processed. _ and J. Nunez, (2001), "Measuring the Impact of Minimum Wages", World Bank Policy Research WP 2597. Marquez, G., (1997), "Proteccion al Empleo y Funcionamiento del Mercado de Trabajo: una aproximacion comparativa", in Empleo, Flexibilidad Laboral y Proteccion Social, Proccedings of the Second Technical Meeting, Circulo de Montevideo, UNDP. Marquez, G., (2000), "Labor Markets and Income Support: What Did We Learn from the Crises?", IDB Working Paper 425. Matusz S. and, D. Tarr, (1997), "Adjusting to Trade Policy Reform", unpublished Manuscript. Mazza J., (2000), "Unemployment Insurance: Case Studies and Lessons for Latin America and Caribbean', IDB Working Paper 411. Neri M., G. Gonzaga and J. M. Camargo, (2000), "Efeitos Informais do Salario Minimo e Pobreza", forthcoming, Revista de Economia Politica. O'Connell L., (1999), "Collective Bargaining Systems in 6 Latin American Countries: Degrees of Autonomy and Decentralization", IDB Working Paper 399. Obstfeld, Maurice and Kenneth Rogoff, (1996), Foundations of Intemational Macroeconomcis, MIT Press, Boston. Okun, A., (1962), "Potential GNP: Its Measurement and Significance", Proceedings of the Business and Economic Statistics Section of the American Statistical Association", pp 89-104. Rodrik, D., (1996), "Understanding Economic Policy Reform", Joumal of Economic Literature, Vol 36.1, pp 941. Peattie, L., (1987), "An Idea in Good Currency and How It Grew: The Informal sector", World Development, 15, pp 851-860. - 35 - Portes R. and R. Schauffler (1993), "Competing Perspective on the Latin American Informal Sector", Population and Development Review, 19, 33-59. Saint-Paul G., (2000), The political economy of labour market institutions', Oxford U. Press. Thomas, J.J., (1992), Informal Economic Activity, LES Hanbooks in Economics, London. Tumham 0. and Erocal D., (1990), "Unemployment in Developing Countries, New Light on an Old Problem", OECD Development Center Technical Paper. Tokman, V.E, (1992), Beyond Regulation, The Informal Economy in Latin America, Lynne Rienner, Boulder. Vijverberg, W. P. M., (1986), "Consistent Estimates of the Wage Equation when Individuals Choose among Income-Earnings Activities", Southem Economic Journal, 52, pp 1028-42. Yamada, G., (1996), "Urban Informal Employment and Self-Employment in Developing Countries: Theory and Evidence". -36 - Appendixi 1: Recursive Cointegration Tests Figure 2.5 Recursive Cointegration Tests Between Real GDP and Employment: The Trmce test 0 68 Z4 0 66 064 0 62 0 60 0058 0086 0 54 0 52 1990 t19 1992 1W3 1994 1995 199 1997 1998 19m 2000 Figure 2.6 Recursive Cointegration Tests Between Real GDP and Real Wages The Trace tests 075 Z4 070 095 060 0.59 0.00 045 040 035 1990 1991 1992 1993 1994 1995 1999 1997 1998 1999 2DDO - 37 - Appendix 2: Cointegration tests between relative size of self-employed to informal, ndnit, and relative wages of self-employed to informal, WN/WIF: The VAR model is estimated over the sample from 1990Q1 to 2000Q1 and includes a constant in the cointegration space and 4 lags for WN/WF and nfnlff and these specification proves sufficient to produce random errors. The X ttests indicate one significant cointegrating vector in model: Null Altemative Lag: 4 90% Hypothesis Hypothesis With Constant Critical Value fA,e test r=0 r>0 19.44* 17.79 r S 1 r>1 5.09 7.50 A hypothesis test of procyclical behaviour between the WN/WIF and nrnff cannot be rejected. (The LR test, * (1) = 0.35, p-value = 0.55). The resulting cointegrating vector, f, is: /3 nr/nff 1.000 WN/WIF -1.000 Constant 0.093 Figure 2.7 Self-Employed versus Informal 0 48 - 120 -.15 0 44 0 40 Os0 00 0 36 -095 0.32 - '-0,90 028 8 024 -..... . . . .. .. .75 1982 1984 1986 1988 1990 1992 1994 1998 1998 - 38 - Figure 2.8 Sel-Employed versus Informal 0.48 - -1.20 027 1 . t_,.~~~~~~~~~~~~~~~~- 1.5 0.44 -15 11.10 0.40 A- 036 r 1 ~~~~~~~~~~~~~~~~~~1 05 0 32'I' 0.28 ~ ~ ~ 09 0.28 ~~~~~~~~~~~~~~~~~~~~~~~0.90 - SELF --- WSEIF 0.24 I- 8 1990 1991 1992 9319 1995 1961997 1998 1999 - 39 - Figure 2.9 Self-Employed versus Formal self-employed versus formal 1 150 -. -2.05 EMPSE WFOSE -2 00 1.125 - -195 1100 - 190 4' ~~~~~~~~~~~~~1.85 1.075 1.60 1990 1991 1992 1993 1994 1995 1996 1997 4£98 180 i 50 - \, ,_ 17 1 25 t1990 t1991 t 992 t1993 t1994 t1995 t1996 t1997 t1998 1 999 45 21~~~~~4 404 4 -44 4 4 190499 102 19 94 195 196 19419 9 - 40 - 30 RIE-1EKAHNNX THE OMFORNAL SECTOR Prepared by William F. Maloney and Wendy V. Cunningham36 3.1. Three decades of research have not yielded consensus either on the definition of the informal sector or its razon de ser.37 Broadly speaking, the small-scale, semi-legal, often low- productivity, frequently family-based, perhaps pre-capitalistic enterprises continue to employ between 30% and 70% of the urban work force in Latin America. A long tradition views informal workers as comprising the less-advantaged sector of a dualistic or segmented labor market.38 Above market- clearing wages force workers to queue for preferred jobs while subsisting in the informal sector, which is characterized by an absence of benefits, irregular work conditions, high turnover and, overall, lower rates of remuneration. A recent variant on the dualism view, albeit with different emphasis, sees informalization as an effort by firms facing international competition to reduce these legislated or union induced rigidities and high labor costs, particularly through subcontracting production out to unprotected workers.39 3.2. This paper argues that such a view does not convincingly describe the Latin American case. First, despite compelling descriptions of the inflexibility, inefficiency and costliness of the labor code,40 the usual sources of wage rigidity that would segment the market seem absent in various Latin American countries with large informal sectors: minimum wages have not been binding for the last decade, unions to date have primarily been concerned about preserving employment rather than raising remuneration, and wages have shown extraordinary downward flexibility during crises.41 Second, the view that emerges from time series data from Latin American countries and global cross sectional data correspond more to an unregulated entrepreneurial sector that behaves like small firm sectors everywhere rather than one comprised of involuntary, disadvantaged, precarious, or underpaid workers. 3.3. This paper challenges the traditional dualism theory by drawing on evidence from Latin America to better understand the structure of the informal sector, why it exists, and the quality of its jobs. It goes through a series of assumption about the informal sector and presents recent evidence 36 This paper is heavily based on Maloney, William F. (2000) 'Informality Revisited', unpublished manuscript. 37 See Peattie (1987), Fields (1990), Tumham & Erocal (1 990), Tokman (1992), Portes and Schauffler, (1993) for excellent overviews. 38 The Harris-Todaro (1970) model is perhaps the traditional statement of this view. See most recently Chandra and Khan (1992), Chandra (1994), and Loayza (1994). 39 See Beneria (1989), Portes, Castells and Benton (1989), Portes and Schauffler (1989), and Roberts (1989) in the Structural Articulation school. Piore and Sabel, (1987), Gordon, Edwards and Reich (1983), more generally in the decentralization school. 40 See Davila Capalleja (1997) and Marquez (1994) 41 See Maloney and Ribeiro (1998) and Maloney (2000) -41 - and new statistics to test the validity of the assumption. Rather than supporting the dualistic view of poor jobs for unfortunate workers, the review shows that the informal sector is a dynamic sector that is an outgrowth of the entrepreneurial sector in any country in the developing or developed world. Its workers are rational people who largely select informal employment due to the many non-wage (and wage, in some cases) benefits derived from it. These findings suggest that policies that attempt to create a single market by eliminating labor market rigidities are perhaps misguided. Instead, the sector should be brought into the formal tax and benefit system, as in developing countries, by improving the public goods such that informal workers and firms choose to pay into them, and improving the institutions that the informal sector substitutes for. 3.4. The informal sector is heterogeneous, but its workers may be Figure 3.1 Self-employment in the Working Population classified into three general groups, (by gender), 1999 which should be treated separately due to the nature of the jobs and the characteristics of the workers. For the Venezuela . - purposes of this paper, the informal Paraguay . sector is defined as those jobs for which Nicaragua U the employer does not pay into the government insurance program on Mexco . . behalf of the employee.42 First, the Honduras informal self-employed may be defined Sad as those individuals who identify E Salvador themselves as self-employed and have Dominican Republic no more than completed secondary Costa Rica l education. Those who have more than l secondary education are likely to be Colombia professionals and thus are not the Chile . "vulnerable" informal. Self-employment Braz constitutes the largest source of employment among men (16-35 Argentina percent) and women (13-15 percent), after formal salaried employment (40-64 0 10 20 30 40 and 28-57 percent, respectively) (Figure onmae Efeile 3.1). 42 Informal sector jobs may be defined from the point of view of the worker or the firm. In this paper, the status of the worker with respect to labor taxes and benefits, defines informality. Altematively, the status of the firm with respect to labor taxes may define an informal sector firm and therefore its employees are informal sector workers. However, it is becoming increasingly common that workers who do not have labor taxes paid in their name work in firms that pay taxes for some of their employees, so the definitions do not necessarily classify workers in the same categories. -42 - 3.5. Second, informal salaried workers are employees who receive a wage but do not receive benefits. They are perhaps the most "vulnerable" since their employer does not pay for, and thus they are not eligible for, government benefits, but more importantly, since the govemment does not recognize the employer-employee relationship, they are not govemed by labor laws. These workers constitute 5-19 percent of the male labor force and 12-40 percent of the female labor force in a nine country sample for Latin America (Figure 3.2). 3.6. Finally, unpaid workers are those individuals who work but do not receive a wage. A large portion of these workers is employees in family firms (particularly women and children) so they do share in the benefits of the firm owner's profits. Thus, unpaid workers are actually paid in-kind. They make up a small part of the labor force with 1-8 percent of males and 2-11 percent of females in the nine country sample (Figure 3.3). Figure 3.2 Informal Salaried by Gender Figure 3.3 Unpaid by Gender (Proportion of the working population, 1999) (Proportion of the working population, 1999) Peru Venezuela Paraguay Paraguay ParagayNcaragua Nicaragua rf Mexico Mexico Honduras El Salvador El Salvador Costa Rica - Dominican Republic Chile ~~~~~~~~~~~Costa Rica Hrazil =l_ ]Chile Brazil B Brazil 'Argentina Argentina 0 10 20 30 40 50 0 10 20 30 40 l jo male o female l I|Omale O femaele 3.7. Formal sector employment is not necessarily preferred to informal sector employment. Table 3.1 shows that of Mexican male workers who started in the formal salaried sector but moved into informal self-employed sector 15 months later,43 two-thirds report moving voluntarily, citing a desire for greater independence or higher pay as the principal motives. In Brazil, over 50 percent of men 43This is done by linking the Encuesta Nacional de Micro-Empresas (ENAMIN) with the Encuesta Nacional de Empleo Urbano (ENEU). The ENEU permits following workers over a 15 month period and hence studying their transHtions among sectors. -43 - and 43 percent of women who were informal stated that they did not want a formal sector job, primarly because they were happy with their current job (Table 3.2). These findings are consistent with the sociologists Balan, Browning and Jelin's (1973) extensive interviews with Monterrey workers who state that being one's own boss was well-regarded and that movements into self-employment from salaried positions often represented an improvement in job status.44 3.8. This is also broadly consistent with interview data from Argentina: a small survey in the province of Jujuy revealed that 80% of the self-employed had no desire to change jobs and less than 18% saw self-employment as a temporary activity before they found a "real" job.45 In Greater Buenos Aires, another survey found that while 36% would have preferred to work more hours, only 26% were looking for other work.46 In Paraguay, only 28% of those in the informal sector stated a desire to change occupations. Among those often thought to be the worst off, informal workers, the percent rose only to 32%.47 Table 3.1 Real Hourly Wage Differential - Formal Salaried to Informal Self Employed (Men Only, Mexico, percentage change) Share of Mean Differential3 Median Respondents Differential National Urban Employment Survey (ENEU)1 100 27.3^** 23.6"** Micro-enterprise Survey (ENAMIN)2 Reason Left Previous Job More Independence 35 21.6"" 19.6* Higher Pay 32 16.7 12.1 Involuntary 29 -14.7 -4.1 Total 100 12.6"* 10.0** I Adjusted for hours worked and taxes. 2Adjusted for hours worked, taxes, capital costs, unpaid workers. 3Mean employs Huber weights to redress non-normality. * = significant at 10%, ** 5%, "' 1% level. Table 3.2 Rationale for not Preferrng to Move to a Formal Sector Job, Brazil (%) Men Women Earn more in current job 14.4 5.3 Household duties 0.2 22.7 Need time for other activities 4.0 6.3 Happy with current job 65.7 53.7 Do not want to fulfill requirements of a formal job 10.2 6.7 Other 5.5 5.2 Sample size 22,845 10,791 Source: PNAD, 1989 44 These results are very close to Gottshalk and Maloney's (1985) finding that roughly 70% of U.S. job changes are voluntary. Put differently, if self-employment given the common earnings differentials, are close substitutes for formal salaried work, the implied rates of involuntary entry would be normal by the standards of a flexible industrialized country market. 45 Micro-enterprise Study for Jujuy, Consultoria Nordeste, Fondo de Capital Social (FONCAP). 46 SIEMPRO/FONCAP (1998) "Perfiles de la Microempresa". 47 Direccion General de Estadistica, Encuestas y Censos, Rpublica de Paraguay, 'Sector Informal," based on the Encuesta Integrada de Hogares 1997/1998. -44 - 3.9. Among women, self-employment also constitutes a large part of the informal sectors, but in several countries in the sample, the informal sector is larger. Women's reasons for becoming self- employed differ from men's. As reported by the Mexican Micro-firm Survey, approximately 10% of women left their last job due involuntary circumstances (dismissed, firrn closed, or the contract ended) while 21% left for more independence or higher pay, the primary reasons cited by men. For women, though, demographic characteristics are the driving factor, as almost half of the women in the sample left their last jobs due to marriage. When extracting this group from the sample, though, over 44 percent cited leaving their last job in search of higher pay and independence (Table 3.3). Evidence from Argentina, Mexico, and Brazil suggest that this sector also serves as a flexible source of employment for women who are also saddled with home responsibilities. Women with young children are more likely to become self-employed rather than formal sector employees, and interview data from Geldstein (2000) for Argentina suggests that women may more easily balance their productive (market) and reproductive (homecare) roles if they work for themselves than if they are employees. Furthermore, Table 3.2 showed that 22.7 percent of Brazilian women in the informal sector prefer these jobs because they allow them to perform their household roles as well, as opposed to 0.2 percent of Brazilian men. Table 3.3 DAotivations for bacoming selfemployed (in %) Women Men All Worked All Worked before before Reasons for leaving the last job Never worked before 37.3% --- 13.8% --- Fired 2.1 3.35 7.7 8.93 Firm closed 3.98 6.35 5.9 6.84 Contract ended 0.1 1.76 4.0 4.63 Low pay 6.85 10.93 18.7 21.68 To be independent 6.41 10.23 31.8 36.89 Retired 0.33 0.53 2.0 2.31 Moved 3.2 5.11 3.5 4.05 Married 28.84 46.03 0.3 0.35 Illness 4.42 7.05 3.7 4.29 Other 4.42 7.05 8.6 9.98 N 905 567 6871 5923 Source: National Survey of Micro-enterprises in Mexico, 1992. 3.10. Panel data from Mexico's National Urban Employment Survey (ENEU) show that, as in industrialized countries, self-employment is not an entry occupation from school and there is little evidence that the sector serves as a holding pattem for young workers. Transitions into self- employment from the other paid sectors occur 4 to 6 years later than transitions into formal or informal salaried work, leaving the mean age 8 years higher than the next closest sector.48 This is supportive of the findings of Balan et. al., and increasingly elsewhere, for a ulife cycle' model where 48 See Maloney (1999), and Aroca and Maloney (1998). - 45 - workers enter into salaried work, accumulate knowledge, capital, and contacts, and then quit to open their own informal businesses.49 3.11. The sociologist Gonzalez de la Rocha (1994) is almost certainly correct that for many older workers, the sector does provide a safety net by offering 'insecure occupations (such as the services) in which their age is not a limitation after they have been squeezed out of the formal manufacturing or formal services." This dynamic may have been of particular relevance during the economic restructurings of the 1990s where older displaced workers may have found their skills obsolete and little demanded in the emerging sectors. She does also, however, suggest some degree of voluntary movement when she says that "Older men may also find the pace of industrial (formal) work too arduous and leave such jobs."50 This more voluntary take is stressed by her anthropologist colleagues Selby, Murphy and Lorenzen (1991) who note the "surprising desirability of informal sector employment as the basis for a household eaming strategy, particularly for poorer, older household with lower educational qualifications."51 3.12. Anthropological evidence also suggests that for poor women, the self-employment sector does serve a sort of safety net, but in a different way than is typically understood. Geldstein (2000) suggests that women simultaneously own a number of small firms at all times. The firm owners recognized these income generating "activities" (rather than "jobs") as extensions of household duties that provide women with pocket money when the household has sufficient income. However, when the household experiences negative income shocks, such as caused by the prolonged 1999-2001 recession in Argentina, these small firms expand to be the primary sources of income for the household, even to the extent of employing male household members. Thus, they are not a safety net in the sense that the worker cannot find employment elsewhere, necessarily, but instead are stretched to protect the household as the household's income needs change. 3.13. On balance, sociological and anthropological studies are consistent with the findings of labor market surveys that most self-employed choose to be so, as well as the view that workers may enter formal sector employment initially partly as a means to accumulate human and physical capital. This leads to an intriguing inversion of the traditional dualist view. If firms must pay "efficiency' or above market clearing wages to dissuade their workers from opening their own firms, this creates a segmented market. It may be the attractiveness of informal self-employment that causes dualism rather than a segmented market causing informality.52 49 See Lopez-Castanio (1990) in Colombia, Fields (1990), and Peattie (1982) who find a tendency for employees of large firms to leave and open their own. This is likely to be due to the same kind of credit constraints found binding in industrialized countries. See Evans and Jovanovic (1989). 50 See Maloney (1999) Labor Markets Regional Study. The anthropology literature also weighs in on this point as well when Selby, Murphy and Lorenzen (1991) argue 'On average, opportunity costs to these worker in terms of foregone earnings in, say, registered blue-collar employment may be quite low (p. 147). 51 Selby, Murphy, Lorenzen (1991), p. 144 52 See Maloney (forthcoming) and Krebs and Maloney (1999) for an elaboration of this view. - 46 - 3.14. Wage comparisons between sectors are fundamentally meaningless so they should probably not be used as measures of segmentation and job quality. In a some countries, especially among men, the median eamings of self-employed workers are higher than the median wages of formal sector employees while in other countries the reverse is true.53 Figure 3.4 shows that in Argentina (men), Chile, and Nicaragua (men), self-employment eamings surpass formal sector wages. The gender differential is perhaps not surprising given that men express a greater expectation than women do that self-employment would offer higher wages. Data from Mexico show that workers who voluntarily entered self-employment gained roughly 25% in their income upon leaving salaried formal employment (Table 3.1). However, a simple comparison of wages does not accurately reflect the welfare benefits to one sector over another since, when selecting a sector, workers not only take into account wages (or earnings) but also all other characteristics and benefits associated with each possible job. Figure 3.4 Ratio of Self-employment Earnings Figure 3.5 Ratio of Informal to Formal Sector to Formal Sector Wages, late 1990s Wages, late 1990s Argentina - Argentina _ Brazil - Brazil _ Chile T _ ,-B L Colombia Costa Rica : 1 , Costa Rica Dominican Republic . El Salvador MeIico -~ Honduras - _ , A Mexico .l f Nicaragua Mexico Paraguay Nicaragua - _ 0.0 0.5 1.0 1.5 2.0 0.0 0.2 0.4 0.6 0.8 1.0 O nale _ fe__le I|E male n female 3.15. Several examples may make this clearer. First, if the formal sector pays benefits - pensions, health insurance, housing subsidies, day care - which the informal sector does not, then in a market wHth no distortions, the wage in the unprotected sector would need to be above the formal sector wage to compensate for the lost benefits. This higher wage would not imply a superior job but only suggest that more of the total remuneration was paid in cash instead of benefits. On the other hand, 53 Using mean income or wages rather than the median income or wages shows that self-employment eamings surpass formal sector wages in even more countries. This is due to the higher variance in self-employment eamings, where eamings may be derived from everything from washing car windows at stoplights to a sophisticated electronics repair shop. - 47 - income taxes support public goods from which tax-avoiding informal workers cannot be excluded (national defense for example), and therefore dictate that formal sector workers need to be paid more to compensate for the taxes they cannot avoid. Further, formal work-places may be cleaner and the jobs may be more secure, but, as in the industrialized world and as Table 3.1 and Table 3.3 suggest, being ones' own boss and having flexibility is also very valuable and is worth taking a wage cut for. Finally, starting a business is risky anywhere so higher incomes should be found among the self- employed to compensate them for this risk. 3.16. In most countries in the Region, informal salaried employees earn approximately half of their formal sector counterparts, with a larger difference among women than for men (Figure 3.5). Although informal salaried workers always earn consistently less, this may be due to the fact that they are often related to the owners of the enterprises where they work and thus may receive unobserved payments in kind (food, lodging). Further, to the degree that the sector appears to play a job training role for young workers, some fraction of the salary may be deducted to cover implicit training costs. 3.17. To summarize, in a market with no distortions and hence no segmentation, workers would equate utility, or the total 'package" of benefits, not just wages. To establish segmentation and show that formal sector workers were better off, the value of all of the factors discussed above should be added up, which is an impossible task since many of them are unmeasurable. This implies that comparisons of wages cannot be tests for segmentation. It also shows that the fact that informal salaried workers may eam less than formal salaried workers does not mean that they are "poorer' when measured in terms of total welfare. IV ISN'TFI.ItE INFORNIAL SECTOR TItIE sAE 1:1 E NEIFOR I I)IlSPLACEI)D FORNMAlSECTOR0 W(O)RKER{S?I .... ... .... 3.18. The informal sector may act as a safety net, but this is not necessarily the rule. The traditional "dualistic" view argues that as the economy enters a recession, workers are forced into the informal sector, thereby driving down these wages relative to those in the formal sector. This does seem to be the case in Colombia after 1995, a period of deep recession due to a financial crisis combined with a very ill-advised but dramatic rise in the minimum wage that accentuated the segmentation in the economy. Figure 3.6 shows that the size of the formal sector relative to the informal self-employed sector fell while its relative wage rose - exactly the pattern predicted by the dualistic view. -48 - Figure 3.6 Colombia, Relative Sector Size and Earnings 3.19. However, Figure (men only) 3.7 suggests that the opposite occurred in Mexico 1.4 4 during the 1987-93 period. 1.2 - A 3.5 Here the share of the 1 ~ JA\/ /At - > v - 3 workforce in self- I J\/t W \ v / J zs employment grew at the a.8 - 2 same time that the self- a.6 -u2 employed went from eaming 0.4 - 1.5 roughly the same amount as Q4 - _ S I formal salaried workers to 0.2 - __- Q5 eaming 30% more than 0 0 l ll ll ll l l l ll ll ll ll l l l ll ll ll ll l l lfommal sector workers. 9002' 02' 02' 02' 9 02'99 099S 99) 99299A ? 3.20. The boom in N s s s construction and other non- tradables offered many good jobs to informal skilled workmen, and this is when they choose to open their businesses.S4 This pattem reverses somewhat going into the crisis of 1995 where there is an increase in the size of the self-employed sector at the same time that relative self-employed eamings are falling. The point here is not to show that the informal sector neverserves as a safety net but instead that most of the time the vast majority of Mexican entrepreneurs want to be in the sector so it should not be treated as inherently inferior. Figure 3.7 Mexico, Relative Sector Size and Earnings 3.21. Perhaps Figure 3.8 (men only) is the most provocative. It 2.3 - 0.75 suggests that in Brazil, 12 - A 02 informal self-employment 2.1 - ( Al-0 -7 rises at the same time that 2 nl\ 7\IA I 0.65 its relative earnings are 1.9 - 0.6 roughly stable or rising. As 1.8 - A a 055 some observers have 1.7 - \ / LA - 0.5argued, during the periods of 1.6 -___0.5 high inflation, self- 1.5 - wage F/SE 0.45 employment is preferred 1.4 -F.Form/SE 0.4 since formal wages are indexed intermittently and &9°° &9i>69i9 099As99As99A99>99 99099 9>9S 0hence lose purchasing power while the self- employed can index daily. With stabilization in 1994, however, a long period of decline in relative formaVinformal incomes is accompanied by a decline in formality. Without question, as the ILO and the Economic Commission on Latin America and the 54 See Maloney (1 99B) - 49 - Caribbean (ECLAC) have noticed, informality is rising across the period. But the critical question is - was it voluntary? Figure 3.8 Brazil, Relative Sector Size and Eamings (Men only) 22 - 15 2 -\AA 14 18- _ 16--V~ O~ 14-~~~~~~~~~~~~~~~~~~~~~1 12 909 I ,, , 008 Table 3.4 Sector of Origin of the Unemployed Argentina' Mexico Sector of Origin All Paid All Paid Informal Self- 18 38 9 40 Employed Salaried Informal 12 26 7 35 Formal Salaried 17 36 5 25 Previously 34 19 Unemployed Previously Out of 6 44 Labor Force School Graduates 13 16 Total 100% 100% 100% 100% 3.22. Women with children do appear to use the informal sector as a form of insurance when the likelihood of a negative income shock threatens the household, but the formal sector is their employer of last resort in Mexico and Argentina. When the unemployment rate increases, women in Mexico show a strong tendency to increase entry to the self-employment sector (Cunningham 2001). The same trend is weakly suggested in Argentina (World Bank 2001). However, it is women with children, i.e. wives and single mothers, who follow this trend, but not women without spouses or children. This again points to the question of whether this sector offers income eamings opportunities to women whose work has a high value in the household, while women without such competing demands prefer the formal sector. 3.23. Another counterintuitive finding appears in Table 3.4. Here panel data from Mexico and Argentina are used to ask the unemployed from what sector they entered unemployment. Surprisingly, 75% of the unemployed in Mexico and 64% in Argentina were informal sector workers before becoming unemployed. Although their unemployment spells are 30% lower than those of - 50 - formal sector workers, it is not the case that they instantly find new informal jobs. Thus, the sector is not simply or even primarily absorbing the unemployed from the formal sector. 3.24. The argument that informal employment can be desirable is most compelling where micro- firms offer remuneration comparable to that eamed in the formal sector, namely among low education workers who unlikely to accumulate much firm specific capital. Observing Mexican informal workers, the anthropologists Selby, Murphy and Lorenzen (1990) confirm that, "On average, the opportunity costs to these workers in terms of foregone eamings in, say, registered blue-collar employment may be quite low."(p. 147) This view is supported statistically by logit analysis of Argentine and Mexican worker movements between sectors that finds that they become less likely to leave formal employment for self-employment, or any other informal sector, as their education level increases. At a global level, Figure 3.9 suggests that as formal sector productivity increases with development, the share of the workforce in self-employment falls. When controlling for productivity (or GDP) and other relevant demographic variables, Latin America does not have an unusually large share of its workforce in informal self-employment: Mexico and Brazil are below average, perhaps suggesting more flexible than average labor markets. Argentina is above average suggesting the reverse. Figure 3.9 Self-Employment vs. Indusrial Productivity 0.46- Peru Bol o _ EIS Col Ven L; Hon Gua Par Uru cn Arg Gre .5i _ Pan Meyurk 18. CR ~~~ ~ ~~~~Bra Kor t Por Spa New Ice Ire A,IBeI | ~~~~~~~~~~~~~~~~~~~A8 Capreft AMP Den Nm 0.058 l Ij_ 6.63 Log of Industrial V.A./ Worker 10.53 3.25. Another way to approach the question is to assume the informal sector exists because of distortions and ask how much remuneration in the formal sector would need to fall to eliminate the -51 - informal sector. Given that own wage labor demand elasticities in manufacturing tend to range from - 0.2 to -0.5,55 a 10 percent increase in the size of the formal sector would require a 20-50 percent decrease in formal sector remuneration. If, as is often the case, the informal sector is half of the workforce, the necessary fall in formal sector wages implied is vast and far above any measured differentials in the two sectors. 3.26. Many of the characteristics associated with informality are natural by-products of the fact that the informal small firm owner or patr6n is fundamentally the owner of a small firm. The industrialized country literature on firm behavior offers two important findings about such firms. First, there is a wide range of sizes among longstanding firms determined by factors such as how efficient or hardworking an entrepreneur is, how well-placed his/her firm is, what the logic of the production process is, etc. This means that the existence of many small firms does not necessarily imply failure of either labor or credit markets. It may be that the reason that 80 percent of micro-firms have only one or two employees and tend to be family based reflects a logic that has roots in the tradition of the family farm, or reflects the sustainable reach of informal contracting relations. This could explain, for instance the finding that only 10 percent of urban Mexican micro-firms report plans for expansion and only 9 percent report that lack of credit is a major business problem.56 3.27. A second finding about small firms everywhere is their extraordinarily high rates of failure. Seeking to explain the US case, the economic theorist Jovanovic (1982) argues that this is due to the fact that entrepreneurs cannot know how good their location is or how good an entrepreneur they are until they actually start the business. Very soon after starting, many find that they are not viable and fail. The sociology literature provides striking confirmation of this insight when Balan, Browning and Jelin (1973) argue that although self-employment is a goal for many Mexican workers: 3.28. Becoming self-employed involves a large risk, especially for those men who had stable and secure jobs. Income is uncertain, in particular during the first perilous years of the business. Often the men lack the financial and administrative skills needed for successful operation of the enterprise. Most men are aware of the fact that many small shops and stores close soon after opening. Some men therefore proceed with much care when they decide to become self-employed. (p 216-217). 3.29. That said, rough calculations from the Mexican micro-enterprise survey suggest that these firms, show high failure rates, but not particulady higher than those in the US. 3.30. If a new view of "formality" is added to this picture, most of the characteristics of the sector can be generated without implying any inferiority or undesired precariousness. Levenson and Maloney (1996) treat "formality" as participation in the numerous formal institutions: federal and local treasuries, govemmental programs such as social security (including pensions and health care), the legal system, the banking system, health inspection, firm censuses, trade organizations, civic 55 Fajnzylber and Maloney (2000) estimate own wage labor demand elasticities for blue collar manufacturing workers to be -0.29 for Chile, -0.49 for Colombia, and -0.18 for Mexico. 56 See Cunningham and Maloney, forthcoming. - 52 - organizations, etc. These, of course, have costs in terms of compliance with legal norms which very small firms can choose to avoid in many lesser developed countries. Small firms are anchored in social networks of family and immediate neighborhood that allow them to enforce implicit contracts and insure against risks. Furthermore, participation in formal institutions is needlessly expensive. However, as firms grow they increasingly need to secure property rights or permit formal contracting mechanisms, pool risk, gain access to credit - all things that become increasingly important as firms grow. De Soto (1989) offers a striking example where informal street vendors in Peru tried to pay their taxes since this would guarantee them some property rights over their pitch and hence offer some security to investments they wanted to make. Statistically, the data from the Mexican micro- enterprise survey suggests that firms do become more formal with age and size.57 3.31. Combining the two characteristics of micro-firms and the notion of formality implies that small firms will have higher costs, are likely to be informal, and will have very high failure rates. Though this corresponds exactly to the standard picture of the stagnant, precarious, unprotected informal worker familiar in the literature, it is, in fact, the opposite. It emerges naturally from the workers trying their luck at entrepreneurship (risk taking), often failing, and not engaging in the formal institutions until they grow. In sum, there may be nothing pathological about the informal sector firm and its existence may be largely unrelated to questions of labor market dualism or even credit market distortions. 3.32. This also may explain the high rates of entry into unemployment found in Table 3.4. If small firms have high mortality rates, both owners and the workers are more likely to find themselves without employment and searching for a new job. 3.33. Gonzalez de la Rocha (1994) provides a compelling explanation of the life cycle pattem discussed earlier based on the household's capacity to manage these risks. Her interviews show that the heads of young families are more likely to be found in manufacturing while heads of uconsolidated" households can move into less onerous but more risky informal service jobs precisely because their mature children provide a hedge against the risk. Further, Balan, Browning and Jelin (1973) argue that it is common for workers contemplating opening a firm to maintain their formal sector job until the micro-enterprise is safely established, perhaps staffed by the wife or mature children, thereby effectively maintaining a diversified portfolio of income streams.58 In sum, for workers desiring to become self-employed, there are informal strategies for managing risk. 57 Levenson and Maloney (1999) 58 'The worker in the Fundidora ....for instance, was extremely cautious when he decided to enter the ranks of the self- employed, after twenty years in the plant. His timing was almost perfect. For five years he had been in the highest position he could expect to attain, so no more mobility could be anticipated within the plant. His youngest son was three years old, and now his wie would be more free of household concems and could help in the store. He had saved a considerable amount of money during his years in the steel plant. In any even, when he first opened a small shoe store, he did not give up his job in the factory until he was sure of the success of his business venture. He moonlighted in his store for three years, his wife tending it when he was not present. During the first year of operation the store lost money, and in the second year it broke even. At the end of the third year it tumed a profit so he decided to leave the factor and devote his full time to the store.' p. 216-217. - 53 - VI 1< - BUIX PROIU)riCi rV (:IT; wwrHO_V 3.34. Lower productivity growth in the informal sector is very likely to be the case, but not because of informality per se. Traditionally, the big gains in productivity occur in the tradables sectors and often (although not exclusively) in manufacturing. This differential in productivity growth underlies Balassa's (1964) theory of why the real exchange rate appreciates as countries develop.59 Informal businesses tend to be concentrated in services, transport and commerce, where productivity gains are hard to achieve: a haircut takes about 20 minutes, whether in New York or in Bogota. However, this does not mean that wages of informal barbers won't rise over time. As productivity rises in the formal tradables sector, workers are slowly pulled from the informal sector, and wages must rise there to keep barbers from moving into formal manufacturing. \V1RI_ - I.SN 'T FORMIA 3.35. Since a small portion of informal workers are found to be affiliated with larger firms,60 clearly sub-contracting relations are not the dominant modality of informal firm behavior. Further, it appears that the path of eamings across time of those who worked in subcontracting relations followed very closely those who were independent. This suggests that common motivations may underlie a workers decision to engage in sub-contracting, and that the sector may not represent inferior work. This is supported by the sociologist Bryan Roberts'(1989) interviews with Guadalajaran workers that suggest that, given the very weak unions and low wages in Mexico, informalization is not primarily a strategy for reducing remuneration and worker control over production: "Market uncertainty and the large number of income opportunities in the city mean that it is useful for both employees and employers to have flexibility in allocating labor."(italics added, p. 48). 3.36. More generally, it is possible that sub-contracting is not so much a way of avoiding labor legislation, as avoiding the inefficiencies in it. The differentials between costs to firms and value to workers of benefits discussed previously offer a benign interpretation of informal subcontracting as a way of reducing firm costs where contract workers gain some of the value of benefits foregone. Some caveats are in order for the end of the 1990s. Here, remuneration to subcontracted workers in Mexico appears to fall relative to self-employed and to formal sector workers suggesting that the quality of employment may have fallen. 59 Productivity growth rises in tradeables pushing up wages in both the tradeables (formal) and non-tradeables (informal) sectors. But since there has been no productivity growth in non-tradeables, their price must rise relative to tradeables prices and the exchange rate defined as Pt/Pn must appreciate. Rogoff and Obstfeld's text book (1996) does a nice job of both laying out the modem view of this as well as providing some empircal evidence 60 Approximately 20 percent of Mexican or Argentine informal sector workers (those who did not eam benefits) are affiliated with 'large" firms, defined as those with more than 15 workers. -54 - 3.37. A larger share of women than men are in the infomnal sector in all countries (Figure 3.10).61 This is partly due a higher propensity for informal wage work, which includes the domestic service category. In Brazil, Chile, and Paraguay, for example, 20 percent of working women are domestic servants. On the one hand, this is a vulnerable job, where employer abuse, isolation, few chances for organization, and even less opportunities for career advancement are a reality. On the other hand, interviews with poor working class Argentine women reveal that it is not uncommon for unskilled women to leave formal sector jobs, that may be better paid, to enter (or return to) domestic service. The non-wage benefits listed by such women centered on fiexibility, in terms of work hours and pay schedules, and security in terms of a source of emergency income, networks to other sources of employment, and a general feeling of connection with the employer (Geldstein 2001). 3.38. The high level of female participation in self-employment has been attributed to discrimination that regulates women to inferior jobs. Cunningham (2001) found that, while it is true that single mothers have high levels of participation in informal self employment and wives are over- represented in the unpaid sector, the fact that single women without children have the highest rate of participation in fonnal jobs of any group, male or female, brngs into question the sex discrmination argument. This does not, of course, preclude discrimination against married women or those with children, if an employer fears that the women may be likely to be absent for work for long periods (Chant 1992), but Cunningham argues that the concentration of these two groups in self-employment may be driven by the need to balance household responsibilities. For women attempting to raise children as well as contribute to household income, the flexibility of the sector may make the sector even more desirable than for, as shown in Table 3.2. 61 The number of men in the informal sector is greater since a higher proportion of men are in the labor force, but a greater share of women, relative to all working women, are in the informal sector compared to a similar proportion for men. - 55 - Figure 3.10 Women's Participation in the informal sector (compared to that of men) Venezuela Paraguay .l Nicaragua Mexico . . El Salvador - - Costa Rica . . Chile 7- :-: -- ' ' , -i, c Argentina , . 0.00 0.50 1.00 1.50 2.00 _~~~~~~~~~~~~ - 3.39. Even if the self-employed benefit from being their own bosses, the mainstream view is that those who work for them are the very worst off of the urban workforce: salaried, yet without benefits. As discussed, their lower wages tells us nothing about whether they are truly worse off. Further, rather than being a stagnant group of disadvantaged workers, the sector appears to serve primarily as the principal, although not exclusive, port of entry for young, poorly educated workers into paid employment. The mean age of informal wage employees in Mexico, for example, is 29 years old, which is five years below that of formal sector and contract workers and 14 years below that of the self-employed; the pattern is broadly similar in Argentina. Mexican and Argentine data show a high degree of mobility between school, unpaid work and, to a lesser extent, unemployment that suggests a pool of workers not yet tracked into regular employment. One scenario is that while in school and just after leaving school, many students help out at the family business, and eventually get paid. Once becoming salaried informal, workers spend an average of two years doing this before moving on to other paid work. The brevity of tenure is the same as that found in Brazil by Sedlacek et. al.(1 995) and is similar to the U.S. where the median tenure for young workers 16 to 24 years of age is only 1.4 years and 25 to 34, 3.4 years.62 Furthermore, if Hemmer and Mannel (1989) are correct that in many countries informal small enterprises train more apprentices and workers than the formal education system and the mostly government job-training schemes together, the informal sector experience may constitute continued schooling. Even if this pattem of graduation from school to unpaid to informal salaried work to other modes of work represents the queuing that the dualistic literature might predict, the wait in informal salaried work is not long. 62 BLS News, USDL 92-386 for 1991. - 56 - 3.40. The sector need not be precarious, for two reasons. First, Balan et. al. (1973) argue that this period of life for young workers is one of 'shopping around" and trying out various possible life choices, and hence they will show short tenure. Second, the vast majority of informal salaried workers are employed by informal micro-firms which, as discussed earlier, have higher turnover rates. 3.41. This paper uses empirical evidence to challenge many of the assumptions about the informal sector. The primary finding is that although the sector is heterogeneous, a substantial portion of it is a result of small firm dynamics, total job valuation by employees, and weak institutions rather than to the labor market rigidities that create an above-marketing clearing wage and force half the labor force into inferior jobs. 3.42. In many Latin American countries, much or most of the sector represents a healthy, voluntary small firm sector that should be treated as such. Since most of the informal sector is composed of small firm owners, they are subject to the same pattems of birth, success, and death as are small firms in any developed country in the world. The observed "precariousness" of the sector is simply the reflection of being bom a small firm. This is not to say that in periods of great recession or in countries with very distortionary minimum wages and regulation that the traditional dualistic view would not be a more appropriate model, but in periods that are not extreme, the dualistic view falls short. Thus, policies to support small firms should be grounded in a solid understanding of the dynamics of these enterprises and of barriers to their growth and absorption of labor. To further bring small firms into the formal system, whether for fiscal or other purposes, the government should increase the value of the public goods enjoyed by firms, such that they are more willing to enroll in and pay into the system. 3.43. The paper also argued that those who work in the informal sector ofteri choose to be there due to constraints that make formal sector work less desirable or due to a high valuation of the non- wage benefits of informal sector work. Wages of individuals who own or work in these firms may not be a good measure of the quality of jobs in the sector since workers have many factors to take into consideration when accepting a job, including time responsibilities outsides the market, independence, pride in being one's own boss, and training opportunities, to name a few. Again, this should not be interpreted as claiming that all informal sector workers are voluntarily in the sector, but a large proportion are. As recommended for the small firms, policy to bring informal sector workers into the formal sector, whether for social protection or other motives, should be grounded in an understanding of the reasons that entrepreneurs choose not to be formal. Again, the quality of the benefits need be increased such that workers prefer to pay into the system than to evade. 3.44. These arguments do not imply that there is no need for reform of labor legislation, but instead, improvement in the quality of public goods and institutions, rather than an elimination of costs to employers, is the more appropriate target. The traditional emphasis on rigid wages driving segmentation is probably not correct. However, informality may arise partially as a response to inefficiencies in the provision of medical benefits or pensions, promotion systems not based on merit, education systems that do not train students for the labor market, or other distortions that make being paid in cash informally more desirable. The recent reforms in Chile, Colombia and Mexico that have sought to bring benefits in line with the implicit taxes that workers pay, for instance individual - 57 - accounts for retirement pay, are important steps to reducing incentives to being informal. More fundamentally, informal employment in firms of relatively low technology and capital intensity can only be attractive if the overall level of labor productivity in the formal sector is low also. To the degree that current legislation impedes investment in physical or human capital, or prevents the efficient organization and operation of firms, it perpetuates the low levels of productivity throughout the economy. 3.45. Finally, the poverty found in among informal sector workers is likely to be a function of low levels of human capital, such that whether the worker is formal or informal is largely incidental. A worker with few skills that would be rewarded in the formal sector may prefer to be independent: S/he may prefer being the master of a lowly repair shop to endlessly repeating assembly tasks in a formal maquila. Neither job will lead to an exit from poverty, but the informal option may actually offer a measure of dignity and autonomy that the formal job does not - 58 - References BaIln, J. and H.L. Browning, E. Jelin (1973), Men in a Developing Society, Institute of Latin American Studies, U.T. Austin. Balassa, B (1964). "The Purchasing Power Parity Doctrine: A Reappraisal" Joumal of Political Economy, 72:584-96. Behar, J., (1988), Trade and Employment in Mexico, Swedish Institute for Social Research, Stockholm. Bell, L., (1994) "The Impact of Minimum Wages in Mexico and Colombia" Labor Market Workshop, The World Bank. Beneria, L.(1989) Subcontracting and Employment Dynamics in Mexico City, in Portes et al., The Informal Economy, Johns Hopkins. Chant, Sylvia (1991) Women and Survival in Mexican Cities (Manchester University Press: Manchester). Cunningham, W. (2001) "Breadwinner versus Caregiver: Labor Force Participation and Sectoral choice over the Mexican Business Cycle," in Correia, Maria and Elizabeth Katz The Economics of Gender in Mexico (World Bank: Washington, DC). Cunningham, W. and W. Maloney (forthcoming), 'Heterogeneity in the Mexican Micro-enterprise Sector," Economic Development and Cultural Change. Davila Capalleja, E. (1994) "Regulaciones laborales y mercado de trabajo en Mexico." in Gustavo Marquez ed., Regulacion del mercado de trabajo en America Latina. San Francisco:Centro Internacional para el Desarrollo Economico. Fajnzylber, P. and W. Maloney, "Labor Demand Elasticities and Inference, Dynamic Panel Estimates from Colombia, Chile and Mexico" mimeo, World Bank. Fields G. S. (1990), 'Labor Market Modelling and the Urban Informal Sector: Theory and Evidence," in OECD, The Informal Sector Revisited, Paris. Geldstein, Rosa (2000) "Non-Labor Market Coping Strategies in Argentina", LCSPG/World Bank, draft. Gonzalez de la Rocha, M. (1994) The Resources of Poverty: Women and Survival in a Mexican City, Blackwell Publishers, Cambridge, MA. Gregory, P. (1986), The Myth of Market Failure: Employment and the Labor Market in Mexico, Baltimore: World Bank and Johns Hopkins. Harris, J. R. and M. P. Todaro (1970), "Migration, Unemployment, and Development: A Two Sector Analysis," American Economic Review, 60:1, 126-142. Hemmer, H. and C. Mannel (1989), "On the Economic Analysis of the Urban Informal Sector," World Development, 17:10. Krebs, T and W.F. Maloney "Quitting and Labor Tumover: Micro-economic Evidence and Macro- economic Consequences." IBRD Working Paper 2068, Latin America and the Caribbean - 59 - Region, Poverty Reduction and Economic Management Unit, World Bank. Wshington, D.C. Processed. Levenson, A.R. and W.F. Maloney (1996), 'Modeling the Informal Sector: Theory and Empirical Evidence from Mexico." Mimeo, University of Illinois and the Milken Institute. Loayza, N.V. (1994) "Labor Regulations and the Informal Economy," Policy Research Working Paper 1335, World Bank, August. Lopez-Castano, H. (1990) "Inestabilidad Laboral y Ciclo de Vida en Colombia," Coyuntura Economica, 20:1, p 173-191. Maloney, William F. (1999). "Does Informality Imply Segmentation in Urban Labor Markets? Evidence from Sectoral Transitions in Mexico, World Bank Economic Review, 13: 275-302. . (1998). "The Structure of Labor Markets in Developing Countries: Time Series Evidence on Competing Views." Working Paper 1940. World Bank, Washington, D.C. Processed. (Forthcoming) "Self-Employment and Labor Tumover: Cross-Country Evidence. Proceedings, World Bank Economic Review. World Bank, Washington, D.C. (2000) Minimum Wages in Latin America, mimeo, IBRD. and E. Ribeiro (1999) 'Efficiency wage and Union Effects in labor Demand and Wage Structure in Mexico" Working paper 2131, World Bank, Washington, D.C. and P. Aroca Gonzalez(1999) "Logit Analysis in a Rotating Panel Context and an Application to Self-Employment Decisions" IBRD working paper 2069. Marquez, C. and J. Ros (1990), "Segmentacion del Mercado de Trabajo y Desarrollo Economico en Mexico," El Trimestre Economico. Abril-Junio. Peattie, L. (1982), "What is to be Done with the 'Informal Sector?: A Case Study of Shoe Manufacturers in Colombia" in H. Safa ed, Towards a Political Economy of Urbanization in Third World Countries, Delhi:Oxford University. Peattie, L. (1987) An Idea in Good Currency and How It Grew:The Informal Sector," World Development, 15:7 851-860. Portes, R. and R. Schauffler (1993), "Competing Perspective on the Latin American Informal Sector,' Population and Development Review, 19:1, 33-59 Secretaria del Trabajo y Prevision Social (1993), El Sector Informal en Mexico, no 2. Sedlacek, G.L. and R. Paes de Barros and S. Varandas (1995), "Segmentacao e Mobilidade no Mercado de Trabalho Brasilerio," Unpublished Manuscript. Selby, H. A., A. D. Murphy, and S.A. Lorenzen (1990), The Mexican Urban Household, Organizing for Self-Defense, University of Texas Press Austin de Soto, H. de (1989), The Other Path, Harper and Row, New York. Tokman, V.E. (1982) "Unequal Development and the Absorption of Labour" in CEPAL Review, No. 17, Santiago, 121-133. Tokman, V.E. (1992) Beyond Regulation, The Informal Economy in Latin America, Lynne Rienner, Boulder. - 60 - Tumham D. and D. Er6cal (1990), 'Unemployment in Developing Countries, New Light on an Old Problem,' Technical Paper, OECD Development Centre. World Bank (2001), Household Risk, Self-Insurance and Coping Strategies in Urban Argentina Report No. 22426-AR (World Bank: Washington, DC), green cover - 61 - 4. UNIONS AND THE LABOR MARKET IN BRAZGL Prepared by Jorge Saba Arbache ____~~~~~~01011 1-MNENK_~ 4.1. Although labor unions are considered one of the most important institutions of modern capitalism (Freeman, 2000), the literature shows that their mode of operation and impact on the economy vary from country to country. In France and Spain, for example, although union density is low, collective bargaining processes determine wages for much of the work force. In China, on the other hand, union density is high, but their effects on wages are believed to be modest. In Nordic countries, both union density and collective bargaining coverage are very high. The impact of labor unions on the economy depends not only on the unions themselves, but also on other institutions that complement or even determine the framework within which they act; these are what explain individual union action strategies at different times and places. Labor codes and other institutions that regulate labor relations and collective bargaining, along with anti-union policies such as the reforms introduced by Margaret Thatcher in Great Britain during the 1980s, are examples of such institutions. 4.2. Although there is a wide variety of systems, institutions, and union action strategies in the different countries, an important stylized fact in the literature is that labor unions reduce wage dispersion through collective bargaining processes. This has led Metcalf et al. (2000) to refer to them as a 'sword of justice." According to Freeman (2000), this phenomenon is actually more widespread than the more thoroughly researched effect of unions on wages, namely their ability to raise the relative wages of their members. The literature shows that: (i) the distribution of wages among unionized workers and/or those covered by collective bargaining is more concentrated than the wage distribution for other workers, even when demographic and productive characteristics are controlled for in the corresponding regressions; and (ii) collective bargaining diminishes the importance of merit in wage formation, thereby narrowing the wage spread between jobs (Freeman, 1980,1992; Hirsch 1982; Card, 1992; DiNardo et al., 1995; Gosling and Machin, 1995; Blau and Kahn, 1996; DiNardo et al. 1997; Metcalf et al., 2000, among others). 4.3. Estimates of the impact of unions on the labor market in Brazil are still very preliminary. The few results available include those obtained by Arbache (1999), who investigates the wage and income-distribution effects of unions among male manufacturing workers; Arbache and Carneiro (1999), who examine the importance and effects of unions on collective bargaining; Arbache (2000), who studies the effects of trade liberalization on unions and collective bargaining; and Amorim (2000), who analyzes the relationship between unions and indirect pay. Little is known about the effect of unions on employment, productivity or labor market rigidity, and especially on wages and income distribution. Knowledge of the impact of unions on pay dispersion is particularly important for Brazil, given its highly unequal income distribution. The study of unions in the economy is therefore challenging, because collective bargaining and union operations are heavily regulated by the law, and history records far-reaching State involvement in such institutions. Research into unions in Brazil may involve issues beyond the confines of economics, pertaining to law, sociology, political science and possibly other areas as well. If the legal framework that regulates labor relations is a decisive -62 - factor in the functioning of unions and collective bargaining processes, then knowing this could help us to understand the effects of unions on the overall economy. 4.4. This paper seeks to answer the following questions: (i) How do employment and union laws affect collective bargaining processes? (ii) Do unions affect wage formation and income distribution? (iii) Do unions increase the rigidity of the labor market? We conclude by making some suggestions for enhancing collective bargaining and unions in Brazil. 4.5. The paper is organized as follows: Section 2 describes labor laws and union legislation in Brazil. Section 3 describes the characteristics of unionized and non-union workers, union density, and the determinants of unionization. Section 4 investigates the effects of unions on wages and income distribution. Section 5 discusses the results obtained and seeks to demonstrate the effects of employment and union laws on collective bargaining processes and union behavior. Section 6 offers some recommendations for improving labor relations in Brazil. 1--- 1 4.6. The institutions that regulate labor relations in Brazil have a huge impact on collective bargaining, given the way they are involved in employment contracts, labor disputes and union activity, and the intensity of that involvement. This section provides a historical overview and describes the characteristics of the main institutions governing the country's labor relations. As we shall see, such institutions contribute to adversarial labor relations and curb the development of collective bargaining. We also show that the nature and functioning of those institutions have potential effects on income distribution and on macroeconomic stability. Labor Laws and Labor Courts 4.7. Relations between workers and employers in Brazil are governed by the Consolidated Labor Laws [Consolidagao das Leis do Trabalho] (CLT). The CLT law is extremely wide ranging and detailed, regulating the most varied aspects of labor relations. The CLT was first introduced in 1943, during the administration of Getulio Vargas, in order to consolidate the labor laws existing at that time. One of its main aims was to create a system to protect workers from exploitation by employers, and to harmonize labor relations with a view to avoiding direct disputes between the parties. Camargo (2001) points out that the CLT was premised on the existence of asymmetric power relations between capital and labor, and that the role of the law was to protect and regulate workers' interests. The all-embracing and paternalistic nature of the law created an atmosphere that was not conducive to the development of collective bargaining. The CLT also created unstable labor relations, generating disequilibrium by providing excessive protection for workers. The law requires disputes to be settled in labor tribunals rather than in the companies involved, so little space is left for direct negotiations between employers and employees; and it discourages the development of a cooperative relationship between the two parties. 4.8. As Camargo (2001) shows, an important feature of Brazil's labor legislation is the coexistence of individual and collective employment contracts. Individual contracts (i.e. contracts involving employment record cards) are entered into between the company and the worker and deal - 63 - with issues such as working conditions and wages. Collective contracts, on the other hand, are concluded between the employer and the workers' union, or between the employers' association and the union, and cover minimum working conditions and minimum wages, among other issues. Workers without an individual contract are not covered by collective contracts in their job category. Legally, the results of collective bargaining processes are extended to all workers and companies in that occupation or industry, respectively, even if the workers or companies involved are not members of the respective unions negotiating the agreement. This provision accords great importance to unions in labor relations. 4.9. In Brazil, the justice system is required to rule on disputes related to compliance with labor laws, and on those relating to individual and collective employment contracts. It is also called upon to promote conciliation and arbitration in collective bargaining processes. If negotiations between workers and employers break down, the labor tribunal is required to rule on the dispute. If the impasse is the result of a failure to comply with the legislation, the tribunal will merely apply the law in reaching its decision. But if the impasse stems from a lack of agreement on other issues, the tribunal may decide according to the judges' point of view, possibly even using political criteria in doing so (Camargo, 2001). The decision of the labor court must be complied with by the parties. Accordingly, labor tribunals have regulatory authority that affords them great power within the collective bargaining and labor relations framework in Brazil. The courts offer many incentives for free-rider behavior by workers, since employers have to disprove allegations of legal and contractual non-compliance, and it is they who have to defray most of the legal costs involved. Thus, the paternalism of the law, together with the tremendous power and involvement of the justice system in labor disputes, act as obstacles to the modemization of labor relations by inhibiting cooperation between the parties and fostering disputes between workers and employers. Union Organization and Collective Bargaining 4.10. In the legal arrangements originally established by the CLT, unions were responsible for: (i) contributing to the harmonization of relations between capital and labor; and (ii) helping to implement the Government's economic policies. Thus, the principles on which union and labor legislation and organization were founded gave labor unions powerful and close links with the State, which compromised their action in support of workers' interests. 4.11. Under the CLT, labor unions are organized by occupational category, but employers associations are organized by economic sector. Job categories and economic sectors are defined by the Ministry of Labor, on the basis of similar characteristics. Up to 1988, different occupations and economic categories were prohibited from grouping together in a single union. This restriction was lifted in the 1988 Constitution, when the formation of nationwide unions and union confederations was authorized. 4.12. In order to control the unions, the law instituted monopoly of representation and the union levy, together with mandatory extension of the results of collective bargaining processes to all workers, including non-union members. Once the union has been recognized by the Ministry of Labor, it then has a monopoly in its predefined geographic jurisdiction. The smallest regional base is the municipio, but unions can also have regional, state, or even national jurisdiction. All collective bargaining processes in a given category have to be carried out with participation from the union holding the monopoly representation in the geographic area concerned. - 64 - 4.13. Although union membership is not compulsory, workers and employers are required to collect a union levy each year; 60 percent of this is passed on to the respective union by the Ministry of Labor, which is the body responsible for collecting it. The remaining funds are divided between the Ministry of Labor and the federation and confederation for the occupational or economic category concemed. By law, funds transferred to labor unions must be used exclusively for purposes such as recreation, social assistance, education, and consumer cooperatives, but never to finance political activities, collective bargaining processes, or strike funds. Only funds obtained through voluntary contributions can be used for such purposes. 4.14. Up to 1988, the Ministry of Labor could intervene in labor unions for reasons such as misuse of the union levy, or for calling unauthorized strikes or lock-outs. The law even enabled the authorities to close down a union if it were judged to have impeded the implementation of government economic policy. The Constitution of 1988 abolished this provision. 4.15. Collective bargaining is compulsory and must take place once a year, during the "base-date' period, in which the workers' union and the employer's organization or individual company negotiate wages and other employment issues. Base dates vary between occupations and categories, and this results in negotiations being spread throughout the year. It is possible, however, for different occupations to sign agreements on the same day in the same company or economic category, thereby resulting in a collective agreement covering a large proportion of the workers in a given industry. In bargaining processes between employer organizations and worker unions, the result is known as a collective contract (convengao coletiva); if bargaining takes place between the workers' union and a single company, the outcome is known as a collective agreement (acordo coletivo). Up to 1988, any contract or agreement contrary to the Govemment's overall economic or wages policy was susceptible to legal rescission. Since then, unions have been free to negotiate agreements without the threat of govemment interference. 4.16. Monopoly representation, the union levy, and the extension of collective bargaining results to all workers gave great power to union leaders, but fostered a lack of responsibility among them with respect to workers' interests and situations. There were two reasons for this: firstly, union finances are guaranteed by an assured income from the union levy, which means union leaders do not have to attract members to finance union activities. Secondly, mandatory union participation in collective bargaining processes in their area of jurisdiction, together with compulsory extension of the results to all workers, gives union leaders great power in labor relations, whether union density is high or low. Consequently, as history shows, union leaders have not always been concerned with attracting and maintaining workers affiliated to their unions, or with working on behalf of the most obvious interests in their category. In many cases, union legislation has spawned bureaucrats that have little interest in workers' situations, but who maintain close relations with the Government in order to maintain their power in the unions. 4.17. Restrictions on the use of the proceeds of the union levy, compounded by the union bureaucracy's relations with the Government, have tumed unions into bodies more concemed with assistance and recreation, but with little interest in promoting and developing collective bargaining processes. Moreover, the monopolistic nature of unions has resulted in collective bargaining taking place on a fragmented basis with no inter-union coordination across the different occupations and economic categories. - 65 - 4.18. When the military dictatorship took power in 1964, amendments were made to the laws goveming union organization and pay-setting processes. One of the most important changes was the introduction of a hardline law on strikes, aimed at regulating the right to strike and creating new criteria for strikes to be considered legal. Highly restrictive rules inhibited the outbreak of strikes and restricted union action; in practice, the changes were intended to repress movements in support of claims of any type. These and other measures, such as prison for union leaders, caused the union movement to stagnate until the late 1970s. 4.19. In the 1960s, a law on wage adjustments was also introduced with the aim of centralizing and regulating the wage-setting process. This gave the Govemment power to determine the minimum rate of increase for all wages in the economy. The formula and frequency of pay raises changed several times between the mid-1 960s when the law was introduced and the end of the 1980s, but it tended to set pay increases below the rate of inflation and even below productivity gains. The lag in pay raises also tended to generate wage erosion because of the accelerating inflation of that period. The law on pay increases was created as an economic policy tool, with wages in practice being used as adjustment variables in stabilization policies. Any pay increases above those stipulated by law had to be negotiated individually between the company and its workers. Given the weakness of labor unions and the centralization of pay-setting processes, the law not only operated as an instrument of wage control, but also inhibited collective bargaining from the 1960s until the mid-1 990s, when it was shelved. 4.20. A key feature of the period in which the wage law was in force was that, while wages could theoretically be negotiated between workers and employers, in practice the latter would invoke "collective disagreement" during the bargaining process, which meant referring the dispute to the labor tribunal for settlement. As wages were set by law, the courts did no more than apply the corresponding legislation, overriding any attempt to negotiate increases above the correction set by the Govemment. Thus, although wages were supposed to be negotiable, pay raises in practice adhered to the rates of increase set by the Government, leaving little or no space for collective bargaining over pay. As a result, the tremendous sway of the Government over wage setting for more than twenty years seriously undermined the development and modemization of collective bargaining processes in the country. 4.21. Control of wages and repression of the union movement prompted major wage-claims in the late 1970s, and gave rise to a new phase in the organization of the union movement.. Workers sought greater autonomy, freedom of action, and changes in wage policy. The reorganization of the movement was particularly significant in the wealthiest and most industrialized areas such as the ABCD region in Sao Paulo, where many multinationals and large local companies are based. The perception held by workers that companies in the region were extracting monopoly rents possibly contributed to the emergence of this movement. As shown by Booth (1995) and Nickell et al. (I1994), unions tend to demand pay raises where there are quasi-rents to be shared out. In such a context, the law on wage increases placed a straitjacket on pay in the most developed regions and m6st dynamic sectors of the economy, where wage increases above the law were supposedly possible; this led to demonstrations that later culminated in the 'new unionism" movement (see Camargo, 2001; and Amadeo and Camargo, 1989). Following a series of violent strikes in the ABCD region during the late 1970s, the frequency of wage hikes was reduced and the rate of productivity increase to be passed on to wages was left for collective bargaining processes to calculate. - 66 - 4.22. With this additional flexibility, collective bargaining became more important, and the rates of increase determined by law became the floor and no longer the ceiling for pay raises, as it was now possible to negotiate over productivity. Unions now started to bargain over wage increases above the floor set by the law, through negotiations at the company level. Success in bargaining, however, depended on industry characteristics, occupational category, and region. Unions in companies or industres with greater market power, or those that were more concentrated, had stronger bargaining power. Unions in more competitive sectors had little chance of bargaining over productivity, however, and contracts were negotiated in a more individual and less collective fashion. In practice, as pointed out by Camargo (2001), the bargaining system was hybrid: on the one hand it was partly centralized, since unions in some sectors negotiated wages collectively; on the other hand, the weakness of unions elsewhere meant that wages were bargained in a decentralized fashion, with actual pay raises tending to match the adjustments set by law. This system of labor relations and collective bargaining led to scant cooperation between unions, as bargaining power between them varied widely. Unions in companies or sectors that were more inclined to grant pay increases had little interest in the possibility of centralized negotiations. 4.23. Against this backdrop, decentralization and the fragmentation of collective bargaining led to different unions obtaining different rates of pay increase. Wage indexation through the wage-hike law, compounded by fragmentation and scant incentive to coordinate collective bargaining processes, and supported by protectionism, meant that unions had little incentive to build the employment/wage trade-off into their objective functions, thereby raising the costs of stabilization policies. Two potential consequences of this situation were: (i) an increase in wage dispersion; and (ii) problems with macroeconomic coordination. 4.24. Promulgation of the new Constitution led to the overhaul of legislation on collective bargaining, wages, strikes, and unions. Collective bargaining began to be encouraged, the concept of "illegal strike" ceased to exist, prohibition of worker organization at national level was lifted, and unionization of civil servants was allowed, along with other changes making union organization more democratic. Nonetheless, these changes were insufficient to reduce the greatest obstacles to implementing collective bargaining processes in Brazil, since the basic rules governing employment contracts, the powers of labor tribunals, and union organization remained virtually intact. 4.25. Our research on unions draws on data from the National Household Sample Survey (PNAD) conducted by the Brazilian lnstftute of Geography and Statistics (IBGE) for 1986, 1992, 1993, and 1995 to 1999, years for which the "unionization" variable is available. Unions are analyzed by industry at the IBGE two-digit level, which covers economic activities ranging from agriculture and manufacturing industry to services and public administration. 4.26. The data were filtered for individuals of 18 to 65 years of age, economically active, and eaming a positive wage. We took pay eamed in the individual's main activity, and transformed this into a real hourly wage. Although PNAD data are acknowledged to be of high quality, 0.1 percent of cases were suppressed from each end of the real wage distribution as a precaution to avoid outliers arising from measurement errors. As a result of the filter, we have at least 100,000 individuals for each year, about 20 percent of whom are union members. - 67 - 4.27. This section discusses some of the main socioeconomic characteristics of unionized workers and their non-union counterparts. The aim is to determine who the union members are, and where they are located, along with information on union density and whether there is a correlation between socioeconomic profile and union membership. Characteristics of Unionized Workers 4.28. Table 4.1 shows that the vast majority of unionized workers hold an employment record card (carteira de trabalho assinada). While almost 90 percent of unionized workers benefit from the guarantees afforded by a formal work contract, only about 54 percent of non-union workers have this card. As only formal workers are covered by collective bargaining, one might expect that all unionized workers would have a formal contract. The shortfall could arise from measurement error, or unionized workers migrating from formal to informal jobs for some reason. The figures suggest that: (i) workers holding an employment record card have a greater incentive to unionize; (ii) unions are active in guaranteeing workers' rights, thereby reducing the level of informal contracting in the industries where they are strongest; and (iii) unions tend to be concentrated in industries where there is a higher percentage of formal contracting. 4.29. These three hypotheses help to explain the higher level of formalization among unionized workers. It is reasonable to assume that formal workers have a greater incentive to join unions, because the costs of union membership for them are less than the benefits. Informal workers do not have incentives to join unions, however, as they do not benefit from the results of collective bargaining. It is also reasonable to assume that the unions' main demands include upholding workers' rights, which results in the high percentage of formalization in unionized sectors. It is well known that unions are better organized in industry and more sophisticated services, where a lack of formalization is rare. 4.30. Figure 4.1 shows a positive relation between union density and the degree of formalization in the labor force at the industry level, thereby supporting hypotheses (ii) and (iii). It remains to be seen whether there is a causal relationship between the two variables and if so in what direction: i.e. whether greater union activism produces a high level of formalization, or whether greater formalization facilitates union activism. 4.31. There seems to be a correlation between union membership and status in the family. About 62 percent of unionized workers are heads of households, compared to just 52 percent of non- unionists. These figures suggest that union membership is a characteristic of more mature workers, and that young people participate in unions less. In fact, the average age of union members is higher than that of non-unionists. The mean age of union members also rises over time proportionately more than the average age of non-union workers. 4.32. There also seems to be a correlation between union membership and race: about 61 percent of unionized workers are white, compared to 52 percent of non-union members. In addition, the proportion of men among unionized workers is higher than among non-union members. It can therefore be concluded that either: (i) minorities have fewer incentives to join unions; or (ii) unions are concentrated in sectors or regions where these groups are under-represented. -68 - 4.33. The distribution of education among unionized and non-union workers is the variable that most clearly distinguishes the two groups. While at least 60 percent of non-union workers are in the first three education brackets (ranging from no education at all to some primary education), at least 50 percent of union members are in the three highest education brackets - covering primary school completed to higher education. This result seems even more remarkable when we analyze the distribution of workers with higher education. Whereas at least 13.7 percent of unionized workers have completed a higher education course, no more than 5.4 percent of non-union workers have done so. If we group the two higher education brackets together (i.e. secondary and tertiary education both completed) we see that in 1999, for example, 47 percent of unionized workers had this level of schooling, compared to just 26.6 percent of their non-union counterparts. Thus, unlike the situation in developed countries (Freeman, 1982; Booth, 1995; Metcalf, 2000), education is a key characteristic of unionized workers in Brazil. 4.34. The distribution of education between unionized and non-union workers suggests that: (i) more educated workers have a greater incentive to join unions; and/or (ii) unions are concentrated in sectors that use more highly skilled labor.6 4.35. The distribution of workers by size of establishment (a variable not shown in Table 4.1) shows that nearly 85 percent of unionized workers are in establishments with more than 11 employees, while 58 percent of non-union members work in establishments with 11 employees or less.f4 4.36. Table 4.2 shows the distribution of unionized workers by region and job category. The proportion of unionized workers in rural and urban areas is quite similar, while unionization is greater in metropolitan than in non-metropolitan areas. The regional distribution shows that the proportion of unionized workers in the southeast - where most of the country's industry and its financial center are located - is similar to that in other regions. The proportions of unionized workers in the north, north- east, and central-westem regions, are in all cases less than in the other regions combined. In contrast, the level of unionization in the south is considerably higher than in the rest of the country, which suggests that the union movement in this area is relatively more active and has stronger bargaining power. 4.37. The lower panel of Table 4.2 shows the distribution of unionized workers by occupational category. Unionization is a clear characteristic of managers, skilled production workers, office workers, and, in particular, professionals. Levels of union membership are substantially above the average in each of these job categories. Among professionals, a group that encompasses all higher- level staff, at least 42 percent are union members; this is much more than the average for other occupations. One possible explanation for this is the regulation and legal control maintained over the practice of higher-level professions through Regional Councils (for example, the Conselho Regional 63 The rising proportion of educated workers who belong to unions may be the result of growth in the relative demand for workers with higher levels of schooling during the 1990s (Green et al., 2001). 64 For this calculation, we used the PNAD variable that classifies the size of establishments in which individuals work as aup to 11 employees" and wover 11 employees. - 69 - de Economia, Ordem dos Advogados, Conselho Regional de Medicina, etc.), supported by compulsory membership of such bodies as a precondition for professional practice. Often, these bodies operate in traditional areas of union activity, for example regulating fees and establishing wage floors. Nonetheless, it is not uncommon to find unions operating in the same location as the these professional councils, and this facilitates and publicizes union activities, promotes union membership, and ehcourages integration between the two bodies, leading even to some unification of agendas. 4.38. About one-third of managers, technical staff, qualified production workers, and office staff are union members, compared to just 12 percent of unskilled production and manual workers - a group covering over half of the entire workforce. Unionization is even lower among sales staff, at around 10 percent.65 This distribution of union membership by occupational group is surprising, as the literature portrays union membership as a characteristic of low-skilled and production-line workers; moreover, workers with higher qualifications presumably have incentives to demand wages related to productivity, rather than based on pre-established pay scales. The occupational distribution of unionized workers in Brazil suggests that unions have the strongest presence in industries that concentrate higher qualified workers, including executives and office staff. 4.39. Table 4.1 and Table 4.2 therefore suggest that unionization in Brazil is a characteristic of more educated workers, holding the best jobs and working in large establishments; they also tend to be white, male, and heads of households; in addition union members are likely to be in metropolitan regions and to hold an employment record card. This profile suggests that the typical union member is very different from the average Brazilian worker, whose most common characteristic is a low skill level. 4.40. Bearing in mind the different characteristics between unionized workers and their non-union counterparts, it is reasonable to assume socioeconomic variables affect the probability of union membership. To investigate this issue, we estimated models that identified the contribution of individual characteristics to the likelihood of a worker being a member of a union. Binomial logistic regressions were estimated using the union-membership indicator as the dependent variable. The model's explanatory variables were education, experience, experience squared, gender, head of household, urban/rural, metropolitan region, geographic region, and industry classification at the IBGE 1-digit level. The results, given in Table 4.3, show the following: 4.41. The likelihood of union membership rises substantially with schooling; in 1999 for example, an individual who had completed a higher education course was 672 percent more likely than an illiterate worker to be a union member; * Experience raises the probability of union membership; in 1999 this increased by 2.6 percent with each additional year of experience; * Men are more likely to be union members than women; 65 Not surprisingly, the distribution of occupations between unionized and non-union workers matches the distribution of education. - 70 - o Heads of households are more likely to be members of unions than other family members; o Workers in urban areas are less likely to be unionized than those from rural areas; o Workers in metropolitan areas are more likely to be union members; o Workers from the south-east, north, and central-western regions are less likely to be union members than workers from the north-east; only workers from the southem region are more likely to be union members than those in the north-east; o Economic activity seems to have a major influence on the probability of union membership, as the estimates vary substantially between sectors. Workers in manufacturing industry, other industrial activities, transport and communication services, and even agriculture show high probabilities of union membership; while workers in civil construction and services are unlikely to be unionized. 4.42. The results therefore suggest that union membership in Brazil is closely related to workers' qualifications; and being male, head of household, holder an employment record card, and resident in a metropolitan region, are characteristics that also help explain the likelihood of union membership. Union Density 4.43. Table 4.4 (bottom line) and Figure 4.2 show the proportion of unionized workers weighted by the share of each industry in total employment. The figures indicate that union density fell uniformly from 22.2 percent to 17.4 percent between 1986 and 1999, a 27 percent decline during the period. This decline in unionization may be associated with the changes that have taken place in the economy and in the labor market over the last two decades, and particularly in the 1990s when trade liberalization policies were introduced. Other potential explanatory factors include market deregulation, including the labor market, and the privatization of state-owned companies (a sector where unionization was traditionally active), compounded by deep recession in 1990-1992 and a consequent rise in the open unemployment rate. These changes increased competition, undermined profits, and forced companies to adopt cost-cutting measures, potentially affecting workers through a decrease in both the number of jobs and the quasi-rents available to be shared out. A drop in union density was an expected consequence of this situation, resulting from higher unemployment and less chance of success in wage claims. Hay (1999) shows that while trade liberalization led to an increase in productivity, the profits of Brazilian companies declined. Arbache (2000) shows that trade liberalization eroded the union wage premium, especially in sectors most affected by imports. 4.44. In order to analyze the level of unionization in Brazil, union density was compared with that of other countries for 1994.66 Blau and Kahn (1999, table 2) report that average union density in a group of developed countries excluding the United States was 42 percent that year, whereas the figure in Brazil was 19.5 percent. Density in Brazil only exceeded the figures for the United States (16 percent) and France (9 percent). Differences in union density between countries may be a reflection 66 As there was no PNAD for 1994, we use the average union density for 1993 and 1995 to compare with density in developed countries. -71 - of the institutions governing unions and collective bargaining processes. In Brazil, legal extension of the results of collective bargaining to all formal workers in the category, even non-union members, must partly explain the relatively low rate of unionization, because it encourages free-rider behavior. In fact, Nickell and Layard (1999) show that union density is relatively low in countries where the results of collective bargaining are extended to non-union workers, as in France, Spain and Holland.67 4.45. Table 4.5 shows union density at the two-digit industry level. Industries are classified into three union-density groups, using the following procedure: the average density of industry i in the period 1986-1999 was compared with the average density economy-wide during the same period, namely 19.2 percent. On the basis of this comparison, the union density distribution at the industry level was constructed, and this was then split into three groups as follows: * Up to one standard deviation - average or low union density; * Between one and two standard deviations - high union density; * Three or more standard deviations - very high union density. 4.46. The medium or low union-density group included industries with unionization coefficients ranging from 3 percent to 33 percent; so dispersion in that group is very wide. Industries with densities less than or equal to the economy average include civil construction, commerce, agriculture and clothing, among industries traditionally characterized by low levels of technology and capital and low market concentration. Industries with union densities between 19.2 percent and 33 percent, include those with a high technology level, such as the pharmaceutical and chemical industries, together with various activities in the service sector. This interval also includes public administration, where union density is slightly above the average for the economy as a whole. 4.47. The high density category includes industries with unionization rates of between 34.5 percent and 45 percent. Industries in this group are characterized by high levels of technology, capital, and market concentration, and they operate in large establishments. The very high density group contains industries with unionization rates above 52.2 percent, including the financial sector, oil drilling and other fuel mineral extraction industries, together with oil refining. These sectors are essentially dominated by the State oil company, Petrobras, and by large multinationals. 4.48. The distribution of employment by union-density group is highly unequal. In 1999, 93.88 percent of workers were employed in medium or low union-density industries, while 66.47 percent were in industries with densities below the economy average; 4.05 percent of workers were employed in industries in the high density group, while just 2.07 percent of the workforce were in very high union-density industries.68 The very unequal distribution of workers by union-density group is 67 Machin and Gosling (1995) and Fortin and Lemieux (1997) show that the decline in union density in the United Kingdom and the United States caused a decrease in union bargaining power, and this helps to explain the observed increase in wage inequality. 68 The distribution of employment by union-density group in other years is quite similar. - 72 - likely to have significant effects on union behavior at the industry level, and hence on the distribution of the rents extracted by them. 4.49. Assuming a positive relation between union density and bargaining power (Booth, 1995), the heterogeneity of union density across industries should be reflected in the inter-industry wage distribution, since bargaining power varies widely from union to union. 4.50. Wage differentials should also be large within each industry, since collective bargaining processes in a given occupation or industry do not necessarily have national coverage. Thus, if union density and bargaining power varies between regions, there may be a wage differential within that industry. 4.51. Lastly, Figure 4.3 shows the relation between average years of schooling and union density, by industry. The two variables display a clear positive relation, suggesting that industries with a higher rate of union membership have a higher skilled workforce. | -- 4 ll0 8b H \ ' - , l 4.52. How do unions affect wages and the distribution of income in Brazil? Is union bargaining behavior influenced by the instftutions that regulate labor relations? How is income distribution affected by the concentration of unions in more sophisticated industries and higher unionization rates among skilled workers? This section seeks to answer these questions. Union Wage Premium69 4.53. One of the best-known effects of unions in the economy is their ability to raise wages through collective bargaining. The lifterature shows that unionized workers with the same characteristics as their non-union counterparts enjoy a wage premium thanks to the action of their unions in pay negotiations. Booth (1995), Abowd and Lemieux (1993), and Nickell et al. (1994) all claim that the success of union activism in raising rates of pay is directly related to the market power of the company and the degree of concentration of the industry concemed. Highly competitive companies or industries cannot pay wages over the odds, for fear of going out of business. A union wage premium will only be seen in companies or industries that generate monopoly rents. 4.54. To analyze the wage spread between unionized and non-union workers, we calculated the difference between the logarithm of the real hourly wage of the two groups. The results shown in Table 4.6 indicate an average wage differential of 58 percent. 4.55. Given that union members have a very different socioeconomic profile compared to non- members, the wage differential should next be calculated controlling for individual characteristics. 69 We use the terms 'union wage premium", 'union premium", 'union mark-up" and 'union wage differentiar" synonymously. - 73 - This was estimated by three methods, namely ordinary least squares (OLS), the Booth method, (1995, p.164), and the Oaxaca-Blinder decomposition. 4.56. The results in Table 4.7 suggest substantially lower union wage premiums than those indicated in Table 4.6, confirming that individual socioeconomic characteristics are a major factor in explaining the wage gap between unionized and non-union workers. Given that the mean wage differential shown in Table 4.6 is 58 percent, while the average premium estimated by OLS is just 17.9 percent, we can conclude that nearly 70 percent of the uncontrolled wage differential is explained by differences in worker characteristics. 4.57. Average values for the union wage premium are calculated as 17.9 percent, 15.3 percent and 17.2 percent, using ordinary least squares, the Booth method, and the Oaxaca-Blinder method, respectively. The salary differential is greater in Brazil than in several continental European countries, where the premium is around 7 percent, but quite similar to the 18 percent differential seen in the United States (Blanchflower and Freeman, 1992).70 Although the figures for Brazil and the United States are similar, the determinants of the premium in the two countries are not the same, as the labor market and collective bargaining processes operate very differently in the two countries. 4.58. The high union wage premium in Brazil contrasts with generally low union density. One plausible explanation for this is that union mark-ups are very significant in industries of high and very high union density (see Table 4.5), and this raises the average union wage premium for the economy as a whole. This hypothesis is bome out by the high correlation between union density and the inter- industry wage premium (0.798, significant at the 1 percent level), and the results obtained by Arbache and Menezes-Filho (2000) showing evidence of rent sharing in industries that generate greatest value-added and highest retum. 4.59. Strictly speaking, the union premium in any given industry should be zero, since by law all gains resulting from collective bargaining should be passed on to all workers. The empirical verification of union wage premiums suggests that pay raises obtained by unions are not always being extended to non-union workers. This would imply non-compliance with the law, suggesting that the legislation regulating collective bargaining, along with many other employment laws, are not being observed. 4.60. Table 7 shows a declining trend in union wage premiums as from 1993. This result is related to the effects of trade liberalization (Arbache, 2000) and possibly other factors also, such as privatization, market deregulation, and higher unemployment. Unions and wage dispersion 4.61. According to Freeman (2000), the most universal-stylized fact in the literature on unions is that wage dispersion among unionized workers is less than that among non-union members. The reasons for this are as follows: (i) unions depart from the principle that there should be a rate of pay 70 Blanchf lower and Freeman estimated wage models using OLS, with very similar specifications to the model estimated in this paper. - 74 - for the job, and wages should have little or nothing to do with performance or individual characteristics; (ii) unions defend the existence of rigid job and wage structures, with promotion criteria based on seniority and occupation, and a narrow wage spread over the job hierarchy; and (iii) unions strive to ensure their members do not eam less than the minimum wage for their category or for the overall economy, if one exists, thereby truncating the wage distribution and making it more compact. Consequently, individual characteristics that normally contribute to wage formation, especially human capital variables, have scant influence on rates of pay for unionized workers. The marginal return on schooling, experience and training, for example, will therefore be lower for unionized workers than for their non-union counterparts; and discrimination based on gender, race, age, and so forth, also tends to be less among union members. 4.62. The reasons why unions seek rigid pay scales and a narrower wage spread between occupations are as follows: (i) to define clear objectives for collective bargaining processes; (ii) to eliminate competition among worker through wage disputes; (iii) to engender a feeling of solidarity among workers in the search for equal treatment by the employer; and (iv) the power of the median voter, which tends to favor redistribution in favor of the lower paid whenever the median wage rate is below the mean. The most important consequence of union action is that there will be less wage inequality throughout the economy, despite union members enjoying a wage premium (Freeman, 2000; Metcalf et al. 2000). 4.63. Table 4.8 shows the dispersion of the logarithm of real hourly wages for unionized and non- union workers. The spread is always greateramong the former. Figure 4.4 and Figure 4.5 illustrate wage dispersion in 1999, showing that the distribution for non-union workers is a good deal more leptokurtic, with strong positive asymmetry. These results are strange, since they are contrary to the stylized fact that wage dispersion is less among unionized workers than among non-union members. The most plausible explanation for this is that unions in Brazil do not attempt to standardize wages. 4.64. One potential effect of the high wage dispersion among unionized workers is greater wage inequality throughout the economy.7' This effect will be greater the higher the wage dispersion between unionized and non-union workers, and/or the higher the level of union density. To examine the impact of unions on wage inequality, we use counterfactual analysis to simulate what would happen to inequality if the pay of non-unionists adhered to the wage policy in force for unionized workers.72 If we find that wage formation for unionized workers increases wage dispersion among non-unionists, we will have evidence that unions increase inequality in the overall economy. To confirm the result of the simulation, we examine what would happen with the distribution if unionized workers were to eam the same as their non-union counterparts. 4.65. The counterfactual for non-unionized workers was implemented as follows: 71 Blau and Kahn (1996) show that one of the reasons for greater wage dispersion in the United States compared to other developed countries is the high level of wage dispersion among unionized workers, although this is no greater than dispersion among non-union members in that country. 72 This could happen if, in order to avoid unionization, companies decided to follow the unionized workers wage policy. - 75 - in y,s = I a ns 4.66. Where y: is the simulation of the logarithm of the real hourly wage of the Ah non-union worker (the simulation is performed using the ,8 s estimated in the unionized workers wage model); a, is the residual from the non-union wage model; as is the standard deviation of the residual from the unionized workers wage model; and a" is the standard deviation of the residuals from the non- union wage model.73 4.67. Table 4.9 summarizes the effects of unions on the standard deviation of non-union wages, calculated in two stages. Firstly, the standard deviation of the wages of non-union and unionized workers was simulated on the basis of coefficients estimated in the unionized and non-union wage models, respectively. Secondly, we calculated the ratio between the simulated standard deviation and the original standard deviation (calculated without simulation). 4.68. The results presented in line 1 show that, apart from 1995, if the unionized worker wage policy was used to set the pay of non-union members, wage inequality would increase by between 5.1 percent and 37.3 percent, depending on the year. This result is unequivocal evidence that unions increase wage inequality in the Brazilian economy, rather than reduce it. Line 2 reports the reverse exercise, using the wage policy for non-union members to set unionized workers' pay. This produces a reduction in wage dispersion of 12.7 percent to 18.7 percent, thereby corroborating the previous evidence. Thus, the effects of unions on wage dispersion in Brazil are precisely the opposite of those reported in intemational literature. 4.69. A very strong factor that may explain the wider dispersion of unionized wages is that merit plays a greater role in setting union members' pay. If so, the coefficients estimated for the productive variables in the unionized wage models ought to be higher than those estimated in the models for non-union members. To investigate this hypothesis, we compared the schooling and experience coefficients in the unionized and non-union wage models. 4.70. The results in Table 4.10 show that the marginal retums on human capital variables in the unionized worker models are always higher than those obtained in the non-union models, which suggests that unions magnify rather than diminish the effects of personal characteristics on wages.74 These results are counterintuitive and contrary to the literature; they also contradict the idea that unions are sources of wage equalization and standardization. 4.71. There are at least three possible explanations for this result. The first relates to dispersion in unionized and non-union workers' characteristics. This could add to inequality insofar as wages are determined by productivity differences. The larger the varance of characteristics within a given 73 The equation for unionized workers is the same, with the ns superscripts replaced by s and vice versa. 7 The results reported correspond to 1999; qualitative results for other years are similar. - 76 - group, the higher will be the variance of wages in that group, and vice versa. To test this hypothesis, Arbache (1999) decomposes the variance of unionized workers' wages conditional on the variance of non-union workers' characteristics. The result indicates that 37 percent of wage dispersion among unionized workers is caused by the difference in the dispersion of characteristics between union and non-union members. In fact, on the one hand, unionization is highly concentrated in certain industries that account for a small proportion of employment, and in certain occupational categories, notably those requiring higher qualifications, which are in short supply in the economy. On the other hand, unionization is substantially lower in traditional industries and among the low-skilled workers who make up the vast majority of the workforce. Thus, the dispersion of characteristics among unionized workers is likely to be large. 4.72. The second explanation relates to the median voter theorem. As more highly qualified workers are over-represented in union membership, they wield significant power in defining policies that favor retums to human capital, thereby increasing wage inequality. If unions place a higher value on merit, then more highly skilled workers will seek jobs in sectors where unions have greatest influence in pay-setting processes. Thus, the market would respond to higher wages among unionized workers by allocating the best workers to industries where unions are most active.75 4.73. The third explanation relates to the scant or non-existent coordination and centralization of collective bargaining processes.76 This point is discussed below. 4.74. Dispersion of characteristics and pay among unionized workers, together with the relatively greater importance of skill in wage formation, suggest that the retums to human capital variables will differ significantly among unionized workers throughout the distribution. In other words, the marginal return on education is higher among workers further towards the right-hand tail of the distribution. To examine this hypothesis, we compared the marginal retums on schooling in the unionized and non- union wage models, estimated by quantile regression. 4.75. The results, shown in Table 4.1 1, reveal three phenomena. Firstly, marginal retums on education rise throughout the wage distribution. A worker in the 90th percentile obtains a greater return from a higher education course than a worker attaining the same level of schooling in any lower income percentile. Secondly, the retums to education are always higher for unionized workers. Thirdly, marginal retums among unionized workers grow proportionately more as the level of schooling rises. These resufts are evidence that unions contribute to greater wage inequality, by providing greater relative benefits to workers further to the right in the distribution, as well as workers with more years of schooling. 75 This would create a potential sample selection problem. At the limit, selection would tend to reduce wage dispersion as workers would have increasingly similar characteristics. 76 By "centralization' we refer to the level at which collective bargaining takes place, namely plant, company, industry or economy. - 77 - Unions and Collective Bargaining 4.76. One important aspect of collective bargaining processes is their potential economic effect, which goes beyond wage distribution. In the most centralized and coordinated bargaining systems, wage determination tends to be more sensitive to the general conditions of the labor market (Layard et al., 1991) and income distribution tends to be more equal. At the other end of the scale, fragmented bargaining systems that operate at the establishment level tend to be more sensitive to the economic conditions facing the company. In both cases, union demands are tempered by disincentives arising from the effects of inconsistent demands on the workers themselves, such as unemployment, inflation, or a lack of formalization of the employment contract. Discussion on this topic initially analyzed the effects of the crises of the 1970s on the macroeconomic performance of rich countries, and showed that the institutions that regulate collective bargaining play a decisive role in macroeconomic performance (Tarantelli, 1985; Bruno and Sachs, 1985; Freeman, 1988). 4.77. Calmfors and Driffill (1988) show that collective bargaining organized at the industry level, as in Brazil, is the worst possible structure, because workers have few incentives to incorporate into their objective functions the conditions facing the economy and companies, or the extemal effects on themselves of potentially inconsistent demands. Moreover, the absence of coordination and synchronization among collective bargaining processes in Brazil means that unions are permanently creating pockets of instability, by holding out for wage increases that are inconsistent with macroeconomic balance. 4.78. The literature indicates that the most important consequence of coordinated and centralized collective bargaining processes is low wage dispersion. The empirical literature shows that wage differentials are narrower in countries with centralized bargaining processes than where they are more fragmented (Kahn, 1998). Blau and Khan (1996) find, for a set of developed countries, that wage dispersion is lower where collective bargaining is more centralized. Metcalf et al. (2000) discuss and show that the retreat of collective bargaining coverage in the United Kingdom since the 1980s was accompanied by greater wage inequality. Rowthom (1992) uses the centralization criterion proposed by Calmfors and Driffill (1988) and finds a negative relation between centralization and inter-industry wage dispersion. Thus, the greater the coordination and centralization of bargaining processes, and/or the higher the proportion of workers covered by collective bargaining, the lower the wage dispersion in the economy will be.n 4.79. The structure of labor relations and union organization in Brazil is highly unfavorable to the strengthening, centralization, and coordination of collective bargaining processes. Unions from the most organized and sophisticated industries are encouraged to adopt individualistic behavior to exploit bargaining power associated with the conditions facing the company or industry. As a result, inter-industry wage differentials may be wider among unionized than among non-union workers. As n It should be bome in mind that wages agreed upon in more centralized collective bargaining systems affect the vast majority of workers, including non-union members, either because the result of the bargaining process is legally extended to everyone, or because companies take the negotiated wage rate as a benchmark for constructing their own pay scale. In either situation, wage dispersion among workers covered by the bargaining process should be less than among those who are not covered. - 78 - the rate and composition of unemployment vary from region to region, unions from a single industry but in different localities will have different bargaining power, leading to potentially greater inequality in the wages of unionized workers. 4.80. Industry-level bargaining processes may be advantageous for unions in sectors that earn monopoly rents and are therefore more receptive to wage claims. Such is the case of unions in the most profitable and highly concentrated industries, which use the most advanced technologies. Companies in sectors with these characteristics tend to be more accommodating because stoppages, strikes, and other forms of demonstration can be very costly for them. As we saw above, these industries attract the most highly qualified workers and have the greatest union density. For these unions, coordinated bargaining processes centralized at the economywide level could constrain their wage demands. In addition, centralized bargaining makes sectoral demands more transparent, revealing possible free-rider behavior by unions. Finally, it is easier to achieve intemal worker cohesion around bargaining goals when negotiations take place at the industry level rather than nationwide. On the other hand, unions in companies with little monopoly power, or industries that are not very concentrated, have less bargaining power. These would be interested in "hitching a ride" on centralized collective bargaining processes. Weak coordination and centralization of bargaining processes are therefore the result of the institutional framework that govems collective bargaining and of highly heterogeneous union bargaining power. 4.81. The effects of the structure of collective bargaining processes in Brazil are particularly important, given the country's highly unequal income distribution and the severe macroeconomic instability it has been experiencing for at least two decades. Negotiations at the industry level and by geographic area mean that unions have less incentive to internalize the effects of the trade off between wage/employment and wage/formal employment contract. Consequently, there will be: (i) high dispersion of inter-industry wages - this will rise with the heterogeneity of technology and market power between industries, and the heterogeneity of bargaining power between unions; (ii) a prolongation of and/or increase in the costs of stabilization processes and/or coordination of macroeconomic policies, with potential effects on unemployment and inflation. 4.82. To investigate the effects of the collective bargaining structure on inequality, we compared the standard deviation of the inter-industry wage differential between unionized and non-union workers. Inter-industry wage differentials relate to that part of wages not explained by human capital, demographic characteristics, regional variables, etc., but associated with the industry to which the worker is attached. Thus, the higher the industry wage premium, the higher will be the wage received by the individual in that industry vis-a-vis the representative worker. Insofar as industry wage premiums and profits are positively related (Arbache and Menezes-Filho, 2000), it is reasonable to expect a positive relation between union activism and the industry wage premium, because unions negotiate where there are monopoly rents to be shared out. As collective bargaining processes usually occur at the industry level, union pressure for higher pay could lead to greater inter-industry wage dispersion. 4.83. We estimated inter-industry wage premiums and calculated the respective standard deviation using the Haisken-DeNew and Schmidt (1997) procedure. Table 4.12 shows that the dispersion of industry wage premiums among unionized workers is greater than among non-union members. This confirms that union bargaining power at the industry level is quite heterogeneous, a situation that tends to be boosted by fragmentation and a lack of synchronization among collective bargaining - 79 - processes, and by the way in which labor relations are structured in Brazil. Little cooperation between unions can therefore be expected on wages, and coordination in collective bargaining processes is unlikely. 4.84. This paper has attempted to answer the following questions: (i) How do labor and union laws affect collective bargaining processes? (ii) Do unions affect wage formation and income distribution? (iii) Do unions increase rigidity in the labor market? On reaching the end of the paper, we realize we have not found unequivocal answers to these questions. The complexity of the topic and the limitations of relying on economics alone to provide answers may help explain the shortcomings of our findings. Nonetheless, the paper offers several pieces of evidence that elucidate the issues addressed. 4.85. Possibly the most important lesson to be drawn from the study is that institutions are fundamental in explaining union actions, along with their effects on income distribution and macroeconomic stability, and on the current state of collective bargaining in Brazil. The answer to question (i) is closely linked to the institutional apparatus. As we have seen, labor and union legislation - particularly through the union levy, monopoly representation and labor tribunals - create disincentives for the development of collective bargaining and cause problems for macroeconomic coordination. 4.86. The mandatory union levy creates contract problems between the union and the worker. As the union thus has an assured income, the bureaucrat will not always need to obtain additional funds from voluntary affiliations to run the organization. This may result in the relaxation or even disappearance of the contractual relation between the two parties, in which the union member (the principal) would expect "contractual" attitudes and decisions from the union bureaucrat (the agent). Consequently, union leaders may perform their union activities with less commitment, and this may lead to a weakened relationship between the union and its workers. The assured income, which is often sizeable, gives enormous power to the union bureaucracy, and nurtures bitter disputes for leadership of the organization. 4.87. In the period prior to the 1988 Constitution, when the State was given great freedom to meddle in union affairs, the contractual relation between union leaders and workers was often tenuous, and union bureaucracy essentially strove to keep itself in power by forging ever closer relationships with Govemment. It was in these circumstances that the so-called "pelego'concept and culture was formed, which in some ways may still exist in the union movement today. More recently, the distribution of funds by Government for unions to use in training and labor intermediation programs has fostered a further strengthening of relations between union leaders and government authorities. 4.88. The contract problem created by the union levy means that union leaders do not always feel committed to the most obvious interests of the worker; this inhibits a strengthening of the relationship between worker and union, thereby impairing the development of collective bargaining. 4.89. Monopoly representation is another institution that may interfere with the actions of unions in collective bargaining processes. The monopoly guarantees union bureaucracy enormous power in - 80 - its geographical jurisdiction, giving it major prerogatives such as mandatory participation in discussions on any issue relating to the interests of the category it represents. A problem of union involvement in all negotiations is that it tends to inhibit bargaining between employers and employees at the company level. In addition, excess power linked to the contract problem may also influence the behavior and actions of union leaders in bargaining processes. 4.90. The regulatory power of the labor tribunal system, together with the patemalism of labor legislation and in many cases the justice system itself, prevent workers and employers from entering into direct negotiations with one another. Employees, protected by legislation and the culture of appealing to the tribunal, are encouraged to use to the justice system to resolve disagreements with employers. Employees frequently have recourse to the justice system even before proposing or presenting their allegations or claims to their employers. The latter, either as a way of delaying payments owed, exploiting the opportunity cost of the worker, or increasing the transaction cost of access to amounts supposedly owed, also has incentives to apply to the courts rather than negotiate with the worker. Direct negotiations are thus replaced by the institutionalization of indirect negotiations through the labor tribunal system. This tortuous means of settling disputes may also be susceptible to political influence in the rulings handed down - a not infrequent occurrence, especially in higher courts. Apart from this, a lack of preparation and technical support on economic issues may lead judges to make spurious decisions. Such was the case during the wage policy period, for example, in defining the productivity increase to be passed on to wages, where courts calculated productivity on the basis of dubious or unknown criteria. 4.91. The wage law - which was in force for many years - and the policies of the military dictatorship explain, in large measure, the backward state of collective bargaining in Brazil. As we saw earlier, employers had incentives during wage negotiations to invoke collective disagreement, which meant applying the wage law and ruling out any chance of negotiation. Throughout the military dictatorship, unionism was subjected to close monitoring and control, which stymied the development, improvement, and modemization of labor relations for over twenty years. 4.92. With regard to question (ii), pertaining to the effects of unions on pay setting and the wage distribution, we find a variety of econometric evidence. Unions clearly contribute to wage formation, by establishing a wage premium that enables union members to eam more than their non-union counterparts with the same individual characteristics and industrial affiliation. In principle, the existence of the wage premium is unexpected, since the law requires the results of collective bargaining processes to be extended to all workers in the occupational category concerned. Thus, any pay raise negotiated by the union must be given to all workers in that category. Possible explanations for the empirical verification of union wage premiums include the following: (i) automatic extension of gains may not have taken place, meaning non-compliance with the law; (ii) wage negotiations may predominantly be carried out at the company rather than the industry level, as we investigated in this study, which would not require extension of bargained results to other workers in the industry; (iii) unions may essentially be negotiating pay floors rather than actual wage rates, thereby giving space for companies to negotiate wages above the floor; and (iv) there may be a wage differential between workers in a given category, but in different municipios or jurisdictions of unions or their federations. Unfortunately, shortcomings in union data prevent us from testing these hypotheses. Nonetheless, we would not be surprised if future calculations using company and/or municipal-level data produced lower union wage premiums. -81 - 4.93. We find evidence that unions do affect income distribution; but contrary to the view prevailing in the literature, unions in Brazil are associated with more rather than less wage inequality. Several factors appear to explain this. Firstly, higher-skilled workers and those in better jobs are proportionately more unionized than less skilled workers. This is counterintuitive, since better qualified workers would not have"an incentive to be paid standardized wages. On the contrary, as they are more skilled they ought to prefer their pay to be performance-related rather than predetermined. 4.94. Our results suggest that the higher rates of unionization among more qualified workers encourage unions to adopt strategies favoring this group. This is consistent with the median voter theorem, the greater mobilization power of like-minded people, or insider power. The finding that marginal returns on human capital variables are higher among unionized workers and those to the right of the wage distribution, compared to non-union members, is rernarkable. Given that the country is so lacking in skilled labor, unions would therefore seem to be exploiting labor scarcity to magnify the returns to skill, thus benefiting workers with more human capital and/or those holding better posts in the job hierarchy. This issue becomes even more important if unions are insensitive to the wage/employment and wage/informal sector trade-offs. 4.95. One possible explanation for the higher rate of unionization among better qualified workers may be the regulation of higher-level professions by regional councils, which, to some extent, behave or end up behaving in a similar way to unions. As we showed above, an example of this is the publication of fee scales and the establishment of pay floors by these institutions. On the other hand, the unions in the respective categories, which often work in conjunction with and in the same place as the regional councils, can take advantage of extemalities arising from the actions of these bodies, to promote and disseminate unionization among professional people and thereby increase union membership among them. 4.96. Secondly, the dispersion of characteristics among unionized workers is greater than among their non-union counterparts, resulting in higher wage dispersion among union members. 4.97. Thirdly, the resurgence of unionism at the end of the 1 970s in the industrialized and wealthy ABCD area of Sao Paulo, does not seem to be mere coincidence. Clearly, it is associated with the nature of the companies and industries in that region, which include multinationals and large domestic companies, mostly operating in concentrated, high technology industries. The conjunction of these two phenomena, namely the extraction of monopoly rents by companies in the region, and the strict law on pay increases, which limited wage hikes, would have encouraged workers to organize and lobby for greater flexibility to facilitate rent sharing. Thus, the reorganization of the union movement in the wealthiest part of the country and in industries using sophisticated technologies and moderately more qualified workers seems to have molded the "new unionism" as an activity associated with workers who are more rather than less qualified.78 78 It should be noted that the concept of skill in a country with a high proportion of illiterate or functionally illiterate workers, such as Brazil, is different from that of a country with a higher average level of education. Thus, even an individual who has not completed second grade, for example, could be considered to have formal education well above the national average. - 82 - 4.98. Fourthly, the fact that occupational categories' base dates are spread throughout the year, compounded by the fragmentation of collective bargaining by region, reduces the chances of centralization and/or coordination of bargaining processes and encourages free-rider behavior. In this context, unions in sectors that extract monopoly rents have incentives to act in a decentralized fashion, exploiting circumstances that are favorable to them, in accordance with the characteristics of the industry and/or region in which they are located. On the other hand, the weakness of unions in more competitive sectors means they have less chance of bargaining success. As union density is higher in the more concentrated and sophisticated industries, we consequently find greater inter- industry wage inequality associated with unions. 4.99. Another major consequence of the fragmentation of bargaining processes is the difficulty this raises for macroeconomic coordination. A fragmented system of negotiations, with highly variable bargaining power between unions, encourages free riding and inconsistent wage demands, constantly generating potential pressure on prices. But the effects of fragmentation become even more acute as unions in more organized sectors show scant concern for the wage/employment and wage/informal sector trade-offs, provoking potentially negative consequences for the level and quality of employment, and harming outsiders in particular. 4.100. As for the relation between unions and market-market rigidity posed in question (iii), we find evidence that union action on wages tends to increase rigidity and segmentation in the labor market. The union wage premium, together with an over-valuation of human capital, raise labor costs, creating entry barriers for less skilled workers in certain industries and encouraging segmentation. Broadly speaking, this produces one group of industries with higher union density, higher wages, and more skilled workers; and another group of industries with lower union density, lower wages, and less skilled workers. lo l 4.101. In the light of the above discussion, we now make some recommendations for encouraging and strengthening collective bargaining processes, ideally turning them into the key instrument of labor relations in Brazil. We believe the following measures should be adopted: (i) abolish the union monopoly; (ii) abolish the mandatory union levy; (iii) abolish the regulatory power of the labor tribunal system; (iv) organize unions by economic activity rather than occupation; (v) strengthen representation at the workplace; and (vi) change collective bargaining base dates so as to reduce fragmentation and encourage centralization and/or cooperation between unions in bargaining processes. We believe these measures would enhance unions and give them greater independence; the measures would also strengthen them and elicit more committed engagement from them on issues pertaining to workers' interests. The current structure of collective bargaining in Brazil reflects and reveals scant consensus among the agents involved in improving and modernizing the country's labor relations. 4.102. Given the paucity of research on unions in Brazil, investigation of their impact on employment, productivity, profits, and working conditions, together with studies of the differences between unions in the public and private sectors, and at the company and regional level, would be particularly welcome in order to shed further light on the issues raised in this paper. - 83 - References Abowd, J.M. and Lemieux, T. (1993), 'The effects of product market competition on collective bargaining agreements: the case of foreign competition in Canada', Quarterly Joumal of Economics, 108: 983- 1014. Amorim, B.M. (2000), Salarios indiretos e sindicatos no Brasil, Campinas: Anais do XXVIII Encontro Nacional de Economia. Arbache, J.S. (1999), 'Do unions always decrease wage dispersion? The case of Brazilian manufacturing', Joumal of Labor Research, 20: 425-436. Arbache, J.S. (2000), 'Does trade liberalization always decrease unions bargaining power?', Proceedings of the 2000 EALE/SOLE Word Conference, Milao. Arbache, J.S. and Carneiro, F.G. (1999), 'Unions and wage differentials', World Development, 27: 1875- 1883. Arbache, J.S. and Menezes-Filho, N., (2000), 'Rent-sharing in Brazil: Using trade liberalization as a natural experiment', Rio de Janeiro: Annals of the V Annual Meeting of the Latin American and Caribbean Economic Association. Blanchflower, D.G. and Freeman, R.B. (1992), 'Unionism in the United States and other advanced OECD countries', Industnal Relations, 31: 57-79. Blau, F.D. and Kahn, L.M. (1996), 'Intemational differences in male wage inequality: institutions versus market forces', Joumal of Political Economy, 104: 791-837. Blau, F.D. and Kahn, L.M. (1999), 'Institutions and laws in the labor market, in 0. Ashenfelter and D. Card (eds.), Handbook of Labor Economics, Amsterdam: Elsevier. Booth, A. (1995), 'The economics of the trade unions', Cambridge: Cambridge University Press. Bruno, M. and Sachs, J. (1985), 'The economics of worldwide stagflation', Cambridge, MA: Harvard University Press. Camargo, J.M. (2001), 'Sindicatos e justi,a do trabalho no Brasil', mimeo, Departamento de Economia, PUC- Rio. Caimfors, L. and Driffill, J. (1988), 'Centralization and wage bargaining, Economic Policy, 3: 14-61. Card, D. (1992), 'The effect of unions on the distribution of wages: redistribution or relabelling?', NBER Working Paper 4195. DiNardo, J., Fortin, N.M. and Lemieux, T. (1995), 'Labor market institutions and the distribution of wages, 1973-1992: a semiparametric approach', NBER Working Paper 5093. DiNardo, J., Hallock, K. and Pischke, J.S. (1997), 'Unions and managerial pay, NBER Working Paper 6318. Fortin, N.M. and Lemieux T. (1997), 'Institutional changes and rising wage inequality: is there a linkage?', Joumal of Economic Perspectives, 11: 75-96. Freeman, R.B. (1980), 'Unionism and the dispersion of wages', Industrial and Labor Relations Review, 34: 3- 23. Freeman, R.B. (1982), 'Union wage practices and wage dispersion within establishments', Industrial and Labor Relations Review, 36: 3-21. Freeman, R.B. (1988), 'Labor market institutions and economic performance', NBER Working Paper 2560. Freeman, R.B. (1992), 'How much has de-unionization contributed to the rise in male earnings inequality?', in S. Danzinger and P. Gottschalk (eds.), Uneven tides, New York: Sage Press. - 84 - Freeman, R.B. (2000), 'Single peaked vs. diversified capitalism: the relation between economic institutions and outcomes', NBER Working Paper 7556. Gosling, A. and Machin, S. (1995), 'Trade unions and the dispersion of earnings in British establishments, 1980-90', Oxford Bulletin of Economics and Statistics, 57: 167-184. Green, F., Dickerson, A., and Arbache, J.S. (2001), 'A picture of wage inequality and the allocation of labor through a period of trade liberalization: the case of Brazil', World Development, forthcoming. Haisken-DeNew, J.P. and Schmidt, C.M. (1997), 'Inter-industry and inter-region differentials: mechanics and interpretation', Review of Economics and Statistics, 79: 516-521. Hirsch, B.T. (1982), 'The interindustry structure of unionism, eamings and eamings dispersion', Industrial and Labor Relations Review, 36: 22-39. Kahn, L.M. (1998), 'Collective bargaining and the interindustry wage structure: international evidence', Economica, 65: 507-534. Layard, R., Nickell, S. and Jackman, R. (1991), 'Unemployment macroeconomic perfonnance and the labor market, Oxford: Oxford University Press. Metcalf, D., Hansen, K., and Charlwood, A. (2000), 'Unions and the sword of justice: unions and pay systems, pay inequality, pay discrimination and low pay', mimeo, Center for Economic Performance, London School of Economics. Nickell, S.J., Vainiomaki, J. and Wadhwani, S. (1994), 'Wages, unions and product market power', Economica, 61: 457-474 Nickell, S.J. and Layard, R. (1999), 'Labor market institutions and economic performance', in 0. Ashenfelter and D. Card (eds.), Handbook of Labor Economics, Amsterdam: Elsevier. Rowthorn, R.E. (1992), 'Centralization, employment and wage dispersion', Economic Joumal, 102: 506-523. - 85 - Table 4.1 Characteristics Of Unionized And Non-Union Workers (Percentages) 1986 1992 1993 1995 1996 1997 1998 1999 Union Non- Union Non- Union Non- Union Non- Union Non- Union Non- Union Non- Union Non- union union union union union union union union General characteristics Employment record 82.3 57.5 91.3 55.2 91.0 54.1 91.4 53.5 90.4 53.8 90.4 53.4 90.5 53.3 89.7 52.4 card I I _I I Head of household 66.8 48.6 64.3 50.7 64.5 50.9 62.8 50.2 62.0 49.7 61.6 50.0 61.3 50.2 61.2 50.2 White 61.5 51.0 61.5 52.0 60.9 51.8 61.1 51.8 62.1 52.5 60.5 51.8 61.2 51.5 59.9 51.7 Male 75.6 63.9 70.5 62.3 69.5 62.4 68.1 61.0 66.7 61.0 66.6 61.5 65.3 61.1 65.2 60.6 Experience (absolute 22.9 22.1 22.4 21.8 22.3 21.9 22.6 21.9 22.5 21.7 22.5 21.7 22.6 21.7 22.9 21.7 value) Age (absolute value) 36.0 33.8 37.0 34.6 37.2 34.7 37.7 34.8 37.9 35.0 38.0 35.0 38.3 35.2 38.5 35.3 Education Illiterate 11.7 17.0 9.8 15.3 9.1 14.1 7.9 13.0 7.2 12.5 7.7 11.9 6.9 11.1 7.5 10.6 Some elementary 13.7 19.8 11.5 17.3 10.9 17.0 10.3 16.5 9.2 15.0 9.1 15.4 8.5 14.8 8.4 14.3 education _ Elementary or some 26.4 32.7 26.2 33.0 25.5 33.6 25.4 33.9 23.6 32.7 22.9 32.7 22.3 32.3 21.8 32.0 primary education I Primary or some 13.2 13.1 14.0 14.1 13.7 14.4 13.9 14.7 15.3 16.3 14.8 15.7 14.7 16.3 15.0 16.5 secondary education II_ Secondary or some 21.4 13.9 25.0 16.0 25.9 16.5 26.8 17.3 28.1 18.6 28.4 19.2 29.8 20.2 29.6 21.2 higher education I I I Higher educaton 13.7 3.4 13.6 4.4 15.0 4.4 15.7 4.6 16.6 4.9 17.0 5.1 17.9 5.3 17.7 5.4 Years of schooling 7.2 5.2 7.7 5.7 7.9 5.8 8.1 6.0 8.4 6.2 8.4 6.3 8.7 6.4 8.6 6.6 (absolute value) I_ I I_ __I_I I -86 - Table 4.2 Proportion Of Unionized Workers By Region And Occupation (Percentages) 1986 1992 1993 1995 19s 1997 1998 1999 Geographic region Urban 22.1 20.2 20.0 19.1 18.7 18.2 17.8 17.4 Rural 22.1 20.0 19.6 17.8 17.7 17.1 16.5 17.2 Metropolitan region 24.2 22.0 22.1 21.2 20.9 20.3 20.0 18.8 Non-metropolitan 20.6 18.8 18.4 17.3 16.8 16.4 15.9 16.4 South-east 22.9 19.1 18.4 17.9 18.6 17.9 17.4 16.7 Other regions 21.7 20.8 20.8 19.5 18.5 18.1 17.7 17.7 North 22.3 16.4 16.2 14.6 15.5 15.4 14.3 14.7 Other regions 22.2 20.4 20.2 19.3 18.8 18.2 17.9 17.6 North-east 19.1 20.8 20.2 18.4 16.6 16.7 16.6 17.2 Other regions 23.4 18.6 19.2 19.2 19.3 18.5 18.0 17.5 Central-westem 16.9 17.8 18.7 15.7 16.6 15.2 15.1 15.0 Other regions 23.0 20.5 20.1 19.4 18.8 18.4 18.0 17.7 South 29.8 27.4 26.4 25.5 23.6 23.1 25.5 21.6 Other regions 20.7 18.5 18.5 17.5 17.4 16.9 16.6 16.5 Occupation Managers 40.2 34.5 36.0 34.7 33.4 32.2 31.4 30.2 Other occupations 21.6 19.7 19.5 18.5 18.1 17.6 17.3 17.0 Higher-level 52.5 44.8 45.2 45.0 42.3 42.4 43.0 42.2 professionals I Other occupations 20.1 19.1 18.8 17.7 17.4 16.8 16.4 16.1 Technical workers 25.2 27.9 29.9 28.4 28.5 28.8 27.4 27.7 Other occupations 22.0 19.7 19.3 18.3 17.9 17.3 17.0 16.7 Office staff 30.8 31.6 32.7 31.1 29.7 29.1 28.7 27.7 Other occupations 21.3 19.2 18.9 18.0 17.7 17.2 16.8 16.6 Sales staff 12.9 8.9 10.1 9.7 10.3 10.0 10.2 9.6 Other occupations 23.8 11.3 21.8 20.8 20.2 19.6 19.2 19.0 Skilled producton 29.3 19.5 28.3 26.8 26.0 25.3 24.9 24.8 workers I Other occupations 20.6 18.5 18.5 17.6 17.3 16.8 16.5 16.2 Unskilled production 15.3 13.9 16.3 12.9 12.3 11.9 11.5 11.4 workers I I Other occupations 27.2 25.0 24.9 23.6 23.4 22.8 22.6 22.2 - 87 - Table 4.3 Probability of Union Membership Marginal effects (%) 1986 1992 1993 1995 1996 1997 1998 1999 Some elementary education 18.59 24.27 21.23 26.43 31.21 15.04 16.07 8.81 Elementary or some primary education 62.60 76.04 71.74 81.50 85.27 64.87 68.87 60.22 Primary or some secondary education 149.75 161.31 157.50 174.43 186.89 162.11 164.81 165.31 Secondary or some higher education 287.43 308.45 304.18 327.67 354.47 291.27 312.00 292.32 Higher education 835.15 624.02 680.75 721.92 760.35 633.53 691.90 671.78 Experience 1.38 1.97 1.92 2.25 2.31 2.32 2.44 2.62 Male 24.50 21.91 15.68 18.20 14.74 17.68 12.99 15.82 Head of household 89.43 47.36 51.98 41.20 41.79 39.31 37.09 34.10 Urban region -13.86 -15.51 -12.41 -9.24 -12.99 -11.87 -9.92 -10.70 Metropolitan region 9.87 12.41 16.32 17.39 21.64 21.87 21.89 12.01 South-east 11.06 -10.12 -19.28 -19.08 1.89 -2.94 -6.76 -13.45 North 19.83 -18.75 -20.74 -27.73 -7.39 -10.23 -16.57 -20.56 Central-westem -20.55 -7.89 -9.23 -21.91 -1.69 -14.45 -13.77 -18.28 South 58.77 45.90 31.47 28.31 37.63 33.18 28.36 15.88 Agriculture 84.91 92.94 119.83 109.82 114.95 107.71 106.25 125.68 Manufacturing 121.24 164.18 178.35 193.99 151.76 134.72 120.43 130.49 Civil construction -30.66 -19.09 -16.82 -14.54 -26.05 -28.75 -30.20 -33.23 Other industrial activities 187.18 343.72 369.61 442.29 355.15 304.29 310.65 317.70 Services -48.39 -37.48 -37.16 -38.46 -33.12 -35.44 -37.32 -32.21 Other services 30.16 44.84 44.97 30.16 33.34 31.52 29.09 29.59 Transport and communication services 133.32 255.47 286.31 253.98 244.85 201.06 190.42 218.62 Social services 32.26 108.21 141.79 155.10 133.87 150.52 129.10 150.60 Public administration -12.02 28.39 49.39 74.81 53.48 75.89 66.99 93.66 Other activities 164.23 284.65 328.61 330.09 229.34 237.05 166.07 183.97 N 99 305 104 154 106 285 114 377 111 358 118 135 117 045 120 685 Note: All coefficients are significant at the 5% level, except those shown in italics. Reference categories are: illiterate, north east region, and commerce. The models were estimated by logistic regression. The standard errors of the estimators are robust. -88 - Table 4.4 Union Density _ ercentages_ Industry 1986 1992 1993 1995 1996 1997 1998 1999 Mean Agriculture 23.6 20.6 20.7 18.5 18.9 18.4 18.1 18.9 19.7 Mining 30.1 22.0 19.6 24.2 15.9 17.0 16.6 16.5 20.2 Oil, gas, and coal 68.3 67.1 71.8 71.7 49.3 64.8 61.1 66.7 65.1 Non-metallic 22.1 19.0 17.5 19.5 17.2 20.1 16.2 15.1 18.3 products I_I Metallurgical 41.0 35.5 38.8 34.9 36.4 30.3 29.7 29.2 34.5 products . Mechanical products 43.5 40.1 41.7 40.2 31.1 36.6 31.7 31.3 37.0 Electrical and 47.6 42.4 36.1 38.2 33.6 33.3 33.0 27.1 36.4 electronic products _ _ _ Vehicles and 43.3 47.2 46.5 48.9 46.8 43.2 41.8 42.5 45.0 autoparts I_I Wood and fumiture 18.2 15.0 10.2 11.2 11.0 9.1 11.2 10.0 12.0 Paper and publishing 36.1 35.1 32.9 32.9 36.0 27.1 27.4 27.2 31.8 Rubber 30.2 28.2 42.9 40.8 28.9 24.6 34.2 31.0 32.6 Chemicals 35.3 30.3 31.1 30.5 31.2 28.2 29.3 34.4 31.3 Oil refining 53.4 63.8 48.7 57.7 49.7 54.6 40.9 48.7 52.2 Perfumes and 31.8 23.9 29.0 27.2 26.7 24.0 23.9 26.4 26.6 pharmaceuticals Plastic products 32.3 28.7 28.4 27.7 31.2 29.4 24.1 27.0 28.6 Textiles 33.1 40.7 38.9 43.3 31.7 33.6 35.3 30.5 35.9 Clothing 8.9 9.4 9.4 9.5 8.2 8.0 8.4 8.7 8.8 Footwear 32.1 30.4 31.5 27.9 22.8 24.2 18.4 21.0 26.0 Foodstuff 26.2 24.1 22.9 22.6 20.9 20.3 20.0 18.1 21.9 Other manufactured 22.6 16.9 14.6 21.6 20.1 16.2 17.3 13.5 17.9 products Industrial services 50.6 66.2 63.3 65.3 60.6 56.6 55.0 54.2 59.0 Civil construction 11.9 10.2 9.5 9.2 8.2 8.0 7.6 6.9 8.9 Commerce 17.9 14.3 13.2 12.2 12.7 12.6 12.7 11.7 13.4 Transport services 34.9 37.1 36.1 32.4 33.1 30.3 29.8 29.4 32.9 Communication 43.3 45.9 52.1 52.2 45.6 41.3 38.9 38.5 44.7 Financial institutions 54.1 60.8 58.7 61.8 53.8 53.8 48.9 49.9 55.2 Social services 14.1 13.9 13.7 12.8 14.0 13.2 13.5 13.0 13.5 Business services 32.1 28.0 28.0 26.1 26.9 25.1 25.4 24.8 27.1 Public administration 24.4 21.1 25.1 19.7 18.0 22.3 20.5 20.2 21.4 Rental and leasing 23.9 26.0 27.9 29.0 28.1 29.5 28.8 29.9 27.9 Other services 4.0 2.7 2.7 2.7 2.8 2.8 2.9 2.7 2.9 Total 22.2 20.2 20.0 19.0 18.6 18.1 17.7 17.4 19.2 - 89 - Table 4.5 Union Density By Group (Percentages) Group Mean 1. Medium or low-density (up to one standard deviation) Other services 2.9 Clothing 8.8 Civil construction 8.9 Wood znd furniture 12.0 Commerce 13.4 Social services 13.5 Other manufactured products 17.9 Non-metallic products 18.3 Agricufture 19.7 Mining 20.2 Public administration 21.4 Foodstuff 21.9 Footwear 26.0 Perfumes and pharmaceuticals 26.6 Other industrial services 27.1 Rental and leasing 27.9 Plastic products 28.6 Chemicals 31.3 Paper and publishing 31.8 Rubber 32.6 Transport services 32.9 2. High-density (up to two standard deviations) Metallurgical products 34.5 Textiles 35.9 Electric and electronic products 36.4 Mechanical products 37.0 Communications 44.7 Vehicles and autoparts 45.0 3. Very high-density (three or more standard deviations) Oil refining 52.2 Financial institutions 55.2 Industrial services 59.0 Oil, gas, and coal 65.1 Notes: Standard deviation = 15%; overall mean density = 19.2%; mean values between 1986 and 1999. Table 4.6 Differential Of Log. Of Real Hourly Wages (Percentages) 1986 55.8 1992 59.1 1993 63.0 1995 57.5 1996 57.0 1997 57.7 1998 57.8 1999 54.2 - 90 - Table 4.7 Union Wage iUHark-Up Year Union dummy (OLS) Booth method* Oaxaca-Blinder method*" 1986 0.1670 0.1365 0.1281 - 0.1662 1992 0.1772 0.1381 0.1203 - 0.2057 1993 0.2008 0.1682 0.1516 - 0.2316 1995 0.1855 0.1698 0.1581 - 0.2190 1996 0.1813 0.1596 0.1469 - 0.2135 1997 0.1881 0.1702 0.1606 - 0.2134 1998 0.1774 0.1580 0.1479 - 0.2040 1999 0.1574 0.1228 0.1106 - 0.1778 Notes: The explanatory variables in the wage equation model are: education, experience, experience squared, employment record card, urban region, metropolitan region, head of household, race, gender, five geographic regions, and 31 industries. * Booth (1995). ** Results vary according to the means vector used in the calculations, of non-union and unionized workers, respectively. Table 4.8 Wage Dispersion Among Unionized And Non-Union Workers Standard deviation of log. of real hourly wage Year Unionized Non-union Total 1986 0.9740 0.8914 0.9397 1992 0.9412 0.9076 0.9448 1993 0.9910 0.9229 0.9702 1995 0.9631 0.8936 0.9349 1996 0.9629 0.8949 0.9352 1997 0.9667 0.8926 0.9332 1998 0.9620 0.8751 0.9181 1999 0.9485 0.8636 0.9028 Table 4.9 Simulation Of Standard Deviation Of Unionized And Non-Union Workers' Wages (simulated value/oniginal value %) |1986 1992 | 1993 11995 1996 1997 1998 1999 Non-union 13.57 5.08 20.82 -4.64 14.43 11.88 15.79 37.33 l Unionized -13.45 -13.77 1-17.45 -12.66 -14.05 -11.97 -14.99 -18.75 Notes: The figures shown in the table represent the ratio of the simulated standard deviation divided by the original standard deviation, expressed as a percentage. The values shown for non-union members refer to the simulated standard deviation obtained using coefficients estimated from the unionized workers wage model. The values for unionized workers refer to the simulated standard deviation obtained with coefficients esflmated from the non-union workers wage model. -91 - Table 4.10 Returns To Human Capital Variables In Unionized And Non-Union Workers Wage equations, 1999 Variable Unionized Non-union Experience 0.0347 0.0331 Experience squared -0.0004 -0.0005 Some elementary education 0.1573 0.1151 Elementary or some primary 0.3678 0.2888 education Primary or some secondary 0.6414 0.5288 education Secondary or some higher education 1.0850 0.9380 Higher education 1.8080 1.7260 Note: All coefficients are significant at the 1 % level Table 4.11 Schooling Coefficients Among Unionized And Non-Union Workers, 1999 Percentile 0.1 0.25 0.50 0.75 0.9 Union Non- Union Non- Union Non- Union Non- Union Non- union union union union union Sornie 0.0589 0.0394 0.0961 0.0416 0.1055 0.0465 0.1591 0.0703 0.1560 0.1017 elementary education Elementary 0.1625 0.1211 0.2180 0.1146 0.2488 0.1175 0.2781 0.1340 0.2656 0.1709 or some primary education Primary or 0.2976 0.2060 0.3428 0.1915 0.3731 0.2244 0.4411 0.2756 0.4790 0.3498 some secondary education ____ _ _ Secondary or 0.5479 0.4123 0.6510 0.4581 0.7549 0.5674 0.8901 0.7118 0.9513 0.8657 some higher education Higher 1.3916 1.1577 1.5364 1.3185 1.6765 1.5176 1.7903 1.6751 1.7487 1.7513 education I_I_I_I Notes: Models were estimated by quantile regression. All coefficients are significant at the 1% level. Standard errors were estimated by the bootstrap method. The models' explanatory vanables are: gender, urban region, metropolitan region, 5 geographic regions, 31 industries, head of household, employment record card, and schooling. - 92 - Table 4.12 Standard deviation of the interAindustr" wage differential Year Unionized Non-union Total 1986 0.2491 0.1827 0.2023 1992 0.2640 0.1713 0.2061 1993 0.2741 0.1619 0.2002 1995 0.2631 0.1614 0.1934 1996 0.2456 0.1652 0.1892 1997 0.2375 0.1685 0.1898 1998 0.2310 0.1605 0.1796 1999 0.2313 0.1625 0.1818 Notes: (1) The inter-industry wage differential and standard deviation were estimated by the Haisken-DeNew and Schmidt (1997) method. (2) The models' explanatory variables are: education, gender, experience, experience squared, employment record card, head of household, urban region, metropolitan region, regions, and 31 industries (IBGE two-digit level). - 93 - Figure 4.1 Union Density vs. Formalization, 1999 .7 0 .60 0 .5 .34 * . 2 o0 0 to .2 o3 °4 ° 0.23~~~~~~~~~~~~~~~~~ 0.22 -~~~~~~~~~~~~~~ 0.21 0.0 00 -~~~~~~~~~ o 0.1 .10 0.1 .2 .3 .4 .5 .6 .7 .8 9 1.0 Formalization Figure 4.2 Union Density (%) 0.17 0.16 0.15 - - 94 - Figure 4.3 Schooling and union density - industry, 1999 70 60 o 60~~~~~~~~~~~~~ 50 40 a a ° ° 40~~~~~~~~~~~ 30o 30 n0 ° ° 6000~~~~ ~ aa 20. a 5000~~ ~~ a 10000 10 0~a 0 0.02.0 4.0 6 0 8.0 10. 12. 12 14 Average years of schooling Figure 4.4 Distribution of Real Hourly Wages - Union Members, 1999 6000 i 5000 a 4000 a 3000 a 2000 a 1000a Std. Dev =4.10 0 ~~~~~~~~~~~~~~~~Mean =4.0 0.0 2.0 4.0 6 0 8.0 10. 12. 14. 16. 18. 20. 22 0 0 0 0 0 0 0 Real hourly wage - 95 - Figure 4.5 Distribution of Real Hourly Wage - Non-union Members, 1999 50000 40000 30000 20000 10000 Std. Dev = 2.77 Mean = 2.3 0 _ _ _ _N = 100176.00 0.0 2 0 4 0 6.0 8.0 10. 12. 14. 16. 18. 20 22. 0 0 0 0 0 0 0 Real hourly wage - 96 - 5. 1ECONOWC MPRAC7 OF MEONS AND COLLECMHE BARGAMMO:G LESSONS FRCMv M7ERMAMuOMAL Prepared by Amit Dar 4D [ACKmGMDCI k.M C(ED aswo MGMOm 5.1. The intemational debate on labor standards has focused on a set of intemationally agreed upon "core labor standards" including: (1) the freedom of association (consisting of the right to unionize and the right to collective bargaining); (2) the elimination of all forms of forced labor; (3) the effective abolition of child labor; and (4) the elimination of discrimination in employment and occupation. 5.2. These core labor standards have been defined in a number of ILO conventions dating from 1930 on. However, not all countries have ratified them and even where ratification has taken place, sometimes implementation and enforcement have been problematic. In June 1998, the ILO adopted the Declaration on Fundamental Principles and Rights at Work in order to revive interest in adopting and ratifying these conventions and to signal a strategic shift towards more flexible implementation and a greater emphasis on technical assistance. The Declaration includes an obligation on the part of ILO members to promote and realize the four core labor standards identified above. It recognizes the obligation of the ILO to assist its member countries to attain the objectives of the Declaration. It also encourages other intemational organizations to support these efforts by offering technical assistance and by helping member countries to create a climate for economic and social development. The intent of the Declaration is to ensure that the trend towards globalization is accompanied by social justice. However, it acknowledges that the comparative advantage of low-wage countries should not be called into question. 5.3. Debate on core labor standards has also become more prominent in the international arena and remains controversial. For example, some (largely developed) countries have expressed the view that if these standards were linked to trade regulations through the WTO, this would represent a powerful incentive for members to improve labor conditions. By contrast, many developing and some developed member states have maintained that the introduction of labor standards in trade negotiations would amount to protectionism and would undermine the comparative advantage of lower-wage trading partners. 5.4. This paper focuses on one of these standards by shedding light on the economic facets of the freedom of association. While the human rights argument in support of freedom of association and the right to collective bargaining is compelling, the question from an economic point of view is to what extent these standards contribute to economic development and performance. Answering this question would, in principle, require a comparison between countries with and without unions. This is, however, impossible as in virtually all countries - even those which suppress labor - unions and employers' organizations play a (sometimes important) role in industrial relations. 5.5. In this paper, we consider two related questions that indirectly can help us disentangle the economic effects associated with freedom of association and the right to collective bargaining. These questions are: (a) what are the microeconomic effects of unions and employers' organizations (unionism); and (b) how do different institutional approaches to collective bargaining affect macroeconomic performance? - 97 - 5.6. The majority of the evidence that addresses various aspects of these questions comes from OECD countries, though available evidence from developing countries will also be presented. This evidence will inform us about the economic consequences of different types of unionism as well as the interaction between unionism and the economic, political and institutional environment in which collective bargaining takes place. The aim oi this review is to summarize the empirical evidence and draw some lessons of experience. 5.7. The study will be organized in the following sections: (a) presenting some data on unionism; (b) providing a brief theoretical overview of unionism and collective bargaining; (c) reviewing evidence derived from microeconometric analysis of the economic impact of unions; (d) reviewing evidence derived from econometric studies focusing on the relationship between different systems of collective bargaining and macroeconomic performance; and (e) conclusion and lessons. The evidence presented in this paper draws heavily on a recent comprehensive study of the issues Core Labor Standards and the Freedom of Association: Economic Aspects by Aidt and Tzannatos (2001, forthcoming). 5.8. A recent development in the economics of the labor market was the appearance during the 1980s of a wide range of studies on the behavior of trade unions and their impact on the process of wage and employment determination. In the specific case of wage determination, it has been now widely suggested that wage moderation is usually associated with a more coordinated bargaining structure, or what has been termed corporatist arrangements. On the other extreme, however, explosive wage demands have been usually associated both with a lack of coordination and with synchronization in wage bargaining. 5.9. The discussion on the role of institutions and wage bargaining structures in determining wage moderation cannot be dissociated from the debate on labor market flexibility. In the competitive model of the labor market, for example, the balance of supply and demand in the whole market dictates wages. Hicks's Theory of Wages (1932) was the starting point for this argument, but even Hicks himself became skeptical about the overall applicability of the competitive model, as he made clear in the preface to Hicks (1963). There has been much debate whether wages can be considered as fixed by non-competitive pressures. A common argument often used is that in most cases, wages are determined by collective bargaining and also that other non-competitive factors might be important (e.g., efficiency wage and rent- sharing theories). 5.10. Two indicators are often used to measure the extent of unionization in a country. These are (a) union density; and (b) bargaining coverage. Union density is defined as the number of workers who are members of a union, as a percentage of all workers (unionized and non-unionized). Bargaining coverage is defined as the number of workers, unionized or not, who have.their pay and employment conditions determined by a collective agreement, as a percentage of all workers, unionized and.non-unionized.' 5.11. Table 5.1 shows union density and bargaining coverage for some non-Latin American OECD countries in the 1970s, 1980s and 1990s and for some Latin American countries where data is available for the latter two decades. Average union density in the OECD countries.declined during the 1 980s (47%) to 40 percent in the 1990s. The average movement, however, hides a lot of variation. Some countries, such as the US, the UK, Japan, and the Netherlands, have experienced a significant reduction in union density. Other countries, such as Finland and Sweden, have experienced a significant increase in union density ' over the three decades. Also, the cross-country variation is significant. Countries such as France, the US, - 98 - and Spain have very low union density rates (less than 30 percent). On the other hand, the Scandinavian countries have very high rates (all above 50 percent, some around 80 percent). 5.12. Some of the same trends are true for Latin American countries as shown in Table 5.1- union density seems to have decreased sharply in the 1990s. While it is still high in some countries (e.g. Brazil 32%, Argentina 25%, Guyana 25%), in many countries union density is below 10%. Another interesting aspect is that union density is significantly lower, on average, in Latin America than it is in many non Latin OECD countries. Table 5.1 Union Density And Bargaining Coverage Union density Bargaining Coverage Country 1970 1980 1994 1990 1994 OECD... Australia 50 48 41 80 80 Austria 62 56 42 98 98 Canada 31 36 38 38 38 Denmark' 60 76 76 69 69 Finland 51 70 81 95 95 France 22 18 9 92 95 Germany 33 36 29 90 92 Italy 36 49 39 83 82 Japan' 35 31 24 23 21 Netherlands 38 35 26 71 81 New Zealand n.a. 56 30 67 31 Spain 27 19 19 76 78 Sweden' 68 80 91 86 89 Switzerland 30 31 27 53 50 UK 45 50 34 47 47 US 23 22 16 18 18 Latin America and Caribbean... Argentna n.a. 49 25 n.a. 73 Brazil n.a. n.a. 32 n.a. n.a. Chile n.a. 12 16 n.a. 13 Colombia n.a. 11 7 n.a. n.a. Costa Rica n.a. 23 13 n.a. n.a. Guatemala n.a. 8 4 n.a. n.a. Guyana n.a. n.a. 25 n.a. 27 Mexico n.a. 54 31 n.a. n.a. Peru n.a. n.a. 8 n.a. n.a. Uruguay n.a. 20 12 n.a. 22 Venezuela n.a. 26 15 n.a. n.a. 5.13. Bargaining coverage is on average much higher than union density in the OECD (and in the few Latin American countries for which this data is available). While high union density leads to high coverage of collective bargaining, the converse is not true. Countries such as Spain and France have very low union density, yet the coverage of collective agreements is very high. The difference between union density and coverage of collective bargaining is largely attributed to mandatory extensions of collective agreements to non-unionized sectors (OECD, 1994). Brazil's collective bargaining system can be characterized as an intermediate level of collective bargaining. Union density is over 30 percent - among the highest in Latin America (Box 5.1). - 99 - Box 5.1 Trade Unions and Collective Bargaining in Brazil The labor, market in Brazil has developed wage setting institutions.'which allow its classificatipri as an intermediate-centralized . 'economy in terms of international comparisons, i Prior t61 964, collective wage adjustments were defined as part of a coilective labor contract between workers' and em,ployers' unions. For each, professional category represented by a local union, negotiations took,effect on a different day of the year, called 'base date" Trade unions were defined in.terms of industry affiliation and geographical location., After the 1964 'rilitary coup, collective bargaining was mornoe or less replaced by centralized- state wage indexation, and this situaton prevailed until the 1980s. The State wage indexaion system in operabon between 1964 and ,1985 provided for a state minimum,wage to be annually indexed based on the average real wage of the previous 24,months (shortened to'12 months in 1975). This system represented a compulsory incomes policy, and for certain periods between 1964 and 1979 due to systematic underindexation it was moderately successful in lowering inflation. By the end of the 1970s, inflation was growing rapidly in Brazil and this provoked the upsurge of strike activities in the most organized sectors of the economy. The- consequences of the increased labor rribilization 'were two-fold" (i) an immediate reduction in the period between salary adjustments from,12 months, unbl 1979,.to_six months thereafter; and (ii) the retum of direct negotiations between unions'and employers,' which opened the door for wage adjustments'above the past rate of inflation, The re-emergence of politcal democracy in the 1980s combined with labor dissatisfaction with the State ,wage policy led to the development of a new unionism. Unions have slowly increased their role in the wage determinaton process and have experienced an enhancement in their bargaining ability at both the regional and industrial levels.' Union density is quite high - it is estmated to be over 30 percent. In terms of degree of bargaining centralizaton, what has emerged is an intermediate level of collective bargaining. In the late 1980s and early 1990s, control of inflation became much more problematic in. Brazil. One consequence of the generally unsuccessful counter-infationary policy shocks was an increased defensiveness on the part of labor unions, keen to protect real wage levels in the face of deflationary,shocks. This contributed to thebreakdown in coordination and synchronization of wage determinabon, cand'the development of a structure for collective bargainring mu'ch more akin to that prevailing in'westem Europe. .Sburce: Cameiro (1998, 1999); Cameiro and Henley (1998) What Unions Do 5.14. Unions are engaged in many different activities. There is a distinction between two different aspects of union behavior: monopoly and participatory. 5.15. Economists have traditionally argued that unions want to increase wages and "share rents" with employers while preserving employment of their membership. To the extent that rent is available for sharing, however, unions can force firms to give up some profits only if they can monopolize labor supply. This is because firms are willing to give up some of their profits to avoid industrial conflict. However, if nonunion workers can readily replace union workers, the union's bargaining position is substantially weakened. Hence the success of unions depends significantly on the economic environment, as measured by competitive pressure from product markets and nonunion labor markets in which they operate. If unions are successful in getting a wage mark-up, they impose a number of (extemal) costs on the rest of the society, known as the monopoly cost of unions (Medoff and Freeman, 1984). These could include firms trying to pass costs onto the consumer or the fact that more senior union members, who typically have a disproportional influence on the decisions of the union, may institutionalize a seniority principle in relation to - 100 - layoffs and other aspects of deployment such as promotion, recall and training. This can create insider/outsider dynamics that can lead to persistently high levels of unemployment. 5.16. However, these costs are based on the assumption that the labor market would be competitive in the absence of unionism. The reality facing policy makers is far less clear cut than this for several reasons. First, the 'removal' of unions may not expose an underlying perfectly competitive situation in the labor market: instead it may reveal market imperfections on the labor demand side in the form of monopsony.79 Second, when job-specific skills are important (or turnover costs are otherwise high), individual workers have sufficient bargaining power to get a wage mark-up, even in the absence of unions. Finally, it is possible that unions and firms can bargain over wages and employment and enter an efficient contract. This can reduce transaction costs which would be involved in dealing with individual employees.80 5.17. Another viewpoint focuses on the economic (participatorv) benefit of unions. Unions facilitate worker-participation and worker-manager cooperation at the workplace. This can have efficiency-enhancing effects to the benefit of workers and management. These benefits can arise from a number of sources: (a) unions can play an important and beneficial role in communicating the preferences of workers directly to the management, plus participating in the establishment of work rules and seniority provisions in the intemal labor market; (b) unions facilitate procedural arrangements and other agency services that help reduce the likelihood of costly disputes about wage and employment conditions; and (c) unions can increase productivity by providing a channel through which labor can draw to management's attention changes in working methods or production techniques that may be beneficial to both parties. Bargaining Coordination 5.18. Some economists argue that the capacity of unions to determine wages increased over time because unemployment insurance and other labor market institutions which favor welfare-state measures led to reduced wage pressure from the unemployed. Thus, a higher rate of unemployment would be required if a noninflationary path of economic development were to be achieved. The effect of unemployment on the behavior of unions as regards wage determination generally depends on the objective function of the union, which may be strongly influenced by the size and composition of the labor force share a union represents. Two effects of the extent of representation on collective bargaining can be broadly identified: (i) with increasing comprehensiveness, unions gain market power that may be used to push up wages and working standards. At the same time, however, (ii) as organizations encompass larger groups, the negative effects of wage bargaining become endogenous, as suggested by Olson (1982)81. 79Employer can derive monopsony power from the fact that it is costly for a worker to leave the firm (because of the firm-specific human capital they have accumulated) and move to another city to get a new similar job. Therefore, once a worker is hired, he or she would be reluctant to quit and the employer can exploit that. 80 This outcome is based on the so-called 'efficient bargain' model which is termed efficient by assuming that neither the union nor the firm will be made better off without making the other worse off, should a change in the agreed contract be effected; i.e., bargained levels of wages and employment will be on the contract curve and the outcome will be Pareto efficient. 81 As argued by Mancur Olson (1982), special interest groups are most damaging once they have gained a certain amount of power, but little responsibility. -101 - 5.19. The economic effects of unions depend on the way collective bargaining is organized. Of particular interest is the degree of bargaining coordination. Here we outline from a theoretical perspective three aspects of bargaining coordination, namely, centralization, concentration and informal coordination.82 5.20. Centralization of collective bargaining refers to the capacity of the peak union confederation and employers organization to influence wage levels/patterns across the economy. This capacity depends on two aspects. First, at what level bargaining takes place, that is, plant, industry or national level. Second, whether or not the peak organization(s) can control the behavior of their constituent organizations. 5.21. To see the implications of this, we need to take a closer look at what is likely to happen when, say, a company union demands a higher wage. Clearly, to avoid an increase in the product wage, the firm will try to pass on the wage demand as a higher product price. This reaction has two implications. First, a higher product price reduces the real consumption wage of all workers (the externality effect) including those who hold membership of the union. Second, a higher product prce leads to a reduction in the demand for goods produced by the firm and a reduction in employment among unionized workers (the competitive pressure effect). The union, of course, dislikes both of these effects. Accordingly, it moderates its wage demands to the extent that they do not pay off in terms of a higher real consumption wage for the membership or/and to the extent that they put the jobs of too many union members at stake. In other words, if a union and an individual firm bargain over wages, the trade-off between a wage increase and employment greatly depends on the price elasticity of demand for the firm's product, and in the extreme, the power of unions at the company level will be limited by competition in the product market. 5.22. On the other hand, for unions whose utility function comprises the entire labor force, as in the case of centralized national unions, all extemal effects of the union's action enter directly into the union's utility function and, therefore, the negative effects of its actions become endogenous to these unions. As examples of this intemalization mechanism, in an entirely centralized system, if workers lose their jobs and become dependent on benefits, these benefits will be financed by members of the union; thus, excessively high wage increases directly affect the members net wage. Similarly, if wage increases induce inflation, the real wage of the union's members will be directly affected. 5.23. Such considerations of political science arguments have led some economists to suggest that wage bargaining at the company level (decentralized) or at the national level (centralized), leads to favorable macroeconomic outcomes, such as low unemployment, low inflation and high employment rates. Economies with an intermediate-centralized level of wage bargaining, however, suffer from union power that does not internalize negative effects and are, therefore, expected to experience less favorable macroeconomic outcomes (Calmfors and Driff ill, 1988). 5.24. In general, the relationship between centralization of collective bargaining and economic performance is determined by the interaction between the externality effect and the competitive pressure effect. If the extemality effect is important even at the industry level and the competitive pressure effect is unimportant, then the relationship is monotonic, i.e., a more centralized bargaining system produces better economic outcomes. However, if the externality effect is unimportant at the industry level and the 82 The relevant, theoretical iterature has been surveyed in great detail by Calmfors (1993), Moene and Wallerstein (1993a), Layard et al. (1991, ch. 2) and Henley and Tsakalotos (1993). - 102 - competitive pressure effect is important, the relationship may be U-shaped, i.e., national and firm level bargaining produces better outcomes than industry-based collective bargaining. In addition to openness to intemational trade, the relationship between bargaining centralization and economic performance also depends on competition from nonunion labor markets. The general point is that the nature of the relationship depends significantly on the context in which collective bargaining takes place.83 5.25. Union concentration refers to the number and type of unions at the relevant level of bargaining. An industrial relations system based on multi-unionism and closed unions is usually characterized by low concentration because of rivalry between unions that results in less coordinated wage bargaining. Multi- unionism refers to a situation in which many unions offer to organize the same workers. A closed union refers to the situation when workers can only join a particular union of his or her craft/profession/education/job. In contrast, an open union can have members with different education and from different trades. It is not, a priori, clear that bargaining coordination in the form of a concentrated union movement based on single-unionism and open unions leads to more desirable economic outcomes. This will depend on several factors such as whether workers in closed unions are complements or substitutes in production, individual preferences of union members etc.. 5.26. Bargaining coordination needs not be embodied in the formal institutional framework of collective bargaining. It can be informal. Informal coordination, typically, takes two forms. The first form is intemal coordination among employers and/or the employees. At the employer side this involves coordination between industry-based employers' organizations or individual firms. This plays an important role in Japan, Austria, and Switzerland (Soskice, 1990; OECD, 1994). At the employee side, intemal coordination, typically, involves coordination between company- and industry-based unions. The second form of informal coordination is pattem bargaining. Here, a dominate industry or company enters a collective agreement that is followed by other firms and industries. This has been important in, e.g., Germany, where the metal industry, traditionally, has acted as the leader. 5.27. This section reviews a large body of empirical evidence about the economic effects of unions derived from microeconomic data on individual workers and establishments. A significant portion of the evidence relates to OECD countries (mainly the U.S. and the U.K.) but there is also some evidence from some developing countries. 5.28. This section is organized as follows. First, evidence is presented on the union-nonunion wage mark-up. This aspect of union behavior is, by far, the most well researched aspect with over 200 studies in the US alone. Cross-country differences in the average wage mark-up as well as the variation in the wage mark-up across skills, gender, occupation, ethnicity as well as with respect to the underlying economic and institutional environment are examined. This is followed by a review of the effect of unions on other 83 Freeman (1988) argues for care on this issue as centralized corporatist labor markets are primarily found in small countries while decentralized markets characterize larger economies. In this context, any policy recommendation should not concentrate only on a particular set of institutions which work better but rather try to identify specific policies which can succeed in particular environments. - 103- economic variables such as employment growth, hours worked, productivity, job mobility, implementation of new technology etc.. Union Non-union Wage Mark-up 5.29. The union-nonunion wage mark-up (the "wage mark-up") is defined as the difference between the average (nominal) wage of unionized and nonunionized workers with similar individual and workplace characteristics divided by the average wage of a nonunionized worker.84 The estimation of the wage mark- up can be based on data related to wages of individual workers or average sector-wide or occupational wage rates (for example, average wages in different industries or broad sectors - such as manufacturing versus services or manual versus nonmanual workers). As it is generally agreed that the mark-up calculated using sector-wide data are biased upwards, we focus below on studies that have used individual cross-section data. 5.30. A major econometric problem involved in estimating the wage mark-up is to control for all other factors, besides unionization, that affect wages. These factors have typically been proxied by variables such as education, work experience, gender, family status, hours worked, firm size, industry, occupation and so on. However, some of these variables may be highly correlated with union status. For example, the fact that a unionized worker works longer hours may be the result of union negotiated overtime or, to take another example, working in a particular industry may be a critical factor for belonging to a union (such as in the mining sector). Another problem is union-status selectivity. This problem arises because the union- status selection process is not random. For example, workers with high productivity can decide not to join a union or decide to work in the nonunion sector because they hope to get a higher wage than collectively agreed. This phenomenon gives rise to selectivity bias and reduces the reliability of the estimated mark- up.85 Bearing these remarks in mind, we present below a summary of country-specific studies on the wage mark-up as well as cross-country comparisons.86 5.31. In all countries where the wage mark-up has been estimated, it has been found to be non-negative. There is, however, significant cross-country variation as well as variation of estimates within countries. There is also some evidence, albeit weak, that the wage mark-up is, on average, lower in high-income countries than in low- and middle-income ones. 84 In principle, we are interested in comparing the union wage with the wage that would have prevailed in the absence of unions (this is called the "wage gain"). However, the wage gain is unobservable and the literature has typically focused on the wage mark-up. 850ne way to deal with the union-status selectivity problem is to estimate the wage equation and a union-status selection equation simultaneously. Another is to use a panel data set that contains time series of cross-sections for the same set of workers. This makes it possible to control for unobserved heterogeneity. For example, if unobserved characteristics of workers stay constant over time, they can be taken into account by comparing the wage of each worker at one date with that at a later date. 86Extensive discussion on methodological issues on the estimation of the wage mark-up can be found in Lewis (1986), Booth (1995) or Pencavel (1991). -104 - 5.32. More speciically, Table 5.2 presents summary estimates of the wage mark-up for six high-income economies (the US, the UK, Japan, Canada, Australia and West Germany), six middle-income economies (South Korea, Malaysia, South Africa, Brazil and Mexico), and one low-income economy (Ghana). These estimates are based on averages of regression analysis estimates in various studies. Table 5.2 Union-nonunion Wage Mark-up for Selected Countries Country PeriodNear Wage mark-up(%) USA 1963-95 15 UK 1969-95 10 Japan 5 Canada 1969-86 10-20 Australia 1984-87 7-17 West Germany 1985-87 0-6 South Korea 1988 2-4 Malaysia 1988 15-20 South Africa 1993-95 10-24 Mexico 1989 10 Brazil 1992-95 7-11 Ghana 1992-94 21-28 5.33. The evidence is sparser for other OECD countries but some generalizations are possible. The wage mark-up for Australia has been estimated to vary between 7 percent and 17. Similary, the Canadian wage mark-up has been estimated to be in the range of 10 to 20 percent, though most estimates seem to fall in the 10-15 percentage range. In Germany, where most unions are industry unions and work and pay conditions contained in collective agreements are largely extended to nonunionized workers, the wage- mark-up is found to be small, especially for male workers. Similarly, a study on Japan has found a small average wage mark-up of about 5 percent. 5.34. There is much less evidence for developing economies. The mark-up in South Africa has been estimated in three studies from anywhere between 10-24% (Moll, 1993; Dabalen, 1998; and Rouse, 1999). Standing (1992) estimates the wage mark-up for Malaysia to be in the range between 15 and 20 percent, depending on the type of union. These effects are somewhat larger than in most industrialized countries. Standing attributes the high mark-up to the fact that Malaysia non-unionized workers can, in the absence of minimum wage legislation, be vulnerable to very low wages. He concludes that the mark-up can reasonably be in the estimated range despite the fact that the political and economic environment in Malaysia is difficult for unions. 5.35. For Ghana, estimates using three surveys of manufacturing firms in 1992, 1993 and 1994, find that the wage mark-up varies between 21-28%. Again, these estimates are significantly higher than those from the industrialized countries. For Mexico, Panagides and Patrinos (1994) find a 10 percent membership wage mark-up after having controlled for an large number of income generating characteristics in a cross- section sample of unionized and non-unionized non-agricultural workers in 1989. They attribute the relative low wage mark-up to the fact that the 1980s were a particular difficult time for unions in Mexico due to recession coupled with government austerity measures. It is, however, unclear if this can explain the result. First, recession and austerity measures should affect unionized sections and nonunionized sectors alike. Second, one could argue that unions would be in a better position than nonunionized workers to resist downwards pressure on wages, thereby increasing the wage mark-up. - 105 - 5.36. Evidence for the case of Brazil is presented by Arbache and Carneiro (1999), based on household surveys for 1992 and 1995. Their results are suggestive that trade unions do have an important role in wage determination, as they found union wage mark-up that varied from 6.7% to 11.3%. Furthermore, these authors have found that a share of between 18.8% and 29.6% of the raw wage differentials between unionized and nonunionized workers were not explained by differences in observable productive endowments, occupation and other variables. This suggests that if a worker changes his/her union membership status from nonunion to union s/he is liable to perceive a wage increase in Brazil. 5.37. Gender wage gaps. Unions are one among many determinants of the gender wage gap. In his survey of the U.S. literature, Lewis (1986) concludes that there is very little, if any, difference between the mark-up for female and male workers. This has been further confirmed by more recent studies and the same result emerges from Australian studies (Christie, 1992; Mulvey, 1986). The evidence from other OECD and middle-income countries, however, unambiguously supports the view that the wage mark-up is greater for women. Nakumura et al. (1988), in their study of Japan, find a wage mark-up of 10% for women but fails to find any for men. Likewise, Schmidt (1995) shows that the small average wage mark-up in West German is mainly due to a wage mark-up among unionized female workers. The findings for Mexico suggest that the mark-up for women is 9.8 percentage points higher than that of men with similar characteristics. Similarly, in South Africa, the wage mark-up among black blue-collar workers in the mid 1980's was about 11 percentage points higher for women than for men. A comparison of the wage ratio of male and female workers in non-unionized and unionized firms in Malaysia suggest that the presence of a union reduces the ratio of male to female wages, leading to the conclusion that women gain more than proportionately from unionization (Standing, 1992). 5.38. Skill wage gaps. Collective bargaining can insert a wedge between worker productivity and wages. Though this can be desirable as part of avoiding wage inequality from a societal point of view, it can also distort the relative wage of skilled and unskilled workers or the relative rewards to different types of jobs. There is some evidence to back up this claim. In the U.S. and the U.K., evidence shows that manual workers get a larger mark-up than non-manual workers. In South Africa, the wage mark-up to workers with different skills varies between different ethnic groups. For example, the average wage mark-up for unskilled non-white workers is 19 percent, while the wage mark-up to skilled workers is practically zero. On the other hand, semi-skilled and skilled white workers got a wage mark-up of 13 percent in 1985 (Moll, 1993) implying that black unions tended to compress wages by skill level. Though Dabalen (1998) finds estimates that are somewhat smaller, his study confirms the general pattern found by Moll (1993). In Malaysia, unions reduce the intra-firm wage differential between skilled and unskilled workers. In particular, industry unions tend to reduce the differential between workers with different skills more than company unions do (Standing, 1992). 5.39. Collective bargaining characteristics. Some studies have also tried to estimate the wage mark-up in a cross-country context. These studies have used comparable individual data to estimate the wage differential for a set of OECD countries. These analysis, for the most part, come up with similar results to the ones stated above. Using these data, analysis which has tried to correlate the wage mark-up across countries with characteristics of collective bargaining87 come up with interesting findings. This analysis 87 These characteristics are - degree of centralization/coordination of collective bargaining, union density, and bargaining coverage. - 106 - shows that there is a tendency for countries with centralized/coordinated or decentralized/uncoordinated bargaining systems to have a higher mark-up than countries with semi-coordinated systems. Union density appears to be largely uncorrelated with to the wage mark-up (correlation coefficient 0.02) while bargaining coverage is negatively correlated with the mark-up (correlation coefficient -0.58). This seems to point to the conclusion that the more workers become unionized/covered by collective agreements, the lower the mark- up they can secure.88 This may be so because the labor supply in the non-covered sector decreases when more workers become covered, thereby pushing the nonunion wage up. Somewhat related, using a time- series comparison, this analysis also shows that the wage markup seems to be lower for countries where the density rates have been increasing or constant between 1970-1993 (an average mark-up of 4.5%) as compared to mark-ups in countries where the union density rates have declined over the same time period (an average mark-up of 11.8%). Other Union Effects 5.40. This section reviews the evidence on the impact of unions on employment growth, productivity, technology and profitability. 5.41. Employment Growth. Available evidence from the US, the UK, Canada, Malaysia and Jamaica suggests that employment generally grows more slowly in unionized firms than in nonunionized firms. Studies from the U.S., the UK and Canada typically find a growth differential in the range of 3-5 percentage points per year in favor of nonunionized firms.89 The evidence on the employment growth differential is mixed in Malaysia. The employment growth differential is about 5 percentage points per year in firms that bargain with industrial unions but insignificant in firms that deal with a company union (Standing, 1992). Rama (1998) estimates the employment differential between sectors with high and low union membership rates for Jamaica over the period 1986-93 to be 2-5 percentage points per year. Rama (1998) provides a number of potential explanations for the observed employment growth differential including - (a) it takes time and effort to organize a union. As a consequence, at a given point in time, old firms are more likely to be covered by unions than young firms are. If young firms expand faster than old firms, we would expect to observe higher employment growth in the young, nonunionized firms; (b) unions are more likely to be concentrated in sectors that enjoy larger rents. If these sectors are less dynamic and their activities limited by the size of the domestic market, employment would tend to grow slower in these sectors; and (c) labor costs grow faster and productivity slower in unionized firms than in nonunionized firms. 5.42. Voluntary tumovers, tenure and layoffs. Evidence from the U.K., the U.S., Japan, Australia, and Malaysia unanimously shows that voluntary tumovers (measured by the quit rate) are lower and that job tenure is longer in unionized firms than in nonunionized ones. Freeman and Medoff (1984) estimate the welfare gain associated with a reduction in labor turnover to be by equivalent to a 0.2-0.3 percent increase in GDP in the U.S. in the 1980s. They base their estimate on a calculation of the how much the reduction 88 As argued by Arbache and Cameiro (1999), the case of Brazil seems to contradict the intemational evidence, as in the Brazilian context, the more workers become unionized the larger is the mark-up they can secure. 89 Boal and Pencavel (1994), however, find, in their study of coal-mining data from West Virginia over the period 1897-1938, that the employment growth differential is approximately zero if the variation in working days is explicitly taken into account. Machin and Wadhwani (1 991) find, using UK data, that employment grows faster in unionized establishments in the 1970s. Blanchf lower and Burgess (1996) find a negative effect of unions in the UK, but not in Australia. -107 - in exit is worth to firms and to workers. For unionized firms, they estimate the gain to be equivalent to a 1-2 percent reduction in costs. While voluntary tumover tends to be lower in unionized firms than elsewhere, unions increase the use of layoffs, in particular temporary layoffs. Unions also significantly alter the firm's choice between layoffs, wage and hours adjustment in response to the business cycle fluctuations. Unionized firms adjust through temporary layoffs rather than reductions in weekly hours (work sharing) or wages.90 In particular, unionized (blue-collar) workers are 50 to 60 percent more likely to be laid off temporarily than nonunionized workers. An explanation for this relates to the fact that junior workers can be more easily laid off. Senior workers typically have more influence on the union's policy than junior workers. Faced with the choice between a reduction in their eamings or temporary layoff of junior workers, unions are likely to prefer layoffs. Another explanation, especially in OECD economies, with well- developed income support schemes for the unemployed, is that the cost of temporary layoffs can be shifted on to the unemployment benefit system. As long as there is less than 100 percent experience rating, i.e., as long as firms contribute to unemployment benefits an amount less than the costs of the unemployment they generate, those firms with above-average layoffs are subsidized at the expense of firms with below- average layoffs. 5.43. Profitability. It is a commonly held view that unions reduce profitability of firms because they appropriate part of the rent that would otherwise have been available to shareholders. The fact that unions are able to get a wage mark-up supports this view. It is, however unwise to deduce the effect of unions on profitability by looking at the wage mark-up alone. By increasing moral and job-satisfaction among workers and facilitate worker-employer cooperation, unions can contribute positively to profitability. Therefore, instead of trying to capture a given rent, unions may help create profits from which they can achieve wage gains. The impact of unions on profitability has been estimated in a large number of studies. Bellman (1992) surveys 14 studies from the U.S.. All studies find that unions have a negative impact on profitability. The impact tends to be larger in industries or firms that have some monopoly power in the product market. Booth (1995) surveys seven British studies. While a few studies find that unionism has no impact on profitability, the general impression is that unions have a negative, often significant, impact on profitability in manufacturing firms. The adverse impact is larger when firms have product market power. A study on Japan (Brunello, 1992) finds that unions reduce the rate of return on equity by 20-25 percent and the ratio of profits to sales by about 40 percent. 5.44. Technology. The available empirical evidence suggests that new technology is adopted as fast in unionized firms as in nonunionized ones and that unions have little impact either way on technological innovations in firms. Keefe (1992) surveys research on the relationship between unions and technological change in the U.S., U.K. and Canada in the 1980s. He concludes that unions have no effect on firm's use of advanced manufacturing and microelectronic technologies and that "in most cases unions welcome technological modernization. Betcherman (1991) reaches a similar conclusion in his study of the impact of unions on technological change in Canada in 1980-85 but adds the observation that unions do have an impact on the waytechnological change is being implemented. In particular, he finds that unionized firms were more likely to introduce technological changes that nonunionized firms for of cost-cutting or production control reasons. Likewise, Machin and Wadhwani (1991) and Latreille (1992) find a small positive impact of unions on the introduction of new microelectronic equipment in U.K. firms in the mid- 90Temporary layoffs refer to a situation in which a worker is laid-off for a shorter period of time (less than a mouth) and is recalled or rehired by the same firm. - 108 - 1980s. Finally, Standing (1992) in his study of industrial relations in Malaysia concludes that unions actually stimulate capital, product, and labor process innovations. - v Im.0 M%, ok 4 I - 5.45. In this section, we review studies that aim at capturing the macroeconomic impact of the institutional framework of collective bargaining. While there are differences in bargaining institutions across sectors and changes do take place in the long-term within a given economy, the institutional framework is, by and large, unchanged in the short- and medium run. To investigate the impact of collective bargaining on macroeconomic performance, it is, therefore, necessary to take a comparative approach and look at cross-country evidence. Though there is scattered evidence from developing countries, most of the reviewed studies focus on the group of OECD countries. Union Density and Bargaining Coverage 5.46. The relationship between union density and bargaining coverage and a variety of economic performance indicators has been examined extensively, especially for OECD countries. From a theoretical perspective, the impact of union density on economic performance is unclear. Union density determines the number of (unionized) workers that can be call upon to strike and thereby it is a proxy of the bargaining power of unions. In countries with high union density, unions are more likely to succeed in pushing (union) wages up, leading to less employment, more unemployment and inflation. The negative impact of unionization on economic performance may, however, be reduced to the extent that unions participate in productivity-enhancing activities (by giving voice to dissatisfied workers) at the firm level. 5.47. Table 5.3 summarizes the findings of the studies from OECD countries that have investigated the relationship between union density and economic performance. Union density perse (i.e., for given bargaining coverage and for given level of bargaining coordination) appears to have little or no impact on comparative labor market performance measured by the unemployment rate, inflation, the employment rate, real eamings growth, the level of compensation, labor supply, adjustment speed to wage shocks, real wage rigidity, labor and total factor productivity.91 91 Blanchflower (1996), who uses country-specific microeconomic data to analyze OECD countries, find similar results. - 109 - Table 5.3 Union Density and Economic Performance in the OECD Countries Study and years Performance indicator Result OECD (1997) Unemployment rate Union density increases the employment rate but 1980-94 Inflation has no effect on the unemployment rate, inflation, Employment rate and real eamings growth. Union density reduces Real eamings growth earnings inequality. Earnings inequality OECD (1997) Unemployment rate Union density reduces eamings inequality in 1990 1980-94 Inflation and 1994. Weak indication of a positive relationship Employment rate between union density and the employment rate and Real eamings growth a negative relationship between union density and Eamings inequality real earnings growth in 1980 but not in other years. Freeman (1988) Unemployment rate Union density has no effect on the unemployment 1979-85. Employment rate rate, the employment rate, and compensation. Compensation Scarpetta (1996) Unemployment rate Union density increases unemployment, in particular 1983-93 youth and long-term unemployment but no control for bargaining coverage is made. Nickell and Layard Unemployment Union density increases total unemployment but has (1997) and Nickell Labor supply no separate effect on short- and long-term (1997), 1983-88, Productivity growth unemployment. Union density has no effect on labor 1989-94 supply and productivity growth. Bean et al. (1986) Adjustment speed Union density has no effect on either adjustment 1956-85 Real wage rigidity speed (to wage shocks), or on real wage rigidity. Layard et al. (1991) Real wage rigidity Union density has no effect on real wage rigidity. 1980-94 Scarpetta (1996) Hysteresis in unemployment Union density increases unemployment persistence 1970-93 but no control for bargaining coverage is made. 5.48. There is, however, one significant exception to the general result that the association between union density and economic performance is weak: Union density compresses the wage distribution and reduces earnings inequality.92 This confirms the evidence from micro data discussed earlier. 5.49. The picture looks quite different when we consider the association between union density and economic performance in developing countries. Evidence (Rama, 1995; McGuire, 1996; Fields and Wan, 1989) from developing countries in Latin America, the Caribbean, and South East Asia suggests that union density has a negative impact on output and employment growth. Rama (1 997a) argues that the difference between the impact of union density in developing and OECD countries is caused by differences in the general economic and political environment. That is, the adverse effect of unions in developing countries may be caused not so much by what unions do but rather because of the context in which they are doing it. Hence, if unions operate in an environment of generally ill designed labor and product market regulation in which rent-seeking is a profitable business (also for unions), it is no wonder that the correlation between union density and economic performance is negative. Likewise, if unions operate in the context of an unstable political environment, the incentive to "invest" in real wage restraint in exchange for (expected) future retums is low and union militancy comes at no surprise. 92 Freeman (1988) also shows that wage dispersion tends to be lower in corporatist economies as opposed to economies with a decentralized labor market. -110 - 5.50. While union density relates to the number of unionized workers, bargaining coverage relates to the total number of workers that have their wage and employment conditions determined by collective agreements. In Table 5.4 we review the studies that have investigated the association between bargaining coverage and macroeconomic performance. 5.51. After controlling for union density and bargaining coordination, countries with high bargaining coverage (such as Austria, France and Finland), ceters panibus, experience higher unemployment rates, lower employment rates, and more inflation than countries with low bargaining coverage (such as USA, Japan and Canada). Moreover, high bargaining coverage seems to increase the supply of labor but has no effect on labor and total factor productivity (Nickell and Layard, 1997). Finally, high bargaining coverage is associated with higher real earnings growth and a reduction in eamings inequality. 5.52. These findings suggest that an increase in coverage at a given level of union density has a greater impact than an increase in density at a given level of coverage (at least in OECD countries). One explanation of this result is as follows. In those parts of the economy to which bargaining results are extended, only the monopoly effect of unions is present. The economic effects of the wage mark-up is, therefore, not compensated by worker/management cooperation or other institutional factors that could lead to productivity gains. On average, bargaining coverage can, therefore, affect unemployment, employment and inflation adversely, while the impact of unionization perse can be less significant. However, there is little evidence available on this issue from developing countries. Table 5.4 Bargaining Coverage And Economic Performance: A Summary Of The Relevant Studies Study and years Performance indicator Result Henley and Tsakalotos Unemployment rate The performance of highly centralized or 'corporatist' (1993) Inflation economies, such as Sweden and Austria, have in the Real eamings growth past been superior to those of the EC members. Eamings inequality Appelbaum and Unemployment rate Finds evidence of a u-shaped relation between the Schettkat (1996) Inflation degree of centralization of wage bargaining and labor Employment rate market performance. Countries with intermediate degrees of centralization performed worse than those at the extremes. OECD (1997) Unemployment rate Bargaining coverage increases unemployment, 1980-94 Inflation inflation and real eamings growth, and reduces the Employment rate employment rate and eamings inequality. Real eamings growth Eamings inequality OECD (1997) Unemployment rate Bargaining coverage increases unemployment only in 1980-94 Inflation 1994, reduces the employment rate in only 1990 and Employment rate 1994 and eamings inequality in 1994. Otherwise it has Real eamings growth no impact on economic performance. Earings inequality Siebert (1997) Unemployment rate Shows the adverse impact on employment creation of Inflation institutional rigiddies, centralized collective bargaining, Employment rate high coverage rates, and generous benefits and Real eamings growth protection. Jackman (1993) Unemployment rate Bargaining coverage increases unemployment 1983-88. ___ Nickell and Layard Unemployment rate Bargaining coverage increases both short- and long- (1997), Nickell (1997) Labor supply term unemployment and labor supply but has no effect 1989-94 Productivity growth on productivity growth. -111 - Bargaining Coordination and Comparative Economic Performance 5.53. Aidt and Tzannatos (2001) have reviewed 26 studies93 that have examined the relationship between bargaining coordination and economic outcomes in a cross-country context since 1970 - again most of this evidence is from OECD countries. The empirical literature on bargaining coordination and economic performance has focused on two hypotheses: (a) Coordinated collective bargaining leads to better economic outcomes compared to semi-coordinated collective bargaining which, in turn, performs better than uncoordinated collective bargaining; and (b) The hump hypothesis - Semi-coordinated collective bargaining leads to worse economic outcomes than both coordinated and uncoordinated collective bargaining. 5.54. According to their findings: (i) countries with coordinated collective bargaining systems tend, ceteris parnbus, to have lower unemployment rates than other countries. Studies that use composite measures of unemployment (such as Okun's index, which controls for inflation, and the open economy index, which controls for current account deficits) show the same tendency; (ii) countries with a high level of bargaining coordination tend to have a more compressed wage distribution. This seems to be the most robust result. It can be attributed to a number of causes, including egalitarian bargaining; a reflection of the fact that centralized bargaining reduces the scope for firm- and/or industry-specific factors to enter the wage bargaining; or a reflection of a concem for social insurance. 5.55. The hump hypothesis has been tested, explicitly, in a number of studies - accounting for nearly half of the sub-studies. The evidence in favor of the hump hypothesis is weak. Overall, only 21 out of the reviewed sub-studies (around 20%) statistically "confirm" the hump hypothesis. To test whether this may just be an artifact of the differences in the underlying econometric methodology and test procedure, Aidt and Tzannatos have tested whether the estimation approach used affects the results. Though there are some differences, the main conclusions remain: irrespective of the test used, most sub-studies do not provide statistical evidence of the hump hypothesis. 5.56. Bargaining Coordination and Labor Market Flexibility. Studies have also attempted to examine the relationship of bargaining coordination with labor market flexibility (e.g. real wage rigidity, adjustment to wage shocks etc.). Most of the evidence in this regard comes from OECD countries (Aidt and Tzannatos, 2001) and seems to support the relationship that the higher the level of bargaining coordination, the less flexible are the real wages. 5.57. Hysteresis (or persistence) in unemployment, arises when high unemployment in the current period tends to cause high unemployment in the future. Persistence can arise because of union membership (insider-induced hysteresis)94 or due to loss of skills or because capital depreciated during recession takes a long time to recover (outsider-induced hysteresis). Layard et al (1991) find that higher level of worker coordination increases persistence while higher level of employer coordination reduced persistence. 93 These 26 studies have been broken up into 125 sub-studies which define a relationship between a specific indicator of bargaining coordination vis-a-vis a specific economic indicator. 94 Only the group of insiders (e.g. union members and employed workers) counts in the wage bargaining. When the insiders are reduced in number (e.g., after layoffs in a recession), they can push for higher wages in the next bargaining round and cause unemployment to remain persistently high (insider-induced hysteresis). - 112 - Further research has also shown that the employer effect is greater. In addition, evidence tends to show that countries with a high or low level of coordination are associated with low persistence, while those with semi-coordinated systems have higher persistence. 5.58. Cameiro (1999) characterizes the Brazilian bargaining structure as semi-coordinated (intermediate- centralized). Cameiro's analysis of wage determination in Brazil between 1985-1993, conforms with the insider-outsider hypothesis in that high levels of current employment lead to subsequent wage moderation - evidence of insider induced hysteresis. His results seem to show that unions may be more interested in protecting the incumbent workforce rather than defending the interests of the unemployed and favorable demand shocks are likely to be converted into wage gains rather than employment gains. Furthermore his results show a positive ucatch-up" effect - suggesting that strong trade unions have the capacity to use their bargaining power to appropriate rents. 5.59. This paper has attempted to layout the evidence on the economic aspects of freedom of association by reviewing the intemational evidence derived from the microeconometric analysis of the economic impact of unions as well as evidence focusing on the relationship between different systems of collective bargaining and macroeconomic and overall labor market performance. Most of the evidence presented is from OECD countres, though there is some analysis of these issues for developing countries too. 5.60. The microeconomic impact of unionism consistently points towards the existence of wage mark- ups. This evidence also points towards the existence of wage compression - the male-female wage gap is usually lower when unions are present - implying less discrimination, but at the same time, the wage gap between unskilled and skilled workers is lower - implying a distortion in relative wages. While employment growth and profitability is usually lower in unionized firms, they are also associated with more rapid adoption of technological innovations. Based on the evidence, it is clear that the impacts are context- specific and no general conclusions about the net costs of unionism can be reached. Depending on the economic, institutional and political environment in which unions and employers interact, unionism can contribute negatively or positively to economic performance. 5.61. Similarly, the macroeconomic impact of collective bargaining is hard to disentangle. In OECD countries, evidence points to union density having little or no impact on indicators of comparative labor market performance - other than wage distribution. Increased bargaining coverage, on the other hand, seem to affect unemployment and employment adversely. However in developing countries, evidence suggests a negative relationship between union density and employment growth - it is possible that these results could be more due to the environment that unions operate within rather than what unions do. 5.62. With respect to bargaining coordination - most evidence from OECD countries seems to show that countries with a high or low level of coordination are associated with low persistence in unemployment, while those with semi-coordinated systems have higher persistence (hysteresis). Evidence presented from Brazil on this issue give evidence of insider-induced hysteresis (e.g. due to union membership). These results seem to show that unions may be more interested in protecting the incumbent workforce rather than supporting interests of the unemployed. Similar to the micro argument, the macro argument shows that the impact of collective bargaining on various aspects of economic performance depends on the economic and political environment in which collective bargaining takes place. -113 - References Aidt, T. and Z. Tzannatos (2001 forthcoming). Core Labor Standards and the Freedom of Association: Economic Aspects. Working Paper. The World Bank. Appelbaum, E. and R. Schettkat (1996). The Importance of Wage-Bargaining Institutions for Employment Performance, in: G. Shmid, J. O'Reilly, and K. Sch6mann (eds.), Intemational Handbook of Labour Market Policy and Evaluation, Chelteham: Edward Elgar. Arbache, J. and F.G. Cameiro (1999). Unions and Interindustry Wage Differentials. World DeveloPment vol.(27), pp. 1875-83. Bean, C., P. Layard and S. Nickell (1986). The Rise in Unemployment: a Multi-Country Study. Economica vol. (53), pp. 1-22. Bellman, D., 1992, Unions, the Quality Of Labor Relations, and Firm Performance, in: L. Mishel and P. B. Voos (eds.), Unions and Economic Competitiveness. New York: M. E. Sharpe. Betcherman, G. (1991). The Effect of Unions on the Innovative Behavior of Firms in Canada. Industrial Relations Joumal vol. 22(2), pp. 142-151. Blanchflower, D. (1996). Product Market Competition, Wages And Productivity: Intemational Evidence From Establishment Level Data. Annals d'Economie et de Statistique vol. 41/42, pp. 219-254. Blanchflower, D. (1997). Changes Over Time In Union Relative Wage Effects In Great Britain and the United States. Working Paper no. 6100, NBER, Cambridge, MA. Blanchflower, D.G., and S. Burgess, 1996, New Technology and Jobs: Comparative Evidence From a Two Country Study, in: B. Hall, M. Doms, and F. Kramarz (eds.) Economics of Innovation and New Technology. National Academy Press. Booth, A. (1995). The Economics of the Trade Union. Cambridge University Press, Cambridge, U.K.. Brunello, G., 1992, The Effect of Unions on Firms In Japanese Manufacturing. Industrial and Labor Relations Review vol. 45, pp. 471-487. Calmfors, L. (1993). Centralization of Wage Bargaining and Macroeconomics Performance: A Survey. Seminar Paper No. 536, Institute for Intemational Economic Studies, Stockholm University. Calmfors, L. and J. Driff ill (1988). Bargaining Structure, Corporatism and Macroeconomic Performance. Economic Policy, vol. 9, pp. 14-61. Cameiro, F. (1998). Productivity Effects in Brazilian Wage Determination. World Development vol. 26(1), pp. 139-153. Cameiro, F. and A. Henley (1998). Wage Determination in Brazil: The Growth of Union Bargaining Power and Informal Employment The Joumal of Develonment Studies vol. 34(4), pp. 117-138. Cameiro, F. (1999). Insider Power in Wage Determination: Evidence from Brazilian Data. Review of Development Economics vol 3(2), pp. 55-169. Christie, V., 1992, Union Wage Effects and the Probability of Union Membership. Economic Record vol. 68(200), pp. 43-56. Dabalen, A. (1998). The Effect of Unions on Wages in South Africa: Repeated Cross-Section Estimates. Working Paper, The University of California, Berkeley. -114 - Daniels, W. (1987). Workplace Industrial Relations and Technological Change. Frances Printer, London. Fields, G., and H. Wan, Jr. (1989). Wage-Setting Institutions and Economic Growth. World DeveloPment vol. (17), pp. 1471-1483. Freeman, R. and J. Medoff (1984). What Do Unions Do? Basic Books, New York. Freeman, R. (1988). Labor Market Institutions and Economic Performance. Economic Policy vol. (3), pp. 64-78. Henley, A., and Euclid Tsakalotos (1993). Corporatism and Economic Performance. Edward Elgar, U.K. Hicks, R. (1963), The Theory of Wages, (1st edn. 1932), London: Macmillan. Jackman, R. (1993). Mass Unemployment: Intemational Experience and Lessons for Policy. CEPR, LSE, Discussion Paper no. 152. Keefe, J. (1992). Do Unions Hinder Technological Change? in: L. Mishel and P. Voos (eds.), Unions and Economic Competitiveness. M. E. Sharpe, New York. Layard, R., S. Nickell, and R. Jackman (1991). Unemployment. Oxford University Press. Latreille, P. (1992). Unions and the Inter-Establishment Adoption of New Microelectronic Technologies in the British Private Manufacturing Sector. Oxford Bulletin of Economics and Statistics vol. 54(1), pp. 31-51. Lewis, H. (1986). Union Relative Wage Effects: A Survey. University of Chicago Press. Machin, S., and S. Wadhwani (1991). The Effect of Unions on Organizational Change and Employment. Economic Joumal vol. (101), pp. 835-54. Moene, K.O., M. Wallerstein (1993). Technical Progress, Wage Compression and the Centralization of Collective Bargaining. Unpublished paper, University of Califomia, Los Angeles. Mulvey, C. (1986). Wage Levels: Do Unions Make a Difference? In J. Niland, (eds.) Wage Fixation in Australia. Allen and Urwin Press, Sydney. Moll, P. (1993). Black, South African Unions: Relative Wage Effects In Intemational Perspective. Industrial and Labor Relations Review vol. 46(2), pp. 245-262. Nakumura, K, Sato, H. and Kamiya, T. (1988). Do Labor Unions Really Have a Useful Role? Tokyo: Sago Rodo Kenkyujo. Nickell, S. (1997). Unemployment and Labor Market Rigidities: Europe versus North America. Joumal of Economic Perspectives vol. 11 (3), pp. 55-74. Nickell, S., and R. Layard (1997). Labor Market Institutions and Economic Performance. Center for Economic Performance, Discussion paper series no. 23, Oxford University. O' Connell, L. (1999). Collective Bargaining Systems in Six Latin American Countries: Degrees of Autonomy and Decentralization. Working Paper, Inter-American Development Bank. OECD (1994). Employment Outlook. Paris. OECD (1997). Employment Outlook. Paris. Olson, M. (1982). The Rise and Decline of Nations. Economic Growth, Stagflation, and Social Rigidities. New Haven/London: Yale University Press. - 115 - Panagides, A., and Patrinos, H. (1994). Union-Nonunion Wage Differentials In the Developing World: A Case Study of Mexico. Policy Research Working Paper 1269, The World Bank, Washington D.C.. Pencavel, J. (1991). The Origins of Trade Union Power. Oxford University Press. Rama, M. (1995). Do Labor Market Policies and Institutions Matter? The Adjustment Experience in Latin America and the Caribbean. Labour, special issue, pp. 243-268. Rama, M., 1997a, Trade Unions and Economic Performance: East Asia And Latin America, in J. McGuire, (ed.), Rethinking Development In East Asia and Latin America. Center for Intemational Studies, University of Southern Califomia. Rama, M. (1998). Unions and Employment Growth: Evidence From Jamaica. Working Paper, World Bank, Washington D.C.. Sapsford, D. and Z. Tzannatos (1993). The Economics of the Labour Market. MacMillan Books, London. Scarpetta, S. (1996). Assessing the Role of Labor Market Policies and Institutional Settings on Unemployment: A Cross-Country Study. OECD Economic Studies, vol. (26), pp. 43-98. Schmidt, C. (1995). Relative Wage Effects of German Unions. Memo, Selapo, University of Munch. Siebert, H. (1997). Labor Rigidities: At the Root of Unemployment in Europe. Journal of Economic Perspectives, vol. 11 (3), pp. 37-54. Soskice, D. (1990). Wage determination: The Changing Role of Institutions in Advanced Industrialized Countries. Oxford Review of Economic Policy Vol. 6 (4), pp. 36-61. Standing, G. (1992). Do Unions Impede or Accelerate Structural Adjustment? Industrial Versus Company Unions In an Industrializing Labour Market. Cambridge Journal of Economics vol. 16, pp. 327-354. - 116 - Appendix An Analytical Framework Assuming a closed economy, Figure 5.1 summarizes the labor market outcomes for different labor market structures, conditioned on the govemment's macroeconomic constraints regarding output and price stabilization targets. There are three possible situations on each axis. Along the horizontal axis, the govemment 'loss function" (see Carneiro et. al., 1999 for details) - assumed to represent the government's macroeconomic constraints - is typified. This affects labor market outcomes in different ways, depending on whether the economy is undergoing a stabilization process, fiscal adjustment, or has reached a path of sustained growth. The impact of govemment actions on labor market outcomes will depend also on the structure of the labor market, especially how wages are set and employment determined. These are broadly classified as: decentralized, intermediate-centralized, or centralized. Finally, the extent to which the economy is closed or open to intemational flows of goods and capital will interact with the labor market structure in influencing labor market outcomes. Decentralized bargaining is undertaken at the firm level, with no unions (or many small unions) so that employment and wage levels are at their equilibrium levels. With bargaining power depending on worker productivity and wages on offer, labor markets function as in the textbook case of perfect competition. Eamings dispersion tends to be moderate; sector affiliation may be an important factor affecting the eamings of workers. Centralized bargaining is at the other extreme, in terms of labor market structure though not, as the table illustrates, in terms of labor market outcomes. Wage and employment determination in the formal sector occurs at the national level, with a few central unions representing all formal sector workers (unionized and non-unionized), bargaining with a few employer federations, under the mediation of the govemment. As agents are powerful in this case, the macroeconomic consequences of their demands will be promptly visible and there is a tendency towards the market solution. Nevertheless, as wage restraint tends to be agreed on, there is also scope for some social fairness and wage and employment levels can be higher than in the decentralized case, with redistribution of rewards from capital to labor. Earnings dispersion tends to be lower basically because of this sense of social faimess that tends to arise in the centralized bargaining. Political scientists usually name this scenario as the corporatist solution. The intermediate-centralized bargaining structure emerges as the worst scenario. In this case, bargaining takes place at the industry level with deleterious consequences in terms of macroeconomic performance. The explanation for the poorer performance relative to the two polar cases is usually the existence of poody coordinated monopolistic power in the labor market that constrains the successful operation of either competitive market forces or corporatist coordination. In this context, trade unions' bargaining power may be quite strong if high hiring and firing costs restrict rapid labor tumover. Insider bargaining power is also high meaning that workers perceive that by pressing for higher wages the firm's product price may increase without however affecting the aggregate price level and therefore their real take- home pay. As bargaining is at the industry level, an individual firm will not achieve a competitive advantage over the others and, assuming that industry demand is relatively inelastic, all the firms believe that the overall employment consequences will be insignificant. Overall, therefore, the situation of intermediate collective bargaining with strong trade unions may be quite conducive to poor employment and inflation performance, with rent-sharing being a pervasive characteristic. Eamings inequality is usually high in this scenario. - 117 - The changes in labor market indicators in any country undergoing macroeconomic adjustment without labor reforms - as appears to have been the case with Brazil and Argentina in the 1990s - is illustrated by moving across columns for any given row. In general it appears that regardless of labor market structure, stabilizing and adjusting countries will experience improvements in the labor indicators under consideration. Decentralized Wage Bargaining: During stabilization, average earnings, inequality, employment, and unemployment rates tend to be at moderate levels. As the undergoes fiscal adjustment, these indicators reach their worst levels due to the contraction in aggregate demand. Once the sustained growth path is achieved, economic indicators reach optimal levels, with the exception of earnings inequality due to a possible sector affiliation factor (viz., intersectoral wage differentials). Intermediate-Centralized Bargaining: Under this scenario, a move from stabilization towards sustained growth will yield modest results at best. The most obvious consequence will be the persistence of eamings inequality as the consequence of pervasive rent-sharing in the economy. Moving from stabilization to fiscal adjustment will still mean high average wages, high unemployment and low employment growth. Earnings inequality remains significant, but due to the reduction in the surplus of rents in the economy it will not be very high. Once the path of sustained growth is reached, rent-sharing practices retum, provoking increasing eamings inequality. Centralized Bargaining: In the corporatist solution, optimal economic indicators are reached in the case of sustained growth. The difference with regard to the decentralized case is in the profile of eamings inequality, which tends to be lowest under centralized collective bargaining. Movements down rows. In line with the framework we have proposed, desirable moves would be those all the way down from decentralized to centralized collective bargaining, since i tends to generate optimal solutions with social faimess. A second best solution would be the reverse move all the way up from centralized to decentralized collective bargaining; this usually happens because of declining public funds directed to social welfare. And the worst scenario would be a move towards the intermediate-centralized collective bargaining structure. In the case of an open economy under the same circumstances, foreign competition will affect labor market outcomes mostly in the intermediate-centralized collective bargaining situation. The main consequence is that industry demand will become relatively more elastic, reducing insider bargaining power and the capacity of firms to pass on to prices any wage increases that are allowed to unions. Nonetheless, there will still prevail a situation in which unions have relatively more bargaining power than in the two polar cases of centralization or decentralization. Insider workers will continue to be able to manipulate tumover costs and obtain wage increases in excess of the equilibrium wages. The overall consequence is low employment growth and high unemployment, with the degree of eamings inequality remaining quite high. The worst scenario, therefore, will be during fiscal adjustment under intermediate-centralized collective bargaining, in the open economy framework. -118 - Figure 5.1 lUacro-economic Constraints and Labor Market Structure W | Scenarios | Stabilization 1 | Fiscal 1 I Sustained I ______________ I Adiustment J Growth 4 72° Firm-Level eo: Moderate co: Lowest (o: Optimal . | Wage and ow: Moderate aw: High ow: Moderate Employment F: Moderate e Lowest e :Optimal { Setting 0: Moderate u: Moderate u : Optimal . Closed oa: Highest c: High o: High a {yiJt Economy: ow: Highest ow: Medium ow: High . ,bA Industry-Level £ : Lowest F_: Low F_ Moderate O3 tR Setting ri Hinhpt _ =H_inh _ * Hinh --fBRA2 IL, 1995199 1 CHILE, 1995-99 0 ODen (o: High _ : High o): High Economy: Yw: High a,: Highest a,: High @ Industry-level :: Low £| Moderate t Setting a: High u: Highest | High ,R |Nation-Level [ co: Moderate | : Low co : Optimal gE |Wage and a,,: Low a,,w: Low aY o: Lowest u, |Employment I : Moderate : Low e: Optimal | | Setting l | il * I nw t_ Ar Ontimrnl | ARGENTINA, 1909 RETN,19- |Note that:in competitive (firm-level bargaining) and centralized (national level bargaining) settings, labor market outcomes do not differ across closed and open economies. co Average wages or eamings, Ows. , _ . Earnings 'nequality. - - Employment growth. - - - I u - ~. - . .Average uner,ployment rate Dashed arrows depict currant Ltnda'ncixs | v; Bold a,rrows depict recant movements - - 119- Figure 5.2 Outcomes in an Open Economy, by Stage of Adjustment and Market Structure !., --- - 4-. -. MACROECONOMIC CONSTRAINTS - SCENARIO Stabilization Fiscal 1 Sustained IL [ J [~~~~~~~~~~~~ ~~Adjustment j Growth IL: 'A, Flexible .A. Wagxible ado Moderate w : Lowest X :Optimal O Employment (Y: Moderate caw: High o.: Moderate . - Settinq E Moderate E Lowest E Optimal R u: Moderate 0 Moderate 0 Optimal A. _ HILE _ IR. Intermediate w: High o:High U:High K: Wageand o,:High cr,: Highest a.: High E andoc :Low c : Low Moderate T; Employment 0 :High I :Highest 0 High Setting__ _ _ _ _ _ _ ,,~ __ _ _ _ __ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ sT - I .BR A Z IL R I U Centralized w Moderate o: Low w Optimal I T Wage and (b: Low ab: Low a.: Lowest ; Employment c Moderate E Low E Optimal .U. Settinq 0 Low 0 Moderate 0 Optimal RI R ! X - n ; , --; | ~~~~~~ARGENTINAi Notes and Legend . o, 1 = ; Average, wages or eamings Earnings inequality Empl6yment growth'; . ' I0b'' -- _: ' Average unemployment rate,, !:; W Arrows depict current tendericies or recent movements - 120 - (S. FAKE CONTRACTS: JUSTICE AND LABOR CONTRACTS OM BRAZOL Prepared by Jose Marcio Camargo _7-x,,_-ri T 'j,-~ D - - m9 l 6.1. The role of the labor justice in Brazil has been the subject of intense debate among those who study the Brazilian labor market performance. Opinions differ on the impact of this institution on the performance of the labor market, but in general, analysis suggests that i has an important role in protecting the less organized workers. However, few analysts have verified the effect of the institution's role on the labor market as a whole and how it affects workers in general, especially the less qualified and organized workers in the long run. 6.2. In this article I propose an alternative approach, already presented in other papers95, to understand how the functioning of the labor justice can interfere on worker's well being. The basic idea is that the functioning of the labor justice generates big distortions in the labor market by sending the wrong incentives to workers and employers, which tends to develop a low quality work relationship and makes productivity gains and real wage gains almost impossible, especially for the unskilled workers. 6.3. In this article, we illustrate some of these points by presenting findings based on a sample of labor justice cases in Minas Gerais. Once the main points are explained, a proposal on how to address this issue is presented. Our proposal indirectly affects the functioning of the labor justice by proposing that the institution concentrates its efforts on its core functions within the judiciary branch. 6.4. These functions are mentioned in the next section. After that we report how the Brazilian labor justice operates (section 3) and how this operation creates incentives that tend to guide the behavior of workers and firms (section 4). Section 5 presents and details the proposal above, while conclusions are drawn in section 6. 6.5. The labor justice has three important functions in the Brazilian labor relations system. First, all disputes regarding law enforcement must be resolved through the Labor Courts. Second, the Labor Courts are responsible for the solution of all disputes regarding individual and collective labor contracts fulfillment. Third, the Labor Courts are also responsible for individual and collective conciliation, arbitration and judgment. These three extremely important roles make the Labor Courts a key element in the Brazilian labor relations system. 95 See Camargo 1996 and Camargo 1997, among others. - 121 - 6.6. At the individual level, all settlements between employees and employers regarding disputes on individual and/or law contracts fulfillment are only valid if done through the Labor Justice. This implies that a dismissal must be done in the presence of a judicial authority, if the employer wants to ensure that the employee will not claim rights that were not received during the period of work. 6.7. Since 1988, the employee has a period of five years to place a claim in the Labor Justice. If this happens, it is the employer's duty to prove that the contract and/or law was abided. If it can't be done, justice promotes a "conciliation" between the parties and in case of no consensus, the judge is free to decide if the claim is legitimate. If the judge rules in favor of the employee, the employer will have to pay the employee's demands. Note that only monetary demands are accepted and that the employee can not demand for his/her job back. 6.8. On collective bargaining the main role of the Courts is to promote conciliation and give judgment in disputes within its jurisdiction and, moreover, to verify that signed collective contracts are complied with. In collective bargaining, when there is an impasse, either parties can unilaterally call for a "Diss(dio" or arbitration. The dispute is then sent to the Labor Justice for the area where the negotiation is taking place and this becomes the new locus of the negotiation. In case there is an impasse here, it is up to the judge to decide the terms to the contract. 6.9. For disputes regarding collective contracts fulfillment, not only the Labor Courts are involved but also the labor union and employer. In this case it is also the employer's duty to prove that the law and the collective contracts established have been followed. 6.10. Arbitration doesn't follow pre-defined rules or principles. When the dispute is about the non- fulfillment of the law, arbitration is based on the law. However legislation on collective contracts is not as detailed as legislation on individual contracts. Because of this, many times the Brazilian Labor Justice adopts norms that are the equivalent of laws. This prerogative is known as the normative power of the labor justice, i.e., in contrast with every other branch of the Brazilian justice system, the Labor Courts can create a law, instead of being limited to applying an existing law. 6.11. In this case it can be said that previous decisions may be used as a guide but many times the trial is based on political factors. This means that the labor justice may interfere in private contracts between employers and employees, having them modified through its rulings. 6.12. The objective of this section is to discuss in more detail the functioning of the labor courts. A brief explanation on how the institution is structured will be presented. Following that, the functioning will be analyzed, emphasizing negotiations, as explained above. At the end, costs of accessing labor justice for all parties will be analyzed. The Structure of the Courts 6.13. The Brazilian Labor Courts system has three hierarchical levels, organized as follows: a Conciliation and Judgment Board; -122- o Regional Labor Courts; o Superior Labor Court. 6.14. Until recently, the Conciliation and Judgment Board was composed by a labor lawyer, one representative from employees, and one representative from employers, also known as class representatives. These last two members were nominated by the president of the Superior Labor Court where the Board is to function. In 1999, the class representatives were abolished. 6.15. The Regional Labor Court was composed largely by labor lawyers and by a minority of representatives of employees and employers with an equal number on each side. The representatives for employees and employers were chosen by the President of Brazil. At this level, the practice of having representatives for employees and employers was abolished in 1999. The Regional Labor Courts rule on demands made by employees and employers and pronounce judgment. Employees and employers may appeal these sentences to the Superior Labor Court 6.16. The members from the Superior Labor Court are nominated by the President but must be approved by the Federal Senate. Its composition was as follows: three representatives of workers, and three representatives of employers, and eleven labor lawyers who have lifelong tenure. The class representatives were abolished in 1999. The decisions from the Superior Labor Court are final, except if the dispute is related to a constitutional Principle, in which case an appeal to the Federal Supreme Court can be made. The functioning of the labor justice and negotiations practices 6.17. All individual disputes start with an employee or his/her labor union filing a claim at the Conciliation and Judgment Board. The employer is notified and is asked to provide documents proving he/she is not guilty. At this level the process is completely bureaucratic. At the hearing the judge asks if the employer would like to make a counter-proposal to the employee. If one is made, the judge asks the employee if the counteroffer is satisfactory. If so the dispute is over. If not, the judge tries to make the parties reach an agreement. If the employer does not make a counteroffer or if the conciliation has no results, the hearing is closed. The demand is then analyzed by the judge who is in charge of issuing a sentence. 6.18. It is worth emphasizing one implicit but extremely relevant point above. There are always negotiations! Therefore, clauses in an individual contract protected by law are included. In reality, most of the demands and, therefore, of the rights negotiated, refer to individual rights enrolled in the CLT and/or the Constitution. Table 6.1 demonstrates this, based on a sample of processes from Minas Gerais' labor courts. - 123 - Table 6.1 Rights Contested in Labor Courts 1995 Rights Percent FGTS payment 30 Payment of 40% fine to FGTS for dismissal 35 Over time payment 65 Commission 5 Payment of 13t salary 65 Vacation 60 Unpaid or late salaries 25 Non fulfillment of clauses in collective contracts 20 Payment for notification before dismissal 62 Night shift additional payment 12 Family pension payment 10 Employees demanding a written contract 10 Gratuities 22 Other rights 10 Source: Study of 200 cases in the labor justice system of Minas Gerais state. Note: In the same process the employee can demand for more than one right and usually does it. 6.19. The most important aspect of the argument above is that only 20 percent of the processes demand the fulfillment of any clause of a collective contract; the other 80 percent are'related to the non-fulfillment of a individual employee right foreseen in the law. 6.20. The most frequently contested rights were payment of overtime wages, thirteenth salary payment and paid vacation, followed by claims of non-payment of FGTS and the 40 percent fine of FGTS balances in case of dismissal without just cause.96 6.21. What the data show is that there is a large fraction (80 percent or so) of disputes that is taken to labor courts for clauses already foreseen in the law. As mentioned before, the regular procedure is to come up with an agreement between parties. In other words, this procedure means that the possibility of negotiation of individual clauses foreseen by law is made available to employees and employers.Table 6.2 Table 6.2 below confirms that many of these clauses are in fact negotiated, since more than half of the disputes taken to court are resolved through settlements at the Judgment and Conciliation Board.97 Table 6.2 Percentage of Resolved Disputes in each Hierarchical Level Year First Instance: Second Instance: Settlements Local judge Regional - Regional Labor (%) (%) Court (%) 1991 28.7 6.6 56 1992 29.2 11.7 56 1993 29.8 14.9 56 1994 32.5 9.1 55 Source: Study of 200 cases in the labor justice system of Minas Gerais state. 96 It is worth mentioning that In the same process the employee can demand for more than one claim and usually does so. 97 The Ministry of Labor has tried to expand the capacity of these councils -124- 6.22. The table also shows that nearly 30 percent of the disputes are settled at the first local instance. In general, less than 10 percent of the cases reach the second instance (the Regional Labor Court) and a similar percentage reach the Superior Labor Court. Finally, one must note that in the same sample of court data, as a rule, the settlements signed at the conciliation and judgement boards, paid 40% of the amounts demanded by employees. In other words, if the amount of the workers' demands correspond to what is actually owed to them, taking in account the legislation, employers would seem to be paying less than half of what the law determines.98 6.23. In the case of collective contracts, the negotiation is done through the labor unions and the companies, or between the labor unions and the employer federations. The clauses in the collective contracts must be more favorable to the workers than the individual rights accorded in the legislation. However, no individual contract can stipulate less rights than those stipulated in collective contracts signed by the union where the work is being done. At any given time along the negotiation of a collective contract between a group of workers and one company or a group of companies, any of the parties may unilaterally summon a "dissidlo" or agreement, having the negotiation transferred to the court room, giving it the power to arbitrate the dispute. 6.24. Other relevant information about the functioning of justice is the existence or the non-existence of a viMs pr6 one of the parties involved. 6.25. Table 6.3 presents the distribution of judgments according to the "winning" party. The data shows that more than 60 percent of the disputes that are not decided at the conciliation stage, are decided in part in favor of the workers and in part in favor of the company. A lower and apparently declining percentage is resolved entirely in favor of the worker (between 10 and 20 percent of the cases),while the percentage of cases entirely resolved in favor of the employers is between 14 and 32 percent. 6.26. Since there is no formal rule to be followed by the judge to pronounce the sentences, the results above can have at least two interpretations: first, it might suggest that the judge tends to make a division considered "fair" by him of the demands, in a way that both agents, employers and employees, receive a share of the result; a second interpretation would be that workers tend to demand more than they deserve and that the judge is simply applying the law. With the available data it is impossible to decide which interpretation is correct. But it seems that the data do not support the idea that is common among employers that everything demanded in labor courts by workers is accepted by judges. 4. It cannot be precisely determined whether the value of the workers' demands effectively corresponds to what is stipulated by law. In reality, since the workers know that a negotiation will take place at the labor court, the incentive is for them to demand the maximum possible because they know that most likely the judge will attempt a conciliation by splitting the difference, viz., use an intermediate value. - 125 - Table 6.3 Acceptance of Claims by the Labor Justice (local level in Minas Gerais) Year/Region Claims completely Claims partially Claims accepted accepted Not accepted (%) (%) (%) Capital - Belo Horizonte 16.7 69.0 14.4 1991 Province 19.7 63.8 16.5 Total 20.3 64.2 15.5 Capital - Belo Horizonte 15.2 67.2 12.4 1992 Province 16.6 64.2 19.3 Total 15.8 65.2 18.6 Capital - Belo Horizonte 12.0 67.7 20.3 1993 Province 15.1 63.4 21.5 Total 13.8 64.7 21.6 Capital - Belo Horizonte 10.6 68.1 21.3 1994 Province 12.1 55.8 32.0 Total 14.5 65.3 20.2 Source: Labor Justice annual report, Third Region, Minas Gerais The cost of accessing the labor justice 6.27. Finally, it is worth describing the costs incurred by the parties which are involved in disputes in the labor justice. Costs may be divided into two components: monetary expenses and the time costs of waiting for the final sentence. In relation to the monetary expenses, if found guilty, the employer pays a 2% fine on the final value agreed 99 in the process. For employers this cost tends to be very low. From the collected data in the research and from discussions with specialists, it is obvious that the employee never pays anything. 6.28. Regarding the waiting time, there is usually a 20 days wait between the initial date of the process and the day of the first hearing. If the process is not decided by conciliation, the period between the first audience and the sentence of judge from the conciliation and judgment board is usually 700 days. In other words, if the negotiation intermediated by the judge from the conciliation and judgment board is not effective, which means that either the employer doesn't make a counterproposal to the employee or that the employee doesn't accept any of the proposals made by the employer in the first instance, the worker will have to usually wait for two years for the first sentence from the labor justice. 6.29. In case the employer (or the worker) decides to appeal to the sentence of the judge from the conciliation and judgment board to the Regional Labor Court, the usual time for the dispute to be resolved is around four years. 9 Since the usual value of the agreements is R$1,000.00, it represents a cost of R$20.00 per process, if resolved during the conciliation stage, R$26.00 per process resolved in the local justice , and R$46,00 for those resolved at the Regional Labor Court. -126- .Lma W 1¶U-IE 6.30. The description above shows that the functioning of the labor courts in Brazil tends to affect the relation between employer and worker. On the one hand, employers have no incentive to comply with the law and pay workers' rights. If they do not do it, they can negotiate the amount of the benefits at the Labor Justice and-in the worst case-they will have to pay the same amount they would have paid along the working relation. 6.31. Therefore, if the objective is to minimize costs, the right strategy for the firm is not to pay and wait until the worker files a complaint when fired. Since it is known that the time until the sentence is very long100, the business man hopes that the workers tend to accept the counter-proposal even if the worker believes these to be less then the entitled amount. 6.32. This strategy tends to be more effective when the work contracts are informal since that with a formal contract it is easier to solve the dispute favoring the worker. This way, one of the main problems of the functioning of the Brazilian labor justice is to induce informal work contracts. 6.33. On the other hand, the worker knows that a liability exists and that it has been accumulated by the employer and that some of this liability will be taken by him/her when demanded in court. This means that as the working relation grows, the incentive for the worker to go to court increases too. If it is done while he/she is working, he/she is dismissed with difficulty. This way, instead of going to court while working, the worker starts to force a dismissal, becoming increasingly uncooperative in the working relation. Therefore, many times, the demand occurs when the working relation has already been disrupted. 6.34. One additional aspect is that, since the cost for the worker to take the employer to court is zero, except the costs for the hearings'01, there is a permanent incentive for the employee to go to court even if the employer has fulfilled all his contractual obligations. Having nothing to lose, he/she can win the process.102 This way, from the workers' point of view, justice provides good protection against illegal practices but creates incentives for him/her to have a free-rider attitude towards the employer. 6.35. Since the employer knows that as the working relation grows, the incentive for the employee to have a free-rider attitude, shirk and become less cooperative increases, the employer has no incentive to invest in training since there is a growing possibility of losing the investment when it is time to dismiss him/her, resulting in low levels of productivity and therefore real wages. 6.36. For workers with a low bargain power there is a tendency to use the labor justice as tool for compensation. Given the reduced probability of having a demand accepted by the employer, which in many instances is aggravated by an absence of trade unions, it is left for these workers to take their demands to 100 In Minas Gerais, it usually takes a worker 100 days to go to court after being dismissed. Note that here is no impediment for the worker to go before he/she is dismissed but in real Iffe it doesn't happen. 101 In general, labor lawyers determine their fees as a proportion of the value of the process, if the sentence is favorable to the worker. 102 Besides this, knowing that his/her demands will tend to be negotiated, the worker may try to exaggerate his/her rights. - 127 - court. Table 6.4 reveals that, in fact, the profile of workers who mostly go to court is in accordance with the profile of the worker with low bargaining power. Table 6.4 Workers Going to Labor Courts by Wage Level Wage interval (minimum wage) Workers at Labor Justice (%) All workers in Southeast Region (%) 1993 1 to 2 70 47.0 2 to 5 26 26.9 5 to 10 2.5 9.7 More than 10 1.5 6.4 Average 2.1 minimum wage 3.7 minimum wage Source: Research done with 200 processes at Labor Courts in Minas Gerais and in the PNAD survey, 1993 6.37. Something similar happens with collective contracts. There is an incentive for the weaker party to summon an agreement and to have the negotiation transferred to the labor justice, transforming the result of any collective dispute between employers and the workers union in a matter of justice and not the result of a dispute settled by the bargaining power between the parties. The main consequence is the atrophying of the collective negotiation process and a reduced scope for trade unions leading to the latter playing a weak role in the wage-negotiation process. 6.38. From the analysis of the previous sections, it is clear that the functioning of labor justice generates perverted incentives with relevant implications in the performance of the labor market. In order to minimize these incentives, it is fundamental that legislators keep in mind the behavior of the economic agents involved, that is, workers and employers. This way, two aspects must be kept in mind in proposals for reform: * Despite the large volume of regulations and restrictions, the system of working relations in Brazil is extremely "negotiable" and negotiated. However, the locus of the negotiation is the Labor Court (not the workplace), when the worker is dismissed (not while the worker is still employed by the company). * Negotiation may indeed reduce the effective cost of work in relation to what is stipulated by law. 6.39. The main consequences are a congested Labor Justice system and adversarial employer/employee relation, non-cooperative behavior and many times ruled by informal contracts. 6.40. The first point is to have negotiations removed from the labor court and let it happen within the companies, along the working relation. In other words, the guiding principle for reform would be to make negotiations more attractive at company level than at court, for both the workers and employers. This can be reached by: * Increased flexibility of the legislation regarding the prohibition to negotiate many clauses in individual contracts outside the labor justice, and/or, * raising costs of accessing the labor justice system. - 128 - 6.41. In reference to the first point, the best way to implement it would be to allow the negotiation of individual rights through collective contracts at firm level with the participation of the union. From the point of view of the employer this enhances the attractiveness of a formal contract as it avoids the uncertainty of decisions in the labor justice. The imposition that the unions participate in the negotiation and that it cannot be done individually by the worker is to avoid situations where the low bargaining power of workers may result in them signing contracts which are excessively unfavorable to them. 6.42. Note that what is being proposed is not that the rights be negotiated since these rights are indeed negotiated at the Labor courts when the worker is dismissed. What is being proposed is that the negotiation takes place within the firm through the union while the employee is working. 6.43. But it also implies that the recognition that negotiations between employers and workers are a very good way to resolve the conflict between the parties. In this context the Labor Courts must have a much less dominant role to play. In other words, another system much less based on the idea of 'justice" and based more on the idea of 'dispute" would have to be outlined, where the existence of conflict and negotiation at the firm level would prevail. 6.44. In relation to the second point, the suggestion is that all firms involved in processes at the labor justice have its duties to the state (social security, other taxes, etc) verified. 103 This would create an incentive for businesses avoid the labor courts, being induced to pay the benefits negotiated between parties. On the other hand, it would create an incentive to reduce informality by penalizing at least one side of this relation, in this case the employer. This way it would raise the bargain power of the worker when claiming his/her rights. 6.45. This proposal reduces the role of the labor justice as negotiators of clauses in individual work contracts and it raises fts functions as the inspection agency and effective implementation of these contracts. In respect to the collective contract, the proposal is that conflicts be effectively resolved by dispute between workers and firms and that the result depends on the bargain power between parties, therefore not being a matter of justice. 6.46. This article analyses how the role of the Brazilian labor justice affects the performance of the labor market in the country. The argument developed in this article suggests that the main problem originating from this institutional arrangement is that it allows a renegotiation of work contracts a posteiorn, i.e., after the contract has expired. As a result, none of the parties have any incentive to honor the signed contract or even to sign a work contract. 6.47. In other words, the work contracts in Brazil have little value since they are not signed to be respected but to be renegotiated after they have ended. Consequently a free-rider attitude is developed by the workers, and a tendency to illegality by employers, generating working relations with increasing tension, 103 Traditionally the labor justice does not interfere on the additional contributions, which are not demanded in the process (which are generally appropriated by the worker). Recently the social security service has been following the processes in the labor justice in order to locate workers that are not contributing to the social security. - 129 - which is reflected in a large number of processes that reach the labor justice every year (in 2001, nearly 3 million), less cooperation and little commitment to work. From an economic point of view, this result generates little investment in the working relation from both parties, little investment in qualification and training, and consequently low productivity and low real salaries. 6.48. Since work relations are one of the main determinants of productivity and competitiveness in a given economy, a reform of this institutional arrangement is vital for the country to increase productivity and competitiveness, making viable a higher rate of sustainable growth in the future. - 130- References Abramo, L. (1986) 0 Resgate da Dignidade, Dissertac,o de Mestrado, USP. Amadeo, E. R.P. de Barros, J.M. Camargo, R. Mendon,a, V. Perro e A. Urani (1993) "Human Resources in the Adjustment Process", BID. Amadeo, E. e J.M. Camargo (1989) "Polftica Salarial e Negocia,6es: perspectivas para o futuro", mimeo, OIT/Minist6rio do Trabalho. Amadeo, E. e J.M. Camargo (1990) 'The Brazilian Labour Market in an Era of Adjustment', Texto para Discussao, Departamento de Economia, PUC-Rio. Consolidacao das Leis do Trabalho, 1989, EDUSP. Camargo, J.M. (1996) Brasilian Labour Justice: who demands, how it works and who pays, mimeo, PUC- Rio. Camargo, J.M. (1997) Revista Dados Camargo, J.M. (1990) "Salarios e Negocia,6es Coletivas", Pesquisa e Planejamento Econ6mico, agosto. Castro, M. (1988) ParticipagAo ou Controle: o dilema da atuagco operaria nos locais de trabalho. IPE/USP. Colier, R.B. e Collier, D. (1979) Inducement versus Constraints: disaggregatin corporatism", The American Political Science Review, vol. 73. Katz, C.H. (1993) 'The Descentralization of Collective Bargaining: a literature review and comparative analysis", Insdustrial Relations Review, october. La Rocque. E. (1989) "Sindicalismo Brasileiro", mimeo, PUC-Rio. Pastore, J. (1993) "A Flexibiliza,co como Estrat6gia de Competi,co", CNI, mimeo. Pastore, J. E H. Zysberstajn (1988) A Administracao do Conflito Trabalhista no Brasil, IPE/USP, Sao Paulo,. Simonsen, M.H. (1983) 'Indexation: current theory and the Brazilian experience" in R. Dombush e M.H. Simonsen,(eds.) Inflation. Debt and Indexation. Soskice, D. (1990) "Wage Determination: the changing role of institutions in advanced industrialized countries", Oxford Review of Economic Policy, winter. Tavares, M.H. (1988) "Diffcil Caminho: sindicatos e politica na construc,o da democracia" in F.W. Reis e G. O'Donnel (eds.) A Democracia no Brasil, Wrtice, Sao Paulo. - 131 - 7. AN OVERVOEW OF THE EFFECTS OF THE MINIMUM WAGE ON THE BRAZILIAN LABOR MARKET Prepared by Francisco Galrao Carneiro 7.1. The debate concerning the impact on employment, wages, and poverty of the introduction of a minimum wage, or even changes in the minimum wage, into an economy is the subject of a separate chapter in the literature on labor markets. In the case of Brazil, this debate dates back to the 1970s, when a controversy arose as to whether or not changes in the minimum wage could affect all wages in the economy [e.g., Souza and Baltar (1979)]. Now this debate has been revived, in terms not only of the possible impact of the minimum wage on other wages, but also of its effects on the levels of employment, informal labor, and poverty in this country. The debate becomes more relevant when discussions are held regarding possible future trends in the minimum wage policy in Brazil as a result of its impact on govemment accounts, and the feasibility of adopting a system that combines regional minimum wages with a national wage floor. 7.2. The theoretical debate suggests that the effects of the minimum wage on the labor market depend on the structure of that market. In competitive markets with a homogenous labor force, the minimum wage tends to create unemployment if it is set above the equilibrium wage as determined by the market. Under imperfect competition, in the presence of monopsonies, the introduction of a minimum wage may help raise workers' wages without having deleterious effects on employment. In segmented labor markets, the minimum wage tends to generate informality and a certain amount of unemployment. 7.3. In general, there are two clearly dichotomic views as to the normative impact of a minimum wage policy for developing countries. On the one hand, there are those who argue that the minimum wage fosters a better distribution of resources in the economy, so as to improve the population's general welfare by helping reduce poverty, increase productivity, and stimulate economic growth. On the other hand, we hear holders of the distortionist view argue that the minimum wage produces an inefficient allocation of labor and encourages rent-seeking behaviors, thus negatively affecting investments and contributing to a reduction in the rate of economic growth [Freeman (1996)]. 7.4. In this article, we review the literature on the impact of the minimum wage on the Brazilian labor market. First, we will examine the key results pertaining to the employment-effect of the minimum wage in Brazil, and contrast the results that were based on data at the family level and at the individual level. Similarly, we will analyze the principal results with respect to the impact on other wages and on poverty levels. Since in Brazil there is a direct connection between the minimum wage and the benefits paid by the social security system, we will also present an analysis of the impact of changes in the minimum wage on the government accounts. Lastly, we will attempt a critical analysis of some options for minimum wage policy in Brazil. 7.5. The article is structured in the following manner: after this Introduction, we will discuss briefly the history of minimum wage policy in Brazil (Section 2), current minimum wage policy (Section 3), and the profile of the workers who eam minimum wage (Section 4). Then, in Section 5, we present the review of the results relative to the impact of the minimum wage on employment and wages. The results relative to the - 132 - effects on poverty are reviewed in Section 6. Section 7 analyzes the effect on the government accounts, in particular the impact of the minimum wage on payrolls at all three levels of govemment (federal, state, and local). Section 8 discusses policy altematives for the future of the minimum wage in Brazil. Lastly, Section 9 summarizes the results and conclusions of the article. 7.6. Until the 1930s, there was no specific legislation on labor relations in Brazil. Unions were not recognized by the Govemment, and wage negotiations depended on each social group's bargaining power. The main demands in that era were higher wages and a reduction in the number of hours worked. It was not until 1930 that the Brazilian Govemment began to institutionalize these demands. The Ministry of Labor, Industry and Commerce was established in 1931, and both employee and employer organizations were recognized at that time. 7.7. The first official minimum wage began to take shape in 1936, when Law 185 (of January 14, 1936) was passed. Article 1 of that law provided that 'every worker shall have the right, in payment for the service rendered, to a minimum wage capable of meeting, in a given region of the country and in a given time period, his normal needs for food, housing, clothing, and transportation." Initially, the legislation ordered that the value of the minimum wage, to be set by a Wage Commission, should be sufficient to provide remuneration to an adult worker for a normal working day. (Not until 1946 was the law changed to recognize that the minimum wage should be high enough to satisfy the needs of both the worker and his family). 7.8. Although the law that established the lowest wage to be paid to an active worker was enacted in 1936, the first such wage was not established until 1940, when it was initially set at Cr$ 0.24 (in nominal terms), to remain in effect for three years (until 1943). In 1943, a nominal increase of 25 percent occurred in July, followed by another 26 percent in December. After that, the minimum wage remained frozen until 1952. As a result, there was a substantial decline in its value. In 1962, an attempt was made to reconcile its value with the objectives set forth in the Constitution, when President Goulart created the 'lamily wage," a kind of bonus equivalent to 5 percent of the minimum wage for every child age 14 or under. 7.9. A period of successive changes in wage policy legislation began in 1979. Between 1979 and 1985, the Govemment modified wage policy seven times, sparking a series of debates about the impact of the wage increases on the inflation rate. The changes adopted altered the percentage of inflation that could be passed along to wages, and the frequency of the readjustments (which were now occurring every six months). Therefore, a formula has been developed to calculate the wage increases that could be applied to all the policies in effect during the 1979-85 period: W,(,) = {I1+ (A + B, 6 O)Rs}OS(o) (1) where: 0i = W(oVS(o) W(t) = nominal wage of the it wage bracket at the beginning of the six-month period; Rs = change in prices during the six-month period, measured by the INPC [Consumer Price Index]; S(o) = the highest minimum wage in effect during the month of the adjustment; A, and B, = correction coefficient per wage bracket, fumished by the wage policy. - 133 - 7.10. As we can see, this formula shows that the minimum wage played a key role in the determination of the size of the wage readjustments. It should also be noted that any real increase in the minimum wage would have repercussions for the entire structure of wages that are govemed by the wage policy, since it would change the composition of the wage brackets and hence the magnitude of inflation that each wage would receive in the form of an adjustment. 7.11. With respect to the minimum wage in particular, the most significant change occurred in 1984, when its value was standardized nationwide. The minimum wage for all regions was raised to the level in effect in southem and southeastem Brazil. Prior to that change, there was a wide variety of regional minimum wages. When the minimum wage was originally created, there were 14 different regional minimum wages, but by 1963, there were 39 of these. There were at least two in effect immediately prior to the unification. 7.12. In 1987, the role of the minimum wage as a benchmark for the determination of other wages in the labor market was officially revoked in order to permit minimum wage policy to be implemented more flexibly without generating multiplier effects on the other wages in the economy. As a replacement for the role of the minimum wage as a benchmark for other wages and benefits, the Govemment created the National Wage Floor, the real value of which was eroded considerably in subsequent months due to the prevailing inflationary environment and the absence of policies for restoring its value. 7.13. Current minimum wage policy was defined by the new Constitution of 1988, which stipulated that the minimum wage should be the same nationwide and sufficient to meet the basic needs of a worker, as well as those of his family, in terms of housing, education, health, recreation, clothing, hygiene, transportation, and social security. The wage is to be adjusted periodically to preserve its purchasing power, and there is an explicit prohibition against its use as an indexing tool. 7.14. Several aspects related to the treatment of the minimum wage in the new Constitution are worth mentioning: (i) the new Constituton lengthened the list of needs to be met by the minimum wage, by adding health, recreation, and social security to the existing needs; (ii) the nationwide standardization of the minimum wage was introduced into the body of the Constitution; (iii) periodic adjustments of its value were made mandatory, in order to preserve its purchasing power, (iv) the 1988 Constitution ratified the prohibition against using the minimum wage as an indexing tool in the economy; (v) the normal work week was cut from 48 to 44 hours, which meant an increase of about 10 percent in the hourly value of the minimum wage. 7.15. The most important change introduced by the 1988 Constitution was the determination that social security benefits could not be lower than the equivalent of one minimum wage. As Foguel, Ramos, and Cameiro (2000) point out, this rule ultimately proved to be the main stumbling block for policies that sought - 134 - to raise the real value of the minimum wage, since those benefits have a heavy and significant impact on the govemment accounts. 7.16. The profile of the workers who earn minimum wage was examined in Foguel, Ramos, and Cameiro (2001) using data from the 1998 National Household Sample Survey (PNAD). Knowledge of the characteristics of the labor force that is directly affected by the minimum wage is important in determining the scope and effectiveness of a minimum wage policy. The data presented pertain to the economically- active population (EAP), equivalent to 40.2 million workers. Of this total, about 22.5 million workers, 55 percent of total EAP, are in the formal sector. The number of workers receiving up to one minimum wage is 2.4 million, equivalent to 6 percent of EAP. 7.17. The characteristics of the workers who earn minimum wage are summarized in . About 42 percent of such workers are female, 52 percent are 10-29 years of age, and 40.6 percent are heads of households. Comparing those figures with the composition of the total EAP, we see that those who earn only one minimum wage have less education than average and are young, female, and predominately non-white. Also with respect to total EAP, workers between the ages of 30 and 49 and heads of households are under- represented. 7.18. Table 7.2 enables us to study the profile of minimum wage eamers in terms of the kind of jobs they hold. In regional terms, there is a clear distinction between the state of Sao Paulo and the Northeast. For example, while Sao Paulo is home to about 30 percent of total EAP and the formal sector, only 8.2 percent of those who earn minimum wage are found in that state. In the Northeast, on the other hand, the local work force does not represent even 20 percent of the EAP, but about 40 percent of the workers who eam minimum wage are found in this region. 7.19. Similarly, the Sao Paulo labor market is much more formalized than the labor market in the Northeast, which may indicate that the imposition of an official minimum wage is becoming less restrictive in more dynamic labor markets. Thus the adoption of a national minimum wage may have different effects in each region, to the extent that the nature and composition of their labor markets are not homogenous. In principle, therefore, one could argue that the adoption of regional minimum wages could be part of a more effective strategy, since it may potentially permit a better "marriage" between the level of the minimum wage and the characteristics of each labor market. 7.20. Government intervention in the labor market via minimum wage policy has been the subject of intense debate in many countries [OECD (1988)]. Many recent studies have been devoted to investigating the impact of the minimum wage on the labor market in different countries. For the United Kingdom, for example, Bell and Wright (1996) analyzed the impact of the so-called Wage Boards and Councils and found that the minimum wage had not raised the wages of workers in the formal sector above the level of the wages paid in the informal sector. Thus their work suggested that there existed only minor effects on the levels of wages and employment in the economy. Machin and Manning (1994) concluded that the decline in the level of the minimum wage relative to the average wage in the economy contributed significantly to broadening wage dispersion during the 1980s in the United Kingdom. Furthermore, Machin - 135 - and Manning (1996) emphasized the fact that abolition of the Wage Boards and Councils in the United Kingdom resulted in a reduction in new jobs and did not generate any gains in terms of wages (p. 672). 7.21. Studies also exist that associate increases in the minimum wage with the number of small business bankruptcies. Waltman eta! (1998) and Fischer (1997), for example, examined whether increases in the minimum wage were responsible for the departure of small companies from the market.'04 The conclusion reached in these studies was that in the case of the United States, increases in the minimum wage did not cause small businesses to go bankrupt at a higher rate than observed in periods when the minimum wage was not rising. 7.22. In the case of developing countries, there are conflicting results regarding the impact of the minimum wage on the labor market. Bell (1997), for example, found significant negative effects on employment in the case of Colombia, and insignificant effects on the labor market in Mexico. Maloney (2000) presents a summary of the literature on minimum wage in Latin America that indicates that while the minimum wage has a positive effect on other wages in Latin American economies, the effect on employment and poverty is unclear. In other words, while it has been observed that increases in the minimum wage tend to contribute to a reduction in poverty, negative effects on employment have also been noted. Furthermore, Maloney (2000) states that the minimum wage is also an important determinant of employment and wages in the informal sector in most Latin American economies. 7.23. In the case of Brazil, Carneiro and Faria (1997) and Cameiro and Henley (1998) concluded that the minimum wage was an important determinant of the average level of other wages during the 1980s, but that this importance gradually declined during the 1990s. Lemos (1997) showed that the minimum wage has a positive effect on other wages in the economy for a period of five quarters, after an initial shock. Emphasizing the bicausality, the author also shows that the average wage positively affects the minimum wage for a period of three trimesters after the shock. Soares (1998), furthermore, concluded that the minimum wage was affected by the labor market during the 1990s. Finally, Lemos (2000) estimates that increases in the minimum wage tend to compress the distribution of wages and have moderately adverse effects on the level of employment. 7.24. The classic textbook description indicates that under perfect competition, setting a minimum wage above the equilibrium level of the market will reduce the demand for labor and cause unemployment. More commonly, however, the negative effects of the minimum wage will depend on a series of factors, among which are the level at which it is fixed (its absolute value relative to worker productivity), the elasticity of the demand for labor (the more elastic the demand, the greater the negative effect), the elasticity of the supply of labor (the more inelastic, the greater the negative effect), and the responses in terms of investments by firms and individuals (the smaller the investment, the greater the negative effect). The greater the elasticity of the substitution between skilled and less-skilled workers, the greater the negative effect tends to be for the less-skilled. The size and indication of the effect of the minimum wage on employment may therefore differ among firms, individuals, and geographical areas, and depending on its value. 104 The argument underlying this point is that since small companies presumably employ their resources at the point of maximum efficiency, increases in labor costs must somehow be absorbed. Since demand may be inelastic for certain industries and the replacement of labor with capital may require expensive investments, some firms might indeed be forced out of the market. -136 - 7.25. However, some aftemative models suggest that the introduction of a minimum wage in an economy will not significantly affect the level of employment, and may even have a slightly positive impact on the demand for labor [Card and Krueger (1995)]. The simplest model that reached this conclusion is the one that examined the labor market under the assumption of a monopsonistic employer. In this model, workers have little or no bargaining power since they cannot easily find jobs with other employers. This enables an employer to set the wage level below the marginal product of the labor. In general, imposing a minimum wage above the level that would be determined by an employer in a monopsonistic market may, therefore, raise the level of employment. 7.26. Other models that predict that the minimum wage will have a positive effect on the level of employment are the models associated with the theory of the efficiency wage, the theory of human capital, and the theoretical framework of job search. In efficiency wage models, it is assumed that employers will set the wages of their employees above the market equilibrium level in order to increase their productivity, reduce production softness, and cut the costs associated with labor tumover. In that context, the minimum wage can result in an increase in employment. The evidence presented by Rebitzer and Taylor (1995), however, shows that in a context of efficiency wages, the positive effect of the minimum wage on employment may dissipate over the long term, depending on the position of the firms along their profit curves and on subsequent changes in the price of the products and the number of firms operating in the market. 7.27. Models based on the theory of endogenous growth and elements of the theory of human capital also generate forecasts that the minimum wage will have a positive effect on employment. The principal hypothesis here is that the minimum wage creates incentives for workers whose productivity is low to invest in more training or education in order to boost their productivity and, therefore, their paychecks. The resulting increment in human capital will have a positive impact on economic growth and, consequently, on employment. In this respect, Cahuc and Michel (1996) show that a reduction in the minimum wage may bring about a reduction in economic growth. Cubit and Hargreaves-Heap (1996) argue that the net loss of employment expected with the introduction of the minimum wage may be zero for a given interval of minimum wage values, since its introduction will raise the investment in physical capital by firms and in human capital by the workers. Similarly, Acemoglu and Pischke (1998) also argue that the minimum wage may increase the amount of training that firms offer their less-skilled workers. 7.28. Considering the framework of the job search theory, the indication of the effect of the minimum wage on employment will depend on the level of the minimum wage and its impact on the intensity of job search, the level of the acceptance wage, and the probability of an offer of a new job. In this regard, Swinnerton (1996) presents an equilibrium search model under which the firms have a negatively-sloped labor demand curve, labor productivity varies from firm to firm, and the unemployed have imperfect information and seek employment in a random, sequential manner. Under these assumptions, the author demonstrates that, because of an increase in the average productivity of labor, positive effects on well- being can be obtained by the introduction of a minimum wage, even in the presence of a negative impact on the level of employment. Estimates of the Employment-Effect of the Miinimum Wage 7.29. There is little empirical evidence available in literature regarding the impact of changes in the value of the minimum wage on the level of employment in the case of Brazil. Some contributions on this point are found in Foguel (1997,1998), Corseuil and Morgado (2000), Foguel, Ramos, and Cameiro (2000, 2001), - 137 - Lemos (2000), Carneiro (2000), and Cameiro and Corseuil (2001). All these articles will be reviewed later in this paper, and we will attempt to present a summary of the principal results conceming the impact of changes in the minimum wage on the level of employment in Brazil. 7.30. Before moving on to the specific case of Brazil, let us review the information available for the other countries in Latin America. Maloney and Nuiez (2001) attempted to evaluate the impact of the minimum wage on the labor market for a group of Latin American countries. The authors provide estimates of the employment-effect of the minimum wage based on analyses of kemel functions and panel data indicating that changes in the minimum wage tend to generate reductions in the levels of employment in Latin American countries. In the case of Puerto Rico, for example, when the minimum wage was raised to a level equivalent to 63 percent of the average wage in manufacturing, the employment-elasticity of the minimum wage became 0.91 and there were significant reductions in the number of jobs in that country. 7.31. The same general picture is reported by Bell (1997) in a study on the impact of the minimum wage in Mexico and Colombia. For Mexico, the effects on the level of employment were almost non-existent, but for Colombia, the employment-elasticity of the minimum wage for workers whose incomes were close to the minimum wage ranged from 0.55 to 1.22. Bell (1997) concluded that during the period 1981-87, an increase of 10 percent in the minimum wage had reduced employment among the less skilled workers by about 2 to 12 percent, in Colombia. 7.32. Maloney et al. (2001) pursued the analysis of the Colombian case more deeply, using a method suggested by Neumark (2000) that consisted of using a rotating panel of data to analyze the impact of an increase in the minimum wage on wages and on the probability that a worker would become unemployed. This estimated model uses a limited dependent variable that assumes the value of "1" if the individual keeps his job after the minimum wage increase and the value of "0" if the individual becomes unemployed. The explanatory variables of the model, interpreted as the factors that would affect the likelihood that the individual would keep (or lose) his job, are past values of the minimum wage, individual characteristics (such as sex, race, education), and position in the wage distribution (whether close to or distant from the minimum wage). 7.33. The results indicate that an increase in the minimum wage has a statistically significant impact on the probability that an individual will become unemployed. However, that probability is reduced for the higher positions in the wage distribution; i.e., the negative impact of the minimum wage tends to affect more severely those whose income is close to the minimum. The employment-elasticity of the minimum wage estimated by Maloney et al. (2001), according to the Neumark (2000) methodology, was 0.15, indicating that an increase of 10 percent in the minimum wage would have the effect of reducing employment by about 1.5 percent in Colombia. 7.34. Foguel (1998) used data from the Brazilian Institute of Geography and Statistics (IBGE) Monthly Employment Survey (PME) to conduct a natural experiment considering the period of convergence of minimum wage values in Brazil and its impact on the labor market. The idea of a natural experiment is borrowed from such fields as the biological, physical, chemical, and medical sciences and consists of testing the effectiveness of a certain medication in fighting a specific disease. Randomly selected patients are therefore divided, also randomly, into two groups: (a) the treatment group, which is given the medicine; and (b) the control group, which does not get the medicine. Since both groups are selected randomly, there is no reason to believe that, were they not to be given the medicine, the average behavior of the people in the treatment group would be different from the average behavior of the control group. The effects of the - 138 - medicine can therefore be evaluated when the evolution of the state of health of the two groups is compared. The control group ends up serving as a counterfactual for the treatment group, which makes it possible to assess the nature and extent of the effect of the medication.'°s 7.35. Therefore, in order to apply the natural experiment methodology for purposes of analysis of the impact of the unification of the minimum wage on the labor market, it was necessary to select one date associated with the period of convergence of the values of the minimum wage and a second date corresponding to the post-convergence period. Since the regional values of the minimum wage were unified in May 1984, Foguel (1998) selected a pre-convergence period prior to this date, and a post- convergence period after 1984. The pre-convergence period corresponded to the 24 months preceding the unification of the minimum (May 1982 to April 1984). The post-convergence period covered the 24 months between May 1985 and April 1987. Note that the period between May 1984 and April 1985 was excluded from the study in order allow for an interval of market adjustment to the convergence of the minimum before evaluating the effect of the unHfied minimum wage. 7.36. The author also selected two groups of metropolitan regions for which the values of the minimum wage were convergent during the pre-convergence period: (a) Recife and Salvador, and (b) Belo Horizonte, Rio de Janeiro, Sao Paulo, and Porto Alegre. The results of the exercise in comparison between the pre- and post-convergence periods permitted analysis of the impact of the unification of the minimum wage for a set of labor market indicators that included the activity rate and the percentage of employed and unemployed, the unemployment rate, the degree of informality, and the sectoral structure of employment. Since the author worked with the level of those indicators and the logarithm of the minimum wage, the estimates obtained should be interpreted as semi-elasticities of the minimum wage with respect to those indicators. 7.37. The results estimated by Foguel (1998) indicate that convergence of the regional values of the minimum wage led to a decline in the percentage of employed. It also led to increases in the proportion of both the unemployed and the inactive, a higher rate of open unemployment, an increase in the presence of workers whose jobs are registered on their employment record cards among the total employed, and a decline in the percentage of employed persons in industry and commerce, with a corresponding increase in their participation in the services sector and other activities. With regard to the unemployment rate in particular, Foguel (1998) suggests that an increase of about 10 percent in the minimum wage could raise the open unemployment rate by 0.5 percentage points. 7.38. The analysis by Foguel (1998) made an important methodological contribution to the discussion of the question of the impact of the minimum wage on the Brazilian labor market. The idea of treating the process of unification of minimum wage values as a natural experiment was a creative altemative adopted to attempt to explain how, in fact, the minimum wage impacts key labor market indicators. The responses generated, however, were limited to a specific period when the concern for achieving economic stabilization and cutting the inflation rate was still quite intense. We can speculate whether the results found might not have been contaminated by the high inflation rates observed in the Brazilian economy during the period considered. We would therefore need to evaluate the extent to which those results hold up in a different economic environment, even if later studies use different methodologies. 105 See Foguel (1988) for further details concerning this explanation. - 139 - 7.39. Lemos (2000) also used figures from the Monthly Employment Survey to evaluate the impact of the minimum wage on levels of employment in Brazil. Using monthly data for the period 1982-98, the author estimated wage equations by following the traditional time-series analysis methodology. Four different specifications were tested: group against group; first differences of the ordinary least squares (OLS); the twelfth differences; and the first difference of the twelfth difference (to obtain the rate of variation of the annual variation in the value of the minimum wage). 7.40. In all the specifications, lags of the dependent variable were introduced into the model, since changes in the minimum wage may not have immediate contemporaneous effects on employment, but rather future effects [Brown (1982); Neumark and Washer (1992)]. Other explanatory variables included were lags of the inflation rate, dummy variables for different periods, as well as the percentages of workers who are young, female, students, retired persons, civil servants, people whose jobs are not registered on their employment record card, and those employed in the construction sector and in manufacturing. 7.41. Lemos (2000) makes no mention at all of the statistical properties of her variables, thus failing to address an extremely important source of problems for time-series models, i.e., problems related to the possibility that the statistical series used in her estimates are not stationary. Despite this potential source of problems, which could lead to spurious regressions devoid of economic significance, we shall proceed to analyze, with the appropriate caution, the elasticities presented by the author. According to the estimated coefficients, the elasticities of employment with respect to changes in the minimum wage ranged between - 0.006 and 0.005, but in many cases were not statistically different from zero. 7.42. The author also used altemative methodologies to analyze in greater depth the impact of changes in the minimum wage on employment. One of these alternatives was to use instrumental variables to verify whether the results obtained were being affected by problems of endogenicity of the explanatory variables. The instrumental variables were values that lagged by 12 months, a proxy for the election cycle, and a proxy variable for capacity for political intervention. On this point, the author attempts to capture the importance of the election cycle in the process of determination of employment, together with the impact of changes in the minimum wage, by introducing into the model a variable that incorporates the percentage of politicians considered as being "leftist," the political cycle, and the minimum wage multiplicatively. The justification for this is that the incentives for politicians on the left to seek increases in the minimum wage are greater not only during election years, but also when the value of the minimum wage is low. 7.43. The estimation procedure was the least squares method, in three stages. The elasticities obtained this time ranged from -0.020 to 0.035. The results obtained earlier suggested that a 10 percent increase in the minimum wage would tend to curb employment by about 0.06 percent, while the estimates obtained by the instrumental variables method could generate a decline in employment of about 0.20 percent. Once again, we must be careful in reading these results, which may be spurious owing to the possible existence of unit roots in the time series used for the study. 7.44. Furthermore, the introduction of variables such as the percentage of "leftist" politicians, the election cycle, and the capacity for political intervention really needs additional detail and justification; absence of this casts doubt as to their actual explanatory capacity. In general, however, Lemos (2000) presents a creative way of analyzing the impact of the minimum wage on the labor market and presents additional evidence that this impact can be adverse in aggregate terms. - 140 - 7.45. Other estimates of the employment-effect of the minimum wage that took into consideration the problem arising from the possibility that the variables in the estimated model are not stationary were presented by Cameiro (2000), Foguel, Ramos, and Cameiro (2000), Foguel, Ramos, and Carneiro (2001), and Cameiro and Corseuil (2001). In general, the empirical modeling strategy present in those articles adopts the more general idea with respect to the existence of a relationship of long-term equilibrium between the level of employment, the product, and the minimum wage. Therefore, using as our basis the methodology known as co-integration analysis of time series, it is possible to separate the long-term structure among those variables from their short-term dynamics. This means we can identify the speed of adjustment with which potential deviations from the common trajectory of equilibrium among the variables are corrected, as well as the long- and short-term elasticities of employment with respect to the minimum wage and the product. 7.46. The estimated econometric model is based on employment equations that are commonly found in time-series literature, and takes the following form: E= f(m, Y, t) (2) where E is employment, m is the minimum wage, Y the aggregate product, and t represents a deterministic tendency. This formula also enables us to test the exogenicity of the minimum wage variable in order to check whether the equation (i) is correctly specified (weak exogenicity), (ii) can be used for forecasts (strong exogenicity), and (iii) appears structurally stable and survives the Lucas Criticism (super- exogenicity). 7.47. Another general characteristic of those articles is that they usually cover approximately two decades for which time series on employment are available. Since they were based on data from the IBGE Monthly Employment Survey, the sample period started in 1982 and covers the first years of the new millennium. 7.48. The long-term elasticities of employment with regard to changes in the minimum wage found in those studies are situated in the interval between -0.001 and -0.020 for workers in the formal sector, and between 0.0004 and 0.003 for workers in the informal sector. Although these elasticities are low in absolute values, they suggest important long-term trends in the adjustment process of employment for both sectors of the labor market. In the formal sector, changes in the minimum wage tend to affect employment negatively, while in the informal sector such changes tend to affect employment positively. One explanation for this might be that the informal sector ends up serving as a temporary refuge for workers who lose their jobs in the formal sector. 7.49. Another interesting result of those studies pertains to the way in which the level of employment in both sectors behaves throughout the economic cycle. In the formal sector, employment tends to react pro- cyclically to changes in economic activity, while employment in the informal sector reacts anti-cyclically to fluctuations in the product. The perception underlying this process is simple, and suggests that economic growth tends to create more jobs, to stimulate the emergence of formal occupations and, therefore, to discourage informality. 7.50. With regard to short-term dynamics, the more general pattem encountered for the long term remains the same. Changes in the minimum wage tend to have a negative effect on formal employment and a positive effect on informal employment. Similarly, note the pro-cyclical behavior of formal employment in the short term, and observe that the opposite occurs in the informal sector. Meanwhile, the -141 - adjustment speed of the models stood at 5 percent to 9 percent per month, suggesting that deviations from the long-term trajectory of equilibrium among employment, product, and minimum wage are fully corrected within a period of approximately one year. This seems to be consistent with a slow adjustment by employment. 7.51. One limitation of these studies is that, although they allow us to separate the long-term and short- term effects of the minimum wage on the level of employment, they cannot tell us precisely who loses and who wins when the minimum wage changes. The analysis was done with aggregate data and fumishes only an indication of the net effect on the labor market caused by a change in the value of the minimum wage. In general, time-series studies assume that the impact of the minimum wage on the labor market tend to be concentrated among young, inexperienced workers whose income is close to the minimum. The only fact that one can glean from those studies, however, is that the minimum wage does tend to generate a certain amount of unemployment. 7.52. Hoping to fill this gap and try to answer the question as to which workers end up losing their jobs when the minimum wage is raised, Corseuil and Morgado (2001) and Cameiro and Corseuil (2001) used an altemative methodology based on the idea of natural experiments and the differences-in-differences method (Angrist and Krueger (1999)]. The treatment group consisted of workers whose employment was affected by the minimum wage, and the control group was comprised of workers whose employment was not affected by the minimum wage, but by other factors that had also affected the employment of the first group. The difference in the change in employment of the treatment group in relation to the change in employment of the control group can be seen as an estimate of the change in employment of the treatment group that would occur had this group been affected solely by a change in the minimum wage. 7.53. In order to identify the treatment and control groups, the authors used as a tool the results of an analysis of the impact of changes in the minimum wage on wage distribution that will be discussed later in this article. Since changes in the minimum wage tend to have a more obvious effect on the incomes closest to the minimum wage, i.e., the wages that are on the left tail of the wage distribution, the treatment group chosen was composed of workers who were eaming the equivalent of one minimum wage between April and May. The control group chosen consisted of workers who had incomes equivalent to a value that lay between one minimum wage at the April level and twice the minimum wage at the May level, on the assumption that the characteristics of those workers are the same as those in the treatment group, but that their jobs were affected primarily by factors other than the change in minimum wage. 7.54. After the treatment and control groups were defined, implementation of the differences-in- differences method requires the use of the following equation: Y = (4EI/E) - (&Eq/E) (3) where E' and E' denote the levels of employment of the treatment and control groups, respectively, and A denotes the change in each variable before (April) and after (May) the increase in minimum wage. 7.55. The results obtained using this methodology showed that increases in the minimum wage cause reductions in the level of employment. The elasticities reported in Cameiro and Corseuil (2001) indicate that a 10 percent increase in the minimum wage had contributed to a 3 percent decline in employment among the treatment group in 1995 and a drop of about 13 percent in 1999. Then in the case of the informal sector, the elasticities indicate that a 10 percent increase in the minimum wage had been responsible for a 2.2 percent growth in informal employment in 1995, and about 15 percent in 1999. - 142 - 7.56. The authors offer no explanation for the difference in the magnitude of the impact of changes in the minimum wage between 1995 and 1999, but we can speculate that the greater sensitivity observed in 1999 is associated with an economic context of lower inflation rates and greater competitiveness in the market for products. Therefore, the effects of any variations in cost would be much more marked in 1999 than in 1995. When relative prices are aligned, it is much more difficult to pass along cost increases than it is in situations of high inflation, where the price system loses much of its signaling ability. 7.57. Fajnzylber (2001), using as a basis the relative information on the effect of the minimum on wage distribution, also found that the minimum wage has a negative impact on employment in the formal sector. His estimates were lower for the formal sector (elasticity of -0.10) than for the informal sector (elasticities between -0.25 and -0.35). According to the author, those results are consistent with a movement of informal workers into the formal sector after a rise in the minimum wage, or even their departure from the labor market because of (i) low prospects for employment or (ii) increases in family income brought about by rises in income experienced by other members of the family. 7.58. Note, therefore, that the results obtained using different methodologies, for different periods and contexts, appear to converge toward a more traditional view of the impact of the minimum wage on the labor market. Consequently, after the minimum wage is raised, one can expect some reduction in aggregate employment, with a greater impact on workers who eam income at levels close to the minimum wage, as well as growth in informal occupations. While the aggregate results suggest that the net effect is small, the disaggregated analysis seems to suggest that the people most affected by job loss, after an increase in the minimum wage, are the lower-income workers. The Ompact of the Minimum Wage on Olher Wages 7.59. The debate as to whether the minimum wage impacts other wages in Brazil began in the 1970s. Macedo and Garcia (1978) were the first to argue that changes in the value of the minimum wage had only a limited impact on the current wage rate. Their argument was based on estimates of an elasticity less than unity between the minimum wage and the wage rate during the period 1964-74. Souza and Baltar (1979), on the other hand, considered a longer sampling period and offered a rather different interpretation of the data. Those authors argued that minimum wage policy was an important determinant of other wages in Brazil. Their argument was based on the fact that changes in the real value of the minimum wage were followed closely by the real wages of workers in the state of Sao Paulo between 1961 and 1976, which they concluded was consistent with unitary elasticity. This finding gave rise to the term "beacon effect," inasmuch as it suggested that changes in the minimum wage would induce increases in all wages, even among less-skilled workers in the informal sector. 7.60. The debate as to whether or not the beacon effect is valid persisted throughout the 1980s. Countless studies were done on the question of whether wage policy in Brazil was effective. Since wage policy expressed all wages as multiples of the minimum wage for purposes of indexation, the way to test its effectiveness was to make a comparison between an institutional wage index constructed by strictly applying wage policy principles, and an average market wage. If the institutional wage were to remain below the market wage for a long time, the conclusion would be that wage policy was ineffective and, consequently, the theory of the beacon effect would lose some credibility. Cameiro (1995) researched this question. - 143 - 7.61. Velloso (1990) considered an additional explanatory variable and introduced the business cycle as a new explanatory variable in the wages equation. He estimated the impact of changes in the real value of the minimum wage and changes in the business cycle (measured by the rate of open unemployment) on the real value of the wages in the formal and informal sectors throughout the 1976-86 period. For workers in the formal sector, a 10 percent increase in the real value of the minimum wage would mean increases of between 3.6 percent and 6.3 percent in real wages of workers whose hold an employment record card. For workers in the informal sector, a 10 percent increase in the minimum wage would raise their earnings by 4.3 percent to 6 percent. The proximity between the elasticities found would suggest that wages and earnings by both formal and informal workers responded in a fairly similar manner to changes in the minimum wage. The effect of changes in the business cycle, however, was more pronounced for the informal workers. A 10 percent increase in open unemployment is thought to have caused a negative impact of about 0.4 percent on wages in the formal sector, while it negatively impacted earnings in the informal sector by approximately 0.8 percent. 7.62. However, all the literature of the 1980s was later considered to have been contaminated by the problem of spurious regression, since the time series used in the estimated regressions invariably possessed a unit root and did not receive appropriate statistical treatment. So once again, an avenue opens for testing the robustness of the earlier results as to the validity or invalidity of the beacon effect and the effectiveness of the minimum wage as an important determinant of the other wages in the economy. 7.63. A methodology of analysis different from those used prior to the early 1990s was implemented by Cameiro and Faria (1997), Lemos (1997), and Soares (1998). Instead of calculating elasticities, these authors investigated the temporal precedence between the minimum wage and the other wages in the economy. Using aggregate monthly data on the trend in the average industrial wage, considered as a proxy for the market wage, and data on the minimum wage for the period 1980-93, Cameiro and Faria (1997) identified the temporal precedence of changes in the official minimum wage over changes in the average market wage between 1980 and 1985. For the subsequent period, 1986-93, however, the authors found that the two wages were determined simultaneously. The results for the first sub-period corroborated the so-called "beacon theory," under which the minimum wage is believed to be an important determinant of the other wages [Souza and Baltar (1979)]. For the second sub-period, the simultaneous determination of the minimum wage and the market wage was interpreted as an indicator of (i) the minimum wage's loss of effectiveness and (ii) the increased bargaining power of unions beginning in the second half of the 1980s [Carneiro and Henley (1998)]. 7.64. Adopting a similar methodology, Lemos (1997) showed that the minimum wage has a positive effect on the other wages in the economy for five trimesters, after an initial shock. Emphasizing the simultaneity between the minimum wage and the other wages, the author also showed that the average wage positively affects the minimum wage for a period of three quarters after the shock. Soares (1998), furthermore, concluded that the minimum wage behaved reactively toward the labor market during the 1990s. 7.65. A limitation of this methodology is that because it does not supply elasticities, it is hard to quantify the extent to which a change in the minimum wage impacts the other wages in the economy. Furthermore, the fact that there is temporal precedence does not mean that there is a cause-and-effect relationship; therefore, there also exists a limitation in terms of the causal relationship between the minimum wage and the other wages. Compensating for this methodological limitation, Foguel, Ramos, and Carneiro (2000) present econometric results based on the methodology known as co-integration analysis of time series in -144- order to investigate the impact of changes in minimum wage on the other wages in the Brazilian economy during the period 1983-99. The equation estimated by the authors takes the following general form: W=f(m, q - 1, s¢, b, )r) (4) where all the variables are expressed in their logarithmic form, W is the average nominal wage, m is the minimum wage, q - I represents the productivity of labor, u is unemployment, h is a proxy for labor costs, and D is the rate of inflation.106 Just as in the case of the examination of the employment-effect of the minimum wage, it was possible here to obtain short- and long-term estimates for the wage-elasticity of the minimum wage. Over the long term, the model presented a unitary elasticity of wages with respect to the minimum wage and positive coefficients for labor productivity, cost of labor and inflation, with the unemployment rate attracting the expected negative signal. 7.66. Furthermore, the same behavioral pattem was observed for the formal and informal sectors, indicating that the two sectors adapt to demand shocks in a fairly similar manner. For workers whose jobs are registered on their employment record cards, a 10 percent increase in labor productivity would lead to a 4 percent increase in their nominal wages, but a 10 percent increase in unemployment would provoke a decline of about 6 percent in their wages. The price-elasticity of the nominal wages for the workers in the formal sector varied between 0.3 and 0.5, indicating that those workers are no longer able to recover all their inflation-related wage losses. For workers in the informal sector, the impact of changes in the unemployment rate on their income was much more pronounced than in the formal sector, since the unemployment-elasticity of the informal eamings ranged from -0.38 to -0.89. On the other hand, the workers in the informal sector appear to be able to adjust their earnings more effectively with respect to inflation, since the price-elasticity for that sector was higher than unity. In both sectors, increases in labor productivity result in wage increases, suggesting that in both the formal and informal sectors, workers are able to convert positive demand shocks into wage gains. 7.67. With regard to the finding of a short-term dynamic, we note that the minimum wage has only limited power to influence the other wages. The minimum wage-elasticity declines considerably over the short term and does not exceed 0.10 for the formal sector and 0.24 for the informal sector. This result seems to indicate that changes in the minimum wage have a more pronounced impact on the eamings of informal workers. In addition, eamings by informal workers are also more sensitive to short-term fluctuations in economic activity. The employment-elasticity of nominal wages for informal workers was -0.12, while it was only -0.09 for formal workers. 7.68. Despite presenting elasticities that enable us to measure the impact of changes in the minimum wage on the other wages in the economy, this methodology of analysis does not allow us to identify which wages would be most directly affected by changes in the minimum wage. To accomplish this, we would need to study the behavior of the wage distribution in the economy in the presence of changes in the minimum wage. On this point, the work of Lemos (2000), Corseuil and Morgado (2000), Maloney et al. 106 In the authors' formulation, 'q" represents aggregate industrial product, 1 the total employed, and the ratio between them a measure of labor productivity. Then labor costs are measured by the ratio between the real wage from the standpoint of the producer and the real wage received by the worker; the numerator incorporates the taxes paid when a formal worker is hired and the denominator deducts the income tax on worker wages [see Layard et al. (1993) for further details]. - 145 - (2000), Cameiro and Corseuil (2001), and Fajnzylber (2001) presents empirical evidence of the impact of changes in the minimum wage on wage distribution. 7.69. Lemos (2000), Corseuil and Morgado (2000), Cameiro and Corseuil (2001), and Maloney (2000) present evidence that suggests that the wages most affected by changes in the minimum wage are precisely those that are close to the minimum. Indeed, Corseuil and Morgado (2000) and Cameiro and Corseuil (2001) show that the wage distribution changes little after a change in the minimum wage for wages above the equivalent of two minimum wages. This behavior pattem remained practically the same throughout the 1990s, according to the authors. 7.70. The study by Fajnzylber (2001), however, deviates from this rule by presenting evidence that the wages farthest from the minimum are also influenced by changes in the minimum wage, thus corroborating the so-called "beacon effect." The author argues that this result can be observed via the wage distribution if we acknowledge that the use of the minimum wage as a unit of measure is a widespread practice in the labor market. Furthermore, the author postulates that in an environment of perfect competition, one should expect that employers would respond to increases in the minimum wage by replacing workers whose productivity is deemed to be lower than the minimum wage with workers whose initial earnings and productivity are above the minimum. 7.71. In contrast to earlier works, Fajnzylber (2001) does not limit his analysis to those workers who earn one minimum wage or less than that, or to those workers who are paid multiples of the minimum wage. Instead, the author adopts the methodology proposed by Neumark et al. (2000) and estimates the impact of changes in the minimum wage on different points in the wage distribution, calculating both the contemporaneous effects and the lagged effects of the minimum wage. The analysis is developed using data from the Monthly Employment Survey for the period 1982 to 1999. Simply put, the procedure followed involves estimating the impact of changes in the real minimum wage on real monthly eamings, thus permitting the occurrence of differentiated effects throughout the wage distribution, as well as lagged effects. Furthermore, they are controlled for individual characteristics (race and years of schooling), as well as for the survey month (May and September) and interactions between the metropolitan area and some years for which there were significant interventions in the economy (in order to see whether different regions respond differently to political intervention). 7.72. The results obtained by Fajnzylber (2001) suggest that the impact of changes in the minimum wage is not restricted to those workers who eam up to one minimum wage in the formal sector. Instead, the author finds that the minimum wage has a significant impact on the entire wage distribution in the formal sector, as well as the eamings of workers in the informal sector, thus corroborating the so-called "beacon effecf'. Moreover, the author reports results that indicate that the income of adult male workers and of heads of households are the ones most affected by the changes in the minimum wage, vis a vis the income of women, young people, and workers who are not heads of households. 7.73. Possible extensions of that study and the agenda for research on the impact of the minimum wage on employment would cover an analysis of the behavior of hours worked and the hourly wage in the presence of increases in the minimum wage. Those points have not yet been explored in Brazilian literature and would certainly add relevant information to the debate on the behavior of the labor market after a change in the value of the minimum wage. - 146 - 7.74. In the next section, we will discuss the principal results available in the literature relating to the impact of the minimum wage on poverty, inasmuch as the primary objective of a minimum wage policy is directly associated with protection of the income of the lowest-income workers, the ones closest to the poverty line. 7.75. In principle, considenng the main objective of a minimum wage policy, we would say that real increases in that wage tend to reduce poverty. However, this cause-and-effect relationship is not always that clear. Considering a change from the traditional Harris and Todaro model, for example, one might argue that when the economy includes a large informal sector, an increase in the minimum wage in the formal sector can lead to an increase in informal employment and a reduction in formal employment. Given the surplus of labor in the informal sector, eamings by the lowest-income workers who find work in that sector will tend to fall, and this can manifest itself, in reality, as an increase in poverty [Lustig and McLeod (1996)]. 7.76. Altematively, one might argue that the effects of the minimum wage on poverty will depend on the elasticities in the formal and informal sectors. If, for example, demand for formal labor is inelastic, wages in both sectors will rise if the wages in the formal sector rise. Consequently, if the elasticity of the other wages with respect to the minimum wage is positive and close to unity, any increase in the minimum wage will be transmifted to the other wages in the economy, and the ultimate effect will be to reduce poverty levels, because of the inelasticity of the employment in the formal sector [Hamermesh (1993)]. 7.77. One can also argue that in small open economies, increases in the wages of the formal sector always lead to an increase the wages in the formal and informal sectors, while at the same time reducing the rate of return on capital. This is because an increase in the minimum wage cannot be passed along to prices, hence profits decline and we see a migration of capital, instead of labor, in the formal sector. Capital would therefore move to the informal sector, boosting both the eamings and employment of the informal workers [Carruth and Oswald (1981) and Leamer (1985)]. 7.78. Furthermore, there are some factors that would condition an increase in the income of poor families following an increase in the minimum wage. First, the increase in the income of poor families resulting from the increase in the minimum wage must exceed the decline in family income that occurs as a result of the loss of eamings of the lower-income workers in the family who become unemployed after the minimum wage is raised. Second, it is important to know the direction and magnitude of the effect of the minimum wage on the supply of labor by the other members of the family. And, lastly, since changes in the minimum wage affect family income-for example, through the loss of eamings by some members who become unemployed-it is also important to consider what types of monetary compensation are received through mechanisms such as government transfers, as in the case of the unemployment insurance program. 7.79. In general, there is a certain bias in the literature in crediting increases in the minimum wage with reducing poverty in developing countries. [Lustig and McLeod (1996)]. Some studies have attempted to research this matter specifically for the case of Brazil. Corseuil et al. (2000), Neri et al. (2000), and Barros et al. (2000) are recent examples. In the study by Corseuil et aL. (2000), the poverty gap for six metropolitan regions of the country is broken down so as to isolate the impact of the minimum wage on poverty throughout the period 1995 to 1998. More specifically, this breakdown attempts to capture the effects of the - 147 - minimum wage on the incomes of workers who would probably be affected by the minimum wage in the formal and informal sectors, i.e., those who had an income situated between the old minimum wage and the new one. The data used are from the Monthly Employment Survey. 7.80. It is important to note that the methodology captures only the impact on the earnings of those workers who did not lose their jobs after the increase in the minimum wage. This way, the results should be viewed as an upper limit of the impact of the minimum wage on poverty, since they do not consider the possible negative effect on employment of an increase in the minimum wage. The results by Corseuil et al. (2000) show that the impact of increases in the minimum wage on poverty during the period considered was positive: the elasticity of the poverty gap with respect to the minimum wage found by the authors was 0.4, which means that a 10 percent increase in the minimum wage would reduce poverty by 4 percent. The authors also concluded that about two-thirds of the reduction in poverty associated with the minimum wage is attributable to increases in the income of workers in the informal sector. 7.81. The study by Neri et aL (2000) used a different methodology. Instead of breaking down the effects of the minimum wage on poverty, the authors simulated the effect of real adjustments in the minimum wage on the incomes of individuals from different segments of the labor market for which the minimum wage policy is effective. The analysis was based on data from the 1996 National Household Sample Survey. The simulations, however, do not take into account the effects of the minimum wage on employment, since the authors assume that the wage-employment elasticity is zero (p. 6)107. Simulating a real 43 percent adjustment in the minimum wage, the authors found that there would be about a 6 percent reduction in the proportion of poor people. 7.82. The methodology adopted by Barros et al. (2000) consists of breaking down the observed changes in the level of poverty into a set of changes associated with different effects of the minimum wage and another set not associated with the minimum wage. This procedure enabled them to determine, in the change in poverty associated with the minimum wage, which portion is due to the earnings tied to the minimum wage and which portion is due to changes in the level of employment linked to the minimum wage. The period analyzed fell between 1995 and 1998 and the authors used longitudinal data from the Monthly Employment Survey in their analysis. The strategy used to identify the impact of the increases in the minimum wage consisted of comparing the level of poverty recorded at a given moment prior to the raising of the minimum wage with a simulated level of poverty corresponding to some point after that increase. 7.83. In general, the study concluded that there is a direct relationship between increases in the minimum wage and poverty reduction. However, the results indicate that the bigger the adjustments in the value of the minimum wage, the smaller its effects on poverty will tend to be. That relationship persists even when one considers the change in the poverty that is associated with changes both in earnings and in jobs linked to the minimum. The authors further found that the reduction in poverty related to the increment in earnings associated with the minimum is more significant for informal earnings. 107 Using the 43 percent increase in the minimum wage approved in May 1995, Neri (1997) demonstrated that the probabilities that groups of formal employees affected by the minimum wage would become unemployed or informal workers are greater than for unaffected groups. Amadeo and Neri (1998), however, present evidence that the month of May 1995 represents an inflection in the poverty series in Brazil. - 148- 7.84. In the study by the Institute for Applied Economic Research (IPEA) (2000), however, it was argued that there are three ways to measure the impact of the minimum wage on poverty. In the first, one measures the direct effect of the increase in the level of the minimum wage on poverty, without considering the impact that such increase will have on the level of employment and the cost of living. In the second, one takes into account the fact that raising the minimum wage should have negative effects on the demand for labor, thus reducing the level of employment. And, lastly, a third way would be to consider a computable general equilibrium model in which it is possible to examine indirect effects on poverty, such as changes in the degree of informality of the labor market and increases in social security benefits. 7.85. The IPEA study presents some simulations with respect to the impact of the minimum wage on poverty, based on indexation scenarios and the ability of the minimum wage to impact the other wages in the formal and informal sectors. The simulations were developed according to the following expressions: W,, = (1++ X)W. se WM where M is the old level of the minimum wage, Wn represents the new wages and Wa the old wage. The parameter aindicates the percentage by which the minimum wage is raised and the parameter A its ability to impact the other wages. The larger this parameter, the more rapidly will the impact of the minimum wage increase on higher wages lessen. Therefore, when X = 2, for example, a 15 percent increase in the minimum wage would lead to a 2.0 percent increase for someone who earns 2 minimum wages. When X = 1, a 15 percent increase in the minimum wage would lead to a 5.5 percent increase for someone who eams 2 minimum wages. 7.86. Looking at a case in which only those workers whose wages are below the new minimum wage benefit from an increase in the minimum wage, the effect of an increase in the latter on poverty would be rather limited. Under this scenario, the IPEA estimates that a 15 percent increase in the minimum wage would reduce poverty by less than one percentage point, even considering that the minimum wage impacts employees without employment record cards and the self-employed. Considering a scenario under which changes in the minimum wage tend to affect not only wages below that level but also the higher wages, a 15 percent increase in the minimum wage could reduce poverty by about 3.3 percentage points. 7.87. However, these simulations still do not take into account the possible negative impact of the minimum wage on the level of employment. Taking this aspect into consideration, the effect of increases in the minimum wage on poverty would depend, additionally, on the price-elasticity of the demand for unskilled labor in the formal sector. Based on estimates calculated by Ramos and Reis (1995), the IPEA study simulated several different scenarios for the impact of the minimum wage on poverty, taking into account the effect of the minimum wage on employment. In general, the impact on poverty does not seem to be very significant when the employment-effect of the minimum wage is considered. Leaving aside the so-called "beacon effect," for example, a hypothetical increase of 25 percent in the minimum wage would result in a reduction of 1.4 percentage points in the proportion of poor people, adopting the hypothesis that the demand for unskilled labor is inelastic. For the case in which the demand for labor is unitary, the effect of a 25 percent increase in the minimum wage would result in a reduction of less than 1 percentage point in the proportion of poor people. - 149- 7.88. Lastly, the IPEA study estimates the effect of increases in the minimum wage on the poverty level in Brazil, taking into account its various effects on the economy, in addition to wage changes, based on a general equilibrium model developed by Cury, Barros, and Corseuil (1999). The model makes it possible to estimate what the level of poverty would be if the minimum wage were the only parameter to be changed in the economy. The exercise consists of permitting wage changes for workers whose wages are tied to the minimum. For purposes of illustration only, we will cite the results of one of the simulations. Assuming a 20 percent increase in the minimum wage and recognizing that all the wages in the economy are affected by this increase, an increase of 0.1 percentage point in the proportion of poor people was registered. Several altematives were considered, and in all of them, the impact of increases in the minimum wage on poverty was quite insignificant. Therefore, the study concluded that minimum wage policy appears to be irrelevant in combating poverty in Brazil. 7.89. Clearly, the debate about the effects of the minimum wage on poverty in Brazil is far from reaching a consensus. Some studies found results that indicated that increases in the minimum wage reduce the percentage of poor people, but the strongest criticism of those results is that they do not take into consideration the possible negative effects of the minimum wage on employment. When those effects are taken into account, the effectiveness of minimum wage policy in relieving poverty is substantially reduced. The intemational evidence seems to be similarly ambiguous. Studies that use the family, rather than the individual, as a unit of analysis tend to obtain results that indicate that the minimum wage has a fairly modest impact on the poverty level. Neumark (1999, 2000), for example, suggests that increases in the minimum wage merely cause a rearrangement of the individuals who find themselves close to the poverty line in the United States. But Lustig and McLeod (1996) found a reduction in poverty following increases in the minimum wage for a cross-section of developing countries. For this reason, unequivocal results in this respect have yet to be generated in the literature. 7.90. In the next section, we will review the principal results of the impact of increases in the minimum wage on the govemment accounts in Brazil. Our examination of this issue is relevant, since it can indicate whether the mechanisms for combating poverty, such as minimum wage policy, compromise the country's fiscal situation. 7.91. In this section, we examine the impact of changes in the minimum wage on two important components of the fiscal deficit in Brazil: (i) the payroll at all three levels of government (federal, state, and local), and (ii) the social security system budget. Foguel, Ramos, and Cameiro (2001) are credited with developing the evidence in this respect, which is based on simulations of the impact that the minimum wage would have on the government accounts when that wage assumes certain arbitrary values. Impact on Payroll 7.92. Increases in the minimum wage can have a direct impact on the payroll at the three levels of govemment, since the salaries of those civil servants who earn the equivalent of a minimum wage must be - 150 - automatically adjusted when the minimum wage rises.108 Identifying which of the three levels of government suffers the heaviest impact due to changes in the minimum wage can make an important contribution to the future of minimum wage policy in this country. 7.93. The analysis by Foguel, Ramos, and Cameiro (2001) was done using data from the Annual Social Information Report (RAIS) compiled by the Ministry of Labor and Employment for 1997.109 This database contains information on the number of workers in the formal sector and their wages and salaries as of December of that year. The sample was constructed using only civil servants who were assigned to units of the "direct" govemment administration. More specifically, only civil servants who were employed in the executive, legislative, and judiciary branches of the federal, state, and local government were considered, including those who were working in semi-autonomous govemment agencies, the armed forces, and the police. It is important to observe that workers whose employment contracts are governed by both the Single Legal Regime (RJU) and the Consolidated Labor Laws (CLT) were represented in the sample, and that the sample did not include employees of state-owned companies. 7.94. The methodology was based on simulations in which certain arbitrary values were assigned to the minimum wage and used to calculate the change in the payroll at each level. The base value of the minimum wage used for the calculations was R$1136, the level in effect in April 2000. The figures used in the simulations were R$151 (the value established in May 2000), R$163 (equivalent to a 20 percent increase), R$177 (a 30 percent increase), and R$204 (a 50 percent increase). 7.95. The results of the simulations by Foguel, Ramos, and Cameiro (2001) appear in Table 7.3. The figures are on an annual basis, including the year-end bonus (the u1 3t month"). The results indicate that changes in the minimum wage do not seem to have an important impact on the federal government payroll, since only 0.2 percent of federal government workers have eamings equivalent to one minimum wage or less. The impact on the payroll of state governments is much more pronounced, since about 1.9 percent of state government employees receive the equivalent of one minimum wage. Local govemments experience the biggest impact on payroll, since about 13 percent of civil servants at that level have eamings equivalent to the value of up to one minimum wage. In relative terms, the elasticities of the payroll at the three levels of government with respect to changes in the minimum wage are placed at 0.003, 0.045, and 0.134, respectively. This means that a 10 percent increase in the minimum wage would increase the payrolls at the three levels of govemment by 0.03 percent, 0.45 percent, and 1.34 percent, respectively. Impact on the Social Security System B3udget 7.96. With the adoption of the 1988 Constitution, the minimum wage became the floor for social security benefits. This means that every adjustment in the minimum wage is passed along to the pensions and other benefits paid by the social security system. Foguel, Ramos, and Cameiro (2000) present estimates of the effect of the minimum wage on the social security budget, obtained through an exercise similar to the one they performed with the govemment payroll-by assuming different values for the minimum wage, ranging from R$151 to R$177. 108 According to the 1988 Constitution, no public wage or benefit can be less than the minimum wage. 109 The 1997 RAIS is the most recent available at present. - 151 - 7.97. The authors investigated the impact of changes in the wage on both the revenues and expenses of the social security system. The data used were obtained from the Brazilian Institute of Geography and Statistics (IBGE) National Household Sample Survey, and from the Statistical Bulletin and Statistical Yearbook published by the social security system, for 1996, 1997, and 1998. The impact on the social security budget was calculated by subtracting the aggregate increase in revenues from the aggregate increase in expenses, and the results appear in Table 7.4. The figures indicate that the increase in the expenses that is derived from the increases in the minimum wage is approximately 13 times greater than the increase in revenues. In general, each R$1.00 increase in the minimum wage generates an additional deficit of about R$160 million in the social security budget. This finding demonstrates how vulnerable the govemment's fiscal situation is to minimum wage policy. IMI 7.98. In this section, we discuss policy options that may contribute constructively to the debate on the future of minimum wage policy in Brazil. The intention here is not to offer a preferred altemative, but rather to identify the features of each policy option, along with its advantages, disadvantages, and prospects for the Brazilian situation. We discuss in general terms the issue of regionalization vis a vis the maintenance of a national minimum wage, the sectoral approach to the minimum wage, differentiated coverage, criteria for adjustment, and alternatives for managing minimum wage policy. All the options discussed here are summarized in Table 7.6. National vs. Regional Minimum Wages 7.99. In general, there are four possible types of minimum wage: national, regional, occupational, or by industry. Those systems are not necessarily mutually exclusive, and in several countries, a national or regional minimum wage can coexist with minimums that are occupation-based or differentiated by industry. The first two have already been used in Brazil. Initially, from its creation in 1940 until 1984, the minimum wage was differentiated by region. From 1984 to 2000, the minimum wage was standardized at a single nationwide value. More recently, the country has had a combination of a national minimum wage defined as the "floor" and regional minimum wages set by the local governments. 7.100. One of the principal advantages of the national minimum wage is ease of implementation and monitoring. Additionally, if we consider that wage differentials are a potential source of migration, the establishment of a single level of a minimum wage for the entire country would tend not to encourage migration by rural workers to densely-populated urban areas. 7.101. The main disadvantage of this system, however, is that it presupposes both the existence of a homogenous labor market and the presence of few regional economic disparities. When those conditions do not exist, the fixing of a single wage floor for an entire country can produce significant negative effects. If labor productivity is significantly differentiated among the various regions of the country, it is to be expected that regions with lower average labor productivity will tend not to respect the official minimum wage. We 110 This section closely follows the discussion of minimum wage policy options presented in Foguel, Ramos, and Carneiro (2001). - 152 - might then see a growth in informality in the labor market, or even a rise in open unemployment in the less developed regions. Furthermore, in the case of Brazil, the national minimum wage system tends to impact significantly the govemment deficit because of its relationship with the pensions and other benefits paid by the social security system. 7.102. With regard to the option for a regional minimum wage, there are two altemative configurations. In the first, there is a differentiated minimum wage for each unit in the federation, as was the case in Brazil until 1984. In the second, a system can be adopted that is a combination of the national and regional systems, with the Federal Government defining a minimum floor to be observed throughout the country, and the local govemments establishing minimum wages above that floor, depending on their ability to pay and the nature of their particular labor markets. 7.103. The main advantage of having minimum wages that are differentiated by region is that this system can reduce the negative effects in terms of informality and unemployment in the less developed regions, situations observed in the case of the national system. Furthermore, establishing regional wages may be more appropriate since the minimum wage of each region would necessarily reflect more accurately the differences in terms of cost of living and configuration of the different labor markets. The principal disadvantage, however, is the potential encouragement of migratory flows to more dynamic and more densely populated regions. Since the flow of low-skilled workers increases the labor supply in regions that have a higher average labor productivity, the expected result would be an increase in informality and under- employment. 7.104. Moreover, minimum wages differentiated by region can encourage the migration of the more highly-qualified workers from the less developed regions to areas where labor productivity is higher. Therefore, the average productivity in the less developed regions would tend to decline, reducing per capita income in the less developed regions, thus delaying their development even further. 7.105. For this reason, implementation of a combination of national and/or regional minimum wage systems demands a careful review of the economic conditions and the degree of heterogeneity among the various regions of the country. In the implementation of that combination of polices, the functions of labor supply and demand, distributions of productivity, degree of worker aversion to risk, and the ability of state and local govemment to pay, as well as political factors, must be evaluated. An Occupational or Sectoral M inimum Wage 7.106. There is yet another altemative, one that does not necessarily substitute for the options of a regional minimum wage and/or nationally-unified minimum wage. It is the system that differentiates the official minimum remuneration by industry sector or certain occupation. In most of the countries where this system has been applied (Germany, Austria, Sweden, and Denmark), the value of the minimum wage is determined by collective agreements at the industry level. The occupational minimum wage is defined in a similar way and is based on fairly specific occupations. It has been applied in countries like Costa Rica, Colombia, Spain, Luxembourg, and some African and East European countries. 7.107. The chief advantage of the industrial system is related to the heterogeneity observed among industrial sectors. To the extent that different sectors present different productivity levels, it seems reasonable to differentiate the wage floor for each of them. In the case of occupation-based minimum - 153 - wages, the primary objective is to protect groups of less well-organized workers, who do not have enough bargaining power to negotiate their own minimum wage rates. 7.108. The disadvantage of the industrial or occupational-based systems has to do with the complexity of implementing them and monitoring compliance, which requires a major coordination effort. In order to function properly, the system also presupposes a decentralized wage bargaining structure, to prevent better-organized groups from setting higher wage floors. In the case of Brazil, the prevailing intermediate bargaining structure, along with the high geographical concentration of industry and the heterogeneity of the labor market, tend to make this altemative unfeasible. Differentiated Coverage 7.109. Under the differentiated-coverage system, there is a choice of two alternative strategies. The first would be to allow certain types of industrial establishments, employing small numbers of workers, to be exempt from paying the official minimum wage. Under the second, wages lower than the minimum can be adopted for young people and trainees. The main argument to justify these procedures has to do with small business' limited ability to pay. It is harder for them to get credit, and they have fewer means of surviving periods of lower economic activity. In the case of young people and trainees, the argument is that their productivity is lower than that of adult workers. 7.110. In weighing the advantages and disadvantages, one might argue that lower wages for young workers tend to encourage them to remain in school longer, thus improving their future prospects and the allocative efficiency of the labor market. On the other hand, in a country with a large informal sector, the lower the minimum wage is for young people, the greater will be the incentive for small businesses to hire young workers. Thus the incentive to staying in school will be stronger in the more developed regions, but the opportunity cost of doing so may perhaps be too high in the less developed regions, which have a large informal sector and higher poverty rates. We would therefore have to know precisely the extent of the demand for young workers and its function in the labor supply before thinking about implementing the altemative of minimum wages differentiated by age bracket. Criteria for Adjustment and Management of the Minimum Wage Policy 7.111. With regard to the determination of the criteria for adjusting the minimum wage, there are independent or combined criteria involving: (a) basic needs of workers; (b) their impact on employment and wages; and (c) macroeconomic factors. A strategy that combines all three criteria would seem to be better than one that takes into account only one or the other. 7.112. However, combining the three criteria would involve a certain amount of coordination among the interests of workers, companies, and the Govemment. Certainly such a policy would be more easily implemented in countries that have centralized collective bargaining structures. Its viability presupposes the existence of tripartite negotiating committees that would meet periodically to study and decide on changes in the minimum wage. 7.113. In Brazil, where the bargaining structure is intermediate, there is no single entity that can be considered capable of representing the interests of the working class as occurs in countries where the bargaining structure is centralized. Furthermore, fiscal difficulties would always tend to force the - 154 - Govemment to depart from the initial objectives of the minimum wage policy. Therefore, this is not seen as a viable alternative for the Brazilian situation. 7.114. This paper has presented a comprehensive review of the literature on the subject of the impact of the minimum wage on the Brazilian economy. It is particularly important under the present circumstances to be aware of the principal effects of the minimum wage on the economy, since academicians and policymakers have expressed a growing interest in debating alternative approaches for the future of minimum wage policy in Brazil. Therefore, the aspects examined in this article involve the impact of the minimum wage on the level of employment--both formal and informal--on the other wages, on the percentage of the population that is poor, and on the govemment accounts. Also briefly reviewed were the first minimum wage policies implemented in this country, and the profile of the workers whose eamings are close to the minimum. A summary table (Table 7.6) presenting the main results reviewed in the text appears in the Appendix. 7.115. With regard to the impact of the minimum wage on levels of formal and informal employment, all the studies reviewed here point to negative effects on formal employment. On the other hand, some results suggest that increases in the minimum wage generate increments in informal employment, a situation compatible with the existence of dual labor markets in which the impact of the minimum wage on the unemployment rate is attenuated by the absorption of workers in the informal areas of the economy. The article by Fajnzylber (2001), however, found a negative impact for both the formal and informal sectors, thereby confirming the classic prescriptions of textbook models of competition, in which the minimum wage tends to lead to involuntary unemployment. 7.116. With regard to the impact on the other wages in the economy, the hypothesis usually tested by the studies we reviewed was on validity, rather than on the so-called beacon effect. The debate on this point was initiated in this country in the 1970s, with Baltar and Souza (1979) presenting empirical evidence that suggested that changes in the minimum wage impact all the other wages in an economy in the same proportion. Those results were the opposite of those found by Macedo and Garcia (1978), who found elasticities lower than unity for a different period of time than was analyzed by Souza and Baltar (1979). More recent evidence based on time-series analyses prove that the minimum wage has some effect on the other wages, but there is no consensus in the literature on the existence of a unitary elasticity of the other wages in relation to the minimum wage. 7.117. Some methodological advances were introduced into the debate by studies that examined the effects of the minimum wage, not on an average of other wages in the economy, but on the distribution of wages. This made it possible to examine which wage bracket is most affected by changes in the minimum wage. Here too, the evidence does not help us reach a consensus, since there are studies that conclude that the minimum wage tends to affect only those wages that are closest to it [see Corseuil and Morgado (2001), Cameiro and Corseuil (2001), and Lemos (2000)], and findings that suggest that the entire distribution of wages in the economy is affected by changes in the minimum wage [see Fajnzylber (2001)]. 7.118. One point on which there existed a certain amount of consensus in the literature pertains to the impact of the minimum wage on the percentage of poor people in a country. In general, all the studies reviewed seem to suggest that increases in the minimum wage have a fairly limited impact on poverty levels. That finding comes as an unpleasant surprise, to some extent. Because of the specific objectives of - 155 - the minimum wage policy, one would expect to find a direct relationship between increases in the minimum wage and a reduction in the number of poor people. Indeed, such a correlation does exist for Brazil, but what the studies reviewed here show is that the increases in the minimum wage in real terms would have to be fairly significant before we could observe significant reductions in poverty levels. 7.119. The frustration with the success of wage policy in negatively affecting poverty intensifies when we consider its effects on the govemment accounts. The results presented in this article show that, although a different impact was observed for each level of govemment (federal, state and local), the biggest and most significant impact is on the social security accounts. In general, each R$1.00 increase in the minimum wage adds about R$160 million to the deficit in the social security budget. This situation demonstrates the vulnerability of the Govemment's fiscal situation in terms of minimum wage policy and makes clear the limitation in the Government's ability to use minimum wage policy as a meaningful tool in a more aggressive poverty-reduction strategy. 7.120. Lastly, we presented a critical discussion in this article conceming several possible altematives for the future of minimum wage policy in this country. We looked at options for regionalization versus unification of the minimum wage, sectoral and occupation-based minimum wages, differentiated coverage, and altematives for managing minimum wage policy. Instead of pointing to one political option as being the best among all those presented, what we showed was that isolated positions tend to have serious limitations. 7.121. In general, what has become clear is that there are certain costs associated with a poverty- reduction strategy that relies heavily on a minimum wage policy. Those costs may manifest themselves most directly in the short run via the impact of an increase in minimum wage on the govemment accounts and a rise in informality and/or unemployment. Furthermore, we have observed from the analysis of the literature on the effects of the minimum wage in Brazil that there are winners and losers, which leads to ambiguous results as to the redistributive effects of minimum wage policy. Therefore, more specific analyses must be done in order to determine whether the expenditure associated with a minimum wage policy can effectively help, over the long term, to lift families out of poverty. -156- References ACEMOGLU, D. e PISCHKE, J. (1998), The Structure of Wages and Investment in General Training, National Bureau of Economic Research, Working Paper, 6357, Boston, Massachusetts. AMADEO, E. e NERI, M. (1998), Politica Macroeconomica y Pobreza en Brasil, in: GANUZA, E. e TAYLOR, L. (org.), Politica Macroeconomica y Pobreza en America Latina y el Caribe. Mexico: Fondo de Cultura. ANGRIST, A. e KRUEGER, A. (1999) Empirical Strategies in Labor Economics, in Ashenfefter, 0. and D. Card (org.) Handbook of Labor Economics, Vol. 3, Elsevier. BARROS, R.P., CORSEUIL, C.H. e LEITE (2000), Labor Market and Poverty in Brazil, Texto para Discussao No. 723, Instituto de Pesquisa Economica e Aplicada, Rio de Janeiro. BARROS, R.P., C.H. CORSEUIL, M.N. FOGUEL, e P.G. LEITE (2000) Uma Avaliacao dos Impactos do Salario Minimo sobre o Nivel de Pobreza Metropolitana no Brasil, IPEA, mimeo. BELL, D.N.F. and WRIGHT, R.W. (1996), The impact of minimum wages on the wages of the low paid: evidence from the Wage Boards and Councils, Economic Joumal 106, 650-656. BELL, L.A. (1997), The Impact of Minimum Wages in Mexico and Colombia, Joumal of Labour Economics 15, S102-S135. BROWN, C., GILROY, C. and KOHEN, A. (1982), The Effect of the Minum Wage on Employment and Unemployment, Joumal of Economic Literature, 20: 487-528. CAHUC, P. and MICHEL, P. (1996), Minimum Wage Unemployment and Growth, European Economic Review, August: 1463-1482. CARD, D. and KRUEGER, A. (1995), Myth and Measurement: The New Economics of the Minimum Wage, Princeton University Press, New Jersey. CARNEIRO, F.G. (1995), Govemment Intervention, Institutional Factors and Market: An Analysis of the Wage Barganining in Brazil, Analise Econ6mica, 23-24:19-36. CARNEIRO, F.G. (2000), Time Series Evidence on the Employment Effects of Minimum Wages in Brazil, Texto para Discussao No. 18, Mestrado em Economia de Empresas, Universidade-Cat6lica de Brasilia. CARNEIRO, F.G. e CORSEUIL, C.H. (2001), The Impact of Minimum Wage Changes on Employment and Wages in Brazil: Evidence from Time Series and Longitudinal Data, Proceedings of the Il/ Intemational Colloquium, pp. 163-189, Brasilia-DF. CARNEIRO, F. G. e HENLEY, A. (1998), Wage Determination in Brazil: The Growth of Inside Power and Informal Employment, Joumal of Development Studies, 34, 117-138. CARNEIRO, F.G. e FARIA, J. R. (1997), Causality between the Minimum Wage and Other Wages, Applied Economics Letters, 4, 507-510.' CARRUTH, A. e OSWALD, A.J. (1981), The Determination of Union and Non-Union Wage Rates, European Economic Review, 16, 285-302. - 157 - CORSEUIL, C.H. e W. MORGADO (2001), Salario Mfnimo, Distribuic,o de Salarios e Emprego no Brasil, Serie de Seminarios Academicos, Departamento de Estudos Sociais - Instituto de Pesquisa Economica Aplicada, Rio de Janeiro, IPEA, mimeo. CUBITT, R.P. e S.P. HARGREAVES-HEAP (1996), Minimum Wage Legislation, Investment and Human Capital, Economics Research Centre, University of East Anglia, mimeo. CURY, S., BARROS, R.P. e CORSEUIL, C.H. (1999), Modelo Aplicado de Equilibrio Geral para Simulac,o de Impacto sobre Pobreza e Desigualdade: Simulacoes Comparativas de Mudancas no Com6rcio Exterior e no Fluxo Extemo entre 95 e 98, Anais do SeminArio Desigualdade e Pobreza, IPEA, Rio de Janeiro. FAJNZYLBER, P. (2001), Minimum Wage Effects Throughout the Wage Distribution: Evidence from Brazil's Formal and Informal Sectors, Departamento de Economia e CEDPLAR, Universidade Federal de Minas Gerais, mimeo. FISCHER, C.C. (1997), A Note on In Defense of the Minimum Wage, Joumal of Economic Issues 31, 261 - 263. FOGUEL, M. (1997), Uma Analise dos Efeitos do Salario Minimo sobre o Mercado de Trabalho no Brasil, Dissertac,o de Mestrado, Departamento de Economia, Pontificia Universidade Cat6lica do Rio de Janeiro, Rio de Janeiro, pp.179. FOGUEL, M. (1998), Uma Avaliagao dos Efeitos do Salario Mfnimo sobre o Mercado de Trabalho no Brasil, Texto para Discussao No. 564, Instituto de Pesquisa Econ6mica Aplicada, Rio de Janeiro. FOGUEL, M., RAMOS, L. e CARNEIRO, F.G. (2000), The Economic Impact of Minimum Wages, XXVIII Encontro Anual de Economia - ANPEC, Campinas. FOGUEL, M., RAMOS, L. e CARNEIRO, F.G. (2001), The Impacts of the Minimum Wage on the Labor Market, Poverty and Fiscal Budget in Brazil, Texto para Discussao - forthcoming, Instituto de Pesquisa Economica Aplicada, Rio de Janeiro. FREEMAN, R.B. (1994), Minimum Wages - Again !, International Joumal of Manpower 15, 8-25. FREEMAN, R.B. (1996), The Minimum Wage as a Redistributive Rool, Economic Joumal 106, 639-649. HAMERMESH, D. (1993), Labor Demand, Princeton University Press. IPEA (2000), 0 Nivel do Salario Minimo no Brasil, Instituto de Pesquisa Economica Aplicada, Diretoria de Estudos Sociais, Rio de Janeiro, mimeo. LAYARD, P., NICKELL, S., e JACKMAN, R. (1993), Unemployment: Macroeconomic Performance and the Labor Market, Oxford University Press. LEAMER, E. (1995), The Heckscher-Ohlin Model in Theory and Practice, Princeton Studies in Intemational Finance No. 77, Princeton University. LEMOS, S. (1997), 0 Efeito da Polftica de Salario Mfnimo na Determinagco do N(vel de Salario M6dio no Brasil no Perfodo de 1970 a 1994, Tese de Mestrado, UFPE/PIMES. LEMOS, S. (2000), The Effects of the Minimum Wage on Eamings and Employment in Brazil, University College London, mimeo. LUSTIG, N. and McLEOD, D. (1996), Minimum Wages and Poverty in Developing Countries: Some Empirical Evidence, Brookings Discussion Papers, 125, The Brookings Institute, Washington. - 158 - MACEDO, R. and GARCIA, M. (1978), Observa,6es sobre a Politica Brasileira de Salario Minimo, Trabalho para Discussao, 27, Instituto de Pesquisa Econ6mica, Universidade de Sao Paulo, Sao Paulo. MACHIN, S. and MANNING, A. (1994), The Effects of Minimum Wages on Wage Dispersion and Employment: Evidence from the U.K. Wages Councils, Industrial and Labour Relations Review 47, 319-329. MALONEY, W. (2000), A Note on Minimum Wages in Latin America, The World Bank, Washington, mimeo. MALONEY, W. e NUNEZ (2001), Minimum Wages in Latin America, The World Bank, Washington, mimeo. NERI, M. (1997), 0 Reajuste do Salario Minimo de Maio de 1995, Anais da Sociedade Brasileira de Econometria, RecHfe. NERI, M., GONZAGA, G. and CAMARGO, J. (2000), Efeitos Informais do Salario Minimo e Pobreza, Texto para Discussao No. 724, Instituto de Pesquisa Economica Aplicada, Rio de Janeiro. NEWMARK, D., SCHWEITZER e WASCHER, W. (1992), Employment Effects of Minimum Wages and Subminimum Wages: Panel Data on State Minimum Wage Laws, Industrial and Labor Relations Review, 46:55-81. NEWMARK, D. e WASCHER, W. (1997), Do Minimum Wages Fight Poverty?, Working Paper No. 6127, National Bureau of Economic Research, Cambridge, Massachussets. NEWMARK, D., SCHWEITZER e WASCHER, W. (2000), The Effects of Minimum Wages Throughout the Wage Distribtuion, Working Paper No. 7519, National Bureau of Economic Research, Cambridge, Massachussets. OECD (1998), Making the Most of the Minimum: Statutory Minimum Wages, Employment and Poverty, Employment Outlook, June, Paris. RAMOS, L. e REIS, J.G. (1995), Salario Minimo, Distribuic,o de Renda e Pobreza no Brasil, Pesquisa e Planejamento Economico, Vol. 25, No. 1. REBITZER, J. and TAYLOR, L. (1995), The Consequences of Minimum Wage Laws: Some New Theoretical Ideas, Joumal of Public Economics, 56: 245-255. SOARES, F.V. (1998), A Existencia e a Dire,ao de Causalidade entre o Rendimento dos Trabalhadores Nao-Qualificados por Posic,o na Ocupac,o e o SalArio Minimo entre 1982 e 1995 - Uma Analise Empirica, Anais do XXVI Econtro Nacional de Economia, pp. 1149-68. SOUZA, P. and BALTAR, P. (1979), Salario Minimo e a Taxa de Salario no Brasil, Pesquisa e Planejamento Econ6mico, 9: 629-660. VELLOSO, R.C. (1990), Salario Minimo e Taxa de Salarios no Brasil, Pesquisa e Planejamento Econ6mico, 20(3). WALTMAN, J.; McBRIDE, A. and CAMHOUT, N. (1998), Minimum Wage Increases and the Business Failure Rate, Joumal of Economic Issues 32, 219-223. -159 - Table 7.1 Profile of Workers who Earn Minimum Wage in Brazil Urban Sector- 1998 (%) EAP Formal Sector Workers who Earn MW Sex Male 66.5 61.0 58.3 Female 33.5 39.0 41.7 Age 10-19 9.6 6.7 18.1 20-29 28.8 32.3 34.9 30-39 27.7 30.3 21.4 40-49 20.5 20.9 15.1 50-64 11.7 9.1 9.5 65+ 1.8 0.7 1.0 Education Illiterate 7.4 4.5 12.8 Elementary 25.9 19.4 31.5 Middle School 27.4 25.6 32.2 High School 27.3 33.8 22.3 University 12.0 16.7 1.2 Position in the Family Head 53.9 51.9 40.6. Non-Head 46.1 48.1 59.4 Race White 58.8 62.4 41.1 Non-White 41.2 37.6 58.9 No. of Workers (Millions) 40.2 22.5 2.4 Source: Foguel, Ramos, and Carneiro (2001), p. 7. - 160 - Table 7.2 Profile of Workers who Earn PGinimum Wage Characteristics of Jobs Held - 1998 - (%) EAP Formal Sector Workers who Eam MW Region South 16.2 17.4 11.7 Sao Paulo 27.6 30.3 8.2 Southeast 22.3 23.7 27.1 Northeast 20.8 17.1 38.1 Other 13.1 11.6 14.8 Area Metropolkan 37.7 41.9 25.2 Non-Metropolitan 62.3 58.1 74.8 Business Sector Manufacturing 15.6 20.2 19.1 Construction 9.5 4.4 6.7 Commerce 17.0 13.8 17.6 Services 27.6 22.5 24.2 Other 30.3 39.1 32.4 No. of Workers (Million) 40.2 22.5 2.4 Source: Foguel, Ramos, and Carneiro (2001), p. 8. - 161 - Table 7.3 Impact of the Minimum Wage on the Payroll, in R$ Million Federal, State, and Local Govemments - Annual Figures Minimum Wage Federal State Local Total 151 3.1 129.6 197.7 330.4 163 4.9 211.0 359.2 575.0 176 9.1 324.2 553.6 886.9' 204 23.6 562.0 1017.2 1602.7 Relative Impact (%) 151 0.02 0.52 1.41 0.64 163 0.04 0.84 2.55 1.11 176 0.07 1.29 3.94 1.71 204 0.19 2.24 7.23 3.09 Source: Foguel, Ramos, and Carneiro (2001), p. 18. Table 7.4 Impact of the Minimum Wage on the Social Security Deficit Annual Figures in R$ Million Minimum Wage Impact on Impact on Net Impact Revenue Expenditure 151 171 2.340 2.169 160 290 3.820 3.530 165 360 4.660 4.300 170 432 5.520 5.088 177 537 6.740 6.203 Source: Foguel. Ramos, and Carneiro (2001). p. 18. -162 - Table 7.5 Mahix of Options for Iinimum Wane Policy in Brazil GENERAL ADVANTAGES DISADVANTAGES PROSPECT FOR THE CHARACTERISTIC BRAZILIAN SITUATION National Single minimum wage Simple to apply and Assumes economic and Sub-optmal option in light Minimum for the whole country. monitor; reducton in labor market homogeneity; of fiscal vulnerability, Wage rural-urban migration. significant fiscal impact major economic unemployment and dispanties, and informality in less heterogeneous labor developed regions. markets. Can be combined with a policy of minimum wages differentated by region or level of govemment. Reglonal Different values for Reflects reglonal Incentive to migrabon; Since April 2000, has Minimum different regions. differences in terms of unemployment/informaiity; been implemented in this Wags cost of living and decline in productivity in country, using a mix of a productvity. less-developed areas. national floor and regional minimum wages. Sector or Different values for Reflects heterogeneity Complicated application Intermediate bargaining Occupation- certain industry of economic sectors and monitoring. Assumes existing in this country based sectors or specific and productivity; decentralized bargaining so makes this altematve Minimum occupations protects the less well- as to avoid favoring better- impractcable. Highly Wage organized worker organized groups. concentrated industrial groups. sector and heterogeneous labor market would make it difficult to apply here. Differentiated Some companies with Takes into account Difficulty in defining the Existence of a significant Coverage a certain number of company ability to pay. exempt groups or those informal sector in the employees may be Can be an incentive for entited to differentated economy reduces exempt from paying young people to stay in minimum wage. effective scope of this minimum wage; school. Increases the altemative. minimum wage ailocative efficiency in differentiated by age the economy. bracket. Adjustment Independent or (a) Meets workers' Political factors may divert Efficient allocative criteria Criteria combined criteria, nutriftonal needs; (b) the minimum wage policy for adjusting the minimum involving: (a) basic considers possible from Its inital objectives. wage will be possible only needs; (b) impact on impact on the labor Links social security if the link with social employment and market; (c) takes into benefits to the minimum security pensions and wages; (c) account impact on wage, and its impact on the other benefits is broken. macroeconomic inflabon, the public sector deficit. factors. govemment deficit and ___________________ economic growth. Centraliled Definition of the Can facilitate Fiscal problems may dictate Difficulty in management Management scope, coverage, and macroeconomic the rules of minimum wage in light of the rules for adjusHng the coordinabon and the policy, resulting in inefficient heterogeneity of the labor minimum wage are monitoring process. allocabon of resources. market and sharp regional established by the Can be used for disparities. federal Govemment. regional or sectoraVoccupatonal minimum wages. Tripartte Govemment, Democrabc system. In In economies that have an Not feasible because of Commisslons employers, and economies with Intermediate bargaining the presence of an employees decide centralized bargaining, structure, may favor the intermediary bargaining jointly on the rules of facilitates better-organized groups, structure. minimum wage policy. macroeconomic which has deleterious coordinabon. consequences for the Govemment can economy and worsens represent the interests income distribution. of the outsiders (the unemployed). - 163 - Table 7.6 Summary of the Results of Studies on the Impact of the Minimum Wage in Brazil Source Methodology Type of Data Sample Effects On Period Employment Wages Poverty Macedo & Time-senes Aggregate 1964-74 Limited Impact Garcia (1978) analvsis time series E< I Souza & Time-series Aggregate 1961-76 Validates the Baltar (1979) analysis bme series beacon effect Velloso (1990) Time-series Aggregate 1976-86 Limited Impact analysis time series Formal Sector 0.36 <<0.63 Informal Sector 0.43 < I < 0.60 Carneiro & Causality Aggregate 1980-97 Granger Faria (1997) analysis tme series Minimum wage Lemos (1997) Causes other Soares (1998) wages Foguel, Time-series Aggregate 1980-99 Validates the Ramos & analysis Ume senes beacon effect in Carneiro Co-integrahon long term (2000) and exogenicity E = 1 Umited impact in the short term ._ _ _ _ _ _ _ _ _ _ < 1 Foguel (1998) Differences-in- Aggregate 1980-87 Negative differences tme series informality Natural experiment Lemos (2000) Time-series Aggregate 1982-98 NegaUve Positive impact analysis time series -0.020 < E < on wages close Impact on wage 0.006 to the minimum distnbution Carneiro Time-series Aggregate 1982-99 Negative formal (2000) analysis tme series sector Co-integraton -0.020 < E < and exogenicity 0.001 Positive Informal sector 0.0004 decrease Addison and (1983-1996) poverty an average of 3.6% for unskilled workers in Blackburn (1999) the 1990s (no dis-employment effects); by 0%in the 1980s (with dis-employment effects). Colombia 10% increase in the minimum wage => decrease Maloney and Nunez (1997-2000) individual poverty of those earning 0-1.3 m.w. by 5- 18%; increase poverty of those earning more than 1.3 m.w. by 2% United States 10 % increase in minimum wage => increase Neumark, Schweitzer (1986-1992) individual poverty of minimum wage earners by 6% and Wascher (2000) because hours worked fell and wages increased very little; zero change for those earning above the minimum wage Family Poverty (Risk Pooling at the Household Level) Asia, Africa, LAC 10% increase in m.w. => decrease in poverty rate Lustig and McLeod (1950-1980) 6-10% (1996) Latin America An increase in the minimum wage is more Morley (1995) (1981-1989) correlated with a decrease in the poverty rate during periods of recession and a decrease in the poverty rate is more correlated with a decrease in the m.w. in periods of recovery United States An increase in the minimum wage moves poor Neumark and (1986-1995) families into a non-poor status and moves non- Wascher (1997) poor families into poverty, with a net increase in poverty, though this is not statistically significant. However, it does boost incomes of families who are below the poverty line both before and after the change in the minimum wage. United States Increase in the minimum wage 1 percent => no Card and Krueger (1990,1992) effect on poverty rates or a decline of the poverty (1995) rate by 1.6 percentage points, but this measure cannot be fully attributed to an increased minimum wage rather than to other changes in the economy at the time. - 176 - Poverty without risk pooling 8.37. Taking the individual as the unit of observation, the impact of a higher minimum wage on net average income is mixed in the five studies reviewed. Evidence is found in the United States that the unemployment effect dominates the wage effect since those who earn near the minimum wage experience a fall in net income of six percent for an 10 percent increase in the minimum wage (Neumark, Schweitzer, and Wascher 2000), but another study shows that poverty decreases by 3.5 percent for a 10 percent increase in the minimum wage (Addison and Blackbum 1999). Mixed results are also found in Brazil. One study finds no correlation between a higher minimum wage and the rate of poverty (IPEA 2000b) while another shows that the minimum wage does reduce poverty across the wage distribution due to indexing of the minimum wage.120 Under the assumption of no dis-employment effects and no risk pooling, a one percent increase in the minimum wage in Brazil is associated with a decrease in poverty of 0.1 percent for those earning the minimum, but a fall of 0.14 percent for workers across the whole wage distribution (Neri, Gonzaga, and Camargo 2000). Similarly to Brazil, an increase in the minimum wage in Colombia is felt across the whole eamings distribution, not just at the minimum, but the change in net income (poverty) effects depends on where the individual was located in the distribution before the change. An increase in Colombia's minimum wage of 10 percent leads to an increase in poverty (net fall of average income) by two percent among those earning more than 1.3 minimum wages, but among the poor, poverty falls 5 percent to 18 percent among those eamings 0-1.3 minimum wages (Maloney and Nunez 2001). Who Earns the Minimum Wage? 8.38. A newer set of papers have examined the poverty impacts of the minimum wage by using the household as the unit of observation, which is perhaps more appropriate for developing countries due to risk pooling at the household level. As pointed out in the developed country literature (Neumark and Wascher 1997, Gramlich 1976, Card and Krueger 1995), poverty at the individual level is not synonymous with poverty at the household level. In the United States, it is found that 40 percent of minimum wage eamers live in households with a total income more than three times the poverty line. Thus, to identify the poverty reduction effectiveness of the minimum wages, it is important to identify who eams the minimum wage and if they are the poor. 8.39. Using data from Mexico and Argentina, Table 8.3 shows that most minimum wage eamers are not the primary income eamer in the household. In Argentina, ten percent of household heads (male or female) eam at or below the minimum wage while four percent of Mexican households heads do. In contrast, 21 percent of Argentine children living with their parents and ten percent of Mexican children in their parents' home (age 12-17) eam the minimum wage. This is supported by the breakdown by age where 70 percent of working people age 12-17 eam the minimum wage or less in Argentina and 26 percent in Mexico, compared to 20 percent and 6 percent, respectively, of young working adults (18-24). The proportions in the United States are similar to those in Mexico 120 The conflicting results from studies in the same country that examine the poverty effects over the same period are due to different assumptions made by the authors and different estimation techniques. -177 - with 29.4 percent of those age 16-19 and 19.8 percent of those age 20-24 earning at or below the minimum (Card and Krueger 1995). 8.40. Table 8.3 and Table 8.4 show that the poor are more likely to earn at or below the minimum wage, but households in a/l wealth quintiles include minimum wage eamers in Argentina, Mexico, and the United States. In Argentina, 24.5 percent of the workers in the first wealth quintile (poorest) earn at or below the minimum wage while 5.6 percent of the top quintile do. In Mexico, there are fewer minimum wage eamers across all wealth quintiles with 14.6 percent of those in the first wealth quintile and 4.1 percent of those in the fifth quintile earning the minimum wage. The United States is similar with 28.8 percent of the lowest decile eaming below the minimum wage but only 2.7 percent of the highest income decile. Minimum wages constitute 17 percent of poor household's income but two percent of wealthy household's income in Argentina and 30 percent in Mexico (Table 8.4) while in the US, they contribute to 51 percent of total family income (Card and Krueger 1995). Thus, in the two Latin American countries selected for the table, minimum wages are a less important source of income than in the United States. Although minimum wages constitute a larger portion of poor households' total income, nearly equal proportions of households have minimum wage earners in Argentina: 22 percent of poor households have at least one minimum wage eamer while 18 percent of wealthy households do. This pattern is not reflected in Mexico, where the poor are three times as likely as the wealthy to have a minimum wage earner in the household. Table 8.3 Wage Earners at or Below The Minimum Wage Argentina' Mdxico2 Argentina Mexico Role in household Sector Head 10.1 4.2 Self-employment 23.7 13.9 Spouse 19.8 14.3 Informal wage 31.7 16.9 Child 20.9 9.9 Formal 4.7 0.5 Age Education 12-17 70 25.6 None --- 22.7 18-24 20.3 6.5 Primary 24.6 11.9 25-64 12.4 6.03 Secondary 12.9 6.5 65+ 31.7 24.4 Higher 6.5 2.2 Sex Wealth quintile Male 11.9 4.6 1 24.5 14.6 Female 19.9 13.2 2 16.4 10.3 3 12.1 8.4 4 10.5 6.9 5 5.6 4.1 'The National Urban Employment Survey (Encuesta Nacional de Empleo Urbano) 1997, quarter two 2 The Permanent Household Survey (Encuesta Permanante de Hogares) 1997, semester one 8.41. Table 8.3 also shows that those who are typically associated with lower wages tend to be minimum wage earners. Argentine women are nearly twice as likely as Argentine men to be minimum wage eamers while Mexican women are three times as likely as Mexican men to eam at or below the minimum wage, despite working women's higher levels of education in both countries. These proportions are very similar to the percentage of spouses (who are likely to be women) who earn at or below the minimum wage. Less educated workers are also more likely to be minimum wage earners, with over 25 percent of those with less than primary education eaming the minimum wage. However, 6.5 percent of Argentines with higher education and 2.2 percent of Mexican who "**==>This document did not complete OCR process. <==**"