Federative Republic of Brazil
Social Insurance and Labor Supply: Assessing Incentives and Redistribution
Technical Report
December 30, 2009
Human Development Sector Management Unit
Latin America and the Caribbean Region
Document of the World Bank
1
Acronyms and AbbreviationsAAA / Analytical and Advisory Activities
BNDES / Banco Nacional de Desenvolvimento Econômico e Social
BPC / Beneficio de PrestaçãoContinuada (Pensions for the Elderly Poor)
CEF / CaixaEconômica Federal (Federal Savings Union)
FGTS / Fundo de Garantia por Tempo de Serviço (Link of. Cerzisse Guarantee Fund)
GDP / Gross Domestic Product
GOB / Government of Brazil
INSS / InstitutoNacional de Seguridade Social (National Institute of Social Security)
IPEA / Instituto for Applied Economic Research
IRR / Internal Rate of Return (on contributions)
IT / Information Technology
LOC / Length of Contribution Pension
MTE / Ministry of Labor and Employment
OECD / Organization for Economic Cooperation and Development
PLOC / Proportional Length of Contribution Pension
RGPS / Regime Geral de Proteção Social(General Scheme for Social Protection)
SI / Social Insurance
SP / Social Protection
MPS / Ministry of Social Security
UI / Unemployment Insurance
TTL / Task Team Leader
Vice-President:Pamela Cox
Country Director:MakhtarDiop
Sector Director:Evangeline Javier
Sector Manager:Helena Ribe
Task Team Leaders:David Robalino/
Bénédicte de la Brière
Table of Contents
1.Motivations, Objectives and Context
1.1. Why this Technical Report?
1.2. Some Stylized Facts About Labor Market Dynamics in Brazil
2. Social Insurance, Redistribution and Economic Incentives
2.1. Financing Mechanisms, Tax Wedge and Employment
2.2. Pensions
The mandate of the system
Retirement decisions
Sector choice and entry into the labor force
Redistribution
2.3. Income Protection System for Workers
The mandate of the system
Incentives to work and sector choice
Redistribution
Other issues that deserve attention
3. Improving Redistribution and Incentives
3.1. Policy Framework for Social Insurance Policy
3.2. Implications for Brazil
Pensions
Income protection
3.3. Potential Effects on Behaviors and Program Costs
References
Tables
Table 1: Summary of Pay-Roll Taxes and Social Security Contributions
Table 2: Expenditures in Social Insurance (excluding health)
Figures
Figure 1: Labor Force Age/Gender Composition and Growth
Figure 2: Labor Force by Occupation and Distribution of Net Jobs
Figure 3: Labor Force by Level of Education
Figure 4: Distribution of Earnings for Workers with and Without Carteira
Figure 5: Monthly Probabilities of Separation in the Formal and Informal Sector
Figure 6: Distribution of a “Typical” Cohort of 25 Year-Old Males by State (Urban Region)
Figure 7: Tax Wedge in Brazil and Selected Countries
Figure 8: Gross Replacement Rates by Retirement Age
Figure 9: Retirement Ages and Life Expectancies Around the World
Figure 10: Distribution of Type of Pension, Vesting Period and Retirement Probabilities
Figure 11: Change in Pension Wealth for Each Additional Year of Contribution
Figure 12: Internal Rates of Return on Contributions by Retirement Age and Level of Income
Figure 13: Informal Sector Productivity, Payroll Taxes, and Incentives for Informality
Figure 14: Expected Net Life-Time Wealth for Low Income Individuals
Figure 15: Relative and Absolute Implicit Subsidies
Figure 16: Replacement Rates and Accumulations from UI and FGTS
Figure 17: Replacement Rates and Benefit Duration in Selected Countries
Figure 18: Relative Speed of Exit from Unemployment by Type of Worker and Sector Choice
Figure 19: Take-up Rates and Share of UI Payments by Income Level
Figure 20: Projected Expenditures on Unemployment Benefits
Figure 21: Designing Subsidies for Individuals with No or Limited Savings Capacity
Figure 22: Probabilities of Contributing to INSS and Retiring
Figure 23: Effects of SI Programs on Contribution Densities and Retirement Ages (Average Earner)
Figure 24: Effects of Matching Contributions
Figure 25: Costs of Matching Contributions and Minimum Pensions
Acknowledgements
This Technical Report is the result of a joint effort between the Government of Brazil, local academic and research institutions, and the World Bank. The report builds on the considerable amount of empirical analysis by Brazilian researchers compiled during the first phase of the PESW on Labor Markets. The new analytical work and the preparation of the Report was conducted by a team including: David A. Robalino (TTL and main author), Benedicte De La Briere (Co-TTL), Wendy Cunningham, Jason Hobbs, Stanislao Maldonado, LerickKebeck, Carla Zardo (World Bank), HélioZylberstajn and Luis Eduardo Afonso(University of Sao Paulo), André Portela, Vladimir Ponczek and Eduardo Zylberstajn (Getúlio Vargas Foundation/São Paulo), and David Margolis and Raul Samponagro (University of Paris I). The team worked under the guidance of Helena Ribe (Sector Manager Social Protection). The team would like to thank Luis Enrique da Silva de Paiva (MPS), Adriana Maria Giubertti (MTE), Antônio Sérgio A. Vidigal (Secretário de Políticas Públicas de Emprego, MTE), Rodolfo Péres Torelly (Seguro-Desemprego - MTE) and Milena Souto Maior de Medeiros (MTE) for helpful discussions and for providing access to data. Also our thanks go to the teams at IPEA Rio de Janeiro and IPEA Brazilia for insightful comments at various stages in the preparation of the analysis.
The peer reviewers for this Technical Report were Bill Maloney, Polly Jones, and Gordon Betcherman.
1
1.Motivations, Objectives and Context
ThisTechnicalReportanalyzes the potential effects of thepensions and income protection systemson labor supply decisions and through this channel the coverage and cost of the programs. The focus is on old-age pensions provided as part of theGeneral Scheme for Social Protection (RGPS) by the National Institute of Social Security (INSS), and the income protection programs managed by the CaixaEconômica Federal (CEF), which include the Unemployment Insurance (UI) system and the Length of Service Guarantee Fund (FGTS) -- essentially funded unemployment individual savings accounts. The Report is part of Pillar I of the Brazil PESW on Labor Markets, a three year program of analytical and advisory services aiming at improving the design of social protection programs to promote the creation of high quality jobs and help low income individuals graduate from welfare. Pillar I investigates issues related to the effects of public transfers on labor supply, in particular, whether there are reforms that need to be considered to improve incentives for work and stimulate formal vs. informal employment.
The Report addresses fivepolicy questions: (i) how is the pension system influencing decisions regarding entry into the labor market, sector choice (formal/informal), and retirement? (ii) How are current income protection programs affecting turnover; job search efforts and sector choice?; (iii) Are there interactions between the two systems that aggravate or mitigate incentive effects?;(iv) What is the role of redistributive policies in determining observed outcomes?; and (v) Are there policy interventions that could be considered to correct incentives, while securing adequate income protection for different population groups (including workers in the informal sector)?
The Report is organized in three sections. This first section discusses the motivations for the study and sets the context by briefly describing some key stylized facts about the Brazilian labor market. The focus is on the composition of the labor force, its distribution by occupational categories, unemployment risks, and transitions between employment states -- including workers flows between the formal and informal sectors. The second section analyzes the Brazilian social insurance system in terms of incentives and redistribution. It starts with an overview of institutional arrangements, programs, and financing mechanisms, which flags the problem of a high tax-wedge. The Report then analyzes the rules of the pensions and income protection systems and their potential effects on behaviors. The final section proposes a policy framework to guide reforms that could “correct” incentives by making redistribution more transparent and progressive. A behavioral life-cycle model estimated for Brazil is used to illustrate how the application of this framework could affect contribution densities (and therefore the time spent working in formal sector jobs), retirement ages, savings, and programs cost. A companion Policy Note on Social Insurance and Labor Supply summarizes the main conclusions and recommendations from the analysis.
1.1. Why thisTechnical Report?
The main motivation for preparing this Report is a general concern in Brazilwith the unintended economic consequences of public transfers. Indeed, transfers through social insurance, active labor markets and social assistance programs are estimated at 13 percent of GDP.[1] Some of these transfers are explicit, like thoserelated to social assistance programs; others are implicit, like in the case of pensions and unemployment insurance wheninternal rates of return on contributions for certain population groups are above market rates. In general, there is evidence that the various programs have played an important role in helping individuals manage risks such as unemployment and inflation, and have contributed to reduce poverty.[2] At the same time, however, there are concerns among Brazilian policymakers about the potential effects of these programs on incentives to work and saveand the associated economic and fiscal costs. Indeed, experiences elsewhere show that some of these effects can be important, including in the case of insurance programs.[3] In OECD countries, for instance, mandatory earnings related to pension systems are shown to encourage early retirement, which costs the economies around 7 percent of GDP. In Brazil,several studies reviewed throughout the Report also flagproblems with incentives in the pensions and income protection systems.[4]
This Report argues that incentives problems depend, to a large extent, on the type of redistribution embedded in the various programs. In principle, programs that do not redistribute income are less likely to affect behaviors. By design, however, social insurance programs are redistributive. Redistributive in the sense that, systematically, some population groups “get out” of the programs more than what they “put in,” which means that others get less. In essence, the programs involve subsidies and taxes which depend not only onintrinsic characteristics of the individuals but also on their behaviors. Individuals might attempt to exploit subsidies and avoid taxes. And as a general rule, the less explicit the form of redistribution, the more prone the system is to abuse or strategic gaming and the more likely that it will induce undesirable behaviors. For instance, individuals might have incentives to avoid the formal sector, strategically manipulate wages, retire early, reduce job search efforts, and, in general, reduce labor supply and savings. This type of behaviors increase the fiscal costs of the programs, can reduce aggregate productivity and output, and at the end also generate adverse redistribution.
In countries like Brazil, where the formal and informal sectors are reasonably integrated, problems with the design of the social insurance system can be amplified. Indeed, in a country where the labor market is segmented into a formal and informal sector and workers flows among the two are thin or non-existent, benefit formulas and eligibility conditions in the Social Insurance System (SI) system will do little to affect the relative sizes of the two sectors[5] and/or contribution densities. The Social Insurance (SI) system can still influence when individuals retire, how much they save, and how much effort they put into exiting unemployment, but it is unlikely to affect flows between the two sectors or relative job finding rates. Things are more complex in a country like Brazil where the empirical evidence suggeststhat the informal sector, at least in part, can be seen as an unregulated, voluntary self-employed sector.[6] In this case, workers seem to make explicit choices about formal vs. informal jobs based on an assessment of expected costs and benefits. The SI system can affect these costs and benefits and thus influence relative separation and job finding rates and the size of the formal sector. There is evidence, for instance, that the constitutional reforms implemented in 1988 which increased the direct and indirect costs of labor, contributed to the contraction of the formal sector during the 90s.[7] In part, this can be explained by lower hiring rates in the formal sector, but as this Report will argue, reduced efforts to find/keep jobs in the formal sector can also be part of the story.
TheReport will illustratehow current SI programs can be upgraded to improve both equityand economic efficiency by making redistribution explicit. Redistribution per se is not a problem. Having redistributive public programs is a choice that most societies make to different degrees. Some losses in efficiency are to be expected, but these are compensated by the social benefits that redistribution brings. The problem is when redistribution brings excessive and unnecessary losses. TheReport argues that this can be the case when the insurance and redistributive functions of the programs are not well defined. Improving efficiency then implies separating these two functions and ensuring,to the extent possible, that the risk-pooling/savings programs are incentive neutral[8]and that public subsidies are then used to “top up” the benefits or contributions of individuals with limited or no savings capacity. These top-ups would still affect incentives for those with incomes close to the eligibility line, but the effects would be more localized and would concern fewer and less productive workers.[9] Moreover, negative effects can be minimized by appropriately designing the “claw-back” for the transfer (i.e., the marginal tax on the transfer) and/or introducing conditionalities.
TheReport is based on a series of analytical pieces looking at how benefit formulas and eligibility conditions in the pensions and income protection systems canaffect individual behaviors. In the case of pensions, the analysis is first based on the calculation of four indicators that provide information about incentives (not actual behaviors) to participate in the labor force, save and contribute to the social security. These are: the pension wealth, life-time net earnings, internal rates of return on contributions, and implicit taxes and subsidies received by different plan members. For the income protection system, an econometric model was estimated looking at the potential effect of the unemployment insurance system and the unemployment individualFGTS savings on employment duration and sector choice in Metropolitan areas.[10] A review of previous studies was also conducted to better understand the impact that FGTS might be having on hiring and turn-over rates. Finally,the interactions between the pensions and income protection systems and the potential impacts of changes in benefit formulas and eligibility conditionson individuals work choices were analyzed on the basis of a life cycle behavioral model. The model takes into account a large range of possible behavioral responses that could affect contribution densities, the distribution of retirement ages, savings, and ultimately program costs.[11] The focus is on how different features of the programs (i.e., benefit formulas and eligibility conditions) affect job search/job preservation efforts in the formal sector, which in turn affect separation and job finding rates. This is in contrast to the more traditional analyses looking at how the presence of a given SI program (e.g., the unemployment system) affects outcomes such as the duration of unemployment spells or separation/job finding rates in the formal sector, without taking into account, explicitly, individual choices.
1.2. Some Stylized Facts About Labor Market Dynamics in Brazil
Today, there are around 88 million people in the Brazilian labor force, a number that has been growing quite rapidly over the past decade. Between years 2001 and 2006 the labor force grew at an average of 3.4 percent per year, among the fastest expansions observed in the world. If current trends continue, by year 2010 the number of people working or looking for jobs could surpass 95 million. The labor force remains very young with more than 90 percent of its members below age 50, compared to 74 percent in OECD or 64 percent in Japan. There has also been a considerable increase in the share of women who now represent 35 percent ofits members. Their participation rate, however, is still considerably lower than in OECD countries. Only around 50 percent of Brazilian women between ages 25 and 50 participate in the labor market compared to an average of 80 percent in OECD.
Like in other countries in the region, at any point in time, a large share of the labor force is outside the formal sector. In Brazil, formal sector workers are classified in the household survey as those with “carteira.” The “carteira” is a document indicating that the contract was registered by the employer with the Ministry of Labor. Thus, the majority of workers with carteirahave social security coverage; most workers without carteira do not. Workers with carteira represented in 2006 only 33.7 percent of the labor force down from 40 percent in 1990. Workerswithout carteiraand the self-employed, on the other hand, represented respectively 28and 8percent of the labor force. This group of informal sector workers has been expanding. Indeed, between 1990 and 2006 the economy created, net, 31.2 million jobs. A large part was jobs without carteira(27 percent) andself-employment (14 percent).
Figure 1: Labor Force Age/Gender Composition and Growth
Source: Authors’ calculations based on the PNADs 1990 and 2006.