This paper analyzes several issues regarding fiscal sustainability and fiscal adjustment in Brazil during 1990 and searches for econometric evidence of a monetary dominant regime during some subperiods. The following statistical data are also presented in detail: macroeconomic flows and balances, industrial production, consumer price index, relative public sector prices and tariffs, minimum wage statistics, financial system loans, monetary aggregates, exports by principal commodity groups, direction of trade, detailed balance of payments, total external debt, central government operations, and so on.


This paper analyzes several issues regarding fiscal sustainability and fiscal adjustment in Brazil during 1990 and searches for econometric evidence of a monetary dominant regime during some subperiods. The following statistical data are also presented in detail: macroeconomic flows and balances, industrial production, consumer price index, relative public sector prices and tariffs, minimum wage statistics, financial system loans, monetary aggregates, exports by principal commodity groups, direction of trade, detailed balance of payments, total external debt, central government operations, and so on.

III. Social Spending in Brazil: Recent Trends in Social Assistance and Insurance1

A. Introduction

1. Brazil has a broad array of social insurance and assistance programs. As discussed in Section IV, social security benefits are Brazil’s main social safety nets and account for nearly half of total public social spending, or approximately 10 percent of GDP. Excluding pensions and other social security benefits, unemployment insurance is the main social insurance program in Brazil. Only a small share of public outlays on social programs (approximately 1 percent of GDP) is devoted to social assistance. Most of these programs (for instance, old-age and disability benefits) are targeted and in general pro-poor.

2. In recent years, progress in strengthening Brazil’s social insurance and assistance programs has been twofold. First, efforts have been focused on improving program design and service delivery; and on distinguishing clearly social assistance and social insurance programs, as well as their sources of finance, while at the same time preventing shortfalls in finance for other untargeted social programs, such as education and health care. Second, social assistance policies are being integrated into broader human development initiatives. A case in point is the recently-launched Alvorada Program.

3. This section is structured as follows. Subsection A provides an overview of the existing social assistance programs. Subsection B focuses on the effectiveness of social assistance programs. Subsection C discusses the regional dimension of human development and the recently-launched Alvorada program. Subsection D concludes.

B. Social Assistance Programs: An Overview

4. Most social assistance and insurance programs in Brazil are provided according to the Social Assistance Law (LOAS). These programs comprise rural pensions,2 pensions to elderly and disabled persons,3 and income support programs.4 Some subnational governments also have their own income support programs.5 These programs are in general well targeted.

5. Excluding social security benefits, the main social safety nets in Brazil are labor protection programs, some of which are financed by the private sector. In particular:

  • There are two unemployment insurance programs in Brazil paid through FAT and FGTS.6 FAT is an unemployment fund financed by taxes on enterprises’ payroll and gross earnings, whereas FGTS is a government-run fund of individual accounts for formal sector workers. Access to the FAT unemployment insurance is universal among formal sector workers, but FGTS benefits are restricted to those formal sector workers with individual accounts.

  • The salary bonus consists of a monthly salary paid on an annual basis to formal sector workers earning up to two minimum wages.7 Other labor programs include job creation, retraining, and on-the-job training, as well as nutrition benefits provided to low-income workers.8

6. FGTS unemployment insurance resources are also used to finance infrastructure development programs, particularly housing, urbanization and sanitation, given Brazil’s low private savings rates and limited sources of long-term financing for infrastructure projects in the private sector. These programs benefit the poor by providing housing for low-income households. Because they are labor-intensive, spending under these programs also generates earnings opportunities for the working poor.

7. FGTS has suffered from financial imbalances. Net inflows to FGTS are falling with rising informality in the labor market and withdrawals due to the increase in registered unemployment in recent years (Table 3.1).9 Moreover, there is a mismatch in the maturity composition of FGTS assets and liabilities: whereas FGTS’s assets have long-term maturities, most of its liabilities are short term.10 Furthermore, despite the statutory minimum remuneration, in the period of chronic inflation, the rate of return on individual accounts was lower than that of other financial assets, including savings accounts (Oliveira and others, 1999).11

Table 3.1.

FGTS Outturn

(In billions of reais)

article image
Sources: Caixa Economica Federal; and IMF staff calculations.

Until March 2000.

8. Core social assistance programs have been protected from fiscal adjustment in 1999 and 2000. In November 1998, 22 core social programs were identified in coordination with the IDB and the World Bank to be preserved from cuts in the ensuing period of fiscal adjustment (Table 3.2). The share of GDP devoted to these programs varied between 1.2–1.3 percent in 1998–00, at the same time that spending on other programs was substantially cut. The main protected programs are the unemployment insurance, the social assistance benefits provided through LOAS, and the equalization components in the publicly-provided education and health care systems, discussed in Section 4. Allocations for 2001 have increased to 1.3 percent of GDP. This is a positive development that reflects the government’s current efforts to strengthen the better targeted social assistance programs.

Table 3.2.

Federal Spending on 22 Core Programs, 1995-01

(In millions of reais)

article image
Sources: Brazilian authorities, and IMF staff estimations

As of July 5, 2000.

In 1997, includes R$100 million relative to FUNDEF outlays.

As reported by the Ministry of Health.

C. The Effectiveness of Social Assistance and Insurance Programs

9. Social security benefits are the most important social safety nets in Brazil. However, these programs suffer from three main shortcomings:

  • There are significant differentials between private and public sector pensions and within these social security regimes. Whereas private sector pensions averaged less than two minimum wages in the second half of the 1990s, public sector pensions in the executive branch of the federal government were, on average, seven times higher. Within the social security regime for private-sector workers (RGPS), nearly 80 percent of pensions are below 1.5 minimum wage. In addition, there are significant differentials within the social security regime for public-sector workers (RJU), where average pensions vary from 15 minimum wages in the executive branch of the federal government to over 54 minimum wages in the federal legislature. These differentials have been reduced over time, but remain sizable.

  • Social security coverage is limited. While all civil servants are covered and benefit from higher pensions, only approximately 58 percent of the working age population in the private sector is covered by social security (Ministry of Social Security and Assistance, 2000). Coverage is lower among informal sector workers and the self-employed. Coverage rates also vary according to occupation and among the states.

  • Household survey data provide mixed results on the incidence of public spending on social security benefits. Most pension recipients are concentrated in the lowest and highest income quintiles.12 There is evidence that these benefits have alleviated poverty among the elderly, as expected.13 It has been argued that rural pensions are probably the best targeted social insurance program in Brazil (World Bank, 2000b).

10. Publicly-funded labor programs are less efficient social protection instruments. This is because:

  • There is some evidence of poor targeting in the case of the unemployment benefit, and the salary bonus (Table 3.3).14 15 The share of these benefits in household income has been shown to be higher in high income households. Information on the recipients of special labor programs, such as the workers’ nutrition program, as well as job creation and training programs, is not readily available. This would allow for a better assessment of the incidence of these outlays (Table 3.4). These special programs are better targeted to low-income workers and therefore their incidence rates are likely to be higher than that of the unemployment insurance paid through FAT and FGTS. It has been argued that the incidence of the workers’ nutrition program has improved by extending the program to smaller enterprises, where labor compensation is typically lower.16

  • The truly needy often do not have access to the unemployment benefits paid through FAT and FGTS because eligibility is based on formal employment and a sizable share of the working poor are engaged in informal activities.17 Also, FGTS account balances are typically low for low-income workers, due to short job tenure in the formal sector and high job turnover. Among the working poor, the value of the unemployment benefit paid through FAT often exceeds their FGTS balances.

  • The government’s ability to engage in counter-cyclical job creation is limited. This would provide additional social protection in periods of economic downturn, when job losses are likely. However, the labor-intensive investment programs financed through FGTS are procyclical because FGTS inflows rise with formal employment and pay.18 Also, Brazil has limited experience with public works programs, despite the effectiveness of the Northeast drought program implemented in 1998.

  • Training and retraining programs need to be more attuned to market needs. Given the disparities in skills among labor market participants and the regional differences in the demand for training, efforts have been made to provide these programs in conjunction with state and municipal job creation agencies. Labor training programs have also been provided by BNDES using FAT resources.

Table 3.3.

The Incidence of Social Assistance and Insurance Programs: A Summary of Recent Empirical Findings

article image

Maternity and disability benefits, unemployment insurance, and education grants.

Table 3.4.

FAT Outturn, 1999-00

(In billions of reais)

article image
Sources: Ministry of Labor and Employment; and IMF staff calculations.

11. Labor legislation offers limited social protection. Given the cost of formal employment and high mobility into the informal sector, severance pay legislation creates an incentive for employers and employees to terminate labor contracts and claim unfair dismissal. In this case, employers are required to pay 40 percent of the workers FGTS balance as compensation and workers can draw their FGTS balances and remain working informally.19 However, minimum wage legislation is a good social protection instrument because of the higher concentration of minimum wage earners among the working poor and because the minimum wage has been shown to be a powerful determinant of pay in the informal sector.20 Recent studies have shown that increases in the minimum wage have had a stronger impact on poverty in the post-1994 period, relative to the period of high inflation (Amadeo and Neri, 1999; World Bank, 2000a). It has been argued that, in the period of high inflation prior to 1994, nominal increases in the minimal wage, although frequent, did not preserve its purchasing power.

12. The housing/sanitation programs financed through FGTS and FAT are poor poverty-reduction instruments. The incidence of spending on the housing programs funded by FGTS is low because loans are restricted to applicants with household income between 4 and 12 minimum wages and the housing deficit is concentrated among the poor. Financing options are limited for low-income households, with income up to 3 minimum wages, and emphasis is placed on government provision of housing, upgrading of the existing housing stock in degraded areas, urbanization, and sanitation. Loans to higher income households are also provided by CEF (Caixa Económica Federal) using its own resources.21 Because of its eligibility conditions, the implicit subsidy in the system is often extended to the middle class.22 It has been shown that the incidence of publicly-funded sanitation programs is also relatively poor.23 Low cost recovery in mortgage payments in the case of low-income households has compromised the financial balance of FGTS without improving the incidence of public spending on housing and urbanization.24

13. Spending on housing, urbanization, and sanitation has been affected adversely by fiscal retrenchment, particularly at the subnational level. These programs can be financed through FGTS or through the budget, when cost-recovery is unlikely and the program has a clear social assistance function, as discussed above.25 The impact of fiscal adjustment on these programs has been twofold. First, stricter restrictions on subnational borrowing and indebtedness have limited the ability of states and municipalities to borrow from FGTS. Prudential regulation on financial institutions’ exposure to subnational government debt has also reduced the ability of public banks to lend to subnational governments. Private banks that are not constrained by these regulations are in general not willing to finance long-term infrastructure development projects. Second, fiscal adjustment at the federal level has limited further the availability of finance for housing/sanitation projects outside FGTS.

14. Public spending on housing has benefited from improved governance in recent years. This has been achieved through increased community participation in housing programs for low-income households. It has been argued that better oversight for the use of the resources transferred to subnational governments, particularly municipalities, has reduced misallocation of funds. Direct lending to higher income borrowers has also reduced transactions costs.26 As in the case of other federal financial institutions, CEF operations have been object of stricter central bank oversight and regulation.

D. Social Policies and Human Development: The Alvorada Program


15. With a score of 0.74 in the UNDP’s human development index (HDI) in 1999, Brazil ranks among the countries with medium human development.27 However, the national average hides sizable differentials in human development indicators across regions (Figure 3.1). Although all regions have had a upward trend in their HDI scores since 1980, those with low human development indicators do not seem to be catching up fast enough with their more developed counterparts, inequality in human development also exists within the states and certain poorer municipalities have lagged behind the regions with higher HDI scores in the states where they are located.

Figure 3.1.
Figure 3.1.

Human Development Indices (HDI), 1980-96

(by Region)

Citation: IMF Staff Country Reports 2001, 010; 10.5089/9781451805901.002.A003

Sources: Data provided by the authorities; and IMF staff calculations.

The Alvorada Program

16. There has been growing recognition that the recent positive trends in human development, as well as a faster catch-up for the least developed states and municipalities, can only be sustained through a concerted policy effort. With the exception of cash transfers such as social security benefits and unemployment insurance, for instance, most existing social programs in Brazil are not designed to address these regional inequalities in human development. Regional development programs often benefit regions that are considered poor within the state where they are located, rather than relative to a national poverty benchmark. As a result, poorer regions in richer states often benefit from these programs although they may be more prosperous than richer regions in poorer states. Initiatives in the past—the creation of regional development agencies and banks, for instance—have been unsuccessful in dealing with regional inequalities in human and economic development.

17. The Alvorada Program consists of focusing social policies and outlays on existing social programs in the states and municipalities with human development indices below the national median. The Alvorada Program was originally named IDH-14 Program, given its focus on the 14 states with HDI scores below the national median.28 The program was subsequently enlarged to benefit poorer regions within richer states that would not be eligible for assistance under the IDH-14 program.29 The priority programs are in the areas of health care, education, and income support (Table 3.5). A large share of spending on priority programs will be devoted to infrastructure development, including sanitation. Funding for these programs has been scarce in light of the ongoing consolidation of fiscal adjustment since late-1998, as discussed above.

Table 3.5.

The Alvorada Program Budget, 2000-02

(In billions of reais)

article image
Sources: Data provided by the authorities; and IMF staff calculations.

18. Most financing needs will be met by the federal budget. The federal government will finance R$7.4 billion out of the total estimated budget of R$ 3.3 billion between 2000-02. A large share of total financing (R$5.0 billion) is already programmed in the multiyear budget framework (PPA). Additional funding will come from Poverty Fund resources in 2001-02. Subnational governments will also contribute, particularly in counterpart funds for foreign-financed projects.30 Extension of the original IDH-14 programs will cost the federal government R$0.2 billion between 2000-02, in addition to the R$0.9 billion already programmed in the multiyear budget.

19. To facilitate monitoring and evaluation, targets have been set for each program, in terms of output indicators such as coverage of the sanitation and water network, school enrollment rates, and number of beneficiaries of income support programs. The impact of these programs on the relevant social and human development indicators is to be more closely monitored than in the past, and the authorities have approached the World Bank and the IDB for assistance in this respect.

E. Conclusions

20. Brazil devotes nearly half of public outlays on social programs to social insurance and assistance. Programs of a clear social assistance nature account for a small share of these outlays, or nearly 1 percent of GDP. These expenditures are not only low relative to total public spending on social programs but, more importantly, most social assistance programs in Brazil do not provide adequate social protection. Almost one-half of spending on these programs is allocated to unemployment insurance, which benefits higher income workers in the formal sector. Other programs are reasonably well targeted and cost-effective, particularly the rural pensions. Leakages may nevertheless occur due to difficulties in income certification and self-targeting in the case of rural pensions and income support. Spending on infrastructure development, including housing and sanitation, has been affected adversely by fiscal consolidation.

21. Against this background, it has often been argued that a gradual expansion of Brazil’s safety nets can be achieved primarily through better targeting, rather than increased spending. The most important gap in social assistance in Brazil is the absence of a program targeted to informal sector workers. In this respect, labor market participation indicators could be used for the purpose of categorical targeting of social assistance program, given the difficulty to reach the poor in the informal sector.31 Much remains to be done if social assistance is to be a powerful social safety net and poverty alleviation instrument in Brazil. Protection of a number of core social assistance programs from cuts in periods of fiscal adjustment does not ensure adequate social protection if the programs in question offer limited assistance to the truly needy.


  • Amadeo, E., and J.M Camargo, 1996, “Instituições e Mercado de Trabalho no Brasil” in J.M. Camargo, Flexibilidade do Mercado de Trabalho no Brasil (Rio de Janeiro: FGV Editora).

    • Search Google Scholar
    • Export Citation
  • Amadeo, E., and M. Neri, 1999, “Macroeconomic Policy and Poverty in Brazil,” mimeo.

  • Carvalho, C.E.; and M.M.S. Pinheiro, 1999, “FGTS: Avaliação das Propostas de Reforma e Extinção,” Discussion Paper No. 686, September 1999, IPEA, Ministry of Planning and Budget, Brazil.

    • Search Google Scholar
    • Export Citation
  • Carneiro, F.G., 2000, “The Impact of Minimum Wages on Earnings, Employment and Informality,” Universidade Católica de Brasília, mimeo.

    • Search Google Scholar
    • Export Citation
  • Gonçalves, R.R., 1998, “O Déficit Habitacional Brasileiro: Um Mapeamento por Unidade da Federação e Por Nivel de Renda Familiar,” Discussion Paper No. 559, April 1998, IPEA, Ministry of Planning and Budget, Brazil.

    • Search Google Scholar
    • Export Citation
  • Gonzaga, G., 1998, “Rotatividade e Qualidade do Emprego no Brasil,” Revista de Economia Política, Vol. 18, p. 69.

  • Lobato, A.L., L. Aquino, and J.A.C. Ribeiro, 1999, “Análise de Registros Administrativos de Programas Sociais,” in Gasto Social: O IPEA Debate, IPEA, Ministry of Planning and Budget, Brazil.

    • Search Google Scholar
    • Export Citation
  • Ministry of Social Insurance and Assistance, 2000, Informe de Previdência Social, Volume 12, No. 3, March 2000.

  • Oliveira, F.E.B., K.I. Beltrão, M.T.M. Pasinato, and M.G. Ferreira, 1999, “A Rentabilidade do FGTS,” Discussion Paper No. 637, April 1999, IPEA, Ministry of Planning and Budget, Brazil.

    • Search Google Scholar
    • Export Citation
  • Paes de Barros, R.; C.H. Corseuil, and M. Bahia, 1999, “Labor Market Regulations and the Duration of Employment in Brazil,” Discussion Paper No. 676, October 1999, IPEA, Ministry of Planning and Budget, Brazil.

    • Search Google Scholar
    • Export Citation
  • Paes de Barros, R.; C.H. Corseuil, and M. Bahia; and Foguel, M., 2000, “Os Incentivos Adversos e a Focalização dos Programas de Proteção ao Trabalhador no Brasil,” IPEA, mimeo.

    • Search Google Scholar
    • Export Citation
  • Paes de Barros, R., R. Mendonça, and D. Santos, 1999, “Incidência e Natureza da Pobreza entre Idosos no Brasil,” Discussion Paper No. 686, December 1999, IPEA, Ministry of Planning and Budget, Brazil.

    • Search Google Scholar
    • Export Citation
  • Ramos, C.A., 1999, “Impacto Distributivo dos Gastos Sociais no Mercado de Trabalho,” in Gasto Social: O IPEA Debate, IPEA, Ministry of Planning and Budget, Brazil.

    • Search Google Scholar
    • Export Citation
  • Santos, C.H.M., 1999, “Politicas Federais de Habitação no Brasil: 1964/1998,” Discussion Paper No. 654, July 1999, IPEA, Ministry of Planning and Budget, Brazil.

    • Search Google Scholar
    • Export Citation
  • Soares, M.C., 1999, “A Mensuração do Impacto Distributivo do Gasto Social: Um Estudo para a Região Metropolitana de São Paulo,” Discussion Paper No. 643, May 1999, IPEA, Ministry of Planning and Budget, Brazil.

    • Search Google Scholar
    • Export Citation
  • Souza, M.M.C., 2000, “A Importancia de se Conhecer Melhor as Familias para a Elaboração de Políticas Sociais na América Latina,” Discussion Paper No. 699, January 2000, IPEA, Ministry of Planning and Budget, Brazil.

    • Search Google Scholar
    • Export Citation
  • SEPURB, 1996, Política Nacional de Habitação, Ministry of Planning, Brasília.

  • SEPURB, 1998, Política de Habitação: Ações do Governo Federal, Jan./95 a Jun./98, Ministry of Planning, Brasília.

  • Von Amsberg, J., 2000, “The Poverty Targeting of Social Spending in Brazil,” World Bank, mimeo.

  • World Bank, 2000a, Brazil: Selected Issues in Social Protection, World Bank Report 20054-BR (Washington, DC: World Bank).

  • World Bank, 2000b, Attacking Brazil’s Poverty: A Framework for Sustainable Poverty Reduction—With a Focus on Urban Areas, World Bank Report 20475-BR (Washington, DC: World Bank).

    • Search Google Scholar
    • Export Citation

Prepared by Luiz de Mello.


Rural pensions are treated as a social insurance program by the Ministry of Social Security and Assistance. However, these benefits have a social assistance nature, given the weak link between contributions and benefits. There were over 6 million recipients of rural pensions, at a total cost to the budget of approximately R$ 10 billion in 1999.


The main such program is BPC (Beneficio de Prestação Continuada), which replaced RMV (Renda Mensal Vitalícia).


A national income support program—in effect since 1997—benefits low-income municipalities, defined as those with revenues and per capita income lower than the state average. For a list of all municipalities in the program, see


A subnational income support program that has obtained widespread public support is Bolsa Escola, consisting of targeted cash transfers to low-income families on the condition of school attendance of all children in the household. The program was implemented in 1995 in the Federal District and in the city of Campinas, and subsequently in a few states and about 60 municipalities. For more information, see World Bank (2000a). Child labor eradication (PETI), a federal government-funded program, consists of a cash transfers to low-income households to keep children in school who would otherwise need to work.


Forty percent of FAT (Fundo de Amparo ao Trabalhador) resources are passed on to the National Development Bank (BNDES) to fund its development loan portfolio. The remaining funds are used to finance unemployment insurance and the salary bonus, as well as labor training and job creation programs, to be discussed below. FGTS (Fundo de Garantía por Tempo de Serviço) was created in 1966 as an unemployment insurance fund financed through employers’ contributions (8 percent of employees monthly earnings). The individual accounts are managed by CEF (Caixa Econômica Federal). See Paes e Barros, Corseuil, and Bahia (1999); and Oliveira and others (1999), for more information.


The salary bonus—in effect since the promulgation of the 1988 Constitution—is paid to private- and public-sector workers. The benefit amounts to a fourteenth salary per year, in addition to the thirteenth salary all workers in the formal sector receive at the end of the year. For more information, see


The worker’s nutrition program (PAT)—in effect since 1976—consists of income tax deductibility for enterprises providing meals to their employees earning up to 5 minimum wages. Benefits can be paid through vouchers to be redeemed in restaurants and supermarkets, or as meals provided in the workplace. At the employer’s discretion, the benefit can be extended to workers earning more than 5 minimum wages.


The share of the labor force engaged in informal activities reached 54 percent in 1998, against 43 percent in 1990. Withdrawals are allowed in the case of unfair dismissal, illness, retirement, death, and/or to purchase a house or finance housing repairs and upgrading. Based on PNAD-96 data, only 12 percent of the poor are employed in the formal sector, against over 33 percent for the nonpoor. According to PPV-96 data, over 80 percent of the household in the lowest income quintile are headed by informal sector workers.


In the past, financial imbalances were also due to the indexation of loan repayments to salary increases. In the period of high inflation, the mismatch in nominal increases in wages and salaries and consumer price inflation reduced loan recovery in the housing projects financed through FGTS.


The system guarantees a minimum real rate of return on individual accounts of 3 percent per annum. Higher statutory rates of return are guaranteed based on the length of employment and contribution level. In 1998, withdrawals amounted to 98 percent of deposits, against 80 percent in 1994 and 85 percent in 1995 (Carvalho and Pinheiro, 1999). In this respect, the Supreme Court has recently ruled in favor of monetary correction of the FGTS account balances in the period of the Collor I and Verão stabilization programs.


Based on data for the metropolitan region of São Paulo, Soares (1999) shows that nearly 33 percent of public pensions accrue to those recipients in the lowest income quintile, against nearly 21 percent in the highest quintile. According to PNAD data for the whole country, the share of public pensions accruing to the lowest quintile falls to nearly 27 percent and that accruing to the highest quintile increases to nearly 30 percent.


According to Paes de Barros, Mendonça, and Santos (1999), pensions and social security benefits account for almost 60 percent of per capita household income of the poor in the over-60 age group, against nearly 47 percent for the nonpoor in the same age group. Poverty incidence and income gaps are also lower among the over-60s than in the population as a whole. Moreover, the incidence of poverty is lower among households with an elderly member, given the impact of old-age pensions on the intra-household distribution of income.


Using POF-96 data, Ramos (1999) shows that 43 percent of unemployment benefit payments accrue to households with income between 20 and 30 minimum wages, against 5.5 percent in the case of households with income between 3 and 5 minimum wages. In the case of former PIS/PASEP and FGTS accounts, 47 percent of withdrawals are made by households with income over 30 minimum wages, against 2.3 percent in the case of households with income between 2 and 5 minimum wages. Using PPV data, Barros, Corseuil, and Foguel (2000) show that only 32 percent of unemployment benefit recipients are poor. The World Bank (2000a) also provides evidence that the unemployment benefit is not pro-poor, with the exception of the Northeast, where its impact is primarily on urban workers in the second consumption quintile.


In the case of the salary bonus, households earning up to 2 minimum wages receive 21 percent of outlays, against 27 percent for those earning more that 30 minimum wages. The high concentration of recipients among high-income households can be attributed to the fact that POF data do not allow for distinguishing current salary bonus receipts from remuneration on former PIS/PASEP account balances. Using PPV data, Paes de Barros, Corseuil, and Foguel (2000) also show that the salary bonus is poorly targeted, with only 36 percent of recipients among the poor.


In recent years, the number of smaller enterprises participating in the program has increased. In 1995, 37 percent of the workers in the formal sector benefited from the program. Most beneficiaries (nearly 5.5 million in 1996) are in the Southeast. However, the average number of beneficiaries per firm fell from nearly 200 in 1990 to less than 130 in 1996, reflecting the rise in the number of smaller enterprises in the program and the increase in informality in the labor market. Also see Lobato, Aquino, and Ribeiro (1999).


A recent survey of low-income communities in the municipality of Rio de Janeiro shows that unemployment in these communities is over three times higher than the average for the entire metropolitan region of Rio de Janeiro (18.5 percent, against 5.4 percent for the metropolitan region in 1998). Youth unemployment has also been shown to be higher in these communities, and employment is concentrated in the informal sector.


In 1997, investment financed by FGTS totaled approximately R$2 billion and created 126 thousand jobs. Annual investments through FGTS averaged 0.65 percent of GDP in the period 1990–98 (Carvalho and Pinheiro, 1999).


Employers are also required to give a month’s notice and grant two hours per day in the month prior to dismissal, with no reduction in compensation for the worker to look for another job. See Amadeo and Camargo (1996), and Gonzaga (1998), for more information.


Carneiro (2000) shows that increases in the minimum wage are associated with higher nominal wages in both the formal and the informal sectors. This can be attributed, at least in part, to high mobility in and out of informality. This empirical finding is confirmed by anecdotal evidence of informal indexation of compensation in the informal sector to nominal adjustments in the minimum wage.


Although FGTS is an important source of finance, most spending on housing is financed outside FGTS. Of the approximately 32 million housing units build in the last 30 years, only 5.6 million units have been financed through FGTS (Gonçalves, 1998). In the absence of a market for long-term housing financing in the private sector, most funding for housing/sanitation projects is provided by public institutions, particularly CEF.


In the absence of benchmarks for the long-term user cost of capital in Brazil, it is difficult to calculate the implicit subsidies in housing loans. Rates are lower on FGTS loans for housing projects for low-income groups (with household income up to 3 minimum wages) than for higher-income households (with income between 4 and 12 minimum wages). CEF loans outside FGTS, typically for households with income above 12 minimum wages, have higher rates but even these tend to be subsidized.


Access to safe water ranges between 36 percent of households in the lowest income decile to 96 percent in the highest decile. In the case of sanitation, access rates vary between 11 percent for households in the lowest income quintile and 84 percent in the highest quintile. The incidence of public spending on these services is poor with 12 percent of water services accruing to the lowest quintile and 26 percent to the highest quintile. Incidence differentials have been shown to be higher for sanitation, where only 4 percent of public spending reaches the lowest quintile relative to 32 percent for the highest quintile. For more information, see von Amsberg (2000). According to PNAD-95, nearly 80 percent of rural households and 27 percent of urban households lack adequate sanitation. See World Bank (2000a), for more information.


The housing deficit, measured as the share of inadequate and informal housing units and/or household co-habitation in the total housing stock, is concentrated among the poor: nearly 87 percent of the housing deficit is concentrated among households earning up to 3 minimum wages. It is estimated that the housing deficit in Brazil is approximately 14 percent of the housing stock, or approximately 5.4 million units in 1996. This share has fallen since 1981, when the deficit was estimated at 17 percent. See Gonçalves (1998), Santos (1999), SEPURB (1996, 1998), and World Bank (2000b) for more information.


Housing programs for low-income households include Habitar-Brasil, funded by the federal budget, and Pro-Moradia, funded by FGTS. According to SEPURB (1998), public outlays on these two programs totaled R$2 billion between 1995 and 1998. The programs benefited 723,000 households, and created 205,000 jobs in 3,176 municipalities.


Through housing credits, applicants borrow directly from CEF to purchase/build new housing, as long as they are not home owners, and that monthly loan repayment does not exceed 30 percent of household income.


Among Latin American countries, Brazil’s score is comparable to that of Peru (0.74) but is lower than those of Chile (0.84), Argentina (0.83), and Uruguay (0.83), for instance. See the UNDP’s 1999 World Development Report, for more information. The 1999 scores are not comparable over time because of a methodological change in 1997. According to the old methodology, Brazil scored 0.83 in 1996, against 0.73 in 1980. In 1996, the scores of more developed states such as Rio Grande do Sul and São Paulo (0.87 each) are comparable with the industrial country average (0.91). Less developed states such as Maranhão (0.55) and Piauí (0.53) fare poorly with HDI scores below the developing country average (0.58).


These states are: Acre, Alagoas, Bahia, Ceará, Maranhão, Pará, Paraíba, Pernambuco, Piauí, Rio Grande do Norte, Rondônia, Roraima, Sergipe, Tocantins. Approximately 31 million people, living in 186 micro regions and 1,797 municipalities, will benefit from the program in these states.


The additional states are Amazonas, Amapá, Espírito Santo, Goiás, Minas Gerais, Mato Grosso, Paraná, Rio Grande do Sul and São Paulo. These states have HDI scores above the national median but, nevertheless, contain 81 micro regions and 389 municipalities with HDI scores below the national median. Extension of the IDH-14 program to these micro regions will benefit an additional 5 million people.


The subnational share in financing may be underestimated in the case of health care because of the recent institutional changes requiring the states and municipalities to earmark 12 percent and 15 percent, respectively, of their revenues (net of intergovernmental transfers) to finance outlays on health care.


Because income certification is often difficult, particularly for the poor in the informal sector, categorical targeting, or proxy targeting, involves the use personal and/or household characteristics that are associated with poverty. In recent years, social assistance programs in Latin America have focused on household characteristics for the purpose of categorical targeting when accurate income-testing is difficult or costly. See Souza (2000), for more information.