Information about Western Hemisphere Hemisferio Occidental
Journal Issue
Share
Article

Brazil: Selected Issues and Statistical Appendix

Author(s):
International Monetary Fund
Published Date:
January 2001
Share
  • ShareShare
Information about Western Hemisphere Hemisferio Occidental
Show Summary Details

IV. Social Spending in Brazil: Education and Health Care1

A. Introduction

1. Brazil spends approximately 20 percent of GDP on social programs. Nevertheless, increases in social spending over time have not led to commensurate improvements in social indicators, particularly in the areas of education and health care. This can be attributed, at least in part, to imbalances in the composition of public social spending, deficiencies in service delivery, and poor targeting. Although many programs are well-targeted and reach the poor, others have a relatively regressive impact on income distribution.

2. Social spending in Brazil has to be assessed in the context of the country’s highly skewed income distribution. The 10 percent richest households own nearly half of national income, while the 50 percent poorest own just above 10 percent of the nation’s income. Based on 1996 data, nearly 23 percent of the population (approximately 35 million people) live below the poverty line and own less than 3 percent of national income. Nevertheless, the income gap—the income shortfall below the poverty line—is only 1.6 percent of GDP.2 Consequently, well-targeted publicly-funded social programs could in principle be used to reduce the incidence and depth of poverty at a relatively low cost to the budget.

3. Against the background of a stringent fiscal adjustment since late-1998, recent pressures have emerged to relax the government’s stance on social spending while, at the same time, improving the quality of publicly-funded social programs. These pressures have highlighted the need for a more indepth analysis of public spending on social programs and performance indicators in the formulation of policy objectives in the social area. This chapter sheds some light on the possible directions for reform over the medium term in order to improve the efficiency and effectiveness of public spending on social programs.

4. This section is organized as follows. Subection B presents the main trends in social spending in the period 1995–99. Subsections C and D deal with, respectively, education and health care. Conclusions are presented in Subsection E.

B. Public Social Spending: An Overview

Background

5. Brazil spends a large share of GDP on publicly-funded social programs. Total social spending of the consolidated general government amounts to approximately 20 percent of GDP (Table 4.1).3 Most social spending is financed by the federal government (nearly 60 percent of the total in 1996). Most subnational spending is on education and health care (approximately 5 percent of GDP in 1996). Federal social spending is skewed towards transfer payments, particularly pensions and other social security benefits (Table 4.2). This share has risen in recent years, thereby putting pressure on other social programs. States and municipalities already spend more on social security than on health care, or housing, urbanization, and sanitation.

Table 4.1.Consolidated Social Spending by Function, 1995-96(In billions of reais)
19951996
Federal

government
States and

municipalities
TotalFederal governmentStates and municipalitiesTotal
Social insurance and assistance56.415.071.368.719.287.8
Social security 1/52.413.065.463.516.680.1
Labor3.00.13.13.80.24.0
Social assistance1.01.92.91.32.43.7
Education, culture, and science7.421.028.47.324.932.2
Health and nutrition14.58.122.613.811.625.3
Housing, urbanization, and sanitation0.67.68.21.79.811.6
Total78.951.7130.591.565.4156.9
(In percent of GDP)
Social insurance and assistance8.72.311.08.82.511.3
Social security 1/8.12.010.18.22.110.3
Labor0.50.00.50.50.00.5
Social assistance0.20.30.5
Education, culture, and science1.13.24.40.93.24.1
Health and nutrition2.21.33.51.81.53.3
Housing, urbanization, and sanitation0.11.21.30.21.31.5
Total12.28.020.211.78.420.1
(In percent of total social spending)
Social insurance and assistance43.211.554.743.812.256.0
Social security 1/40.110.050.140.510.551.0
Labor2.30.02.32.40.12.6
Social assistance0.81.52.4
Education, culture, and science5.716.121.84.715.820.5
Health and nutrition11.16.217.38.87.416.1
Housing, urbanization, and sanitation0.55.86.31.16.37.4
Total60.439.6100.058.341.7100.0
Sources: IPEA; and IMF staff calculations.

Includes civil servants’ benefits, private sector pensions (RGPS), and public sector pensions (RJU).

Sources: IPEA; and IMF staff calculations.

Includes civil servants’ benefits, private sector pensions (RGPS), and public sector pensions (RJU).

Table 4.2.Federal Social Spending by Function, 1995-00(In billions of reais)
19951996199719981999Budget
20002001
Federal social spending79.993.5104.5118.3132.2140.0161.8
Health13.713.715.715.418.117.822.1
Education and culture8.48.79.713.715.517.121
Of which:
FUNDEF 1/0.00.00.14.15.27.99.3
Social assistance0.81.22.13.13.82.73.2
Of which:
LOAS 2/0.00.10.81.11.51.72.4
Social insurance50.261.167.576.284.788.9102.1
Private sector pensions (RGPS)32.941.746.152.158.262.973.2
Public sector pensions (RJU)3/15.417.419.721.924.123.726.1
Other1.92.01.72.22.42.32.8
Benefits to civil servants1.31.72.01.91.91.92.1
Housing and urbanization 4/0.10.30.40.30.32.21.2
Labor5.46.87.17.77.99.410.1
Of which:
Unemployment Insurance3.33.84.04.64.54.95.1
Memorandum items:
Federal social spending
In percent of GDP12.412.011.912.913.812.913.1
In percent of total government spendin62.263.662.964.665.963.763.7
Sources: Ministry of Planning and Budget; and IMF staff calculations.

Federal government transfers to FUNDEF (Fundo Nacional de Desenvolvimento da Educacao e Valorizacao do Magisterio).

Lei Organica da Assistencia Social.

Refers to the federal government only.

Includes R$1.4 billion relative to FCVS outlays in 2000.

Sources: Ministry of Planning and Budget; and IMF staff calculations.

Federal government transfers to FUNDEF (Fundo Nacional de Desenvolvimento da Educacao e Valorizacao do Magisterio).

Lei Organica da Assistencia Social.

Refers to the federal government only.

Includes R$1.4 billion relative to FCVS outlays in 2000.

Social spending and performance indicators

6. Efficiency in the provision of social services can be assessed for a country’s social indicators and level of public spending on social programs.4 This is confirmed by the results of the more formal efficiency analysis presented in Appendix I. Brazil fares poorly compared with other Latin American countries in key social indicators, including health care and particularly education (Table 4.3), despite the country’s relatively high ratio of total social spending to GDP.5

Table 4.3.Public Spending on Health and Education and Indicators: Brazil and Latin America, 1985-97(In units as indicated)
Initial levelMost recent level 1/Annual percent changeSample
BrazilLatin

America
BrazilLatin

America
BrazilLatin

America
size
Indicators
Health
Immunization, DPT (% of children under 12 months)65.064.479.087.41.63.234
Immunization, measles (% of children under 12 months)67.061.999.088.03.33.634
Births attended by health staff (% of total)52.866.83.213
Safe water (% of population with access)75.073.469.078.6-0.80.930
Sanitation (% of population with access)24.065.267.069.710.81.027
Mortality rate, under-5 (per 1,000 live births)62.045.244.034.3-3.4-3.129
Mortality rate, infant (per 1,000 live births)55.036.534.025.0-4.7-3.538
Education
Illiteracy rate, adult total (% of people aged 15 and above)21.717.316.112.9-2.5-3.227
Persistence to grade 5, total (% of cohort)69.979.81.913
School enrollment, primary (% gross)99.6103.1122.8106.41.80.323
School enrollment, primary (% net)81.286.389.790.31.10.519
School enrollment, secondary (% gross)35.452.945.158.02.71.123
School enrollment, secondary (% net)14.343.119.547.43.52.116
Public spending2/
Health
In percent of GDP1.62.31.92.50.30.129
In percent of total government expenditures10.08.310.19.80.01.629
Education
In percent of GDP0.83.61.34.10.50.629
In percent of total government expenditures5.513.86.616.81.03.029
Sources: World Bank, World Development Indicators; and IMF staff calculations.

Refers to 1997 for most countries.

Refers to central government only for all countries, except Argentina (general government); and Peru, St. Kitts and Nevis, and Bolivia (public sector).

Sources: World Bank, World Development Indicators; and IMF staff calculations.

Refers to 1997 for most countries.

Refers to central government only for all countries, except Argentina (general government); and Peru, St. Kitts and Nevis, and Bolivia (public sector).

7. Government spending may be weakly correlated with performance indicators due to poor incidence of social programs and limited access of the poor to social services.6 In Brazil, empirical analysis of the incidence of social spending is in its infancy because of, at least in part, data deficiencies (Box 4.1). Most social spending is untargeted, particularly health care and education. In the case of targeted programs, means-testing is often difficult because the poor typically work in the informal sector, where income certification is inadequate. Subnational governments are important providers of social services, but have few targeted social programs. Some programs are relatively well-targeted (for instance, maternity and disability benefits, daycare, kindergarten, and primary education, among others). Recent studies have suggested that, surprisingly, some untargeted social spending has a stronger positive redistributive impact on household income than some targeted social assistance programs.7 Better incidence of social spending is crucial for poverty reduction because output growth alone has a relatively low impact on poverty in Brazil.8

Box 4.1.Social Spending Data in Brazil: Strengths and Weaknesses

Data on federal government spending on social programs are available from SIAFI (Sistema de Acompanhamento Financeiro) and SIDOR (Sistema Integrado de Dados Orçamentários). For states and municipalities, data are available from SIAFEM (Sistema de Acompanhamento das Finanças de Estados e Municípios). Information on social spending is also available from the IBGE, based on national accounts data.

As in other countries, broad data coverage is typically achieved at the expense of data quality and reliability. Data are available, but infrequently consolidated, for the three levels of government (federal government, states, and municipalities), and for direct expenditures (administração direta), and spending carried out by autonomous agencies (fundações and autarquías). Data collected by the National Treasury (STN) on subnational finances through SIAFEM do not cover spending by autonomous agencies. This is an important limitation because a significant share of social spending is carried out by these agencies. However, national accounts data collected by the IBGE, through DECNA (Departamento de Contas Nacionais), cover disaggregated state and local government spending and are consistent with SIAFI/SIDOR data at the federal government level. Subnational government data on social spending are of poorer quality and typically reported at a higher level of aggregation for most social programs, particularly health care and education. States and municipalities often misreport the functional classification of social spending.

International data comparability has improved in many areas, particularly education. A database (SIGPE, Sistema de Informações sobre os Gastos Públicos da Área de Educação) of education indicators, including public sector spending, was created by IPEA (Instituto de Pesquisa Econômica Aplicada, Ministry of Planning and Budget). Annual data on education spending and indicators, including information on students, teachers and curricula, are available from the School Censuses (Censo Escolar). Also, Brazil has reported education spending following the OECD classification since 1998. The OECD classification excludes culture and sports from education outlays, and includes fringe benefits and pension payments. These fringe benefits are recorded separately for the social sector as a whole in Brazilian accounts. See Abrahão and Fernandes (1999), for more information).

Information on access to social services, as well as on private outlays on education and health care, is available from household expenditure surveys. PNAD (Pesquisa Nacional por Amostra de Domicílios) is the main household survey conducted in Brazil. It has nationwide coverage, has been conducted annually since 1967 (with few exceptions), and reports data on a wide range of household characteristics including employment status, income sources, and education attainment. It does not however provide detailed information on access to social services on a regular basis, and typically does not distinguish between social insurance and social assistance benefits. The PPV (Pesquisa sobre Padrões de Vida), conducted in 1996 and 1997, provides information on access to social services but has a relatively small sample: 5,000 households in the Northeast and Southeast. PPV is methodologically similar to the Living Standard Measurement Surveys (LSMS) supported by the World Bank in many countries. The PCV (Pesquisa de Condi÷ões de Vida) provides very detailed information on household income sources, including access to social services, social security earnings and benefits, and private spending on health and education. Its first wave (1990) covered 5,500 households in the metropolitan region of São Paulo, while the second wave (1994) covered 12,000 households in the state of São Paulo, including those covered in capital’s metropolitan region in the first wave. Other household surveys widely used in the analysis of social spending and indicators include POF and PME. POF (Pesquisa de Or÷amentos Familiares) is a national household income survey conducted in 19S7 and 1996. POF contains detailed information on labor compensation, including severance pay, unemployment insurance, and receipt of labor-related benefits such as the salary bonus. High-frequency data are available from PME (Pesquisa Mensal de Emprego), which provides monthly information on employment and earnings for households in six major metropolitan areas. Given the survey’s focus on employment and earnings, PME does not provide information on most transfer payments, such as pensions and social assistance benefits. RAIS (Relação Annual de Informações Sociais) also provides useful information on transfer payments to households.

8. Minimum wage policy is an important determinant of social spending. Most transfers to individuals and households (income support, disability benefits, and pensions, among others) are linked to the minimum wage and constitute a sizeable share of total social spending. Recent studies have shown that real increases in the minimum wage alleviate poverty, but they also inflate the cost of social assistance and insurance programs.9 This tradeoff affects the composition of social spending across government levels because most minimum wage-linked social assistance and insurance programs are financed by the federal government. With the rising share of social security in subnational social spending, real increases in the minimum wage are likely to put additional pressure on the already high payroll costs facing states and municipalities.

9. There has been increased rigidity in financing for social programs. Revenue-sharing arrangements are known to have reduced flexibility in fiscal policymaking in Brazil (Ter-Minassian, 1997). The federal government has also relied increasingly on taxes that are not shared with states and municipalities, such as social security taxes and contributions, to finance social spending. These are typically levied on enterprise payroll and earnings and are known to have a detrimental impact on formal employment and international competitiveness, among others. More recently, new legislation was passed: (1) introducing a floor for total federal spending on health care programs; and (2) earmarking spending on health care at the subnational level, as in the case of education (to be discussed below). States and municipalities are now required to earmark 12 percent and 15 percent, respectively, of their revenues (net of intergovernmental transfers) to finance outlays on health care.

C. Education

Background

10. Publicly-provided services comprise a wide range of programs, including formal and vocational education, as well as adult and special education. Access to these services is universal. Brazil has a long tradition in decentralized provision of education services. Federal government spending covers primarily higher education and vocational training programs (Table 4.4). The 1988 Constitution decentralized most spending assignments to states and municipalities, but little emphasis was placed on granting policymaking autonomy to subnational governments in program design, service delivery, and resource management. It is estimated that private outlays amount to nearly 25 percent of public spending on education.

Table 4.4.Public Spending on Education by Government Level, 1995-97(In billions of reais)
TotalFederal governmentStatesMunicipalities
19951997199519971995199719951997
Education outlays34.544.510.512.514.919.99.112.1
Pre-school1.52.00.10.30.10.11.41.6
Primary education10.219.42.62.96.28.51.57.9
Secondary education2.43.30.51.71.01.50.80.1
Tertiary education6.87.34.34.62.42.60.10.0
Others 1/13.712.63.12.95.37.25.32.5
Sources: Ministry of Education; and IMF staff calculations.

Includes teachers’ pensions and social security benefits.

Sources: Ministry of Education; and IMF staff calculations.

Includes teachers’ pensions and social security benefits.

11. Education spending also comprises other programs dealing with the procurement and distribution of school lunches (Merenda Escolar program)10 and textbooks, transport for students in poor rural areas, and health care services in schools in poor municipalities. These programs involve joint ventures within the Comunidade Solidaria program.11 Other programs aim at improving students’ access to information technology products in poor municipalities and the provision of education services by television, including those for upgrading teacher skills (IPEA, 2000).

Trends in expenditures and performance indicators

12. The sectoral composition of spending on education is skewed towards higher education. Although nearly one-half of public spending is on primary education, higher education spending per student is almost 16 times as high as outlays on primary education (R$7,321 against R$460 in 1995). In 1998, the average spending per student in primary education was R$565. There are significant variations in teachers’ compensation, reflecting differentials in the pay scale between entry and top-level salaries, between state and municipal schools, and among the states. Typically, state school teachers are better paid than their municipal counterparts. The ratios of students to teacher are also lower in higher education. Indicators are worse in the poorer states of the North and Northeast (Table 4.5), reflecting, to a great extent, discrepancies in teachers’ compensation and skill mix, among others. The implementation of FUNDEF in 1998 (Box 4.2) has ensured better equalization of education spending across and within the states but increased rigidity in intergovernmental fiscal relations.12

Table 4.5.Education Indicators by State(In units as indicated)
Registered students 1/Primary education 2/Secondary education 2/Age-grade gap 3/Enrollment rates 4/
TotalFederalStateMunicipalPrivateCompletion rateLengthCompletion rateLengthprimary schoolSecondary school
schools(percent)(years)(percent)(years)GrossNetGrossNet
Rondônia317,816177,664118,38721,76540.110.953.03.846.8126.990.546.719.5
Acre144,28429790,81546,2846,88830.912.075.33.957.8138.689.855.918.3
Amazonas653,857592341,557274,07737,63148.611.372.23.764.6127.788.152.314.9
Roraima79,2777373,0564,9571,19135.410.071.13.945.4133.692.684.225.1
Pará1,614,7433,848561,310986,27563,31029.412.680.14.764.0130.991.546.112.9
Amapá127,14094,07125,9197,15040.511.283.44.148.1141.391.388.722.8
Tocantins356,149216,825129,25510,06938.011.672.74.064.4165.692.771.316.6
Maranhão1,634,218943418,2131,126,94388,11943.111.481.13.765.2148.588.046.917.0
Piaui781,240116264,259455,72061,14540.112.484.14.363.3133.391.135.710.7
Ccará1,868,119550477,9601,180,636208,97360.710.573.13.760.4140.089.846.917.0
R. G. do Norte656,199180287,524303,73364,76263.211.584.93.756.4146.391.558.119.3
Paraiba896,022332,893481,83981,29047.511.875.53.866.5145.291.940.313.8
Pernambuco1,817,7631,287668,962925,795221,71953.011.777.93.958.6146.591.258.419.4
Alagoas701,64329175,308467,31158,99544.913.182.13.967.8132.486.334.711.5
Sergipe434,225265202,648195,18736,12552.112.180.43.867.5135.590.147.212.5
Bahia3,702,7277421,291,4512,209,254201,28050.211.582.23.868.4161.491.945.212.3
Minas Gerais3,773,2473,1082,062,6931,505,666201,78073.29.378.83.542.2153.697.467.729.2
Espirito Santo614,779310,383234,69969,69760.49.984.03.636.0133.294.581.539.5
Rio de Janeiro2,474,64911,583676,9801,303,228482,85869.810.178.13.841.6131.296.780.036.8
São Paulo6,325,2942074,052,9721,511,184760,93171.48.777.93.426.4126.598.295.351.9
Paraná1,732,395433813,596786,423131,94363.89.974.83.728.4124.297.083.944.0
Santa Catarina981,603615538,634366,75475,60071.99.670.53.726.9125.396.874.544.1
R. G. do Sul1,758,3761,148906,816695,197155,21566.49.872.73.822.5123.195.779.146.0
M. G. do Sul460,031577229,454190,20839,79245.710.866.33.842.4131.594.467.232.2
Mato Grosso604,741315,572250,82738,34241.410.466.73.744.1135.493.757.426.5
Goiás1,140,089485666,626388,61184,36747.611.177.23.851.7148.295.970.926.4
Federal District409,1161,493341,21366,41061.410.477.83.933.3138.497.9101.548.2
Brazil36,059,74228,57116,589,45516,164,3693,277,34765.810.478.53.646.6128.195.368.130.8
Sources: Ministry of Education; and IMF staff calculations.

1999.

1997.

Share of students with age-grade gap, 1998.

1998.

Sources: Ministry of Education; and IMF staff calculations.

1999.

1997.

Share of students with age-grade gap, 1998.

1998.

Education and social development

13. Education spending affects poverty primarily by improving the earnings possibilities of the poor. In Brazil, up to 50 percent of earnings inequality can be explained by differentials in education attainment (Paes de Barros and Mendonça, 1999), as the labor market pays a high premium for skilled workers.13 Quality differentials affect education attainment between the poor and the nonpoor. It has been shown that undereducated workers are more likely to fall into poverty than their more educated counterparts. Mobility in and out of informality in the labor market has also been shown to be lower among the less educated.

14. There are few studies on the incidence of education spending in Brazil, and most empirical evidence is mixed. While public spending on secondary and higher education has been shown to have a regressive impact on income distribution, public spending on preschool (daycare and kindergarten) and primary education tends to be more progressive.14 Overall, the incidence of public spending on education is poor given that public outlays on higher education, which tend to be more regressive, account for a large share of total federal expenditures on education. Private outlays on education are concentrated in the highest income quantile, relfecting differentials in the ability to pay for these services and the country’s skewed income distribution. The incidence of public spending on special programs, such as school lunches and daycare facilities, is in general better than that of spending on other education programs. This is because the special programs are better targeted to low-income households and poor municipalities.

Box 4.2.Spending on Education and the Implementation of FUNDEF

In 1998, a fund—FUNDEF (Fundo Nacional de Desenvolvimento do Ensino Fundamental e Valorização do Magistério)—was created to finance subnational spending on education. Accordingly, states and municipalities are required to earmark 15 percent of their own tax revenues and transfers to primary education. The key objectives of FUNDEF are: (1) to reduce shortfalls in financing at the subnational level; and (2) to ensure better equalization in expenditure capacity across and within states. A national minimum curriculum was also set through FUNDEF.

The 1988 Constitution requires subnational jurisdictions and the federal government to earmark, respectively, 25 percent and 18 percent of total revenues to education. However, these targets were not always met at the subnational level due to financing shortfalls, particularly in poorer states and municipalities. With the creation of FUNDEF, a floor was introduced for municipal outlays per student. More recently, some differentiation has been introduced for the spending floors per student between first/fourth grades and fifth/eighth grades. For 2001, the minimum spending levels will be raised to R$353 per student in first/fourth grades (up from R$333 in 2000), and R$370.65 per student in fifth/eighth grades (up from R$349.65 in 2000). There have been pressures to increase the differentiation rate and for creating different spending floors for rural and urban schools. An important innovation of FUNDEF is that the federal government is required to top up spending in the case where subnational jurisdictions cannot afford the minimum spending requirement. In 2001, the federal government transfer is expected to reach R$0.7 billion, virtually unchanged from 2000, given the expected increase in subnational revenues.

To reduce pay inequality across states and within the education sector, 60 percent of the resources spent on primary education are earmarked to wages and salaries, leaving 40 percent to finance capital outlays and operations and maintenance. In the first year of implementation of FUNDEF, teachers’ salaries rose by over 18 percent in municipal schools—where salaries are typically lower—and by nearly 8 percent in the schools run by state governments. These increases were higher in the poorer municipalities of the North and Northeast. FUNDEF also resulted in an increase in total revenues in poorer municipalities. These increased revenues are highly fungible and misuse of FUNDEF resources may occur in the absence of adequate oversight. There is anecdotal evidence of municipalities spending FUNDEF resources on infrastructure upgrading in schools to improve road paving for the entire street where the school is located. The favor of current spending over capital outlays within the education sector has helped to prevent leakages in funding and misuse of resources in investment projects with low social return. More recently, it has been argued that FUNDEF has crowded out some subnational spending on secondary education and that resources are needed to ensure that there will be no bottlenecks in the system in light of the increased coverage of the primary school system. To deal with these problems, a new program (PROMED) will be implemented to strengthen secondary education.

15. The increase in policymaking autonomy at the subnational level has had a positive impact on performance indicators. Recent research has shown that the states in which public schools are free to manage their budgets and appoint the headmaster, as well as to implement participatory management schemes, tend to have lower repetition and dropout rates, and their students tend to have better test scores (Box 4.3).15 It has been argued that better accountability has also been achieved by granting more managerial autonomy to schools.

16. Emphasis on improving the efficiency of basic education spending has yielded encouraging results. School attendance increased from 90 percent to over 95 percent of the population aged 7–14 years between 1994 and 1999. School enrollment rates increased by 41 percent in secondary education and by 25 percent in tertiary education between 1994 and 1998. Enrollment in graduate courses increased by 16 percent in the case of Master’s degrees and by 28 percent in the case of Doctorate programs. Nevertheless, dropout and repetition rates remain high. Between 1990 and 1997, the repetition rate fell to 34 percent in primary education and 22 percent in secondary education. Expected delay is 2.4 years for pupils in primary education. Expected stay in school is 10.4 years. Dropout rates in higher education fell little between 1990 and 1997, and the average stay rose from 7.2 months to 8.4 months.16

Box 4.3.Recent Trends in Education Policy

In an effort to improve the efficiency of public spending on education, a two-pronged strategy was followed in the 1990s. Emphasis was placed on ensuring access to, and improving service delivery in, primary education, as well as more stringent regulation and standard setting for secondary and tertiary education. The provision of primary education services has increasingly taken place in partnership with subnational governments and civil society. Given the discrepancies in the ability of state and municipal governments to finance education spending, as discussed above, greater coordination with the federal government has been pursued to ensure minimum provision and equalization of primary education spending capacity across and within states.

Subnational governments, particularly municipalities, have enjoyed greater autonomy in program design and implementation. More policymaking autonomy at the local level of government has been exercised with increased use of demand-driven, result-oriented, participatory administration in public schools. The implementation of participatory administration schemes took place primarily after 1995. These entities are involved in the school’s administrative, financial, and educational decision-making processes. The most common participatory administration schemes are the School Council, available in 1997 in over 37 percent of the nearly 42,000 schools funded by state governments, and the Parents and Teachers Association, available in nearly one-third of states schools. See Parente and Lück (1999), for more information. Schools are enjoying more autonomy in the organization of curricula and pedagogical projects (subject to minimum standards set by the federal government); personnel management; teaching planning and methods; and procurement.

In secondary education, there has been greater emphasis on improving access to formal education and vocational training. The reform of vocational education has been carried out in coordination with the Ministry of Labor. Plans involve the creation of community centers for vocational training in partnership with municipal governments, trade unions, and community associations. Also, standard setting has been pursued through the implementation of a national exam in 1998. Unlike the higher education case, to be discussed below, the secondary education exam is optional and aims at providing secondary school leavers with a degree, which may subsequently be used for entry in higher education institutions. As discussed in Box 4.2, a new program will also be implemented to strengthen secondary education (PROMED). Emphasis on standard setting and monitoring has been placed in the case of tertiary education. Since 1996, national exams have been extended to graduate students. These exams focus on the coverage of curricula and aim at assessing the quality of services provided, particularly by private institutions and those public institutions in remote, less developed regions, where services are of poorer quality. Higher education institutions are also recertified periodically on the basis of these exams.

D. Health Care

Background

17. Publicly-provided health care services in Brazil consist essentially of curative and preventive care, immunization, nutrition, and epidemiological and sanitary surveillance, among others. As in the case of education, access to publicly-provided services is universal. The national health care system (SUS) combines central government financing with decentralized service delivery: the federal government reimburses private health care providers and subnational governments, particularly municipalities, for the delivery of health care services and maintenance of public hospitals and clinics.17 Public spending on health care also comprises special programs designed and managed in coordination with subnational governments. Community care programs (PACS and PSF) have also focused on the decentralized provision of preventive care services in poor municipalities.18

Recent trends in spending and main issues

18. More policymaking autonomy has been granted to municipal governments with the implementation of SUS. Emphasis has been placed on increasing the system’s coverage and ensuring access of the population to publicly-funded health care. Nevertheless, spending through SUS has not ensured equalization and significant discrepancies persist in expenditure capacity across states (Table 4.6). Transfers from the federal government to states and municipalities are based primarily on the cost of the services provided at the municipal level, rather than needs, and past trends in state budget allocations. More prosperous states—where a wider range of more sophisticated, costly health care services is provided—receive more transfers on a per capita basis than poorer states. Better equalization has nevertheless been pursued in recent years.

Table 4.6.Health Indicators by State, 1996(In units as indicated)
StatesDoctors 1/Health spendingImmunization rate 2/Health care admittanceAccess to 4/Community Care Program 5/Infant mortality 6/
(per capita)DPT 3/PolioMeaslesIn-patientOut-patientDrinking waterSanitationNumber of householdsShare of pre-natalImmunization rateMalnutrition rate
Acre4.124.851.161.354.938,25825,39331.017.5
Amapá6.324.871.973.772.918,357277,78755.56.1
Amazonas6.225.657.985.071.2114,8441,276,55355.612.8
Parà6.823.191.653.265.6455,0132,360,86736.02.0
Rondônia3.925.777.173.882.6115,564769,26142.22.910,78270.074.59.3
Roraima5.424.289.283.488.29,995175,06552.95.2
Tocantins5.332.671.564.471.59,033665,11230.10.133,87775.074.610.8
Alagoas9.835.240.241.044.1222,3771,670,59046.98.5148,46253.962.219.183.0
Bahia7.129.651.455.856.71,006,6625,773,20045.516.6118,73262.968.211.850.2
Ceará7.039.393.3100.098.4530,1324,639,92339.87.6884,55973.073.040.9
Maranhão3.832.457.238.472.6417,4683,119,85224.77.8233,12459.266.011.963.2
Paraiba9.638.769.675.074.422,7271,985,11051.615.7335,47669.269.011.665.5
Pemambuco11.141.874.8100.087.2612,5225,079,58856.519.1408,49961.175.114.667.0
Piaui5.537.670.864.471.0227,7842,030,88324.71.7209,30959.973.714.549.1
Rio Grande do Norte9.232.772.966.376.8184,0211,551,38748.710.3143,65575.680.37.634.3
Sergipe9.232.581.087.183.7125,5171,504,09059.520.158,40559.062.210.553.0
Espírito Santo13.731.288.080.994.120,4771,064,93177.248.728.2
Minas Gerais12.644.966.469.871.31,339,51115,203,92378.156.831.2
Rio de Janeiro24.347.673.577.581.0890,3158,570,94791.572.827.1
São Paulo18.849.283.288.395.02,266,17122,618,98095.579.326.4
Paraná11.052.892.292.292.275,9819,619,92685.227.829.8
Rio Grande do Sul15.146.478.779.984.1805,5624,062,29385.144.422.2
Santa Catarina9.239.779.683.887.8412,1952,808,21789.138.524.7
Distrito Federal29.547.7100.0100.054.9137,3571,325,60586.275.527.6
Goiás10.241.981.582.479.6321,5952,644,76570.927.512,47986.390.37.026.3
Mato Grosso5.838.872.468.578.6194,0832,116,24958.613.428.5
Mato Grosso do Sul10.436.269.568.777.0155,1711,040,13476.09.029.6
Brazil13.041.875.177.879.911,932,654104,208,88175.444.040.0
Sources: Ministry of Health; and IMF staff calculations.

Per 10,000 population.

In percent.

Children less than 1 year of age.

In percent, 1991.

PACS/PFS.

Per 1,000 population. 1994.

Sources: Ministry of Health; and IMF staff calculations.

Per 10,000 population.

In percent.

Children less than 1 year of age.

In percent, 1991.

PACS/PFS.

Per 1,000 population. 1994.

19. Subnational governments have increased spending on health care. The fees paid for the health care services provided through SUS are in general low and often do not ensure full cost recovery. Due to shortfalls in financing, a number of states and municipalities have increased spending on health care using their own budgetary resources. It is estimated that states and municipalities spend approximately one-half of federal spending through SUS. States and municipalities are now required to earmark 12 percent and 15 percent, respectively, of their revenues (net of intergovernmental transfers) to finance outlays on health care. Also, participation of private health care providers in SUS has fallen slightly over time because of delays in reimbursement, and low fees paid for the services provided. It is estimated that private outlays on health care are nearly as large as public spending.

20. The incidence of health care spending is typically better than that of education. There is some evidence that spending on health care is fairly progressive, but private outlays are concentrated in the highest income quintile, as expected.19 It has been shown that the incidence of public spending is worse for the health care services provided by the private hospitals and clinics associated with SUS (hospitals conveniados) than for the services provided by public health care institutions.20 Information on the recipients of certain special programs, that would allow for a better assessment of the incidence of these outlays, is not readily available.21 However, because these special programs are more targeted to low income households and poor municipalities, it is expected that the incidence of public spending on these programs is better than that on the other, untargeted health care programs.

21. Recent initiatives have addressed some of the limitations of decentralized provision of health care. Local governments are often too small to reap the benefits of economies of scale in health care provision and cannot typically finance the provision of more specialized curative care services. To overcome these difficulties, intermunicipal administrative ventures (Consórcios Intermunicipais de Saúde) have been created within SUS. These ventures also perform functions such as personnel management, including hiring new staff, licensing private health care providers, and procurement. More recently, efforts have been made to strengthen the institutional framework within which these ventures are created (Ribeiro and Costa, 1999). According to the Ministry of Health, as of July 1999, there are 143 intermunicipal ventures in Brazil, covering a total of 1,740 municipalities.

22. More emphasis has been placed on improving performance in service delivery in recent years. Typically, an input-based funding system does not focus on performance targets to encourage efficiency in service delivery. It also distorts spending towards more expensive inpatient care. Specific programs have incorporated explicit targets for coverage and a progressive funding schedule for increased coverage. As a result, access to health care services has improved in poorer regions. A better match between the supply and demand for health care services has also been achieved through greater involvement of civil society in program design and implementation.

23. Equalization of expenditure capacity among the states has been pursued recently. Inequalities have been reduced over time in the transfer system. This has been achieved through increases in budget allocations for poorer states, where coverage has been extended with the implementation of SUS. Funding for basic health care programs and preventive care has also increased (Ministry of Health, 2000). A minimum transfer of R$10 per capita was implemented within PAB (Piso de Atenção Básica) and these resources are transferred to the municipalities to finance the provision of basic health care services, including prenatal care, oral hygiene and immunization. Total allocations for PAB are programmed at R$1.8 billion in 2001.

E. Conclusions

24. Preliminary information based on the most recent national household survey (PNAD-1999), released in late July 2000, has revealed important improvements in health and education indicators in Brazil. These indicators have improved against the background of economic slowdown, contraction in real wages, and rising unemployment following the depreciation of the real in January 1999. However, Brazil’s performance in a number of social indicators, particularly health and education, is not commensurate with the country’s income level and share of social spending in GDP. This is due primarily to inefficiencies in program design and implementation. In particular, the composition of total social outlays is skewed towards social insurance, where coverage is limited and most pensions and benefits are typically poorly targeted. Also, the share of social insurance in total social spending has risen in recent years, thus crowding out expenditures on other social programs. Adequate means-testing is carried out in a relatively small number of social programs, but relatively untargeted outlays still account for a large share of spending within each sector, particularly in health care and education. To overcome these deficiencies, recent initiatives in health care and education have combined increased emphasis on decentralized service delivery and managerial autonomy.

25. Education has been the hallmark of Brazil’s social policies, where success in rationalizing spending has been validated by impressive improvements in social indicators. Efforts have been made to equalize spending capacity across and within states, given the disparities in revenue mobilization capacity at the subnational level, to strengthen minimum standards, to involve civil society in program design and implementation, and to introduce goal-oriented management and incentive schemes for funding. Nevertheless, access to basic education remains limited in certain poorer areas and the incidence of public spending could be improved in secondary and tertiary education. It has been argued that better targeting could be achieved without abandoning universal access to services by, for instance, introducing cost recovery for higher education, which benefits the nonpoor disproportionately more. This would release resources in the budget for financing primary and secondary education programs that benefit the poor, and increasing coverage in poorer areas.

26. In health care, recent efforts have been made to improve access and funding mechanisms for service delivery. These efforts are noteworthy but protection of spending levels will be achieved at the expense of flexibility in budgeting, due to increased earmarking of revenues, particularly at the subnational level. The incidence of health care spending differs between public and private health care facilities. This suggests the need for more careful analysis of differentiated access to health care services depending on the service provider and reallocating funding towards providers and services that have a more beneficial impact on the poor.

27. With the consolidation of fiscal adjustment, emphasis in social policymaking in Brazil, particularly in the health care and education sectors, could be shifted from preserving public spending from further retrenchment to a more indepth assessment of the efficiency and effectiveness of public outlays and the adequacy of the existing programs to alleviate poverty and foster social development. Despite the caveats of the efficiency analysis presented in this section, the results reported suggest that much remains to be done in ensuring that increases in social spending translate into significant improvements in social indicators. Given Brazil’s significant interregional and interpersonal inequalities, it is important that the federal government remain at the forefront of social policymaking.

APPENDIX I Measuring Efficiency in Education and Health

Background

28. Brazil’s performance on education and health indicators compares poorly with other Latin American countries, as discussed above. The combination of relatively poor performance indicators and high total public spending suggests inefficiencies in program design and service delivery. To estimate the effectiveness of government spending on health care and education in Brazil, an efficiency frontier can be constructed using the methodology described in Box 4.4. The analysis is carried out for a sample of Latin American and Caribbean countries. Three output indicators are used in the analysis in the case of education (net primary school enrollment, net secondary school enrollment, and persistence to grade 5) and two output indicators are used in the case of health care (immunization rates for DPT and measles). One input indicator (public spending) is used to construct the efficiency frontiers for both health care and education.22

The results

29. Compared with other Latin American and Caribbean countries, Brazil is relatively efficient in the provision of education services (Appendix Table 4.1).23 In health care, Brazil scores highly in output efficiency and lies on the efficiency frontier for immunization. Nevertheless, in terms of input efficiency, Brazil ranks 21 in the sample of 29 under examination, with efficiency score equal to 0.18. This implies that, other things equal, the same, or higher, level of output can be achieved with only 18 percent of the public funds spent. The results show that countries with higher spending levels are relatively less efficient than countries that yield comparable output with less input. Countries that have education spending patterns skewed towards teachers’ compensation, for instance, tend to be less efficient. Likewise, countries that spend proportionally more resources on expensive curative care programs relative to preventive care will also tend to fare poorly in terms on efficiency scores.

Appendix Table 4.1Efficiency Analysis Scores in Education and Health(Brazil and Latin America)
Net primary school enrollment 1/Net secondary school enrollment 2/Persistency to Grade 53/Immunization 4/
Input efficiencyOutput efficiencyInput efficiencyOutput efficiencyInput efficiencyOutput efficiencyInput efficiencyOutput efficiency
ScoreRankDominatesScoreRankDominatesScoreRankDominatesScoreRankDominatesScoreRankDominatesScoreRankDominatesScoreRankDominatesScoreRankDominates
Argentina0.082900.99110
Bahamas0.95540.97751.00451.00360.112600.94190
Barbados0.182000.771900.331210.97131
Belize0.71710.98610.631600.411500.251600.701300.172240.99104
Bolivia0.102700.9990
Brazil1.00491.00491.00501.00401.00341.00340.182101.0070
Chile0.381540.911040.97680.92761.00281.00280.132320.95182
Colombia0.371620.871630.94750.781050.66830.731130.44940.90234
Costa Rica0.461320.891420.661420.491320.461220.88721.005171.00617
Dominica0.91611.0051
Dominican Republic0.59930.901131.00321.00220.261710.82281
Ecuador1.003101.003110.74540.87850.431110.83271
El Salvador0.541120.841720.92810.94510.96450.781051.004141.00414
Grenada0.421010.96160
Guatemala0.72600.701400.301410.86261
Guyana0.331820.881521.00250.761150.132400.89250
Honduras0.74610.9387130.92600.421310.611510.69820.96172
Jamaica0.521210.99510.551700.741200.112500.91210
Mexico1.00291.002100.891050.80950.62930.86930.092800.98120
Nicaragua0.251900.762000.651500.311700.251500.541600.321300.97140
Panama0.251600.96150
Paraguay0.571030.901230.821220.441420.561120.711220.291500.90220
Peru0.62850.901350.89950.83861.003131.00313
St. kitts1.002101.00210
St. Lucia0.361400.85500.81750.9985
St. Vincent1.00131.0012
Trinidad and Tobago0.391460.92951.00191.00190.70780.98480.181930.93202
Uruguay1.001121.001121.001121.001120.182000.90240
Venezuela0.341720.831820.871100.351600.611040.89640.241800.69290
Memorandum Item:
Brazil 5/0.561220.882130.561700.561700.561220.711320.102601.0070
Sources; IMF staff calculations.

Input: public sector spending on education in percent of GDP. Output net primary school enrollment.

Input: public sector spending on education in percent of GDP. Output net secondary school enrollment.

Input: public sector spending on education in percent of GDP. Output, persistency to grade 5.

Input: public spending on health in percent of GDP Output, immunization rates for DPT and measles.

Efficiency analysis based on total, rather than only central government, spending on health care and education.

Sources; IMF staff calculations.

Input: public sector spending on education in percent of GDP. Output net primary school enrollment.

Input: public sector spending on education in percent of GDP. Output net secondary school enrollment.

Input: public sector spending on education in percent of GDP. Output, persistency to grade 5.

Input: public spending on health in percent of GDP Output, immunization rates for DPT and measles.

Efficiency analysis based on total, rather than only central government, spending on health care and education.

30. The results of the empirical analysis should be assessed with some caution for three main reasons. First, public spending data on health care and education exclude subnational outlays. This underestimation of total public outlays overstates efficiency in the provision of education and health care and the upward bias is likely to be greater in countries where subnational governments are important providers of health care and education services. In the case of Brazil, relative efficiency is more likely to be overstated in the case of education than health care because subnational spending accounts for a lower share of total government spending on health care. Second, performance in social indicators may be affected by factors other than (public and private) spending. For instance, social development is likely to be correlated with variables such as income levels, poverty incidence, and lagged spending levels, among others. Unfortunately, these explanatory variables cannot be taken into account in these nonparametric models. Finally, spending data for all countries in the sample, including Brazil, exclude private outlays on health care and education. The exclusion of private outlays underestimates the use of inputs in the provision of health care and therefore overestimates the efficiency of government spending. This upward bias is obviously greater the higher the share of private outlays in total spending.

31. To test the sensitivity of the efficiency scores to the exclusion of subnational governments in Brazil, the empirical analysis was carried out using total public, rather than only central government, spending in health care and education. Information on total spending for other countries in the sample is not available, but, as suggested above, the upward bias in efficiency is likely to be high in Brazil, given that most spending on health care and education is subnational, particularly in the case of education.24 Brazil remains on the efficiency frontier in the case of immunization but has a lower input efficiency score. The fall in the efficiency scores is worse in the case of education, as expected, particularly when the net secondary school enrollment rate is used as the output indicator.

32. The sensitivity of the efficiency analysis was further tested by using all three output indicators in education jointly using the FDH methodology, as in the case of health care. In this case, the sample size is reduced to 12 countries. Brazil remains on the efficiency frontier when central government spending is used as the input indicator. When total, rather than only central government, spending is used as the input indicator, Brazil’s rank in the sample falls to the ninth position based on the input efficiency score (0.90) and to the eleventh position based on the output efficiency score (0.88).

Box 4.4.Measuring Efficiency in Social Spending: FDH Analysis

The efficiency of public spending can be measured in different ways. Regression analysis offers insights into how efficiently governments provide social services, after controlling for other determinants of social development. However, the elasticities calculated using standard regression analysis suffer from a number of limitations, including the sensitivity of parameter estimates to the functional specification of the reduced-form equations to be estimated. Also, most models from which reduced-from equations are derived are based on assumptions (on utility maximizing behavior, for instance) that are not easily applicable to public goods.

Alternative, nonparametric methods have been developed in recent years to measure efficiency in the provision of public goods and services (Tulkens and Van den Eeckaut, 1995). These methods consist of defining an efficiency frontier for the provision of social services treating public spending as an input in a social production function. Outputs are conventionally proxied by social indicators, such as school enrollment rates, illiteracy rates, life expectancy, among others. By using information on both inputs and outputs, the production frontier defines best practices for the production/provision of social outputs and the use of inputs in the set of producers under examination. The tradeoffs in the choice of inputs and outputs is well documented in the literature (Harbison and Hanushek, 1992; Jimenez and Lockheed, 1995). Unlike standard regression analysis, the calculation of these nonparametric efficiency frontiers does not depend on the assumptions used in the theoretical model or the functional specification of the social production function.

A widely-used nonparametric method is Free Disposal Hull analysis (FDH). In this case, a producer is efficient in the provision of public goods and services if its combination of outputs and inputs lies near the efficiency frontier constructed for the sample of producers. The analysis allows for the ranking of producers according to their efficiency scores, The only assumption made is that inputs and outputs be freely disposed of; in other words, it is possible with the same production technology to lower outputs while maintaining the same level of inputs, and increasing inputs while maintaining the same level of output.

FDH analysis shows that a producer (government, for instance) is relatively inefficient in the provision of, say, education services if another producer uses less input (public spending) to generate as much or more output (education indicator). The degree of efficiency is determined as follows. First, the relatively efficient production results are identified for the sample of countries under examination, based on their public spending levels and output indicators in education and health care. Second, an efficiency score is calculated as the distance of individual production results to the production frontier (FDH). This distance can be calculated from the point of view of inputs and outputs. The input efficiency score is the ratio of inputs used by a given producer A to the inputs used by producer B. This efficiency score indicates the excess use of inputs by the inefficient producer and therefore the extent to which resources are used inefficiently. By the same token, the output efficiency score is the ratio of producer A’s output to that of producer B. This ratio indicates the loss of output relative to the most efficient producer with equal or lower level of inputs. Finally, the producers in the sample are ranked according to their input and output scores. Alternatively, a producer is found to be dominated by other producers that achieve a higher level of output using the same, or lower, level of input (output efficiency); or the same level of output using less inputs (input efficiency). Dominance analysis is useful if the sample of producers is small.

Appendix Figure 4.1.Social Spending and Indicators in Latin America 1/

Sources: World Bank database; and IMF staff calculations.

1/ The lines depict the efficiency frontiers.

References

    Abrahão, J., and M.A.C.Fernandes, 1999, “Sistema de Informaèões sobre os Gastos Públicos da Área de Educaèão—SIGPE: Diagnóstico para 1995,” Discussion Paper No. 674, October1999, IPEA, Ministry of Planning and Budget, Brazil.

    • Search Google Scholar
    • Export Citation

    Clements, B., 1997, “Income Distribution and Social Expenditures in Brazil,” IMF Working Paper No. WP/97/120 (Washington, DC:International Monetary Fund).

    • Search Google Scholar
    • Export Citation

    Fernandes, M.A.C.; M.M.S.Oliveira, D.C.C.Rocha, N.S.Marinho, and J.A.C.Ribeiro, 1998, “Dimensionamento e Acompanhamento so Gasto Social Federal,” Discussion Paper No. 547, February1998, IPEA, Ministry of Planning and Budget, Brazil.

    • Search Google Scholar
    • Export Citation

    Gupta, S. and others, 2000, Social Issues in IMF-Supported Programs, IMF Occasional Paper No. 191 (Washington, DC:International Monetary Fund).

    • Search Google Scholar
    • Export Citation

    Harbison, R. W., and E.A.Hanushek, 1992, Educational Performance of the Poor: Lessons from Rural Northeast Brazil (Oxford:Oxford University Press).

    • Search Google Scholar
    • Export Citation

    Inter-American Development Bank, 1999, Facing Up to Inequality in Latin America (Baltimore:Johns Hopkins University Press).

    IPEA, 2000, Desenvolvimento Económico-Social Brasileiro na Década de 90, Diretoria de Estudos Sociais, IPEA, Ministry of Planning and Budget, Brazil.

    • Search Google Scholar
    • Export Citation

    Jimenez, E., and M.E.Lockheed, 1995, “Public and Private Secondary Education in Developing Countries: A Comparative Study,”World Bank Discussion Paper No. 309 (Washington, DC:World Bank).

    • Crossref
    • Search Google Scholar
    • Export Citation

    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 Health, 2000, Ações Prioritárias na Atenção Básica em Saúde, Brasília, Brazil.

    Neri, M., 1999, “A Note on the Growth Sensitivity of Poverty across Brazilian States,”IPEA, Rio de Janeiro.

    Neri, M., 2000, “Salário Mínimo e Pobreza,” presented at Previdência, Assistência Social, e Combate à Pobreza, Brasilia, March, 2000.

    • Search Google Scholar
    • Export Citation

    Neri, M., and J.M.Camargo, 1999, “Distributive Effects of Brazilian Structural Reforms,” Discussion Paper No. 406, October1999, Pontificia Universidade Católica, Rio de Janeiro, Brazil.

    • Search Google Scholar
    • Export Citation

    Paes de Barros, R.;C.H.Corseuil, and R.Mendonça, 1999, “Uma Análise da Estrutura Salarial Brasileira Baseada na PPV,” Discussion Paper No. 689, December1999, IPEA, Ministry of Planning and Budget, Brazil.

    • Search Google Scholar
    • Export Citation

    Paes de Barros, R.;, and R.Mendonça, 1999, “O Impacto da Três Inovaçães institucionais na Educação Brasileira,” Discussion Paper No. 566, Junho1998, IPEA, Ministry of Planning and Budget, Brazil.

    • Search Google Scholar
    • Export Citation

    Paes de Barros, R.;, and R.Mendonça, and M.Soares, 1998, “Urna Avaliação do Impacto do Programa Curumim Sobre o Desempenho Escolar,” Discussion Paper No. 542, February1998, IPEA, Ministry of Planning and Budget, Brazil.

    • Search Google Scholar
    • Export Citation

    Parente, M.M.A., and H.Lück, 1999, “Mapeamento da Descentralização da Educação Brasileira nas Redes Estaduais do Ensino Fundamental”, Discussion Paper No. 675, October, 1999, IPEA, Ministry of Planning and Budget, Brazil.

    • Search Google Scholar
    • Export Citation

    Ribeiro, J.M., and N.R.Costa, 1999, “Consórcios Municipais no SUS”, Discussion Paper No. 669, September, 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, May1999, IPEA, Ministry of Planning and Budget, Brazil.

    • Search Google Scholar
    • Export Citation

    Ter-Minassian, T., 1997, Fiscal Federalism in Theory and Practice, (Washington, DC:International Monetary Fund).

    Tulkens, H.; and P.Van den Eeckaut, 1995, “Non-Parametric Efficiency, Progress and Regress Measures for Panel Data: Methodological Aspects,”European Journal of Operational Research, Vol. 80, pp. 474–99.

    • Crossref
    • Search Google Scholar
    • Export Citation

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

    World Bank, 1995, Brazil: A Poverty Assessment, World Bank Report 14323-BR (Washington, DC:World Bank).

    World Bank, 1999, Brazil: Consultations with the Poor (Washington, DC:World Bank).

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

    World Bank, 2000b, Brazil: Public Policies to Fight Urban Poverty, World Bank Report 20475-BR (Washington, DC:World Bank).

1

Prepared by Luiz de Mello.

2

Calculation of poverty incidence and the income gap is based on the 1996 household expenditure survey (PNAD) and the widely-used poverty line of R$ 65.00 per capita per month, or approximately half a monthly minimum wage per capita, in 1996. Data from the latest household expenditure survey (PNAD-99), released in June 2000, show a slight increase in poverty incidence.

3

There is no consensus over the definition of social spending in Brazil. IPEA (Instituto de Pesquisa Económica Aplicada, Ministry of Planning and Budget) also treats outlays on public transportation, land reform, and environment protection as social spending. These outlays amount to less than 1 percent of GDP. For the purpose of this paper, social spending includes education and culture; health care and nutrition; housing, urbanization, and sanitation; social security and assistance; and unemployment insurance and labor. More recent data on social spending for the consolidated general government are not available. See Fernandes and others (1998), for more information on federal government social outlays.

4

In principle, governments that achieve better social indicators while spending less public resources on social programs can be considered as more efficient than those that achieve comparable social indicators using more public resources or, alternatively, exhibit worse social indicators for the same level of public spending. See Appendix I for more information.

5

Information on social spending in Latin America and the Caribbean is not readily available for the consolidated public sector. This underestimates total spending in countries, such as Brazil, where subnational jurisdictions are important providers of social services, particularly education and health care, as discussed above. The choice of indicators used to measure the efficiency of public spending on health care and education was guided by their appropriateness as proxies for the education and health status of the population, and the availability of internationally comparable data for a wide range of countries. See Gupta and others (2000), for more information on international social development goals and performance indicators. See World Bank (2000a), for more information.

6

See World Bank (1995) and Clements (1997), for more information. Using 1997 data, the World Bank (2000b) estimates that only 18 percent of total federal social spending (excluding social security) reaches the poorest 20 percent of the population. A lower incidence rate (7.4 percent) is reported for federal social security spending.

7

Using household survey data for the metropolitan region of Sao Paulo, Soares (1999) shows that the income share of the lowest quintile increases from 2.4 percent to 3.7 percent when social spending (comprising education, school lunch, and health care) is imputed in total household income. The redistributive impact of social spending is stronger than that of some targeted cash transfers (comprising maternity and disability benefits, unemployment insurance, and education grants). If these transfers are included in total household income, the income share of the lowest quintile increases from 2.4 percent to 3.3 percent, against 3.7 percent in the case of untargeted social spending.

8

The elasticity of poverty incidence with respect to income is relatively low in Brazil. PNAD-96 data have national coverage and suggest that a one-percent increase in mean consumption reduces the poverty headcount by approximately 1.0 percent. PME data, on the other hand, have narrower coverage (6 metropolitan regions) and suggest a lower elasticity, in the neighborhood of 0.6 (Neri, 1999), against approximately 1.5–2.0 percent on average for developing countries.

9

A recent study (Neri, 2000) using household expenditure survey data (PNAD-98) shows that a 10 percent real increase in the minimum wage reduces poverty by 1.3 percent among formal sector workers. If pensions are included in household income and both formal and informal sector workers are taken into account, a 10 percent real increase in the minimum wage reduces poverty by 4.5 percent.

10

The program aims at ensuring at least 15 percent of the child’s daily calorie intake during the school year. It is implemented on preschool and primary education institutions. A supplementary allocation is provided to those poor municipalities participating in the Comunidade Solidària program towards covering delivery and procurement costs. Coverage increased from 32 million children in 1994 to 35 million in 1997, of which 29.3 million are in primary education. See Lobato, Aquino and Ribeiro (1999), for more information.

11

The program coordinates federal government aid to poor municipalities. Eligibility is based on income per capita and assistance is provided over a three-year period based on progress in improvement of social indicators.

12

In addition to transfers within FUNDEF, education outlays are financed by Salário-Educação, a tax levied on payroll, and special social security contributions. There are special contributions to finance vocational training in industry, commerce, and tourism.

13

The impact of schooling on wage differentials is not homogeneous in Brazil. Paes de Barros, Corseuil, and Mendonça (1999) show that the impact on earnings of an extra year of formal education is higher among the more educated. Income inequality is also affected by schooling, and this association has become stronger in the 1990s (Neri and Camargo, 1999). Studies on the private rate of return on education suggest that these rates are higher for tertiary education. Social rates of return are typically lower given the subsidies in publicly provided services. According to the World Bank (1999), consultations with the poor have revealed that unemployment, followed by lack of schooling and urban services, are considered the most important causes of poverty.

14

Using 1990 data, the World Bank (1995) shows that the two highest quintiles receive 63 percent of public spending on higher education, against 19 percent for the two lowest quintiles. Using household survey data for the metropolitan region of São Paulo, Soares (1999) shows that primary education spending is progressive, with those in the lowest income quintile receiving 30 percent of total spending, against nearly 7 percent in the highest income quintile. Public spending on primary education accounts for nearly 2.5 percent of household income. According to the IDB (1999), only 5 percent of expenditures on secondary education goes to the lowest income quintile, against nearly 25 percent in the case of primary education. In the case of higher education, over 95 percent of spending accrues to the two highest quintiles. See World Bank (2000a), for more information.

15

A recent study on the Curumim Program (Paes de Barros, Mendonça and Soares, 1998), implemented in the state of Minas Gerais, suggests that test scores, enrollment and repetition rates, as well as age-grade gap indicators have improved in a sample of pupils in primary schools that participated in the program, after controlling for other determinants of school performance, such as parents’ educational background and occupation, and number of books at home, among others.

16

See www.mec.gov.br/Destaq/ministro.htm, for more information.

17

SUS (Sistema Único de Saúde) was created by the 1988 Constitution to replace INAMPS, which provided health insurance only to formal sector workers and their families. SUS was implemented in the early 1990s to extend publicly-provided health care services to the poorer states of the North, Northeast, and Center-West, where INAMPS coverage was limited, and to informal sector workers and their families.

18

The National Development Bank (BNDES) also provides loans to finance capital spending on health care. In 1998, BNDES loans in the health care sector totaled R$530 million, or over one-quarter of its social loan portfolio.

19

Soares (1999) shows that spending on health care is progressive, with those in the lowest income quintile receiving over 31 percent of total spending, against nearly 8 percent in the highest income quintile. Public spending on health care accounts for approximately 2 percent of household income. In the case of private outlays on health care, households in the lowest income quintile account for almost 11 percent of total outlays, against nearly 42 percent for households in the highest quintile. This reflects quality differentials between publicly and privately-provided services, and the ability to pay of different income groups. According to the IDB (1999), over 20 percent of health care spending accrues to the lowest income quintile, against only 5 rcent in the case of secondary education.

20

Based on PPV data, the World Bank (2000b) shows that nearly half of all urban users of public hospitals and health care institutions are poor, whereas nearly half of the users of hospitais conveniados are in the highest income quintile.

21

The World Bank (2000b) shows that incidence of the Milk Program is good, with nearly 60 percent of spending accruing to the two lowest income quintiles.

22

In the case of education, the analysis is carried out for one input indicator (public spending) and each output indicator separately (net primary school enrollment, net secondary school enrollment, and persistency to grade 5). This is because, at least in part, in the case of education, information is not available for most countries in the sample for all three output indicators. The methodology requires the elimination of these countries, thereby severely reducing the sample size. The data used in the calculation of the efficiency frontiers are available from the World Bank’s World Development indicators data set. The data refer to latest year for which information is available.

23

Countries with score equal to 1.0 define best practices for the sample of countries under examination. These countries dominate those that produce less output with the same or higher level of input (input efficiency) or use more input to produce the same or lower level of output (output efficiency). These countries lie on the efficiency frontier, as depicted in Appendix Figure 4.1.

24

By using total public, rather than only central government, spending for Brazil, a downward bias is introduced in the Brazilian efficiency scores. In this case, the unbiased efficiency estimates are likely to lie between the scores reported in Appendix Table 4.1 and those computed using total government spending as the input indicators in the FDH analysis.

Other Resources Citing This Publication