Document Details
Conference Paper
Publication Year:
Federalism; Intergovernmental Relations and Social Policies in Brazil

1 FEDERALISM, INTERGOVERNMENTAL RELATIONS AND SOCIAL POLICIES IN BRAZIL Marta Arretche (Political Science Department/USP – University of São Paulo) Text prepared for the International Seminar Comparative analysis of Intergovernmental Management mechanisms and formulation of alternatives for the Brazilian case.  Brasília – September 17 and 18, 2003 Please do not quote or reproduce FEDERALISM, INTERGOVERNMENTAL RELATIONS AND SOCIAL POLICIES IN BRAZIL Marta Arretche[1]             Federal states are seen as inclined to produce relatively lower levels of social spending (Petersen, 1995; Banting, 2003) and to have a more limited reach and coverage of social programmes (Skocpol, Orloff, Weir, 1990; Pierson, 1996). They are also seen as tending to aggravate problems with co-ordinating policy objectives, generating competency overlap and competition among the different levels of government (Weaver and Rockman, 1993), given the negative relation between dispersion of political authority and internal consistency of collective decisions. However, concentration of political authority varies among federal states, depending on the way relations between the executive and the legislative branches of government are structured at the federal level (Stepan, 1999); and varies among policies, depending on how intergovernmental relations are structured in regard to specific policies. The purpose of this work is to demonstrate that the political institutions in Brazil concentrate authority in the federal government, facilitating functions of intergovernmental co-ordination of social policies. Approval of constitutional amendments is comparatively easier in Brazil than in other federations and the distribution of functions in the majority of social policies concentrates decision-making authority in the federal Executive. This work is divided into different sections. The first section provides a very brief description of the Brazilian tax and fiscal systems. The following section summarises intergovernmental distribution of functions in relation to a few selected social policies in order to highlight strategies and resources the federal government has to induce decisions by sub-national governments. Special attention is given to the municipalization of social programmes in the 90s – a sort of second round of the fiscal decentralisation of the 80s. Finally, some recommendations drawn from this analysis are presented. The Fiscal System From its origin in the Federal Constitution of 1891, the Brazilian federation adopted a system of separate tax sources, distinguishing taxes for which the states had the exclusive competence from those of the federal government. Municipal governments only acquired exclusive competence to implement taxes after the 1934 Constitution. Since then, the historical evolution of Brazilian tax structure has been characterized by slow gradual changes, with the two major breaks corresponding to the centralization imposed by the Tax Reform (1965-8) during the military regime and, in the following period, the fiscal decentralisation of the 1988 Constitution (Varsano, 1996; Affonso, 1999). In the current system, tax collection is very concentrated: the five main taxes are accountable for more than 70% of total tax collection – four of them are collected by the federal government[2]. The most important tax is collected by the states: Imposto sobre Circulação de Mercadorias e Serviços – ICMS (State Value Added Tax) (Varsano et alii, 1998). The tax load increase of the last decade – from 25% in 1991 to 34% in 2001 (AFE/BNDES, 2002) resulted primarily from tax efforts by the federal government, complemented by municipal governments (Prado, 2001). In 2002, 70.1% of total tax collection was carried out by the Union; 25.5%, by the states; and 4.3%, by local governments (see chart 1). Tax collection is extremely uneven on the horizontal plane, i.e., among sub-national governments. The ratio between states with largest and smallest own tax collection capacity in 1997 was 9.4 (Prado, 2001: 50). Within each state, same size municipalities differ vastly in tax collection. Except for capital municipalities – which collect up to 10 times more than other municipalities within their own state – larger size municipalities do not appear to perform better than small ones, regardless of the income level of the state where they are located (Prado, 2001: 63 ff.). Horizontal tax collection inequalities have been compensated by a system of fiscal transfers since the 1946 Constitution was adopted. Constitutional transfers distribute part of the revenue collected by the federal government to states and municipalities, and from states to their respective municipalities (Varsano, 1996). Centralization cycles (1965-8 Tax Reform) and decentralisation cycles (1946 and 1988 Federal Constitutions) of the Brazilian fiscal system are directly associated with tax rates applied to compulsorily shared taxes and with spending autonomy by local governments in relation to transferred funds. In 1968, the Union’s constitutional transfers added up to 10% of the collection amount from its two main assessments. Moreover, most transfers were linked to pre-defined expenditures (Medeiros: 1986). In the 1988 Constitution, the Fundo de Participação dos Estados (States Participation Fund) and the Fundo de Participação dos Municípios (Municipalities Participation Fund) have as their revenue source an amount equal to 44% of the revenue of two federal taxes[3]. State governments, on their part, must transfer 25% of collected amounts from their main tax to municipal governments. In addition to that, revenue originating from constitutional transfers may be freely spent, except for constitutionally-bound health and education expenditures.  On the vertical plane, the current fiscal transfer system allows almost all relative income to be appropriated by municipal governments, since their results are neutral for the states (Serra and Afonso, 1999). In 2002, the federal government’s available revenue (own collection + transfers) was 60% of the revenue total, while municipalities took just more than 15% and the states remained at the same level prior to redistribution (see chart 2).  Furthermore, this system favours small size municipalities. Gomes and Mac Dowell (1997) have estimated the participation rate of transfer revenue in municipalities with less than 5,000 inhabitants as 91%. In 22 Brazilian states, the smaller the population of a municipality, the greater its income deriving from constitutional transfers, i.e., the greater the availability of resources per capita with free allocation  (Prado, 2001: 68).  On the horizontal plane, this system redistributes revenue from more developed to less developed states (Rezende and Cunha, 2002: 115 ff.), even though constitutional transfer rules have the consequence of creating new inequalities among Federative Units (Prado, 2001: 54). In spite of such redistribution, there is enormous variation in revenue available to Brazilian municipalities. The average per capita revenue of municipalities with more than 1 million inhabitants in the Southeast region may be up to 46 times that of municipalities with less than 20 thousand inhabitants in the North and Northeast regions (see Gomes and MacDowell, 1997:11). In practice, horizontal revenue inequalities make the notion of municipality in Brazil more of a fiction than a reality. Besides constitutional transfers, the federal government transfers funds to the states and municipalities through voluntary transfers not subject to local distribution rules. The latter have increased in the last years – from a R$ 5 billion level to R$ 13.5 billion in 2000, representing 18% of constitutional transfers in 1995 and 36% in 2000  (Prado, 2001: 37). However, over two-thirds of these transfers are linked to Sistema Único de Saúde – SUS (Unified Health System), which has built a highly-regulated system for fund distribution. Once SUS transfers are deducted, voluntary transfers represent approximately 10% to 15% of constitutional transfers in the last years (Prado, 2001: 37). Therefore, the Brazilian system for tax revenue sharing is essentially a legal system (Prado, 2001), by which tax revenue with no compulsory spending allocation – except for constitutional obligations in health and education spending – is ensured to sub-national governments, notably small size municipalities. The result of this redistribution is a significant horizontal inequality in spending capacity of sub-national governments. Social Policies and Intergovernmental Relations Vertical relations in the Brazilian federation – between federal government and states and municipalities and between state governments and their respective municipalities – are characterized by independence, because states and municipalities are autonomous federal entities. In theory, the constitutional guarantees of a federal state allow local governments to establish their own agenda in the social arena. Given the marked horizontal inequality among sub-national governments, constituents in 1988 chose the concurrent competence format for the majority of Brazilian social policies. Any federal entity was constitutionally authorized to implement programmes in the areas of health, education, welfare, housing and sanitation. Conversely, no federal entity was constitutionally obligated to implement programmes in those areas. From this fact comes the view that the 1988 Constitution decentralised revenue, but did not decentralise responsibilities (Almeida, 1995; Affonso and Silva, 1996; Affonso, 1999; Willis et alii, 1999). In the beginning of the 90s, federal distribution of responsibilities in the social arena derived less from constitutional obligations and more from the way with which these services had been historically organized in the case of each particular policy. Throughout the 90s, however, the decentralisation of responsibilities in the social arena was carried out. The decentralisation process as well as the federal distribution of competencies took a particular path for each sector policy. Health Policy The 1988 Constitution defined that access to free health services is a universal right, creating the Sistema Único de Saúde – SUS (Unified Health System). In May 2002, a total of 5,537 of the 5,560 Brazilian municipalities – 99.6% of the total – had assumed partial or full management of health services. In 2000, Brazilian municipalities were responsible for an average of 89% of total health care service delivery in Brazil, with a standard deviation of 19% (considering all categories of providers). In 2000, 84% of the health care network was under the responsibility of municipalities – such average carrying a decreasing standard deviation. In other words, not only do health service network and delivery belong to municipalities essentially, but also the differences among Brazilian municipalities for this aspect of decentralisation have been diminishing (Marques and Arretche, 2002). Brazilian municipalities were responsible for 9.6% of total consolidated health spending in 1985; 35% in 1996 (Medici, 2002); and 43% in 2000 (Ferreira, 2002). From the standpoint of funding source, municipal participation changed from 9.3% in 1985 to 28% in 1996. State participation varied greatly, but remained around 18% during those years and federal government participation fell from 73% to 53% for the same period (Medici, 2002). In the 90s, the only government level presenting regular increase in health spending was the municipal government (Piola and Biasoto Júnior, 2001). In regard to intergovernmental function distribution, the Union is in charge of funding, formulation of a national health policy, and co-ordination of intergovernmental initiatives. This implies that the federal government – i.e., the Ministry of Health – has the authority to make the major decisions on this policy. So the policies implemented by local governments strongly rely on federal transfers and federal government defined rules. On the other hand, state and municipal participation in the process of health policy formulation has been institutionalized by means of councils with state and municipal representation. The institutionalization of such negotiation arenas has removed from the federal government the possibility of unilaterally establishing operating rules for SUS. Such councils work as a counterweight mechanism to the concentration of the authority granted to the federal Executive. Decentralisation of attention to health did not involve a constitutional reform, but rather getting municipalities to follow constitutional norms that had already been approved in the 1988 Constitution. In this case, the strategy underwent edition by a sequence of ministerial orders, whose content consisted in conditioning federal transfers to state and municipal adherence to federal policy objectives.  In addition, the uncertainty concerning transfer execution was reduced, making it believable such transfers were actually going to be performed. Basic Education Policy In Brazil, basic education provision is the responsibility of both states and municipalities, but the two networks operate in a completely independent manner. Due to the way with which enrolment expansion has evolved historically, participation of the two networks in student places provision in each state is highly variable.  In São Paulo State, the state government offered 87.5% of the student places, while in the states of Alagoas and Maranhão municipalities were responsible for 65% of the student places (Vazquez, 2003: 37). Horizontal and vertical inequalities in the Brazilian federation’s spending capacity implied in career and wage inequalities for teachers, inequalities in the nature and quality of educational services, and inequalities in level of spending/student. In 1996, municipalities of the state of Maranhão spent an average of R$ 100 per year with their students, while the state government spent R$ 385 per year. In São Paulo, on the other hand, municipalities spent R$ 1,039, while the state government spent R$ 569 (Vazquez, 2003: 39).  The 1988 Constitution defined basic education competencies as concurrent, simply establishing that basic education should be offered by municipal governments preferably. Moreover, the 1988 Constitution forces state and municipal governments to spend 25% of their available revenue in education. Therefore, the Constitution did not create mechanisms for reducing asymmetries, tending, in fact, to aggravate them. In order to reduce asymmetries in basic education spending and quality of services, and to promote teacher wage increases, the federal government managed to approve a Constitutional Amendment in 1996 (EC14/96), which established that for a time period of 10 years states and municipalities would have to invest a minimum of 15% of their revenues in basic education exclusively. In addition, 60% of those funds would have to be used exclusively for the payment of active teachers. To decrease spending inequality, the EC14/96 amendment determines that the federal government must complement spending in those states where a minimum national amount is not reached based on local government income. In practice, implementation of such Constitutional Amendment implied in a mini tax reform within a state scope, to the extent that 15% of the state and municipal yearly income is automatically held and booked in a State Fund – the FUNDEF. Its revenues are redistributed – within each state – between the state and municipal governments according to the number of enrolment spots offered each year. FUNDEF binds responsibilities to tax revenue, besides ensuring an effective transfer of funds (Oliveira, 2001), whose effect was to eliminate uneven intrastate spending in basic education (Vazquez, 2003). Its effects on spending inequalities between states were almost null, especially because of a reduction in funds provided by the federal government (Vazquez, 2003).  In this particular policy, as opposed to the health policy case, the federal government is not the main funding agent, performing a supplemental role in financing meal provision programmes for public school students, and school unit building and qualification programmes.  In this case, in order to reach its objectives, the federal government has adopted a strategy of constitutionalizing intrastate transfers, thus eliminating uncertainties regarding receiving of funds connected with enrolment provision. Housing and Sanitation Policies The Brazilian housing and sanitation service provision system was instituted in the 60s and 70s. Institutional reforms carried out in the 80s and 90s did not modify the basic structure of the federal distribution of functions. The federal government collects and redistributes – through loans – resources from the main funding source for these policies: a fund destined to indemnify workers fired with no cause, whose net collected amounts are destined to fund housing and sanitation programmes (FGTS – Government Severance Indemnity Fund for Employees). Basic sanitation services are provided by 27 state companies that control the majority of operations in the sector, based on service licence agreements with more than 4,000 Brazilian municipalities. In other municipalities, services are operated by municipal agencies and bodies. The production of houses for the low-income population is performed by municipal companies, which act within a regional scope[4].  The majority of the Fund’s resources are destined to credit lines for direct acquisition of real estate by the final borrower.  Each state has an annual budget, calculated based on redistribution criteria, i.e., inversely proportional to collecting capacity and directly proportional to housing and sanitation needs. State commissions directly appointed by governors distribute funding within each state. However, the management council for FGTS – an entity under federal jurisdiction without federal representation, which is controlled by the federal government – defines the operating rules for the programmes, and the Fund’s financial agent – a federal bank – has the ultimate authorization power to effect the loans. As in the health sector, therefore, the federal government is in charge of financing and formulating national policy. However, the absence of institutionalized representation by sub-national governments in federal decision-making arenas provides the federal government with reasonable amount of autonomy to unilaterally define distribution rules for federal loans. Monetary income transfer programmes             The trajectory of monetary income transfer programmes in Brazil originated from a proposal by Senator Eduardo Suplicy (PT/SP) for implementing a national universal negative income programme. Barred by conservative forces in Congress, the idea was implemented in some municipalities and the Federal District of Brasilia. Its demonstration effect gave origin to many income transfer programmes – federal, state and municipal – with an institutional design different from the original proposal  (Lavinas, 1999; Lavinas and Barbosa (2000).             Federal programmes are neither universal nor unified. Different programmes of a similar format are managed by different ministries. As a rule, families below the poverty line (monthly income up to R$ 90 or approximately US$ 30) form the target population, but the actual number of families who can benefit is limited by the budget of each programme. The permanence of the benefit to each family is conditioned to compensations, such as school attendance, health centre meetings, etc.             In the federal division of competencies, the federal government is the main funding agent, defining access to benefit and distribution rules. Municipalities, in turn, enrol families based on per capita transfer amounts up to the limit of their respective budgets, which are established based on available resources. Once enrolment in the respective ministry is approved, families receive a magnetic card and start to debit the monthly payment (of R$ 45 maximum or approximately US$ 15) through the banking network of a federal bank. Except for the formation of municipal inspection councils, no compensation is demanded in return from municipalities.                These programmes have rapidly expanded in recent years, reaching millions of beneficiaries. The Bolsa-Escola programme (school bursary programme) managed by the Ministry of Education was created in 2001 and reaches 8.3 million children and 5 million families in 2003 (see Ministry of Education web site). The Bolsa-Alimentação programme (food bursary programme) managed by the Ministry of Health was created in 2000 and reached 3.5 beneficiaries in 2002 (see Ministry of Health web site). Costs for municipal participation are very low, particularly when compared to the political credits derived from the autonomy in enrolling beneficiaries. What lessons can be learned from the experience of the 90s?             Throughout the 90s, a decentralisation of responsibilities in social policies took place. Except for the basic education policy, the concentration of authority in the federal government characterises federal relations in policy management, because the Union exercises the role of main funding agent, and of regulator and co-ordinator of intergovernmental relations. As a general rule, a direct relationship has been established between federal government and municipalities[5], under which the latter operate as service executors. This design arose in great part from an effort to match the tax decentralisation of the 80s with responsibility for provision of social services. The management format that concentrates authority in the federal government offers advantages for co-ordinating policy objectives within a country, because it makes it possible to reduce the risk that the different levels of government will generate conflict between programmes and increase implementation costs – risks that are more likely to occur in federal states (Weaver e Rockman, 1993). In addition, funding concentration in the federal government level would allow redistribution results to be reached (Banting, 2003), reducing horizontal inequalities in spending capacity.  Decentralisation of health policy responsibilities and the expansion of monetary income transfer programmes originated from the federal government’s use of its funding and regulating authority to get municipalities to adhere to a given policy objective. To ensure effectiveness, the federal government has made transfers to local government conditional on adherence to the federal government’s agenda. This move has turned out to be a highly appealing strategy when it comes to the choices of the local government. Once uncertainty about finding guaranteed funding is reduced, the willingness to assume responsibilities for social policies increases.   The agenda for privatization of state sanitation companies in the recent past has also been conditioned based on the federal government’s inducing capacity, derived from its authority over the main source of funding. In this case, the absence of federal representation in federal decision-making levels is added to control over funding. This combination has allowed the Union to unilaterally define rules for loan acquisition, toughening loan granting requirements and therefore de-funding sub-national governments in this policy area. Redistribution results from concentration of authority in the federal government have not turned out to be so evident, though. Municipalization of health services has not been accompanied by a reduction in inter-municipal inequalities regarding provision standards (Marques e Arretche, 2002). Federal government supplements to basic education have hardly affected interstate inequalities in spending standards (Vazquez, 2003). This takes place partly because federal disbursements present reduced redistribution effect (Prado, 2001), but also because the objective of reaching decentralisation was more focused than redistribution objectives in strategies adopted in the 90s.  In this case, the improvement of rules governing such transfers as well as the increase of federal funds destined to supplement FUNDEF state revenues would be advisable. There are no conclusive studies on the redistribution impact of federal monetary transfer programmes, because they are so new. Low individual amounts of transfers to families – partly due to their fragmentation into different ministries – does not allow us to estimate a significant impact on poverty reduction. Added to this is the non-universal nature of the programme, associated with low control over enrolment by local governments. Bringing such programmes to all people below the poverty line would reduce their potential use for political patronage. At the same time, their unification would allow greater control over the distribution of benefits and a likely increase in the amount of individual benefits.             In policy areas in which the federal government does not have institutional resources for changing local government choices, the strategy to constitutionalize responsibilities has turned out to be successful. It has been about approving constitutional amendments that greatly reduce the margin of choice for sub-national governments, forcing them to behave in a manner considered desirable by the federal government. This has been the strategy adopted for levelling out intrastate spending in basic education and raising teachers’ wages. In the health sector, this has been the strategy adopted for binding revenue to health spending levels[6]. In fact, approval of constitutional amendments is relatively easier in Brazil than in other federations. A two-thirds majority in two legislative sessions in each house (House of Representatives and Senate) is the required for their approval. It is not necessary to obtain approval from state legislatures, as in the United States, for instance, even if the issue affects the interests of sub-national governments. In the cases mentioned above, obtaining the president’s support for the legislative initiative that have allowed favourable mobilization of the coalition supporting the government in Congress has been the crucial factor for the success of the Executive in the parliamentary arena. Constitutionalization of responsibilities or spending levels is, however, a strategy limited by horizontal inequalities of the Brazilian federation. Horizontal inequalities among sub-national governments continue to recommend caution in the constitutional definition of exclusive competencies in social policy management, even if fiscal and policy decentralisation has increased sub-national governments’ capacities – administrative, fiscal and service delivery. Binding of expenditures also tends to reproduce pre-existing inequalities in spending capacity on the plane of policy implementation. The problem of inequalities seems to be a much more complex challenge than the co-ordination of intergovernmental relations…             References AFE/BNDES (2002). Informe-se ,no. 40. AFFONSO, Rui (1999). “A Federação na Encruzilhada”, in: Revista Rumos. Os caminhos do Brasil em debate, 1 (2): 29-38. Publication by the National Commission for celebration of the 500th Anniversary of the discovery of Brazil. AFFONSO, Rui and SILVA, Pedro Luiz Barros (1996). Descentralização e Políticas Sociais. São Paulo: FUNDAP. ALMEIDA, Maria Hermínia Tavares de (1995). “Federalismo e Políticas Sociais”, in: Revista Brasileira de Ciências Sociais, 28(10): 88-108. ARRETCHE, Marta, 2002. “Federalismo e Relações Intergovernamentais no Brasil: A Reforma dos Programas Sociais”, in: Dados, 45(3): 431-57. FERREIRA, S. (2002). “Federalismo fiscal, municípios: despesa com saúde e transferências federais”, in: Informe-se 38. Banco Federativo do BNDES. LAVINAS, Lena (1999). “Renda Mínima: práticas e viabilidade”. Novos Estudos CEBRAP, no. 53: 65-84. LAVINAS, Lena and BARBOSA, Maria Ligia de Oliveira (2000). “Combater a Pobreza Estimulando a Freqüência Escolar: O Estudo de Caso do Programa Bolsa-Escola de Recife”. Dados,  43(3): 447-477. GOMES, Gustavo Maia and MACDOWELL, Cristina (1997). Os Elos Frágeis da Descentralização: observação sobre as finanças dos municípios brasileiros. Work presented at the International Seminar on Federalism and Local Governments, La Plata, Argentina. MARQUES, Eduardo and ARRETCHE, Marta (2002). Condicionantes locais da descentralização da política de saúde. Work presented at the XXVII National Meeting of ANPOCS, photocopy. MEDEIROS, Antonio Carlos de (1986). Politics and Intergovernmental Relations in Brazil: 1964-82. New York/London: Garland Publishing Inc. MEDICI, André (2002). El desafío de la descentralización: financiamiento público de la salud en Brasil. BID, Washington. OLIVEIRA, Fabrício Augusto (2001). FUNDEF e Saúde: duas experiências (virtuosas?) de Descentralização. Research Report. Project: Fiscal Decentralization and Intergovernmental Financial Cooperation. EBAP/Konrad Adenauer Foundation PIOLA, Sergio and BIASOTO Jr., J. (2001). “Financiamento do SUS nos Anos 90”, In NEGRI, B. e GIOVANNI, Geraldo. Radiografia da saúde. Instituto de Economia, Campinas. p. 219-232. PRADO, Sergio (2001). Transferências Fiscais e Financiamento Municipal no Brasil. Research Report. Project: Fiscal Decentralization and Intergovernmental Financial Cooperation. EBAP/Konrad Adenauer Foundation REZENDE, Fernando. Globalization, Fiscal Federalism, and Taxation, in: Banco Federativo do BNDES. REZENDE, Fernando and CUNHA, Armando (2002). Contribuintes e Cidadãos. Compreendendo o Orçamento Federal. Rio de Janeiro: FGV. SERRA, José and AFFONSO, José Roberto Rodrigues (1999). “Federalismo Fiscal à Brasileira: Algumas Reflexões”, in: Revista do BNDES, 6 (12): 3-30. VARSANO, Ricardo (1996). A Evolução do Sistema Tributário Brasileiro ao longo do século: anotações e reflexões para futuras reformas. Brasilia: IPEA. Text for discussion no. 405. VARSANO, Ricardo et alli (1998). “Uma Análise da Carga Tributária no Brasil.” Text for Discussion IPEA, no. 583. Brasilia: IPEA.  VAZQUEZ, Daniel Arias (2003). Educação, Descentralização e Desequilíbrios Regionais: os impactos do FUNDEF. Master’s dissertation presented at the Instituto de Economia – UNICAMP. WEAVER, K.R. e ROCKMAN, B (1993). Do institutions matter? Washington: Brookings Institutions. WILLIS, Eliza; GARMAN, Christopher da C. B; HAGGARD, Stephan (1999). “The Politics of Decentralization in Latin América”. Latin American Research Review, 34 (1):7-56. [1] Professor of the Political Science Department of Universidade de São Paulo – USP (University of São Paulo). Email address: [2] Namely, social security contribution, Income Tax (IR), Tax for Social Security Financing (COFINS) and Excise Tax (IPI – Tax on Industrialised Products). [3] The 1965-8 Tax Reform created the Fundo de Participação dos Estados – FPE (States Participation Fund) and Fundo de Participação dos Municípios – FPM (Municipalities Participation Fund), consisting of a percent of federal collection of Income Tax and IPI. Such percents were 5% for each Fund in 1968 increasing to 21.5% and 22,5%, respectively, as of the 1988 Constitution.  The Participation Funds involve about 20% of the total revenue managed by the federal government (Prado, 2001: 54). [4] Between 1995 and 2000, 12 of the 44 municipal housing companies closed due to a policy of de-funding by the federal government (Arretche, 2002). [5] This is a somewhat simplistic statement, because the inter-mediation of the states varies according to policy and it is a lot more complex than what the present verification would allow us to conclude. [6] The constitutional amendment EC 29/2000 establishes that states must spend at least 12% of their revenue in health by 2005. In the case of municipalities, this obligation reaches 15% of revenues; and in the case of the federal government, the increase in health spending must follow GNP growth variation.