Contrary to the once-popular notion of the central bureaucracy withering away as an outcome of decentralization, scholars have shown that in cases of local policy success, national bureaucrats have instead redefined their roles, strengthening their focus on monitoring and accountability. Yet building national capacity for effective oversight presents a challenge within a context of strong subnational autonomy such as Brazil. Comparing the dynamics of decentralization across two areas of health policy, AIDS and tuberculosis, this article presents one strategy utilized by federal bureaucrats to increase their regulatory capacity: seeking resources located outside the formal political arena. Specifically, national bureaucrats utilize international resources to mobilize local civic groups as policy watchdogs, thus increasing the accountability of subnational politicians both to the center and to the public.

You do not currently have access to this article.


Reply to the E-Letter Response from Elize M. Fonseca and Diogo Ferrari
16 May 2013
Jessica A.J. Rich

I want to thank Elize Fonseca and Diogo Ferrari for the opportunity to expand the conversation over the challenge of national regulation in a context of decentralized governance. Ferrari and Diogo claim that the case study of AIDS policy decentralization misses the mark in two ways. First, they claim that the federal government in Brazil has stronger institutional mechanisms for controlling subnational policy development than I suggest. Second, they claim that the central bureaucracy has looked primarily to these institutional powers to regulate subnational AIDS policy. I will take advantage of this chance to delve deeper into the complex process of AIDS policy decentralization in Brazil by addressing each of these arguments in turn.

Challenges to the National Regulation of Local Policy in Brazil

Fonseca and Ferrari start by correctly asserting that the central government in Brazil is not completely subordinated to state and municipal interests, despite decentralized authority over many social-sector policies including public health. As they point out, national bureaucracies in Brazil have the constitutional authority to develop federal norms and guidelines that determine how much states and municipalities must spend on health, as well as how they must allocate that spending. Since the early 1990s, the national AIDS program has stood as one of the strongest central bureaucracies in Brazil with respect to its resources, its human capacity, and its policymaking powers.

The challenge of decentralized governance in Brazil--for health policy in general and for AIDS policy in particular--lies in the central bureaucracy's weak capacity to regulate subnational compliance with these federal norms (Rich 2013). This is where the authors and I disagree. Fonseca and Ferrari claim that the central bureaucracy effectively controls subnational AIDS programs by retaining partial control over their funding--using federal earmarked transfers for local AIDS programs as a sort of carrot and stick method of control. By contrast, my own interviews and observations suggest that this mechanism of using federal funding as a resource for central oversight does not function particularly effectively by itself.

On paper, the mechanisms for central oversight of state and local AIDS programs seem to be robust. In fact, the system for regulating local AIDS program implementation was designed in part to overcome certain weaknesses in federal oversight that had proven to be a problem in Brazil's decentralized public health system. For example, as a condition for receiving federal transfers the federal AIDS bureaucracy requires each state and municipality to develop detailed annual AIDS policy goals. In theory, this stipulation promotes effective subnational planning by allowing local authorities flexibility in setting their AIDS program goals --accounting for vast regional differences in bureaucratic capacity and in population needs. At the same time, this stipulation serves a political purpose, by providing national bureaucrats with an irrefutable baseline to use in evaluating and holding subnational politicians to account for their AIDS program development. Another mechanism the federal AIDS bureaucracy uses for subnational oversight is to deposit federal AIDS transfers into special accounts that are separate from other public health accounts and accessible via internet by the national bureaucrats in charge of monitoring AIDS funding. National bureaucrats check the flow of money out of these accounts daily.

In practice, however, the federal bureaucracy has faced strong challenges in using these formal powers to monitor and regulate policy implementation. The challenge to monitoring AIDS policy implementation stems in part from what we might consider to be a problem of weak state capacity. For example, the maze of complicated rules and procedures that are characteristic of Brazilian bureaucracy sometimes actually prevent state and local bureaucrats from complying with national guidelines. One nationally commissioned evaluation of Brazil's healthcare system noted that state and local healthcare managers pay essentially no attention to financial planning, principally because they are unable to predict how much money will actually fall into their accounts at the start of the year (World Bank 2007). State and local health secretariats thus lack both the capacity and the incentive to comply with the federal guideline of collecting the basic information required for the central government to monitor and evaluate subnational AIDS policy performance. In fact, in two separate national evaluations, government-contracted evaluators were unable to access even basic budgetary information to use in developing their analyses of Brazil AIDS programs and its public-health programs (World Bank 2004, 2007).

A separate challenge for AIDS policy regulation is the federal AIDS bureaucracy's inability to coerce politicians into complying with federal AIDS policy guidelines in states and municipalities where local political will is absent. For example, a significant number of governors and mayors have left earmarked federal AIDS transfers sitting in their bank accounts rather than diverting local resources from other, more politically popular programs to provide the matching funds required for AIDS program implementation (Rich 2013). In a local context in which there is no vocal grassroots constituency interested in AIDS program development, the promise of federal funding for local AIDS program development loses much of its appeal--as well as its coercive power.

Strategies for Increasing Federal Regulatory Capacity

The federal AIDS bureaucracy has used two main strategies to improve the quality of state and local AIDS programs. The first strategy focuses on overcoming weak subnational capacity; the second focuses on overcoming the challenge of weak political will. As the authors correctly noted in their comments, federal AIDS bureaucrats have recently attacked the problem of weak state capacity by attempting to strengthen AIDS policy coordination--among different levels of governance, among different types of state agencies, and among states and municipalities. Specifically, the federal AIDS bureaucracy has developed tri-partite management committees (CITs) so that federal, state, and local bureaucrats can coordinate their human and financial investments and increase their efficiencies of operation. Fonseca and Ferrari claim that I ignore the role of inter- governmental coordination in regulating subnational AIDS program development. This is a fair criticism; due to space constraints, I failed to highlight the national AIDS program's efforts to increase state capacity. Instead, I focused the article on describing the central government's strategy to overcome the challenge of weak local political will--a strategy that has been overlooked in the literature on decentralization--which is to mobilize grassroots groups as policy watchdogs and policy advocates.

The federal AIDS bureaucracy has pursued this strategy of grassroots engagement through both formal channels and informal channels. Fonseca and Ferrari claim that I ignore the role of formal institutions--specifically, participatory governance councils--in promoting local accountability for health policy (and, by implication, AIDS policy). Let me take this criticism as an opportunity to clarify my point about the role of formal versus informal channels for state-society collaboration in Brazil. As I note in the article, federal AIDS bureaucrats have pursued their mobilizing objective in part by engaging civil society in formal participatory governance committees, a subject of a significant body of recent scholarship (647). Civil society groups in Brazil have an official seat at the table in health policy debates through a constitutional mandate that the government must include "participation from the community." This vague mandate was later elaborated through a national law (8142/90) and various subsequent national guidelines to specify a mandate for participatory health councils that allot 50% of its seats to "users" of the health system, 25% to workers, and the remaining 25% split between service providers and government representatives. In theory, these councils serve as institutional channels for civil society to hold politicians accountable for investing in policies that align with citizen priorities. At the national level, health councils have indeed been highly influential in shaping the policy agenda (Barbosa and Abreu 2002, Cortes 2002, Mayka 2013). At the local level, however, the effectiveness of health councils as a mechanism for "societal control" of social policy has been mixed (Coelho 2004). Moreover, several studies attribute the degree of success that participatory governance councils have had in effecting local policy reform to local political will and to the pre-existing presence of a strong and politically mobilized civil society (Avritzer 2009, Fuks and Perissinotto 2007, Wampler and Avritzer 2004, ). My point in the article is that the federal bureaucracy has used a wide variety of other channels of engaging civic associations as AIDS policy advocates in addition to the formal channel of participatory governance councils. In practice, health councils are only one among a wide variety of national-level policy commissions, councils, and working groups that incorporate civil society participation in AIDS policy. Some of these arenas that structure collaboration between bureaucrats and civil society groups on AIDS policy are formal institutions, while others are informal. For example, the National AIDS Council (CNAIDS) debates policy and budgetary priorities that are specific to the AIDS sector (Spink 2003). The Commission for Engagement with Social Movements (CAMS) brings grassroots groups and federal bureaucrats together to discuss the role of social movements in AIDS policy development. Other collaborative AIDS policy institutions include the Vaccine Commission and the Commission for Inter-Sectoral Monitoring of STD/AIDS policy. At the same time, a changing array of informal "working groups" structure state-society cooperation over specific policy concerns; these include the Committee on Therapeutic Consensus, the Committee on Pharmaceutical Assistance, the Committee for Adherence to Medical Regimens, the Committee on Epidemiology, the Committee for Ethics in Research, the Committee on Men who have Sex with Men, and the Committee on Lesbians. As I pointed out in the article, national and local events also constitute critical arenas of regular interaction between federal bureaucrats and local civil society leaders. Together, these formal and informal channels of state-society interaction provide a strong set of new opportunities for local civic groups to act as government watchdogs and policy advocates. Existing studies, which focus narrowly on civil society influence within the official forum of participatory governance councils, overlook a much broader range of new channels that are available for civil society groups to monitor and influence policy development at the local level.


Given the national AIDS program's strategy of encouraging grassroots mobilization around AIDS policy, we would expect to see expanded grassroots mobilization around AIDS over the course of the 2000s. Fonseca and Ferrari provide an example of recent grassroots mobilization around AIDS that actually supports this expectation: the successful grassroots mobilization to protect earmarked transfers for AIDS programs in 2012. (See also Rich and Garrison 2013.) Nor does their example of the failed grassroots efforts in the early 2000s to force subnational authorities to continue funding needle-exchange initiatives refute this expectation. This case is in fact somewhat irrelevant, because it has to do with an early decentralizing initiative that took place before broader authority over AIDS programs was decentralized, and before federal AIDS bureaucrats began responding to the unexpected challenges of decentralization by supporting the mobilization of grassroots groups in new regions of Brazil. Moreover, the fact that grassroots groups did in fact mobilize to oppose subnational spending cuts on needle-exchange programs, although without success, corresponds to our expectation. What this 2007 study highlights instead is that the national AIDS program faced strong challenges to ensuring the quality of subnational AIDS programs in a context of decentralized governance.

Fonseca and Ferrari summarize their critique by claiming that grassroots mobilization "is not [a] sufficient condition to explain subnational government support to the national AIDS policy." I could not agree more with that statement. The policy process is complex and involves and wide variety of actors; whether any given policy initiative will succeed or fail depends on a complex set of causes and will vary over time and space. Nowhere in the paper do I make the claim that grassroots mobilization accounts for the continued success of Brazil's AIDS policy in a context of decentralized governance. Rather, the point of the paper is to a highlight new and overlooked strategy that has been used by the central government to regulate subnational policy in a context of decentralized authority.

Conflict of Interest:

None declared

Submitted on 16/05/2013 8:00 PM GMT
Governing decentralized health policy in Brazil
16 May 2013
Elize M. Fonseca (with Diogo Ferrari)

Comments and Response to Rich, Jessica A., & Eduardo Gómez. 2012. Centralizing decentralized governance in Brazil. Publius: The Journal of Federalism, 42(4), 636-661. We read Jessica Rich and Eduardo Gomez's paper on strategies of Brazilian federal bureaucrats to increase regulatory capacity over decentralized heath policies with great interest. Brazil is renowned for its remarkable response to HIV/AIDS epidemic, but as indicated by the authors, still faces challenges to improve its response to the tuberculosis (TB) crisis. However, Rich and Gomez do not analyze key features of the Brazilian federation, particularly the regulatory role of the Union in decentralization and intergovernmental coordination as well as institutionalized mechanisms of civil society to holding the government accountable. Furthermore, the mechanisms of National AIDS Program (NAP) to control the implementation of AIDS policies over municipalities are less related to the NGOs accountability and international resources than Brazil's earmarked transfers to subnational governments.

Federalism in Brazil

The authors ask what tools have national health bureaucrats used to increase their influence over policy implementation in a context of ongoing political/constitutional commitment to health care decentralization?

Rich and Gomez began their argument suggesting that Brazil is a highly decentralized nation. However, this claim warrants closer exploration; a the distinction between policy responsibilities (right to act) and decision-making authority (right to decide) shows that subnational governments in Brazil are highly constrained and regulated by top-down rules. The authors mention studies, such as Abrucio (1998), claiming that state governors have great influence over federal legislation, and therefore over nationwide policy decisions, and Stepan (2001) who argued that the central government in Brazil is highly constrained due to the authority of subnational unities and the Senate to veto the federal-led policy initiatives. Recent studies on the decision- making process at federal level, and the division of authority between the three levels of government in Brazil question the notion that the subnational units are excessively powerful in social policy, including health and education.

For instance, empirical studies on the content of the legislation on federal issues approved by the parliament as well as on the right to decide over nearly all policy areas in Brazil reveal that the Brazilian federation is much more centralized than usually described (Arretche 2012). The central government has the authority to initiate legislation and regulate local policies, (including health). As a result, local policy choices in health cannot be understood if we do not take seriously into account federal-led norms (ibid).

The 1988 Federal Constitution entitles the Central government to take initiatives which have impacted on the finance, autonomy and public policies of subnational unities. Although the Constitution of 1988 increased the decentralization of fiscal revenue to the subunities and gave autonomy of an important tax on consumption of goods and services (ICMS), the Union still has the authority to legislate on where and how much this resource will be used. For example, local government are bound by the Constitution to spend at least 15% of their revenues on health. Moreover, legislation recently approved set which items of spending can be properly understood as health expenditures. Had the subnational unities veto power, it would be unlikely that these changes would have occurred (Arretche, 2009).

Therefore, the starting premise of the authors, that Brazil is a highly decentralized nation, is only partially right. The federal executive has approved important norms that have impacted on intergovernmental relations (even without the support of subnational unities). To date there is hardly any evidence that regional interests have dominated national policies or that federal government had little governing capacity. The puzzle proposed by the authors disappears once we understand that Brazil is not as decentralized as they mention, but also that there are formal mechanisms to regulate implementation of health policies, as we shall see now.

The regulated decentralization process of the health sector in Brazil

Rich and Gomez fail to note that there are three instruments of intergovernmental regulation of the Brazilian health policy. First the Tripartite Inter-Management Committee (CIT), where the three-tier managers negotiate how policies will be implemented and which is also a space of political negotiation (Ministerial Directive 1180/1991). CIT decides on issues related to the national health policy, the role of each level of government and the decentralization process of health sector. CIT are conducted monthly and the discussions are publicly available for consultation(1). Another important instrument of coordination of health policy implementation is the power of the Minister of Health to issue Ministerial Directives and Norms to regulate states and municipalities (Machado 2007). An emblematic example is the Operational Norms (NOB) issued during the 90s to regulate the decentralization process of the health sector (Levcovitz et al. 2001). These norms attach conditions (such as criteria and responsibility) for federal transfers to subnational units. A failure to comply with the agreements would disqualify the municipality from receiving resources in the following year. Finally, a third and perhaps most important means regulation is the control of resources (Machado 2007). Studies conducted by the Center for Metropolitan Studies in Brazil have demonstrated that an important fraction of the local government budget comes from earmarked transfers (Arretche and Fonseca 2013). The Health Ministry's earmarked transfers toward states and municipalities have been automatic since 1993. It means that they are distributed according to legally and publicly known rules, and so subnational governments are entitled to receive them whenever they comply with such rules. However, transfers are usually broken up into several different programs, and so subnational governments have to accomplish each program's requirements to get the total amount of health care transfers they are entitled to.

Finally, different from the idea of informal channels of accountability proposed by Rich and Gomez, civil society oversight is formally included into SUS norms. States and municipalities must have their own civil-society health councils (CMS) (Law 8080/90 and 8142/90). According to SUS rules, transfers should not be made in the absence of these councils' fully operation. CMS are entitled to supervise health spending and if they do not approve the "health budget", transfers are automatically suspended until the accountability process is concluded.

The centralized AIDS response in Brazil

Finally, the study of Rich and Gomez does not acknowledge important information of the Brazilian response to the HIV/AIDS epidemic, particularly the political process to transfer the resources to subnational government. Since its inception in 1985, one of the pillars of the world-renowned Brazilian responses to AIDS has been a coordinated response at a central level (Nunn 2008). With the first World Bank loan agreement in 1994, the resources were transferred to states and municipalities through agreements called convenios, which have subsequently shown to be limited and complicated, given the legal requirements for their functioning (Taglietta 2006). In 2003, in order to comply with the Unified Health System (SUS) rules for transferring resources to subnational units, the National Aids Program implemented the decentralization of its resources, via earmarked transfers, to the 27 states and 424 municipalities out of the five thousand municipalities where the epidemic was concentrated (Ministerial Directive 2313/2002). This process has less to do with World Bank pressures as suggested by Rich and Gomez, than with Brazilian rules for transferring federal resources(2). The option for decentralization through earmarked transferring allowed agreement, via a Plan of Actions and Goals (PAM) (Ministerial Directive 2314/2002), amongst three levels of government and insulated these resources from local disputes over revenues. To qualify for AIDS transfers, local governments had to integrate their policy goals with the National AIDS Policy.

With the 2006 resolution that created transfer via "health functions" (Ministerial Directives 399/2006 & 699/2006), the possibility of integrating the AIDS earmarked transfer within the transfers for health surveillance function, which would allow local and state governments to decide on the allocation of these resources, was discussed. However, after an extensive negotiation process within the Ministry of Health, officials of the National AIDS program were successful in lobbying for the continuity of funds earmarked for AIDS inside the Health Surveillance Function. They argued that this would force the subnational government to commit AIDS care. More recently, in 2012, the Ministry of Health again attempted to integrate earmarked transfers into general health purpose funds. This decision was highly criticized by AIDS patient groups, as there are reports of local governments that would not invest public funds for people living with HIV/AIDS, given the stigma associated with the disease and low political leverage(3).

A counterfactual makes this point clear: in the absence of the AIDS earmarked transfer, would the states and municipalities dismantle initiatives to prevent HIV/AIDS? Studies suggest that although the Ministry of Health's earmarked transfers limit the autonomy of local governments, in their absence, the capacity of the municipalities to provide health services would be considerably unequal (Arretche 2010). Similarly, a study demonstrated that when the subnational government was authorized to decide on the implementation of the needle exchange programmes (an important strategy to prevent AIDS among IV drug users), there was an expressive reduction on public resources for these initiatives (Fonseca et al. 2007). According to the argument of Rich and Gomez, we could expect NGOs to aggressively oppose those cuts, which indeed happened but was not sufficient to stop these reductions.

Therefore, the policy implementation much depends on a coordinated response among the three levels of government. As suggested by Rich and Gomez, NGOs are indeed important players to keep governments accountable to AIDS treatment and prevention. However, these is not sufficient condition to explain subnational government support to the national AIDS policy. Core decisions of the National AIDS Policy are taken at the center, while earmarked transfers and the SUS formal mechanisms of regulation, constrain the states and municipalities to their implementation. The article by Rich and Gomez fails to take into account the importance of these aspects of the Brazilian federation.

Acknowledgements: the authors thank the Study Group of Federalism from the Center for Metropolitan Studies, Marta Arretche, Scott Greer, Patrick Silva, Nilson Costa and Amy Nunn for their valuable comments.

(1) http://portal.saude.gov.br/portal/saude/profissional/visualizar_texto.cfm?idtxt=28654 (accessed in January 16th, 2013)

(2)Note that an important book on policy diffusion in Latin America did not find that external pressure of World Bank or other international agency was a sufficient condition to explain the health reforms conducted in Brazil in 1988 (Weyland 2007). The reforms resulted from social movement pressure and in fact expanded, rather than retrenched in the way internationally current reforms suggested (Costa 2002).

(3)http://ungassforum.org/media/news/news.php?id=292 (accessed in January 16th, 2013)


Abrucio, F. (1998). Os Bar?es da Federa??o: os governadores e a redemocratiza??o brasileira. Sao Paulo, Hucitec/USP.

Arretche, M. (2010). "Federalism, bicameralism and institutional change: exploring correlations and mechanisms." Brazilian political science review(4): 10-31.

Arretche, M. (2012). Democracia, federalismo e centraliza??o no Brasil. Rio de Janeiro, Funda??o Getulio Vargas/Fiocruz.

Arretche, M. and E. Fonseca (2013). Health system decentralization in Brazil (Manuscript). Federalism and Health Decentralization. T. Bossert and G. Marchildon, Forum of Federations.

Costa, N. (2002). "Pol?tica Social e Ajuste Macroecon?mico." Cadernos de Sa?de P?blica 18(18): 13-21.

Fonseca, E., A. Nunn, P. Souza-Junior, F. Bastos and J. Ribeiro (2007). "Descentraliza??o, AIDS e redu??o de danos: a implementa??o de pol?ticas p?blicas no Rio de Janeiro, Brasil." Cad Saude Publica 23(9): 2134-2144.

Levcovitz, E., L. Lima and C. Machado (2001). "Pol?tica de sa?de nos anos 90: rela??es intergovernamentais e o papel das Normas Operacionais B?sicas." Ci?ncia & Sa?de Coletiva 6(2): 269-291.

Machado, C. (2007). Direito universal, pol?tica nacional: o papel do Minist?rio da Sa?de na pol?tica de sa?de brasileira de 1990 a 2002. Rio de Janeiro, Editora Museu da Rep?blica.

Nunn, A. (2008). The politics and history of AIDS treatment in Brazil, Springer.

Stepan, A. (2001). Arguing Comparative Politics. New York, Oxford University Press.

Taglietta, M. (2006). Avalia??o da Implanta??o do Incentivo no ?mbito do programa Nacional de HIV/Aids e outras DST em dois Munic?pios do Estado de S?o Paulo. Rio de Janeiro, Escola Nacional de Sa?de P?blica, Funda??o Oswaldo Cruz. Disserta??o de Mestrado Profissional.

Weyland, K. (2007). Bounded Rationality and Policy Diffusion. Princeton, Princeton University Press.

Conflict of Interest:

None declared

Submitted on 16/05/2013 8:00 PM GMT