Much of the history of Brazilian politics has been dominated by swings between centralized and decentralized political rule. Brazil's recent transition to democracy⁽¹⁾ was no exception to this legacy. The authoritarian regime installed in 1964 represented only the most recent attempt to centralize political control. With the liberalization of authoritarianism, policy authorities and resources were returned to the states and cities. Brazil took a significant step towards democracy in 1982 with direct elections for governor, the first time such contests were held since 1965. Along with directly elected federal deputies, the governors were strategically placed to extract new political authorities and fiscal resources. These politicians and the local interests they represented became key actors in the drafting of the 1988 Constitution which would further decentralize fiscal and policy resources to subnational government.

This chapter will argue that the relationship between political decentralization and building democracy in Brazil is paradoxical. On the one hand, decentralization helped to fragment, and thereby weaken, the central state. By strengthening subnational governments' authority to tax and spend, political decentralization injected greater chaos into the central state's attempts to manage macroeconomic and social policy. On the other hand, the new federal order revealed a growing capacity to adapt to the problems of Brazil's nascent democracy. Many state and local governments engaged in economic reforms that improved public accounts. In some cases, subnational government developed innovative responses to social and economic problems that national policies failed to address.

I will analyze this paradoxical relationship between democratization and decentralization in Brazil as it evolved in three phases. Each phase was marked by an issue or process that had a lasting impact on the way that decentralization and democratization would affect each other. This periodization is used only as a guide for events that partially overlap in time and, regarding the third, and latest, period, are not completed but have clearly been initiated.

The first phase spanned most of the 1980's and was characterized by strong decentralization. During this period, the sequencing of political decentralization added to democracy's rough start and contributed to some persisting problems. The empowerment of the governors following the 1982 elections made these officials key architects of the new federalism. Before national democratic leaders could install a federal framework from above, the governors constructed one from below in ways that favored local political machines.⁽²⁾ This strengthened local clientelism as a centerpiece of Brazilian politics and weakened the central state by fragmenting its resources and authorities. By weakening the central state, decentralization hindered Brasília's capacity to address crucial macroeconomic and distributional issues that were essential for improving the quality of the country's democracy during the 1980's.⁽³⁾

The second phase occurred during the early to mid 1990's and was a period of reequilibration of intergovernmental authority. The decentralization wave stalled as the central state halted unsustainable patterns of fiscal devolution and subnational spending from the previous decade. In the fight against mega-inflation, for example, the Brazilian central bank had to threaten or actually intervene in failing state banks to recapture control over public spending. Although decentralization created many of the problems such episodes of reequilibration were meant to redress, it also facilitated the central state's efforts to recover authority. Divisions among states and municipalities vying for investments and hoping to correct fiscal accounts ruined by clientelistic politics during the 1980's, shifted bargaining leverage to reformist presidents in Brasília during the 1990's. As a result, central state agencies were able to extract reform concessions from weakened and divided states and municipalities.

Nevertheless, reequilibration was only partial as it did not lead to a coherent national framework for requiring subnational compliance with federal reforms. Shifts in the interests of key subnational leaders consolidated reforms in some cases while others requiring additional federal prodding produced few changes. Overall, the haphazard reequilibration of intergovernmental authority became a prerequisite for confronting sources of central state weakness and economic chaos that threatened the new democracy.

The third phase, partially overlapping the second, was a period of receding decentralization and subnational consolidation of policy innovations. During this phase, the central state acquired access to previously devolved resources and authorities and shifted spending responsibilities to subnational government, reversing the gains made by decentralizers. At the same time, many subnational governments pursued a wider range of policies to consolidate fiscal reforms initiated during the second phase and produce new economic and social policies responsive to local needs. Where the influence of local political machines could be displaced or coopted by reformist political leadership, innovative policy experiments moved forward. These cases demonstrated that political decentralization could maximize federalism's "allocative efficiencies" and even improve on previous national policies.⁽⁴⁾

The periodization outlined above has been referred to in the literature on Brazilian politics as the "new federalism." The term often obscures more than it clarifies since it includes both the ascending and receding phases of decentralization in recent Brazilian history. However, since the main argument of this chapter is that decentralization and democratization are related in both complimentary and conflicting ways, references to the "new federalism" are useful for generalizing about a process that is both flexible and evolving.⁽⁵⁾ This is in contrast to the term "federalism" which assumes the (re)emergence of a coherent, previously defined model of intergovernmental authority. Such a model has not been consolidated in Brazil.

Both the dangers and encouraging possibilities of the new federalism in Brazil are recognized by the theoretical and country-specific literature. However, most of these works tend to see either the negative or positive aspects of political decentralization as endemic. Many scholars see subnational politics as hopelessly clientelistic so political decentralization only reinforces the dominance of patrimonialism. Other scholars see decentralization as a democratizing process, one that takes government "closer to the people," expanding access to policy making and increasing the efficiency of policy implementation.

In this chapter, I argue that neither of these approaches is exclusively correct. After contrasting the two theoretical perspectives, I will analyze the three main aspects of the relationship between democratization and decentralization in Brazil during the three periods outlined above. The first section will focus on the sequencing of decentralization and democratization during the 1980's. The second section will discuss the reequilibration of intergovernmental authorities during the early 1990's. This section is subdivided into a study of federal attempts to reform the financial system and efforts to halt fiscal competition among the Brazilian states. Since the section includes references to successful financial and fiscal reform efforts by the central state and key subnational governments, my analysis crosses over into the third period of the new federalism. The final section focuses on the experiences of some reformist states and cities. In this section I will argue that the third phase of the new federalism was not simply one of consolidating intergovernmental reequilibration but of policy innovations among subnational governments.

Two Contrasting Perspectives on Political Decentralization and Democracy

Does political decentralization enhance Brazilian democracy or does it weaken it? Over the years, a number of scholars have advocated one or the other position. Few, however, have reconciled the arguments into a coherent theory of decentralization and democracy in Brazil.

Arguments for the undemocratic effects of political decentralization in Brazil usually focus on how the dispersal of power fragments political organizations and the state. The most common argument is that decentralization empowers local clientelistic leaders who serve only themselves and not more universal concerns (Samuels & Abrúcio 1997; Hagopian 1996). In exchange for policy access, politicians garner the support of these local patrons (Novaes 1994). Clientelism thus breaks down the state's professional hierarchy and institutional unity (Weyland 1996; Evans 1989). Those seeking to open the political system to opposing views and policy alternatives are excluded.

Brazilian political parties fragment in similar ways. As clientelistic vehicles designed to boost the political influence of their leaders, parties are divided by many local alliances required to sustain politicians in office or allow them to gain power (Souza 1976; Diniz 1982; Mainwaring 1992-93, 682-683 & 700-701). The importance of local ties is reinforced by the fact that national parties cannot override the candidate nominations of state-level party conventions (Ames 1994, 97).

Endemic clientelism in Brazilian politics produces serious implications for economic management as well. Since patrimonial elites seek control over the economic bureaucracy for their own personal gain, the policies they propose will be designed according to a political rationale (the pursuit of "rents") rather than the efficiency rationale of the market (Barzelay 1986). This "rent-seeking" explains why economic growth in Brazil has historically served particular segments of the elite and has failed to be distributed or even maintained for long periods (Kaufman 1990).

Other authors, mostly those working from a comparative perspective in other countries or regions, argue that political decentralization can guarantee the most developed forms of democracy. For these scholars, decentralized government is more accountable, more accessible, more innovative, and generally more effective than centralized government. A common rationale of this work is that by being "closer to the people" subnational policy making reduces the cost for most of gaining access to government (World Bank 1997). As a result, subnational government must perform consistently and according to the expectations of regional economic groups and community leadership. Studies of subnational government show that local governments are often more willing than national leaders to respond to the demands of their constituency. The preeminent proponent of such arguments, Robert Putnam (1993, 50-51), in his study of the Italian regions, even suggests that local constituencies apply a higher standard of performance to their local governments than they do to central government. Therefore, political decentralization enhances the accountability of government by increasing its localization.

Fiscal federalism studies address similar questions but they focus on the efficiency of policy outcomes by specifying the optimal distribution of authorities and resources between national and subnational government. The best model of federalism is one in which allocative policies that require specific information about local demands are decentralized while distributive policies that depend upon a national perspective are centralized (Musgrave & Musgrave 1989; Oates 1973; IDB 1994, 178; IDB 1997, 153). As long as these "correct" principles of decentralization are followed, the efficiency of public functions is preserved.

Brazil's new federalism does not fit neatly into either one of the two perspectives reviewed thus far. The adoption and professional management of fiscal reforms and innovative new policies at the subnational level belies the view that all politics is clientelistic. On the other hand, no coherent, national model of efficient or accountable "good government" has emerged in Brazil. Rather, what is more apparent is that the new federalism is a product of conflicting policy interests at different levels of government. Any determination of how Brazilian democracy has been affected by this process must consider: 1) how the sequence of democratization and decentralization interacted during the regime transition to weaken the central state; 2) how the central state has reasserted its role in areas of critical importance to democracy such as macroeconomic policy and distribution of wealth; and 3) why some subnational governments have developed innovative policies while others remain inept vehicles of patrimonial leadership.

The Politics of the Governors and the Fragmentation of the Brazilian Central State

Political decentralization was both a tool of the larger project to initiate and deepen democratization and it was an outcome of the redistribution of access to power that was at the heart of the transition to democracy.⁽⁶⁾ Both processes, however, did not overlap perfectly. The military initiated significant reforms that decentralized fiscal resources to subnational government well before the generals began liberalizing the regime. Political decentralization accelerated before a democratically elected executive and a national constitutional system could be constructed. Consequently, local politicians and their clientelistic political machines would have an inordinate amount of influence over the content of democratic institutions and national policies. Forced by the need to make alliances with subnational politicians, the first civilian government of President José Sarney (1985-1990), played to this panoply of parochial interests to strengthen his own political position. As a result, the central state's capacity fragmented under the strain of particularistic politics.

The Brazilian military, which embraced a formula of authoritarian centralism after the coup of 1964, decentralized new taxes and intergovernmental transfers through the 1967 fiscal reform. The act benefitted the northern and northeastern states where the military found a base of conservative support. In this way, the 1967 reform served to counterbalance the power of the southern and southeastern states where much of the opposition to authoritarianism first emerged (Selcher 1990). Yet it also set a precedent for expanding subnational autonomy in perverse ways.

The reform allowed the state governments to administer the new Imposto sobre Circulação de Mercadorias (Tax on the Circulation of Goods - ICM) and the Tax on the Exchange of Real Estate.⁽⁷⁾ The federal government retained the other major tax on business, the Imposto sobre Produtos Industrializados (Tax on Industrial Products - IPI), another value-added tax. Since ICM was a value-added tax on goods (and eventually, services, thus ICMS), differences in ICMS rates across states created administrative confusion that aided tax evaders and allowed states to grant tax concessions in competition with each other. The latter tendency was partially controlled by the military government. Although the states retained new fiscal resources, the military governments controlled how the states and cities spent transferred resources, linking transfers to specific budget items (Arretche 1996, 52). States seeking to increase expenditures were forced to depend upon the accumulation of debt to finance local development. By one estimate, in 1981 over one-third of Brazil's total external indebtedness was held by the state governments (Graham 1990, 87).

The combination of the fiscal crisis of the state during the 1980's, persisting disparities among the Brazilian regions and the growing inability of authoritarian institutions to deal with these problems added to the forces behind the gradual transition to democracy. Many of these questions became the focus of distributional conflicts that drove the emerging federal framework (Afonso 1995, 321; Sallum 1996). As a result, the new federal order would be created on an ad hoc basis, led by intergovernmental conflict and divorced from any national program of decentralized democracy with demarcated rules and timetables.

Although the liberalization of the authoritarian regime was initiated by moderates within the Brazilian military, the process was propelled by the emergence of civilian leadership in state government and the federal legislature. One of the defining events of the liberalization process occurred during the 1982 governors' election which saw the stunning defeat of the pro-military party (Partido Democrático Social/Social Democratic Party - PDS) and the resounding victory of opposition parties in nine states including the economically developed states of São Paulo, Rio de Janeiro, Minas Gerais, Espirito Santo and Paraná. Several of these governors had been nationally prominent during the liberalization of the regime in the post-1974 period and were often mentioned as potential presidential candidates. Tancredo Neves in Minas Gerais and Leonel Brizola in Rio de Janeiro headed the list. Without a directly elected president, these subnational politicians were popularly identified as the leaders of the democratic transition (Kugelmas et al. 1989).

The persistence of centralized authoritarian control over the process of democratization from above and the surprising acceleration of civilian, non-PDS, penetration of the process from below in key states created a host of contradictions. Prominent among these contradictions was the continued control that the central state had over taxation and expenditures. Democratic leaders in the state governments could not, on the one hand, claim to be independent of the military government and, on the other hand, continue to depend upon the same government for fiscal resources. This obvious paradox catalyzed a movement at the municipal and state level in 1983 and 1984 to expand the size of and the states' control over the ICMS (Affonso 1988).⁽⁸⁾

In preparation for the fight, the governors and the opposition municipal and federal representatives moved to expand their local bases of support. These politicians mobilized partisans through broad distributional and clientelistic networks at the subnational level (Abrúcio 1994; Samuels & Abrúcio 1997; Hagopian 1996). Leonel Brizola, for instance, was a master in the use of public spending to consolidate his support among the urban poor in Rio de Janeiro.

National politicians strengthened subnational clientelistic networks by soliciting their support through the distribution of patronage. President Sarney attempted to woo the governors, federal deputies, and mayors by increasing non-tax federal transfers to the states and municipalities 23 percent from their 1980 level. In this way, the weak president cultivated support among Brazil's governors and their local patrimonial machines in his campaign for a five-year term and the preservation of presidential rule on the eve of the drafting of the 1988 Constitution (Serra & Afonso 1991, 49).

By the time the Constituent Assembly, the elected body charged with the responsibility of drafting the new Constitution, first convened in 1987, regional interests were all but dominant in Brazilian political society. The Assembly itself became a forum for subnational interests. These regional lobbies (bancadas regionais) were strengthened by the support of the governors and by demands at the local level for fiscal resources needed to redress social and economic problems. The states of the Northeast and North formed political alliances in favor of redistributive tax rules aimed at correcting regional disparities in wealth. Leaders from these states condemned Brasília's policies as reinforcing the continued concentration of industry in São Paulo (Rodriguez 1995, 437). In response, politicians from the Southeastern and Southern states moved to block these efforts and expand their access to fiscal resources (Campos 1991).

Whatever their differences, the aggregate effects of the influence of the bancadas regionais were felt in the Constitution of 1988 which made Brazil a more fiscally decentralized federalism.⁽⁹⁾ As Table 1.1 demonstrates, fiscal decentralization in Brazil decreased under authoritarianism but gradually accelerated during the transition. This phase was marked by the shifting of significant fiscal authorities to subnational government, especially municipalities. Increasing control over ICMS by the states explains the sharp recovery of their administration of tax revenues during the mid to late 1980's. The figures which include revenue sharing also show significant gaps between tax collection responsibilities and public spending. Although the central state spent less than it collected in taxes between 1960 and 1994, the gap widened significantly following the 1988 Constitution. The municipalities were the big winners as they were able to spend several times more than they collected.

The 1988 Constitution served to lock-in these fiscal commitments to the states and municipalities by endowing them as democratic "rights." For the governors and their supporters, the Constituent Assembly, "the representatives of the people," had bestowed upon the states and municipalities greater financial autonomy. This made fiscal decentralization a central aspect of the consolidation of Brazilian democracy (Sola 1995, 40).

Beyond having greater control over public revenues, the 1988 reform gave the states reasons to spend by delegating areas of policy not within the jurisdiction of the national state. The open-endedness of this provision granted the states virtual carte blanche for directing spending at an array of "development" needs. This tendency was reinforced with large flows of uncompromised intergovernmental transfers. More than 90 percent of these fiscal transfers went unearmarked by Brasília (Afonso 1994, 356; Affonso 1995, 66). More than 30 percent of these transfers were considered "free" since they could be applied generically to elementary education and health, areas that could be more specifically defined by state and municipal governments (Afonso 1995, 318; Shah 1991).

The open-endedness of subnational fiscal authority, however, was also accompanied by the declining role of the federal government in social and economic policy making during the 1980's and 1990's. In the wake of flagging federal efforts, states and municipalities were forced to assume greater responsibilities. Some Brazilian specialists have argued that the Constitution of 1988 provided the states and municipalities with more fiscal resources but it failed to also devolve official duties de jure to these levels of government. However, data presented by Affonso (1995, 65) on the de facto effects of decentralization show that the state and municipal governments presently account for close to 80 percent of all public investment and 67 percent of current public consumption, excluding expenditures on national public firms. Medici (1994, 63) notes that while federal spending on social assistance programs declined from 48.1 percent of the total in 1980 to 39.7 percent in 1990, the state governments' commitments in this area rose during the same period from 40.8 percent of the total to 44.6 percent. Municipal spending on social assistance also increased, from 11.1 percent of the total to 15.7 percent. Presently, subnational government now accounts for 44 percent of all spending on health, 69 percent of all spending on education, and represents investments two times the size of those of the federal government (Affonso 1997, 15; Rezende 1995, 242-243; Medici 1995).

Although these figures suggest that de facto policy responsibilities were shifted to subnational governments along with resources, they do not specify whether the increase in spending was indicative of improvements to the quality of services. The figures also do not show whether increased spending did not simply fund unsustainable public payrolls and local rent-seeking. In order for these increases in social and educational spending to have value they would have to be accompanied by subnational fiscal reforms and a political dedication to improve public efficiency.

Given the historical record reviewed thus far, how did political decentralization and democratization affect each other? On one level, political decentralization was a pro-democratic force since it attacked authoritarian centralism. At the same time, democratization itself accelerated political decentralization. With the elections of 1982, subnational politicians were armed with a level of democratic legitimacy that they could effectively use to capture new authorities from the military and defend distributional gains as "rights" in the new democratic regime.

On another level, however, political decentralization created many predictable problems for the new Brazilian democracy. The sequence of regime transition and decentralization strengthened the governors as vanguards of democratization, but it also empowered local clientelism as the base of power in the new democracy. Subnational clientelism fragmented the political order, an outcome most clearly reflected in the influence of the bancadas regionais in the 1987 Constituent Assembly.

Decentralization also fragmented the fiscal order, expanding the tax collection and spending authorities of state and municipal governments. This did not make Brazil an unusual case of fiscal federalism, but it did open a Pandora's box in that subnational government could tax and spend with abandon. Although the states and municipalities assumed greater responsibilities over an array of social and economic policies, the patrimonial politics of the Brazilian governors and their tendency to reward political supporters through public expenditures threatened to create severe problems for the central state's attempts to rein in public spending and address macroeconomic instability. Recovering the capacity of the central state in fiscal management during the 1990's became one of the central struggles in reshaping the state in ways that would reinforce Brazilian democracy.

Reequilibration of Intergovernmental Authorities

As part of the game of distributive conflicts that marked the process of political decentralization and democratization in Brazil, critical areas of macroeconomic and social policy were imprudently devolved to subnational government. Without reequilibration, macroeconomic chaos and persisting social disparities promised to threaten the new democracy. Such reequilibration of intergovernmental fiscal authority depended upon an array of ad hoc political tactics by central state managers, shifting subnational interests, and changing economic contexts.

Distributive conflicts between the federal government, especially the central bank, and the state banks under the administration of the governors, were at the center of larger struggles over which levels of government would control macroeconomic policy. Besides the national public banks in Brazil - the Central Bank, the Banco do Brasil and the Caixa Econômica Federal (CEF) - the states administer their own public banks, some of which are development banks. After the 1988 Constitution, the right of the state governments to administer their banks was consecrated by their own state constitutions. These provisions prohibited the governors from "allowing" the Central Bank or the Conselho Monetário Nacional (National Monetary Council - CMN), the institutions designed to regulate the financial system, from intervening in the management of the state banks. These institutions protected what were, in practice, critical political assets of the Brazilian governors. Any attempt by President Sarney to wrest control of these resources would undermine his own base of political support (Sola 1994, 165).

During the 1980's the governors obligated the state banks to finance the expanding fiscal deficits of the states. They often did so by issuing short-term, high-interest debt paper in domestic financial markets. The poor management of many of these banks and the equally inferior administration of the state budgets by the governors added to the deteriorating fiscal condition of the state governments. These processes worsened with democratization as the governors employed the state banks to distribute public patrimony to political allies.⁽¹⁰⁾ As the fiscal conditions of the state governments worsened, the Central Bank was forced to bail out the states annually in response to political pressures placed on national politicians by the governors and their Secretaries of Economy (Silva & Costa 1995, 270-272; Kugelmas et al. 1989). Thus, during the 1980's, a vicious circle developed in the fiscal order: the central state cut its budget to help deal with the external debt while the states spent irresponsibly and then rolled over their debt to the Central Bank, worsening the central state's fiscal woes.

A secondary macroeconomic contradiction developed as the state banks issued currency to finance state spending (seigniorage), a practice that added to astronomic inflation rates and helped scuttle consecutive national anti-inflation plans. Lourdes Sola (1995, 48) insightfully describes this contradiction as the "regional fragmentation of monetary authority" and the concomitant weakening of the Central Bank and the CMN. This vicious financial and monetary circle within the state (ciranda financeira e monetária) spun out of control in the 1980's, generating a total $57 billion of debt among the states by 1991. This amount was almost half Brazil's total external debt and presented a clear threat to maintaining the financial stability of the new democracy.

Reasserting Central State Authority Over the Financial System

The dilemma presented by the ciranda showed signs of improvement during the early 1990's and only because of a conjunction of different factors. First, the national monetary authorities of the Fernando Collor administration (1990-92) forced the state banks to initiate adjustment programs under threat of central bank intervention.⁽¹¹⁾ In 1990, state governments were prohibited from rolling over their debt to the Banco do Brasil, a restriction that effectively scared away investors from state debt paper and placed tremendous pressures on the governors to either support federal tax reform efforts and state bank restructuring or face bankruptcy (Lagemann 1995, 336).

This game of chicken between central and subnational governments improved the overall liquidity of the state banks after 1991, but it was not enough. Total liquidity minus debts and other obligations remained negative through the mid 1990's. By 1992, the debt of state banks represented almost twice their total liquidity. Financial pressures on state banks increased with the advent of lower annual inflation rates under the Real Plan in 1994. Reduced inflation squeezed the profits of the state banks, making financial adjustment more immediate and increasing the leverage of the Central Bank. In the wake of these events, full enforcement of state constitutional provisions that defended the state banks from Central Bank intervention became politically and fiscally unsustainable.

The election of Fernando Henrique Cardoso and the financial impact of the Real Plan represented the high point for the recovery of central state leverage over the subnational financial system. As many state banks began to falter and state governments became increasingly dependent upon federal bailouts, Cardoso employed the leverage these factors accorded the federal government to force the governors to implement financial and fiscal reforms, privatize their utility companies, restructure and/or sell off their state banks, and initiate civil service reforms in return for federal help (Affonso 1997, 25).

Cardoso's pressures were reinforced by supportive federal legislation designed to coerce and cajole subnational governments into reducing bloated state and municipal payrolls. Empowering legislation called the Lei Camata required that states and municipalities limit payroll expenditures to 60 percent of net revenues.⁽¹²⁾ Under the new law, the federal government could even threaten to withhold federal transfers and loans from recalcitrant states.

Although the 1988 Constitution forbade the old practice of using state banks to finance public expenditures, only the threatened or actual intervention of the Central Bank in the state banks during the 1990's provided a stopgap against an unsustainable subnational spending spree. Other federal efforts kept the 1988 reform from generating additional fiscal problems. Losses of federal shares of tax receipts were partially compensated by increases in federal taxes not subject to transfers to the states or municipalities. Substantial increases in collections of one of these sources of federal revenue - social security taxes - accounted for almost 40 percent of total federal income in 1994. Additionally, tax evaders were prosecuted more vigorously. The federal tax collection bureau (Receita Federal) was restructured and modernized, making it a more efficient state agency. Due to countermeasures such as these, the federal government has actually seen a net increase in its total tax receipts over the last few years.

The Fiscal War

While certain aspects of federal regulation of state spending were accentuated, other aspects of centralized control dwindled. The Constitution of 1988 impeded the federal government from employing state tax incentives as it had in the 1970's and transferred those powers to the state governments. As a result, fiscal incentives (mostly of ICMS) soon became the main weapon in attracting new, large investors away from other states, particularly São Paulo. The use of this mechanism by one state was often seen by neighboring states as an offensive that needed to be answered in kind. The Brazilian popular press described the competition as a "Hobbesian fiscal war" among the states for new investors looking to reduce their start-up and capital costs.

Federal governments operating in the wake of the 1988 Constitution had not been oblivious to the impending problem of the "fiscal war." The Constitution empowered the National Council of Fiscal Policy (Conselho Nacional de Política Fazendária - CONFAZ), an organization under the command of the Ministry of the Economy and composed of all the state secretaries of the economy, to regulate state fiscal incentives. By law no new fiscal incentive scheme could be implemented unless the members of CONFAZ agreed to it unanimously. Central state managers thought that the unanimity requirement would discourage the proliferation of state incentive programs, but, in practice, state governments often ignored CONFAZ guidelines. Some states even bartered their votes on the council to states proposing incentive schemes in exchange for approval of their own programs (Affonso 1995, 60). In 1993, as part of the review of the 1988 Constitution, Fernando Henrique Cardoso, then President Itamar Franco's Minister of Finance, attempted to restructure the taxation powers of the states and municipalities. Specifically, Cardoso sought to reduce constitutionally mandated federal transfers to subnational governments. But the reform ultimately floundered as the governors and mayors successfully defended the constitutional framework (Weyland 1996, 122).

The explosion of the fiscal war during the 1991-1995 period created a series of problems for the tax system, for the internal market and for the fiscal stability of the state governments themselves. The proliferation of different ICMS schemes among the states made the job of regulating each state program more difficult. The situation was made worse by routine flouting of CONFAZ requirements (Afonso 1995, 320). The internal market was distorted by the distinctions among state tax systems. Products of similar quality produced by similar firms in two different states often did not have similar prices due to distinctions in ICMS. Predictably, these distortions also made macroeconomic stabilization difficult in that prices could not be regularized nationally. Finally, many state governments were encouraged to grant ever more attractive fiscal incentives to counteract the schemes of competing states. In some cases, large tax exemptions were granted by states already hampered by fiscal problems.⁽¹³⁾

Despite these alarming trends, the overall fiscally destabilizing effects of the "fiscal war" have been greatly exaggerated. First, the argument that state tax incentives based on ICMS forced down tax revenues is based on mixed evidence. Long-range data show that fiscal incentives based on ICMS did not reduce total tax receipts. According to Lagemann (1995, 340-343), the decline in ICMS receipts during the 1990-93 period was not due to the "fiscal war" but to a decline in the value of retail sales, a condition directly linked to the tremendous rise in inflation rates and the economic recession. In addition, the elimination of indexation, the rise of financial speculation and widespread tax evasion also undercut ICMS receipts. All these conditions emanated from the gradual failure of Collor's inflation control schemes. Conversely, 1994 showed a rise in ICMS receipts precisely because the Real Plan stabilized retail prices, reduced the attractiveness of financial speculation and the Receita Federal and the state governments cracked down on tax evaders.

Additionally, a number of informal constraints limited the "Hobbesian" nature of the fiscal war. During the 1990's, most state governments became unwilling to devastate their own fiscal accounts for want of resources to implement social and economic policies. In fact, the most active "warrior states" in the fiscal war advanced the most in correcting for the poor budget administration of the past, privatizing state banks and utility companies, shrinking the civil service and cracking down on tax evaders. Rio de Janeiro, once a reckless exemptor of ICMS under Governor Leonel Brizola (1990-94), pioneered subnational privatization with the sale of Banerj, Rio's decaying state bank; Cerj, the state's electricity distributor and Companhia Estadual de Gás and Riogás, both gas distributors. Rio also led the way in reducing redundant bureaucratic staffs with forced and voluntary dismissal programs. Minas Gerais followed suit with its own voluntary dismissal system and sales of going concerns such as Credireal (finance) and Cemig (utilities). Fiscal reforms in Minas also stabilized public accounts and led to increases in tax collection. Similar reforms were undertaken in Rio Grande do Sul and São Paulo.

In many of these states, fiscal incentive schemes continued to be used selectively as part of new subnational industrial policies, but their fiscal sustainability became a key factor governing their implementation. There is some preliminary evidence that states involved in the fiscal war improved their tax collection functions to compensate for tax incentives to select businesses (Affonso 1997, 18). Moreover, "warrior" states such as Minas Gerais and Rio de Janeiro became reluctant to provoke their neighbors. These states granted strategic concessions to some firms but backed off from more aggressive policies. Governor Azeredo in Minas, for example, provided Mercedes fiscal incentives in 1995 for a new plant in Juiz de Fora, but was hesitant to extend the policy to automakers wishing to move from São Paulo. Moreover, Azeredo, Vítor Buaiz, the governor of Espirito Santo, and other state leaders became active in the campaign to create new national institutions to regulate fiscal incentive schemes.⁽¹⁴⁾ In this way, the states developed incentives not to consume fiscal resources without regard for the welfare of the national fiscal structure and that of neighboring states. As subnational reforms move forward, these governments will overcome the tendency to embrace recalcitrant postures.⁽¹⁵⁾

Encouraged by growing support among the governors for new national rules, President Cardoso attempted to revisit his 1993 campaign for fiscal reform. This time, however, Cardoso faced state and municipal governments that, although not at Hobbesian odds with each other, were less inclined to form a subnational coalition against national reform after several years of "fiscal war." Once again, the context for reform had changed, allowing Cardoso to aggressively assert his proposals for institutional adjustments. In 1994, Cardoso won passage of the Social Emergency Fund (FSE), an ad hoc measure that allows the federal government to employ discretion over federal transfers constitutionally earmarked to the state and municipal governments.⁽¹⁶⁾ A similar strong arm intervention by federal authorities into subnational fiscal policy was passed in November 1996. The Kandir Law, named after Planning Minister, Antônio Kandir, removed ICMS taxes on exports to reduce Brazil's mounting trade and current account deficits. Through the FEF, the Kandir Law, the Lei Camata and transfers of additional responsibilities in health care, housing and social policy to the state and municipal governments, the Cardoso government imposed a de facto reform of the fiscal structure without reform of the 1988 Constitution. Of course, these were stopgap measures designed to rein in public spending while crucial reforms of the administrative and social security systems slouched through congress.

Social Policy

If the central state began to reassert its authority over monetary and fiscal policy in the wake of the decentralization wave, Brasília clearly forfeited its role in creating a coherent national model of redistribution to counteract historical disparities among Brazil's regions. Despite decades of industrial deconcentration - the movement of firms from urban centers to interior regions and neighboring states - and several national policy experiments,⁽¹⁷⁾ the distribution of economic activity in Brazil remains largely concentrated in the Southeast and South in general, and São Paulo in particular (Diniz 1995). The southeastern states alone account for more than 60 percent of Brazil's GDP and their per capita GDP is three times that of the poor states of the Northeast (Neto 1995, 38).

Such geographic unevenness in income registers in the social indicators. Whereas 15 percent of the population in the Southeast subsists under the poverty line, more than 50 percent of the inhabitants of the Northeast are below that marker. The life expectancy of a nordestino (northeasterner) (58) is nine years below that of a paulista (resident of São Paulo State) (67) and 15 years below that of a gaúcho (resident of Rio Grande do Sul) (73).

Predictably, the decentralization of tax authorities to the states and municipalities did little to reverse this legacy as the underindustrialized northern and northeastern states enjoyed a much smaller tax base than the industrialized Southeast and South (Rezende 1995, 250-253). In 1991, for example, the state governments of the southern regions received more than 85 percent of their revenues from state taxes, whereas the northern states collected less than 35 percent of their income from state taxes (Barrera & Roarelli 1995). The poorer states continued to depend upon federal transfers, yet even here, distribution of such monies within the depressed regions was not equitable. Instead, distribution reflected the political importance of certain states. Thus, in 1988, Maranhão, the home state of the sitting president (Sarney), received 25 percent of negotiated federal transfers to the North (Roarelli 1994).

The continued dependency of the poor northern and northeastern states on federal transfers severely limited their capacity to bargain with the federal government for national resources. More politically powerful states such as Bahia have the leverage of influential leaders like former governor, now federal senator, Antônio Carlos Magalhães, and other key politicians. Yet most of the poor states of the North and Northeast found it difficult to escape the constraints of their dependency on federal transfers and the superior leverage this imparted on Brasília to demand austerity. National reforms that force poorer states to adjust without social welfare compensation will have a regressive effect on the distribution of wealth in these regions in the years to come (Dain 1995).

The decentralization of key social policies failed to improve the financing and overall quality of public services to the poor. One example is the performance of health care policy, the most decentralized social policy in Brazil. The decentralization of health care under the Sistema Unificado e Descentralizado de Saúde (SUDS) led many states to reduce their own expenditures as new federal funds arrived. Municipal and state medical personnel also demanded the higher salaries of federal health care bureaucrats and many local politicians unscrupulously misappropriated SUDS accounts. As a result, the decentralization of health care did little to promote either the quality of preventive medicine or citizens' access to policy making (Weyland 1996, 174-175).

The reequilibration of intergovernmental authority in the wake of the decentralization wave proved ad hoc in method and uncertain in results. The worst case scenarios inspired by the fragmentation of monetary and fiscal authority during the 1980's, did not come to pass only because Brasília employed a number of stopgap measures. The use of threats, bribes, and formal and informal agreements between levels of government, as well as shifts in the interests of some subnational leaders, mellowed the fiscal war and allowed the central state to regain key monetary authorities and resources. These ad hoc adjustments demonstrated that the new federalism in Brazil could be flexible enough to avoid the most somber predictions of economic chaos. Their continued success, however, depend on flimsy foundations.

On one level, the Brazilian states and municipalities are politically powerful, especially the governors of the largest states (Abrúcio 1994), but they are also atomized into multiple states and electoral districts (513). All federal attempts in recent years to regain previously devolved authorities have benefitted from this background factor. The worsening of the financial state of state banks, state coffers, and the erosion deepened by the fiscal war of the regional alliances seen during the Constituent Assembly helped to shift political leverage back to the central state during the 1990's. The Central Bank under Collor and, later, Cardoso's reforms, extracted concessions from individual states precisely because federal authorities faced a divided set of subnational interests during the 1990's.

On another level, however, federal efforts were haphazard and uninstitutionalized. These episodes of reequilibration did not produce a coherent national model of fiscal federalism with any promise of accountable government through decentralization. On the contrary, Brazilian democracy survived the fiscal war and the regional fragmentation of monetary authority due to a reequilibration of intergovernmental authority that was decidedly contingent upon particular political events, shifting interests, and changing contexts.

The ad hoc nature of reequilibration also generated serious costs for democracy. In social policy, the Brazilian central state did little to implement coherent redistributive models at the national level. Decentralization proceeded in this area with poor results. Moreover, once Brasília regained crucial fiscal and monetary authorities, the federal government used its added leverage to impose austerity on the neediest states in the union. The reequilibration of intergovernmental authority, therefore, was selective, redesignating economic policies traditionally managed by the central state, while forfeiting federal responsibilities in distributive policies that only the central state could implement efficiently.⁽¹⁸⁾

Conversely, democratization affected the reequilibration of intergovernmental authorities in a fundamental way. By strengthening subnational governments, Brazilian democratization placed a premium on procuring the compliance of state and municipal governments with national economic policy priorities. Exerting central state leverage over subnational government, however, was only one factor in generating this "compliance." A less analyzed series of factors were based on changes in subnational political interests and in the use of innovative policies designed to address persisting social and economic problems among Brazil's states and municipalities.

After Decentralization: Innovative Policy Experiments among the Brazilian States and Municipalities

As noted above, shifts in the political interests of certain subnational governments were essential for moving national reform efforts forward. These dramatic changes in the subnational games of taxing, spending, and rolling over debt were due to new economic and social priorities in state and municipal government.

To be sure, there was plenty of evidence that patrimonialism persisted in subnational politics during the 1990's. One of the largest political scandals of recent years involved the distribution of kickbacks to important city and state officials from the sale of bond issues.⁽¹⁹⁾ Given events such as these, it is easy to conclude that the enduring rule of Brazilian politics in state and municipal governments is the persistence of clientelism and illegal political transactions.

Despite such problems, many states and municipalities demonstrated a less commonly known trait of subnational government in Brazil: innovative leadership. During the 1990's, the expanding command over policy by the governors, mayors and city councils was, in part, a product of their capacity for producing innovative responses to persisting problems. Many of these so-called "innovations from below" (novidades vindas de baixo) appeared in health, education, housing, and industrial policy, areas that had been substantially decentralized during the 1980's and 1990's. In cases such as SUDS, however, decentralization did not improve the quality of services and only reinforced the clientelistic networks seen by many as endemic in Brazilian politics. Under what conditions did the implementation of these novidades vindas de baixo serve Brazilians efficiently and under what conditions did these experiments decompose into "rent-seeking?"

Recent studies of state and municipal policy in Ceará and Minas Gerais suggest that, in these cases, social and economic policies were implemented in ways that maximized allocative efficiency. Crucial to these experiences was the way that the clientelistic tendencies of Brazilian politics were impeded by institutional innovations.

Judith Tendler's (1997) study of state and municipal programs in preventive health, public procurement, and agricultural productivity in the poor northern state of Ceará found a high level of professional dedication among implementors of these policies. The author outlines a number of reasons for the emergence of "good government" in Ceará. First, reformist governors, specifically Tasso Jereissati (1987-91 & 1995-present) and Ciro Gomes (1991-94), blocked parochial encroachments by political opponents by strengthening meritocracy in the state's civil service. Second, through supportive publicity and the distribution of awards for meritorious performance, municipal workers, those most responsible for implementing many of Ceará's social and economic policies, were given incentives to develop working relationships with their citizen "clients" based on trust. Finally, Ceará's inspired public servants cultivated multiple ties with non-governmental organizations, citizen's groups, labor unions, business specialists, and other groups. These ties established mutual monitoring networks that buttressed trust between the public sector and their citizen clients. Multitask work ("job enlargement") and a flexible problem-solving approach to service delivery increased the frequency of contact between public workers and social groups, expanding the system of mutual surveillance that kept both public and private interests from "shirking" their responsibilities. The result was a subnational policy system that proved efficient in responding to local needs.

In Minas Gerais, a state considered by some an exemplar of subnational clientelistic politics (Hagopian 1996), Montero (1997, Chapters 6-7) found a combination of political and institutional factors similar to those highlighted by Tendler in the state's industrial policies of 1989-1995. Once a prominent beneficiary of national industrial policy, especially in the public steel and mining sectors, Minas' economy fell on hard times during the 1980's as the central state became mired in a fiscal crisis. Patrimonial politics, particularly under the governorship of Newton Cardoso (1986-1990), almost destroyed the state's development agencies. Cardoso's exit and a concerted attempt by professional leaders of the Institute of Industrial Development (INDI), the state development bank (BDMG), Cemig, the state utility company, and other agencies, to redraw Minas Gerais' industrial policy in 1988 and 1989 produced new directives for the new governor, Hélio Garcia.⁽²⁰⁾ Garcia delegated authority over economic policy to a team of technocrats in the state Secretariats of the Economy and Planning. The new team emphasized policies that would reduce Minas' barriers for new investment and increase productivity among firms already in the state.⁽²¹⁾

One of the chief examples of the new mineiro industrial policy was public assistance to the Italian automaker Fiat, Minas Gerais' largest private producer. During the early 1990's Fiat attempted to launch a productivity-enhancing system that would place suppliers close to or inside the company's plant in Betim. In that way, supplying firms could manufacture and install parts directly on the line of production, obviating Fiat's costly system of maintaining an inventory of parts. Since most of Fiat's suppliers were largely inefficient Brazilian small- and medium-sized firms with little capital, the market alone could not guarantee that the multinational's new "just-in-time" system could be assembled. INDI, BDMG, Cemig, and other state agencies mobilized informational, financial, and infrastructural resources to move Fiat's suppliers. After the program had been completed, Fiat's productivity improved markedly, moving the firm from the fourth position in the Brazilian market in 1989 to number one in 1994.

As in Ceará, industrial policy in Minas Gerais was enhanced by significant political support by reformist leadership, but it was also made more efficient by an array of horizontal ties among the state's public economic agencies and secretariats and vertical ties between the public sector and private firms. Horizontal associations produced additional levels of political support by creating a broader constituency for industrial policy. These ties also provided an interdisciplinary approach to policy by linking utility companies with financial and informational resources. Vertical associations fostered mutual monitoring networks that created additional barriers to rent-seeking and reinforced trust between civil servants and firm managers. In subnational cases such as Minas Gerais, where political support was strong, and horizontal and vertical embeddedness were developed, industrial policy achieved a high level of allocative efficiency. Where the mix of political and institutional factors was either missing or malformed in states such as São Paulo and Rio de Janeiro, Montero (1997, Chapter 8) finds that subnational industrial policies failed to achieve allocative efficiencies. These policies collapsed into widespread rent-seeking and added to fiscal problems.

The cases of Ceará and Minas Gerais demonstrate that subnational politics in Brazil is not inevitably or endemically clientelistic. The comparison shows that similar political and institutional factors in different areas of policy in very distinct regions of Brazil are responsible for the successful experiences. Replicating these factors in other regions throughout Brazil will be difficult. Nevertheless, Nylen (1997 & in this volume) suggests that reformist parties such as the Workers' Party (PT) can take experiences with local-level governance and create a larger modo (petista) de governar ([PT] mode of governance), applicable to other jurisdictions. Other studies of progressive local governments in Brazil and Latin America agree that subnational innovations can be powerful models, but these experiences also depend on a sustained process of policy experimentation and learning (Jacobi 1995; Winn and Ferro-Clérico 1997). That would be encouraging news for Brazilian democracy as it would enhance the notion that decentralization creates more accountable government over time.

The role of reformist political leadership is crucial to these experiences. By reinforcing the notion that governors, mayors, and federal deputies were the leaders of the regime transition and the defenders of their regions in the Constituent Assembly, democratization enhanced the political standing of reformist governors (as well as non-reformists). The replication of experiments in "good government" will, in part, depend on whether these will become standards by which voters will judge their elected officials and demand a similar level of responsiveness.

Conclusions

The image of the new Brazilian federalism that emerges from our analysis is that of a mosaic of conflicting, flexible, and sometimes innovative, institutions and political interests. During each of the three phases of decentralization, the process generated a mixed bag of implications for the country's new democracy.

During the liberalization of authoritarianism, decentralization played an important role in accelerating, even leading, the process of regime transition. However, the sequencing of decentralization before democratization reinforced subnational clientelism. This proved especially destabilizing for the new democracy as the fragmentation of political authority and fiscal resources led to irresponsible public spending and weakened the national state's distributive functions. Furthermore, the attack on the central state's monetary and financial authorities added to Brazil's mega-inflationary woes and threatened to push the country into economic chaos.

Reequilibration of intergovernmental authority was the only possible response to the problems generated by decentralization. Yet, consistent with the persisting paradox of the new federalism in Brazil, decentralization facilitated reequilibration. Central state authorities during the 1990's such as the Central Bank and the Cardoso administration made headway on fiscal and monetary problems with a combination of positive and negative sanctions to the states and municipalities. The credibility of these sanctions increased as subnational budgets and state banks fell on hard times. Additionally, the "fiscal war" helped divide subnational interests, eroding regional alliances formed during the 1980's. In this way, the leverage of Brazilian presidents increased, clearing the way for nationally imposed austerity.

In addition to changing contexts, shifting interests among governors also played a role in reequilibrating intergovernmental authorities. Subnational efforts during the 1990's to improve public accounts and end the Hobbesian dangers of the fiscal war were critical to the success of federal attempts to reequilibrate intergovernmental authorities and remove the worst problems created by decentralization in the 1980's.

Continued reequilibration, however, rests on flimsy foundations. Current efforts rely too heavily on federal stopgap measures and continued compliance by governors and mayors. Brazil is far from the thorough reform of the federal system and the Constitution that it very much requires. The weakness of national distributive policies are an indication that the central state must still recover significant authorities and resources to face social problems that subnational government can only partially address. Although the new federalism proved itself adaptable in the face of the worse case scenarios, it has yet to prove itself as a lasting framework for strengthening the central state.

Finally, in the current phase of reversing the gains of decentralizers and passing additional policy responsibilities on to subnational government, evidence of innovative policy experiments is emerging in some parts of Brazil. Although these examples only highlight responses to local concerns and do not represent national solutions, these experiences rest on institutional and political innovations that can be replicated in other jurisdictions. Without denying the existence of patrimonialism in Brazil, these cases correct the all too facile assumption among scholars of Brazil that state and municipal politics is endemically corrupt and clientelistic.

How can the new Brazilian federalism be understood? Endemic theories tend to overstate either the clientelistic tendencies of the country's politics or imagine national systems of efficient federalism that are only partially borne out in practice. Neither approach sheds light on the mixed character of the Brazilian federal structure. This evolving process has both enhanced democracy by limiting authoritarian centralism and promoting the allocative efficiency of some subnational governments, but it has also strengthened clientelistic circles in other subnational governments, weakened national political parties, and forced recent Brazilian presidents to resort to extraordinary measures in order to recovery authorities and resources once under the exclusive purview of the central state. Conversely, democratization has itself reinforced some of these tendencies. By empowering governors and mayors as historical leaders of the Brazilian transition to democracy, these subnational politicians have renewed their command over the country's politics. If the reequilibration of intergovernmental authority rationalizes the distribution of federal powers and resources, the new Brazilian federalism will become a force for strengthening democracy and the central state upon which social equity and macroeconomic stability depend.

Table 1.1 Revenues by Level of Government, Before and After Revenue Sharing

First 4 columns: Tax Revenue Before Revenue Sharing; Last 4: Final Disposition of Revenues after Sharing

Sources: Serra & Afonso (1991, 45); Affonso (1995, 63-64); FGV (1996, 103).

1. For the purposes of this chapter, I define "democracy" as a system of rule guaranteeing the rights of elites to contest for state power, the rights of citizens to participate in and be represented in the polity, and the protection of the civil liberties of the ruled. "Democratization" refers to the process by which these guarantees are established and strengthened de jure and de facto. De facto democracy assumes a level of substantive democracy, that is, social equity and economic rights that, although not essential for establishing the existence of democracy, are crucial to the overall process of democratization.

2. Wherever subnational elections precede national elections in the democratization process, especially if those elections are the "founding" elections of the new democracy, local elites and not national rulers will design the federal structure. Invariably, they do so in ways that favor subnational government and weaken central states. See Linz and Stepan (1996, 34).

3. This chapter assumes that the existence of a weak, fragmented state is a danger to democracy as it favors the power of privileged social groups and debilitates political groups espousing more universalistic motives. For the argument, see Weyland (1996) and Weyland's contribution to this volume.

4. Public goods that must address specific local preferences (e.g., infrastructure, local industrial policy, housing, etc.) are handled more efficiently at the subnational level. The greater the diversity of local preferences across jurisdictions, the larger the benefits from decentralization. See Oates (1972).

5. How this process contrasts with the "Old Federalism" in Brazilian history is intellectually interesting but only of secondary importance to the arguments explored in this chapter.

6. Most of the historical detail of political decentralization in Brazil before, during, and after the transition to democracy cannot be discussed here. For more thorough accounts, see Graham (1987), Selcher (1990), Abrúcio (1994), and Sallum (1996).

7. The ICM improved on the state sales taxes it replaced by charging only on the added value in each stage of production. The reforms also allowed the cities to manage their own taxes, primarily the Imposto sobre Serviços de Qualquer Natureza (Tax on Services - ISS), which did not apply to the services already taxed by the ICM.

8. Although the state governments could set ICMS rates, these powers were governed by certain federal laws. For more on ICMS rates, see Longo (1994).

9. The new fiscal structure of Brazil was not atypical of fiscal federalisms in the most advanced countries. The states and municipal governments in Brazil accounted for 49 percent of public sector expenditures in 1992, a figure comparable to the United States (50.5 percent) and Germany (45.7 percent). Like other fiscal federalisms, the Brazilian national state administered progressive taxes on income while subnational governments controlled taxes on property (Lagemann 1995, 330-331). Only the aforementioned distortion of having two value-added taxes administered by two different levels of government remained a particularity of the Brazilian case.

10. This included both finance and jobs in the upper echelons of bank management. Proof of the escalating politicization of the state banks under democratization is the fact that the deterioration of their liquidity commenced a few months before the first governors' election in 1982 and intensified with the elections of 1986. See Sola (1995, 52) and Loyola (1993, 9-10).

11. For an analysis of the legislation, see IPEA (1992) and Loyola (1993).

12. The administrative reform bill, passed by the Chamber of Deputies in April 1997, gave subnational government the authority to dismiss civil employees in order to meet the Lei Camata standards.

13. The evidence for this conclusion is fragmentary. See Wall Street Journal 29 August, 1997: 1.

14. Diário do Comércio, 9 April, 1996. Some of the key governors in this movement include Cristovam Buarque (PT - Federal District), Jaime Lerner (PDT - Paraná), Dante de Oliveira (PDT - Mato Grosso do Sul), Antônio Britto (PMDB - Rio Grande do Sul) and Maguito Vilela (PMDB - Goiás).

15. This is true of reforming coalitions as initial successes attenuate opposition to change. For the argument, see Berensztein (1996).

16. FSE's name was changed to the Fiscal Stabilization Fund (FEF) after its 1995 renewal. The provision allows the federal government to hold 20 percent of constitutionally mandated transfers to the states and municipalities until "all contingencies are covered." The vagary embodied in the final phrase gives Brasília extraordinary discretion over these funds.

17. For an analysis of these policies, see Cano (1994, 232-233).

18. This is true because mobility limits the redistributive effects of subnational policy. Only national states can solve the coordination problems among subnational governments that hinder distributive policies. For a look at the limits of social policy among municipal governments in Brazil that is based on this problem, see Abrúcio and Couto (1996) and Souza (1996).

19. The full complexity of this still unfolding case cannot be treated here. For a more complete assessment, see Veja 5 March, 1997: 22-29.

20. Garcia had been governor of Minas Gerais briefly between 1984 and 1986 while Tancredo Neves, the elected governor, ran for the presidency.

21. The most important statement of this new policy in Minas Gerais can be found in Minas Gerais (1989).

Abrúcio, Fernando Luiz. Os Barões da Federação: O Poder dos Governadores no Brasil Pós-Autoritário. Unpublished master's thesis. Universidade de São Paulo, 1994.

Abrúcio, Fernando Luiz and Cláudio Gonçalves Couto. "A Redefinição do Papel do Estado no Âmbito Local." São Paulo em Perspectiva 10, no. 3 July/September (1996): 40-47.

Affonso, Rui de Britto Álvares. "Decentralização e Crise Federativa: A Especificidade do Brasil." Paper presented at the XX International Congress of the Latin American Studies Association, Guadalajara, Mexico, April 17-19, 1997.

______. "A Federação no Brasil: Impasses e Perspectivas." In A Federação em Perspectiva: Ensaios Selecionados, eds., Rui de Britto Álvares Affonso and Pedro Luiz Barros Silva. São Paulo, SP: FUNDAP, 1995.

______. Federalismo Tributário e Crise Econômica: Brasil 1975-1985. Masters' Thesis. Campinas, SP: Instituto de Economia, UNICAMP (Universidade Estadual de Campinas), 1988.

Afonso, José Roberto Rodrigues. "A Questão Tributária e o Financiamento dos Diferentes Níveis de Governo." In A Federação em Perspectiva: Ensaios Selecionados, eds., Rui de Britto Álvares Affonso and Pedro Luiz Barros Silva. São Paulo, SP: FUNDAP, 1995.

______. "Descentralização Fiscal: Revendo Idéias." Ensaios FEE 15, no. 2 (1994): 353-389.

Ames, Barry. "The Reverse Coattails Effect: Local Party Organization in the 1989 Brazilian Presidential Election." American Political Science Review88, no. 1 (March 1994): 95-111.

Arretche, Marta T. S. "Mitos da Descentralização: Mais Democracia e Eficiência nas Políticas Públicas?" Revista Brasileira de Ciências Sociais 11, no. 31 (June 1996): 44-66.

Barrera, Aglas Watson and Maria Liz de Medeiros Roarelli. "Relações Fiscais Intergovernamentais." In Reforma Tributária e Federação, eds., Rui de Britto Álvares Affonso and Pedro Luiz Barros Silva. São Paulo, SP: FUNDAP, 1995.

Barzelay, Michael. The Politicized Market Economy: Alcohol in Brazil's Energy Strategy. Berkeley, CA: University of California Press, 1986.

Berensztein, Sergio. "Rebuilding State Capacity in Contemporary Latin America: The Politics of Taxation in Argentina and Mexico." In Latin America in the World Economy, eds., Roberto Patricio Korzeniewicz and William C. Smith. Westport, CT: Greenwood Press, 1996.

Campos, H. L. Os Interesses Regionais na Assembléia Nacional Constituente. Masters' thesis. Campinas, SP: UNICAMP, 1991.

Cano, Wilson. "Concentración, desconcentración y descentralización en Brasil." In Territorios en Transformación: Análisis y Propuestas, eds., José Luis Curbelo, Francisco Alburquerque, Carlos A. de Mattos and Juan Ramón Cuadrado. Madrid, Spain: Fondo Europeo de Desarrollo Regional (FEDER), 1994.

Dain, Sulamis. "Conclusão." In Reforma Tributária e Federação, eds., Rui de Britto Álvares Affonso and Pedro Luiz Barros Silva. São Paulo, SP: FUNDAP, 1995.

Diário do Comércio. 9 April (1996).

Diniz, Clélio Campolina. "Dinâmica Regional Recente e Suas Perspectivas." In A Federação em Perspectiva: Ensaios Selecionados, eds., Rui de Britto Álvares Affonso and Pedro Luiz Barros Silva. São Paulo, SP: FUNDAP, 1995.

Diniz, Eli. Voto e Máquina Política: Patronagem e Clientelismo no Rio de Janeiro. Rio de Janeiro, RJ: Paz e Terra, 1982.

Evans, Peter B. "Predatory, Developmental and Other Apparatuses: A Comparative Political Economy Perspective on the Third World State." Sociological Forum 4, no. 4 (December 1989): 561-587.

FGV (Fundação Getúlio Vargas). A Economia Brasileira em Gráficos. Rio de Janeiro, RJ: FGV, 1996.

Graham, Lawrence S. The State and Policy Outcomes in Latin America. New York, NY: Praeger, 1990.

______. "The Role of the States in the Brazilian Federation." In Subnational Politics in the 1980s: Organization, Reorganization, and Economic Development, eds., Louis A. Picard and Raphael Zariski. New York, NY: Praeger, 1987.

Hagopian, Frances. Traditional Politics and Regime Change in Brazil. New York,

NY: Cambridge University Press, 1996.

Inter-American Development Bank (IDB). Economic and Social Progress in Latin America, 1997 Report. Latin America After a Decade of Reforms. September. Washington, D. C.: IDB/Johns Hopkins University Press, 1997.

______. Economic and Social Progress in Latin America, 1994 Report. Special Report: Fiscal Decentralization. October. Washington, D. C.: IDB/Johns Hopkins University Press, 1994.

IPEA (Instituto de Pesquisa Econômica Aplicada). Analise Econômico-Financeira dos Bancos Estaduais. Working Paper, January, 1992.

Jacobi, Pedro. "Alcances y límites de los gobiernos locales progresistas en Brasil: Las alcadías petistas." Revista Mexicana de Sociología 2 (1995): 143-162.

Kaufman, Robert R. "Stabilization and Adjustment in Argentina, Brazil, and Mexico." In Economic Crisis and Policy Choice: The Politics of Adjustment in the Third World, ed., Joan M. Nelson. Princeton, NJ: Princeton University Press, 1990.

Kugelmas, Eduardo, Brasílio Sallum Jr., and Eduardo Graeff. "Conflito Federativo e Transição Política." São Paulo em Perspectiva 3, no. 3 July/September (1989): 95-102.

Lagemann, Eugênio. "O Federalismo Fiscal Brasileiro em Questão." In A Federação em Perspectiva: Ensaios Selecionados, eds., Rui de Britto Álvares Affonso and Pedro Luiz Barros Silva. São Paulo, SP: FUNDAP, 1995.

Linz, Juan J. and Alfred Stepan. Problems of Democratic Transition and Consolidation: Southern Europe, South America, and Post-Communist Europe. Baltimore, MD: Johns Hopkins University Press, 1996.

Longo, Carlos Alberto. "Federal Problems with VAT in Brazil." Revista Brasileira de Economia 48, no.1 January/March (1994): 85-105.

Loyola, Gustavo. "Bancos Públicos Estaduais - Origens da 'Questão BE's e Perspectivas do Setor." Serie Pronunciamentos Nš 6. Brasília, DF: Banco Central do Brasil, 1993.

Mainwaring, Scott. "Brazilian Party Underdevelopment in Comparative Perspective." Political Science Quarterly 107, no. 4 (1992-93): 677-707.

Medici, André Cezar. "Políticas Sociais e Federalismo." In A Federação em Perspectiva: Ensaios Selecionados, eds., Rui de Britto Álvares Affonso and Pedro Luiz Barros Silva. São Paulo, SP: FUNDAP, 1995.

______. A Dinâmica do Gasto Social no Brasil nas Três Esferas de Governo: Uma Análise do Período 1980-1992. São Paulo, SP: IESP, 1994.

Montero, Alfred P. Shifting States in Uneven Markets: Political Decentralization and Subnational Industrial Policy in Contemporary Brazil and Spain. Ph. D. Dissertation, Political Science, Columbia University, 1997.

Musgrave, Richard A. and Peggy B. Musgrave. Public Finance in Theory and Practice. 5th Ed. New York, NY: McGraw-Hill, 1989.

Neto, Leonardo Guimarães. "Desigualdades Regionais e Federalismo." In Desigualdades Regionais e Desenvolvimento, eds., Rui de Britto Álvares Affonso and Pedro Luiz Barros Silva. São Paulo, SP: FUNDAP, 1995.

Novaes, Carlos A. Marques. "Dinâmica Institucional de Representação." Novos Estudos CEBRAP 38 March (1994): 99-147.

Nylen, William. "Reconstructing the Workers' Party (PT): Lessons from North-Eastern Brazil." In The New Politics of Inequality in Latin America: Rethinking Participation and Representation, eds. Chalmers, Douglas A., Carlos M. Vilas, Katherine Hite, Scott B. Martin, Kerianne Piester, and Monique Segarra. New York, NY: Oxford University Press, 1997.

Oates, Wallace E. Fiscal Federalism. New York, NY: Harcourt Brace Jovanovich, 1972.

Putnam, Robert D. Making Democracy Work: Civic Traditions in Modern Italy. Princeton, NJ: Princeton University Press, 1993.

Rezende, Fernando. "O Financiamento das Políticas Públicas: Problemas Atuais." In A Federação em Perspectiva: Ensaios Selecionados, eds., Rui de Britto Álvares Affonso and Pedro Luiz Barros Silva. São Paulo, SP: FUNDAP, 1995.

Rodriguez, Vicente. "Federalismo e Interesses Regionais." In A Federação em Perspectiva: Ensaios Selecionados, eds., Rui de Britto Álvares Affonso and Pedro Luiz Barros Silva. São Paulo, SP: FUNDAP, 1995.

Roarelli, Maria Liz de Medeiros. Análise das Transferências Negociadas por Órgão e Esfera de Governo. São Paulo, SP: FUNDAP/IESP, 1994.

Samuels, David Julian and Fernando Luiz Abrúcio. "The New Politics of the Governors: Subnational Politics and the Brazilian Transition to Democracy." Paper prepared for annual meeting of the International Political Science Association, Seoul, South Korea, August 1997.

Sallum Jr., Brasílio. "Federação, Autoritarismo, e Democratização." Tempo Social: Revista de Sociologia da USP 8, no. 2 (1996): 27-52.

Selcher, Wayne A. "O Futuro do Federalismo na Nova República." Revista de Administração Pública 24, no. 1 (1990): 165-190.

Serra, José and José Roberto R. Afonso. "Trajetória e Mitos." Conjuntura Econômica October (1991): 44-50.

Shah, Anwar. The New Fiscal Federalism in Brazil. World Bank Discussion Papers Nš 124, 1991.

Silva, Pedro Luiz Barros and Vera Lúcia Cabral Costa. "Decentralização e Crise da Federação." In A Federação em Perspectiva: Ensaios Selecionados, eds., Rui de Britto Álvares Affonso and Pedro Luiz Barros Silva. São Paulo, SP: FUNDAP, 1995.

Sola, Lourdes. "Estado, Regime Fiscal e Ordem Monetária: Qual Estado?" Revista Brasileira de Ciências Sociais 10, no. 27 February (1995): 29-60.

______. "The State, Structural Reform, and Democratization in Brazil." In Democracy, Markets, and Structural Reform in Latin America: Argentina, Bolivia, Brazil, Chile, and Mexico, eds., William C. Smith, Carlos H. Acuña, and Eduardo A. Gamarra. New Brunswick, NJ: Transaction Publishers, 1994.

Souza, Maria do Carmo Campello de. Estado e Partidos Políticos no Brasil (1930 a 1964). São Paulo, SP: Editora Alfa-Omega, 1976.

Souza, Celina. "Reinventando o Poder Local: Limites e Possibilidades do Federalismo e da Descentralização." São Paulo em Perspectiva 10, no. 3 July/September (1996): 103-112.

Tendler, Judith. Good Government in the Tropics. Baltimore, MD: Johns Hopkins University Press, 1997.

Veja. "Número alto em lama profunda." 5 March (1997): 22-29.

Von Mettenheim, Kurt. The Brazilian Voter: Mass Politics in Democratic Transition, 1974-

1986. Pittsburgh, PA: University of Pittsburgh Press, 1995.

Wall Street Journal. "Smokestack Samba: Brazil's Poor Northeast is Attracting Industry, But at a Heavy Price." 29 August (1997): 1.

Weyland, Kurt. Democracy Without Equity: Failures of Reform in Brazil. Pittsburgh, PA: University of Pittsburgh Press, 1996.

Winn, Peter and Lilia Ferro-Clérico. "Can a Leftist Government Make a Difference? The Frente Amplio Administration of Montevideo, 1990-1994." In The New Politics of Inequality in Latin America: Rethinking Participation and Representation, eds. Chalmers, Douglas A., Carlos M. Vilas, Katherine Hite, Scott B. Martin, Kerianne Piester, and Monique Segarra. New York, NY: Oxford University Press, 1997.

World Bank. Decentralization in Latin America: Learning through Experience. Washington, D. C.: The World Bank, 1997.