The Deadlock of Democracy in Brazil (2001) PDF - Introduction

Introduction Imagine the following puzzle: A formally democratic nation confronts, over many years, crises of in×ation, government waste and corruption, pension sys- tem deµcits, inadequate social services, violence, and social inequality. Sub- stantial majorities of the population support proposals dealing with these crises. In the legislature, few parliamentarians oppose the proposals because of prin- ciples or voter pressure. And yet these proposals rarely emerge unscathed from the legislative process. Many, because they have no chance of passage, never arrive at the Congress’s door. Others die in committees. Some proposals ulti- mately win approval, but long delays and substantive concessions weaken their impact. Rarely can the president avoid paying a high price, in pork and patron- age, for legislative support. This puzzle characterizes the past µfteen years of politics and policy-mak- ing in Brazil, Latin America’s largest democracy. Brazil is often described as a nation where governability is a permanent problem. Governability is the sort of hot topic whose meaning is hard to pin down, but at its core lie two political processes. One involves the efµciency of a nation’s executive and legislative branches in the making of programs and policies; the other relates to the gov- ernment’s ability to implement these programs and policies. This book ad- dresses the policy-making aspects of governability in Brazil. More precisely, the book explores the relationship between Brazil’s national political institu- tions, especially the rules and practices of electoral and legislative politics, and the probability that the central government will adopt new programs and poli- cies. Although the empirical analysis centers on the past µfteen years of Brazil- ian politics along with the last years of military rule, the explanatory ideas and theories come from the broader literature of contemporary political science, and the results of the investigation have implications for developed and developing countries alike. To comprehend more concretely Brazil’s governability crisis, consider just the most recent presidential administration. When Fernando Henrique Cardoso assumed Brazil’s presidency in early 1995, his prospects seemed extremely fa- 2 The Deadlock of Democracy in Brazil vorable. The new president was credited with authoring the Plano Real, an eco- nomic program that had stabilized the economy and lifted millions out of poverty. The fµve parties backing Cardoso’s election (some only in the second round) included more than 400 deputies, easily sufµcient to pass regular legis- lation and even enough to amend the Constitution. Leftist opposition to the ad- ministration was in disarray, utterly demoralized and with no credible alterna- tive program. And the president himself was no lightweight; indeed, a distinguished foreign historian claimed that Cardoso would “arguably be the most intellectually sophisticated head of any contemporary state” (Anderson 1994, 3). With such an auspicious—and unusual—beginning, Cardoso’s pro- gram should sail through the Congress, and Brazil could begin an assault on its central problems: an oppressive and costly state apparatus, economic inefµciency, and widespread poverty.1 Has the Cardoso administration, now six years in ofµce, lived up to its heady prospects? Policy successes, mainly in the area of economic reform, surely exist. Liberalization, following the path of Cardoso’s predecessor, Fer- nando Collor de Mello, has proceeded apace, with signiµcant areas of the econ- omy opened to foreign investment, major state enterprises sold, and trade lib- eralized (Kingstone 1999). In other policy areas, however, progress has been slow and uneven. Congress approved a constitutional amendment allowing re- election for presidents, governors, and mayors. Congressional assent, however, came only after the executive branch doled out pork-barrel inducements and pa- tronage to signiµcant numbers of deputies. In addition, revelations about vote buying suggested that some governors had literally bribed deputies to support reelection in exchange for control over crucial executive appointments in their states (Kramer 1997). By the end of 1998, pension and administrative reform had been approved, but both had languished in the Congress for years, and neither passed without substantial concessions from the administration. Tax reform, long regarded as a centerpiece of economic modernization, had disap- peared from the executive agenda. The slow progress of pension and adminis- trative reforms, coupled with the absence of anything resembling a new tax pol- icy, had real consequences. In the massive withdrawals of foreign capital triggered by the Asian crisis in the late summer of 1998, foreign investors and bankers used these policy failures to justify their concerns over Brazil’s eco- nomic program, and their contribution to the public sector deµcit forced the government into an even harsher and more recessionary stabilization program. 1. For a summary of Cardoso’s career and political ideas, see Resende-Santos 1997. Introduction 3 Cardoso’s inability to move his program quickly through the Congress can- not be blamed either on a lack of solid public support or on principled legisla- tive opposition. In all these policy areas (with the possible exception of the re- election amendment), substantial majorities of the population supported the president’s reforms. And as political scientist Bolivar Lamounier pointed out, no alternative proposals competed for the Congress’s support (“Soltando as Amarras” 1997). If Cardoso, playing such favorable cards, has had this much trouble ad- vancing his legislative agenda, imagine the situation confronting a “typical” president. Brazilian executives usually lack even nominal congressional ma- jorities but instead depend on deputies mainly interested in their own fortunes, in local pork, or in the defense of narrow interests, and they face publics dra- matically dissatisµed with governmental performance at all levels. The past µfteen years of Brazilian democratic politics, coupled with the pluralist experience of 1946–64, indicate that the nation’s political institutions create a permanent crisis of governability, devastating in normal times and de- bilitating even for presidents like Cardoso, who seem to hold all the cards. I conclude from this experience that Brazil’s political institutions simply work badly. What does it mean to claim that a nation’s political institutions work badly? Badly for whom? Do they serve only the rich, the economic elite? The people who design political institutions belong to the elite, and institutions can hardly be faulted for serving their creators. The tragedy of the Brazilian system is not that it beneµts elites; the problem is that it primarily beneµts itself—that is, the politicians and civil servants who operate within it. All institutions are biased against change, but Brazil’s institutional matrix makes it particularly difµcult to adopt policies deviating from the status quo. While policymakers have been able to adopt, at least since 1990, macroeconomic programs facilitating the na- tion’s participation in the global economy, they have been unable to push through µscal reforms that would consolidate stabilization. Policymakers have also been unable to design and implement educational and social programs that could raise the population’s productivity and capacity or ameliorate the effects of global competition, and leaders have made little progress in reducing the cost of government itself. The argument that a state apparatus can beneµt mainly those occupying places within it does not imply—contrary to the rhetoric of antigovernment con- servatives—that politicians are intrinsically thieves. To the contrary, many Brazilian politicians and civil servants work long hours and sacriµce private 4 The Deadlock of Democracy in Brazil gain to serve the public good. Rather, the argument begins with the idea that po- litical institutions generate incentives for politicians. These incentives motivate actions that either facilitate or hinder the adoption of public policies likely to improve life for the average citizen. In the Brazilian case, the nation’s political institutions generate incentives that encourage politicians to maximize their own personal gain and to concentrate on delivering pork-barrel programs to nar- row groups of constituents or political benefactors. Some politicians, resisting these incentives, struggle to legislate on national issues, but they face an uphill and usually unsuccessful battle. It is necessary to put some substance in the claim that Brazil’s political in- stitutions function poorly. In terms of formal powers, the country’s presidents rank among Latin America’s most powerful. What they need most, however, is support from a political party commanding a congressional majority, and only rarely do Brazil’s chief executives enjoy such support. Instead, presidential au- thority—even, at times, presidential survival itself—depends on the distribu- tion of construction projects and political jobs to crucial governors, mayors, deputies, and senators. Presidents begin their terms with high-minded pieties about avoiding the troca de favores (exchange of favors) that their predecessors so scandalously pursued. But political necessity soon rears its ugly head. Un- fortunately, even after liberally spreading the pork, the most presidents can ex- pect from the Congress—and perhaps the most presidents want—is a limited acquiescence rather than active participation in the legislative process. Because the legislature cannot respond nimbly to presidential initiatives, some Brazilian presidents have ruled mainly through emergency decrees (me- didas provisórias). Since 1988 more than one thousand emergency decrees have arrived at Congress’s door. Constitutionally, these decrees take immediate ef- fect, but after thirty days they lapse unless Congress approves them. Since the presidency has no monopoly on either wisdom or virtue, the light of day quickly reveals serious legal or substantive ×aws in many emergency decrees, and they are simply allowed to die. In many other cases, the Congress fails to act, and the president simply reissues the decree. Some emergency measures do become permanent laws, but rarely do they survive their legislative voyage unscathed. Typically, µnal versions of these bills include major compromises, sometimes re×ecting the pork-barrel demands of particular legislators or parties, some- times re×ecting the power of deputies beholden to economic interests. Overall, emergency decrees are a way of circumventing congressional obstructionism, but they also put up one more roadblock keeping the Congress from meaning- ful participation in policy-making. On its own initiative, the Congress has been too weak, either in the current Introduction 5 democratic experiment (post-1985) or in its earlier incarnation (1947–64), to legislate on issues of national concern.2 The legislature’s weakness was espe- cially painful in 1988, when the Senate and the Chamber of Deputies joined to- gether as the Constituent Assembly. The assembly produced a 160–page con- stitution that included such bizarre items as a grant of life tenure to bureaucrats and a ceiling on interest rates, but it left major issues in health care and educa- tion for resolution by future legislatures. These subsequent legislatures, to no one’s surprise, resolved nothing, doing little more than react to the many emer- gency decrees of Presidents Collor de Mello, Franco, and Cardoso. In truth, though Brazil’s social services are arguably the worst of any large Latin Amer- ican country, the Congress has passed, on its own initiative, almost nothing af- fecting education, health, or housing since the new constitution went into effect in 1988. Without question, macroeconomic stabilization has been the dominant economic problem in Latin America during the early 1990s. Brazil was the last Latin nation to adopt and stick to a workable stabilization program. It was long understood that in×ation discouraged productive enterprise and foreign invest- ment, but even when it became clear that in×ation hurt the poor most (Caccia- mali 1997, “Um Choque na Desigualdade” 1996), Brazilian politicians found it impossible to reach an accord. Conservative and moderate legislators ulti- mately accepted Cardoso’s program in 1994, but they refused to acquiesce un- til the only alternative was a victory by a truly leftist candidate, Luís Inácio Lula da Silva (Dimenstein and De Souza 1994). Even then, rural politicians in the Chamber of Deputies extracted major concessions in exchange for their votes. These concessions, worth billions of dollars, represented not policy compro- mises but personal µnancial payoffs.3 Why are Brazil’s political institutions so ineffectual? Consider the party system and the legislature. Major, electorally successful parties fall all across the ideological spectrum. Some parties embrace distant, hostile points of view; 2. A number of analysts attempting to explain the military coup of 1964, notably Wander- ley Guilherme dos Santos (1979), have stressed the Congress’s legislative immobility at the end of the earlier pluralist period. Research on congressional elections or congressional behavior during the 1947–64 democratic period includes Amorim Neto and Santos 1997; Benevides 1976, 1981; Soares 1973; and M. Souza 1976. 3. Many of these reforms, including tariff reduction and the beginning of privatization, needed no congressional assent. Others required ordinary majorities rather than supermajorities. According to Peter Kingstone, in a personal communication, some northeastern deputies supported economic opening because it meant access to high quality goods previously restricted by paulis- tas. Kingstone also emphasized the large size of the coalition beneµting from reduction as opposed to the small group of losers, as in the highly protected computer industry. 6 The Deadlock of Democracy in Brazil others shelter deputies sharing no ideas at all. Party leaders have little control over their members, and many, perhaps most, deputies spend the bulk of their time arranging jobs and pork-barrel projects for their constituents.4 Parties in Brazil rarely organize around national-level questions; the Congress, as a re- sult, seldom grapples with serious social and economic issues. Brazil’s presidents beneµt little from Congress’s programmatic weakness. With only a small chance of stable legislative support, the executive faces po- litically independent governors, a crowded electoral calendar, municipalities depending on federal largesse for their survival, and a substantial core of deputies caring about their personal incomes µrst, reelection second, and pub- lic policy a distant third.5 Because in×ation has been the overriding problem since the end of military rule, new presidents often take ofµce with elaborate macroeconomic plans but rarely with programs going much further. And be- cause congressional support must be built on a very wide, multiparty base, the cabinet is likely to include ministers whose loyalties are tied more to their own political careers than to the president’s program. An Institutional Perspective What causes these political failures? How can we understand Brazilian poli- tics? The focus of this book is institutional.6 Douglass North deµnes institutions broadly, as simply “the rules of the game in a society or, more formally... the humanly devised constraints that shape human interaction” (1990, 3). While op- 4. Students of the Brazilian legislative process hotly debate party leaders’ degree of con- trol over their backbenchers. Adherents to the “leadership predominance” thesis point to the rela- tively united behavior of Brazilian parties on roll-call votes (Figueiredo and Limongi 1997b; Limongi and Figueiredo 1995). In chapter 7, however, I will demonstrate that disciplined voting among Brazilian legislators results mainly from constituency pressures, electoral insecurity, and pork-barrel concessions to individual demands. 5. The claim that personal income is relatively more important for Brazilian deputies than for deputies in other systems is based only on my experience and on anecdotal evidence, not on empirical comparison. But the salaries of Brazilian federal and state parliamentarians are roughly equivalent to salaries in the House of Representatives, while per capita incomes in Brazil are a tiny fraction of those in the United States. 6. Examples of institutional analyses of Latin American politics include Michael Coppedge’s study of Venezuelan parties (1994), Matthew Shugart and John Carey’s cross-national examination of presidential and legislative power (1992), Mark Jones’s work on Argentine elec- toral laws (1995), Barbara Geddes’s game-theoretic treatment of presidential coalition building (1994a), Brian Crisp’s study of Venezuelan institutional design (1999), studies of coups in Brazil and Chile by Wanderley Guilherme dos Santos (1979) and Youssef Cohen (1994), and Fabiano Guilherme Santos’s treatment of the microfoundations of clientelism (1995). Introduction 7 erating within North’s deµnition, I focus on institutions in a narrower sense— that is, I seek to illuminate the effects of the formal structures of politics on the behavior of politicians and on the outcomes of the political process. The cen- tral institutions of Brazilian national politics include the electoral system, the presidency, and the legislature.7 As this book will demonstrate, these institu- tions are inextricably linked: the electoral system in×uences—simultane- ously—the kinds of candidates who compete in elections, their campaign strate- gies, and their behavior in ofµce. Electoral rules also affect the number of viable political parties as well as their coherence and discipline. Deputies’ preferences strongly in×uence legislative outcomes, of course, and the president must con- tinually struggle to mobilize support within the legislature. An institutional focus implies that institutions have a life of their own, that they are more than the intentions of the actors who created them. But institu- tions cannot arise from nothing; people must create them. If institutions are works of conscious creation, why not treat them simply as the agents of their creators? Why, for example, is the capitalist state, to employ Marx’s famous re- mark, not merely the executive committee of the bourgeoisie? The claim that institutions are more than agents of their creators µnds sup- port in a variety of research traditions. Scholars have viewed institutions both as organizations, with routinized operating procedures, and as arenas of bu- reaucratic struggle. The old saw “where you stand depends on where you sit” re×ects the tendency of members of bureaucracies, legislatures, and judiciaries to defend purely organizational interests (Allison 1971). Another research tra- dition emphasizes the short time horizons of politicians. Elected politicians concern themselves mainly with the immediate consequences of their acts; fail- ure to do so jeopardizes their future as politicians.8 New institutional arrange- ments have consequences that may become apparent only in the long run, even though such arrangements are the consequences of decisions taken to solve im- mediate political problems. Complex social processes also generate unantici- pated outcomes. The more complex the institution and the larger the number of actors involved, the more likely it is that politicians simply have no way to pre- dict ultimate outcomes. Even when institutions produce results that diverge grossly from their founders’ intentions, change is slow. Institutions exhibit path dependence. In North’s terminology (1990, 94), path dependence means that “the consequence 7. Federalism, discussed subsequently as one of the major in×uences on the functioning of electoral and legislative politics, could itself be regarded as an institution. I treat federalism as a historical antecedent simply because its origins lie farther back in Brazilian history. 8. This theme is investigated intensively in Ames 1987. 8 The Deadlock of Democracy in Brazil of small events and chance circumstances can determine solutions that, once they prevail, lead one to a particular path.” Institutions may have built-in barri- ers to reform, barriers deliberately created by political actors who believed they could constrain their opponents only if they constrained themselves (Pierson 1996). Actors who beneµted from earlier institutional change are likely to re- sist efforts at reform. And as institutions become more established, actors make commitments that generate sunk costs. Because actors become locked into on- going arrangements, exit costs rise. Given these justiµcations for treating institutions as distinct from their cre- ators, I will now consider the variety of ways scholars have undertaken institu- tional analysis. At least four different traditions—here I use Hall and Taylor’s (1994) terminology—have marched under the “institutionalist” banner.9 Orga- nizational theorists, notably March and Olsen (1989), stress institutional roles and routines along with duties and obligations. For March and Olsen, institu- tions themselves are political actors. Treating institutions as actors, of course, presupposes that institutions are coherent. March and Olsen recognize that in- stitutional coherence varies, but they believe that at times collectivities may be viewed as acting coherently (1989, 18). A second group, including economists such as Ronald Coase (1937), Douglass North (1981), and Oliver Williamson (1983), relates transaction costs to economic efµciency and to the organiza- tional form of the µrm. Transaction costs in political or policy-making envi- ronments include the costs incurred in such activities as the negotiation of agreements, the monitoring of compliance, the use of middlemen, the punish- ing of the noncompliant, and the creation of quasi-voluntary compliance (Levi 1988, 23). Rational choice theorists such as Kenneth A. Shepsle (1978) cast in- stitutions as “games in extensive form,” games in which rules constrain the be- havior of actors. In the fourth tradition, historical sociologists, including Theda Skocpol, Peter B. Evans, and Dietrich Rueschemeyer (1985) and Kathleen The- len and Sven Steinmo (Steinmo, Thelen, and Longstreth 1992), concentrate on the timing and sequence of institutional developments as they affect major ar- eas of public policy. These scholars sometimes call themselves “new institu- tionalists” or “historical institutionalists.” Though their work is in part a reac- tion to an earlier tradition emphasizing sociocultural factors and class struggle, these scholars maintain that earlier work’s analytic focus on long-term processes of change. Which approaches are useful in a study of Brazilian institutions? The or- 9. For scholars such as Levi (1988), transaction-cost analysis is a subset of a rational choice approach. Introduction 9 ganizational analysts will play no signiµcant role. The organizational approach does not quite µt Brazil: institutions like the Brazilian legislature are too inco- herent to be viewed as unitary actors, and an electoral system is an institution in a more abstract sense than March and Olsen intend. While transaction-cost approaches can be effective in explaining the development of a particular pol- icy, such as tax collection (Levi 1988), they are too narrow to serve as the the- oretical backbone of my argument. Still, transaction costs clearly play a role in the development of particular institutions, including parties and legislative committees (Weingast and Marshall 1988). This book adopts the perspective of rational choice (RC) theorists, with a bit of historical institutionalism mixed in. To understand this theoretical union, I will begin with the basics of the RC approach. Borrowing Geddes’s (1994b) terminology, RC approaches share four principles. The µrst is methodological individualism—that is, the principle that all social phenomena should be ex- plained in terms of the actions of individuals trying to maximize their goals un- der some set of constraints. Second, actors along with their goals and prefer- ences are explicitly identiµed. Third, institutions and other contextual features determining actors’ options are also explicitly identiµed, along with their costs and beneµts. And fourth, hypotheses are generated by a deductive logic—that is, the theories tested are causal, falsiµable, and internally consistent (King, Keohane, and Verba 1994). RC theorists emphasize the short-term constraints and incentives that structures of politics create for political actors. Without denying the relevance of values, RC theorists suggest that strategic preferences are determined by the formal rules of politics themselves. Most importantly, RC institutionalists con- tend that behavior changes, regardless of underlying cultural attitudes, when in- stitutions change.10 Like any new theoretical approach, RC approaches promise more than they deliver. Assumptions are often capricious and self-serving, and institutions are sometimes described in the sketchiest terms. The oft-heard criticism that RC approaches ignore institutions, however, is wrong. On the contrary, as George Tsebelis (1990, 40) points out, “The rational-choice approach focuses its at- tention on the constraints imposed on rational actors—the institutions of a so- ciety.... The prevailing institutions (the rules of the game) determine the be- havior of the actors, which in turn produces political or social outcomes.” 10. A good example of this approach is Cox 1987. Focusing on English politics after the µrst nineteenth-century Reform Act, Cox shows how an institutional change, the adoption of cabinet government, led to the rise of disciplined parties, a dramatic decline in clientelism among British politicians, and a increase in party voting among the populace. 10 The Deadlock of Democracy in Brazil Now I will turn to historical institutionalism. How does it differ from the RC approach? Thelen and Steinmo (Steinmo, Thelen, and Longstreth 1992) outline points of convergence and divergence. RC approaches share with his- torical institutionalism (HI) a concern with the way institutions shape political strategies and in×uence political outcomes. But for RC scholars, institutions are important as features of a strategic context, imposing constraints on self-inter- ested behavior. Political and economic institutions deµne or constrain the strate- gies that political actors adopt in pursuit of their interests. For partisans of HI, institutions provide “the context in which political actors deµne their strategies and pursue their interests” (7). Because institutions contribute to the shaping of preferences themselves, institutions’ role in politics is much broader in an HI framework than in a narrow RC model. HI’s breadth, of course, all too easily becomes theoretically or empirically vague arguments. A second difference between historical institutionalists and rational choice scholars lies in the area of guiding assumptions. For HI scholars, RC assump- tions are too narrow. Most political actors are not all-knowing, rational maxi- mizers. They do not stop at every choice and ask how they maximize their self- interest. Instead, political actors are rule-following “satisµcers,” obeying societally deµned rules even when such rules may not directly maximize their self-interest. But here HI partisans overstate their critique. As Tsebelis makes clear, the use of rationality as a model of behavior does not imply that all peo- ple act at all times as utility maximizers. Rather, rationality is an appropriate model when the “actors’ identity and goals are established and the rules of the interaction are precise and known to the interacting agents” (1990, 32). Tsebelis offers four arguments in support of the utility-maximizing assumption. Behav- ior, he suggests, conforms to utility-maximization principles when the stakes are higher and when more information is available. Actors able to learn from trial and error are more likely to move toward optimal behavior. Even if only a small percentage of actors maximizes utility, the social outcome will often re- semble the outcome obtained if all actors maximized. And natural selection fa- vors those who maximize, since maximization strategies will be rewarded.11 In sum, utility maximization still provides a useful guiding hypothesis for the be- havior of political actors. Rational choice and historical institutionalism also disagree on the inter- pretation of preferences. For RC scholars, preferences are assumed rather than 11. Tsebelis (1990, 36) also offers a statistical argument: If a few people maximize while most behave randomly, the outcome will re×ect the maximizers’ goals. An RC analysis can be quite inaccurate about a speciµc individual but very accurate regarding the average individual. Introduction 11 explained. Advocates of HI regard preferences, especially those related to deµnitions of self-interest, as requiring explanation. Thelen and Steinmo make this distinction very clear: historical institutionalists argue that institutional contexts shape actors’ goals. Class interests, for example, are more a function of class position, mediated by institutions like parties and unions, than of indi- vidual choice. For historical institutionalists, only historically based analyses can indicate what goals rational actors seek to maximize and why they empha- size certain goals over others. The distinction between the two approaches is not simply the exogeneity or endogeneity of preferences. For historical institutionalists, preferences are affected not just by institutions themselves but by such factors as new ideas (such as Keynesianism) and by leadership. If historical institutionalists and rational choice theorists need not µght over utility maximization, they do differ over the breadth of what they seek to pre- dict and explain. RC theorists are likely to focus on the behavior of generic politicians or generic political organizations such as legislative committees or parties.12 As a result, RC arguments tend to be probabilistic rather than deter- ministic. They might suggest, for example, that parties in plurality systems are more likely to converge programmatically than are parties in proportional sys- tems. HI theorists, by contrast, typically direct their arguments at the broader outcomes that result from the strategic behavior of politicians or parties. This project falls into the branch of choice theory known as soft rational choice. I present no mathematical derivations, and I seek to explain real rather than imaginary political institutions. Along with most RC scholars, I assume that political actors’ central motivations are more likely to emphasize personal goals, including desires for reelection and personal riches, than policy interests. Although in much of my argument the centrality of personal goals is assumed, at various points I offer empirical evidence supporting the assertion and pro- viding a µner distinction between personal and policy goals. Speciµcally, I demonstrate that while some members of Congress seek to maximize their life- time income from politics and other members have broad, national policy goals, many are motivated simply to help particular µrms, discrete economic groups, or even specialized groups of workers. In some parts of the book’s argument, rational choice reasoning contributes much more than just assumptions. A variety of research traditions motivate the 12. Examples include Baron 1991; Cox 1990b; and Shepsle 1988. Major exceptions include Knight (1992), who concentrates on the redistributive quality of political institutions taken as a whole, and North (1981, 1990). 12 The Deadlock of Democracy in Brazil empirical models. In the chapters on the electoral system, for example, the the- oretical discussion begins with the literature on legislative “credit claiming” (Mayhew 1974). The argument also draws on studies of candidate behavior in proportional representation systems (Cox 1990a). In the chapters examining the legislative process, theory begins with the “distributional” versus “informa- tional” controversy in the literature on the U.S. Congress.13 The role in leg- islative bargaining of Brazil’s parties—parties with neither programs nor close ties to voters—is explored with reference to the rise of parties in early U.S. Con- gresses (Aldrich 1995). Although many of my arguments apply to “generic” politicians and par- ties, I focus on illuminating the actions of political actors in a particular coun- try during a particular time. Broader outcomes—individual electoral campaigns succeeding or failing, legislative coalitions forming or dissolving, the Congress adopting or rejecting proposals—all become an intrinsic part of the explana- tion, because the political life of a real country is composed not of generic out- comes but of the actual outcomes of the strategies and struggles of real politi- cal actors. Likewise, a narrative that simply “assumed” preferences would be too limiting. Ideas, leadership, and chance events affect not only actors’ pref- erences about economic and social policy but also ideas about political institu- tions themselves. In effect, preferences cannot be taken as given; rather, it is necessary to uncover all possible information about them. A Perspective on Brazilian Politics: An Excess of Veto Players In any political system, the adoption of a new policy deviating from the status quo requires the agreement of certain political actors. When the absolute num- ber of such crucial actors, or veto players, is large, policy innovation becomes very difµcult. Brazil’s institutional structure, I argue, inherently produces a large number of veto players. As a result, its central government has enormous difµculty producing innovative policies. The idea that the number of veto players signiµcantly affects the chances of adopting new policies is a recent innovation in political science. For its orig- inator, Tsebelis (1995), one of the advantages of the veto-players framework is its ability to subsume a host of typology-centered theories. These alternative theories usually take the form of dichotomies: presidentialism versus parlia- 13. The terms distributional and informational are Krehbiel’s (1991), but the real distinc- tion is between a distributional, pork-barrel orientation and an orientation to some conception of the broader, national public good (see Knight 1992). Introduction 13 mentarism, two-party versus multiparty systems, and so on. Some have policy implications: Presidential regimes, for example, are thought to be more prone to military coups than are parliamentary regimes (Shugart and Carey 1992). But the policy implications of these alternative theories can often be seen only af- ter they are combined with other typologies, for example, two-party systems combined with presidentialism. When comparing small numbers of countries, these combinations lead to overdetermination; in other words, the variables out- number the observations. The veto-players perspective, with its focus on policy change, offers both a general, overarching framework and a clear set of predictions.14 Whether a government is presidential or parliamentary, whether it tends to a two-party or a multiparty system, whether the legislature has one or two houses, the veto- players logic yields an unambiguous prediction as to the likelihood—in com- parison with some alternative system—of policy change. The argument, then, is that a larger number of veto players increases the stability of policy. Consider the fundamental reasoning.15 Whenever a choice must be made over alternative policies, every actor has a preferred policy point. Actors are indifferent to policies that are equal distances from their ideal points, but they prefer policies closer to their ideal points. In the language of spatial modeling, their indifference curves are circular. As µgure 1 (taken from Tse- belis 1995) demonstrates, when these indifference curves are drawn to intersect the position of the status quo and decisions are made by simple majority rule, any two actors can defeat the status quo. These areas (shaded in the µgure) are the winset of the status quo. The larger this winset, the more likely policy change becomes and the more likely it is that policy change will be substantial rather than incremental. If policy change requires the concurrence of more ac- tors, the winset of the status quo cannot increase—and usually decreases—in size. In µgure 1, in fact, a decision requiring the concurrence of all three actors results in no action at all. With increases in the distance (the spread of the ideal points) between the players required to agree, the winset shrinks and policy stability once again in- creases. Ideological disagreements represent one form of spatial distance. As 14. Whether observers prefer political systems facilitating or hindering policy innovation is irrelevant to the logic of the theory, although I do not hide my view that Brazil’s institutional struc- ture prevents the adoption of policies beneµting and preferred by substantial majorities of its citi- zens. 15. The explanation that follows is taken from Tsebelis 1995. Interested readers should con- sult this piece for a fuller treatment. Empirical veriµcation of the theory can be found in Bawn 1997; Franzese 1996; Hallerberg and Basinger 1998; Kreppel 1997; Treisman 1998; and Tsebelis 1999. 14 The Deadlock of Democracy in Brazil Fig. 1. Winset of status quo with three players in two dimensions Tsebelis (2000) shows, the winset of the status quo shrinks as the ideological distance between players grows. Veto players may be individuals, but more often they are parties, factions, or groups. Such collective players typically include individuals with varying ideal points—in other words, the policy positions of collective players are less than perfectly coherent. As the range of positions increases within each of these players—that is, as coherence declines—the winset of the status quo grows. This result is very important, because it means that the chance of adopting a new policy is greater—given a certain number of veto players—when the play- ers are less coherent or united in their policy views.16 How are veto players deµned and counted? In a broad sense, veto players may include the military, industrial capital, or any other group whose concur- 16. If any majority is possible in a legislature, then the legislature is a collective veto player—i.e., no single member is necessary for a majority. If there is a stable majority, or—as is the case in contemporary Brazil—if certain parties are always excluded from the majority, then the players that must be included are veto players. It is more difµcult to change the status quo in the presence of a stable majority, since all the players must agree. Introduction 15 rence is required for policy adoption. In this book I limit the concept to actors in formal legislative and executive politics. “Institutional” players include the president, the Senate, and the Chamber of Deputies. “Partisan” players are the organized political parties in the Congress. Though for my purposes an overall, administration-wide count of veto players serves as a useful simpliµcation, the number of veto players may vary by issue. Interest rates, for example, are set by the executive without legislative participation, while ordinary bills must be approved by both chambers. Equally important, especially in the Brazilian case, is the possibility of supermajority requirements. Because constitutional amend- ments require three-µfths of the total membership of each chamber, while or- dinary legislation requires only an absolute or simple majority, the number of parties necessary for approval on constitutional issues may be higher, depend- ing on the size of the parties.17 The key to counting veto players is the absorption rule. Suppose the analy- sis is limited to two policy dimensions, and suppose one veto player is located in the Pareto set of the others—that is, that player is within the polygon pro- duced by connecting the other players’ ideal points.18 Tsebelis (2000, appen- dix) shows formally that it makes no difference whether this inside player is counted. It follows that if the president comes from a coherent and disciplined party that is part of a stable majority, the president is absorbed as a veto player— that is, the president need not be counted. Likewise, a party can be absorbed in one chamber of a bicameral legislature if it is part of the majority in the other chamber. In a case like that of the United States, where majorities are unstable and where the parties are programmatically incoherent, presidents and parties in each chamber remain distinct veto players. How does Brazil compare with other Latin American countries in terms of the number of veto players typically present in each presidential administra- tion?19 For this comparison I eliminate provisional and authoritarian regimes (both civilian and military), and I count only partisan veto players—parties and presidents. The absorption of parties depends on their programmatic coherence and legislative discipline. Given that the existing research on Latin American parties has no accepted criteria for assessing coherence and discipline, it makes 17. Supermajorities are particularly important in Brazil, because the nation’s extremely de- tailed constitution imposes a wide range of costly obligations on government. 18. The Pareto set is deµned as the area in which one player cannot be made better off with- out harming another. 19. Focusing on legislative parties, Amorim Neto (2001) presents a variety of data support- ing the idea that Brazil’s legislature has been extremely fragmented since 1988. 16 The Deadlock of Democracy in Brazil sense to employ a counting method that establishes upper and lower bounds on the number of veto players.20 The µrst method, complete absorption, absorbs all presidents and similarly named parties in bicameral legislatures—that is, I count them only once. The second counting method, partial absorption, absorbs parties and presidents only in those cases where most scholars regard the par- ties as coherent and disciplined: Argentina, Chile, Colombia (between 1958 and 1974 only), Costa Rica, Mexico, Uruguay, and Venezuela.21 Since 1900 the average number of veto players for all Latin America is 1.95 using the complete absorption method and 2.79 with partial absorption. With complete absorption, Uruguay has the highest country average at 3.05, followed by Brazil at 2.74 and Chile at 2.65.22 With partial absorption, Brazil is far and away the leader at 4.43. Given that Brazilian politics became much more com- petitive after 1945, it is worth repeating the comparison for the postwar period. From 1946 to 1998, the overall regional averages are 2.09 (complete absorp- tion) and 2.93 (partial absorption). By complete absorption, Brazil leads with 3.40 veto players, followed by Ecuador with 2.88 and Chile with 2.75. By par- tial absorption, Brazil leads with 5.13, followed by Ecuador at 4.50 and Bolivia at 3.88. By either standard, then, Brazil is at the top in terms of the average num- ber of veto players in each administration. Moreover, from 1986 until 1999, Brazil averaged 4.6 veto players by complete absorption and 6.5 by partial ab- sorption. Brazil’s competitors offer some instruction in the consequences of high veto-player scores. Uruguay exceeded three veto players only between 1967 and 1971, when it reached four. Uruguay’s democracy began to collapse dur- ing this period, µnally falling victim to a military coup in 1973. Chile exceeded three veto players only during the presidency of Carlos Ibañez (1952–58), who 20. Any counting scheme requires arbitrary choices. Reasonable people will disagree. This scheme characterizes all the parties in a system as either disciplined or not; in reality, Chile’s Rad- ical Party was undisciplined in a mostly disciplined system, and Brazil’s Workers’ Party is disci- plined in an undisciplined system. I have tried to avoid counting parties in permanent opposition, like the Workers’ Party, but there are undoubtedly some errors. Nothing is assumed about the lin- earity of the relationship between veto players and innovation. Generally, however, the jump from two to three veto players must be more paralyzing than the jump from four to µve. Such powerful µgures as committee chairs should not be counted as veto players, because they can usually be over- ridden by ×oor majorities. 21. In Uruguay, party factions (sublemas) rather than parties were absorbed. I am indebted to Scott Morgenstern for noting that legislative discipline is high among factions but low among parties. 22. Three veto players were counted during the Sarney administration (1986–90), even though the PSDB split off from the PMDB during this legislature. Introduction 17 attempted to govern as an independent with no party backing. Faced with ris- ing in×ation, Ibañez tried to implement the recommendations of U.S. consult- ants, but the Chilean Congress refused to pass tax increases on the wealthy. By the end of the Ibañez administration, as Stallings (1978, 33) puts it, “the initial indecision of the Ibañez government and the later attempts to implement the [consultants’] recommendations apparently convinced Chilean voters that the solution of Chile’s problems was not to be found in a leader ‘above politics.’”23 Bolivia is a case in which the number of veto players has varied extensively. Before 1982 Bolivia never had more than two. With the restoration of democ- racy in 1982, Bolivia has averaged about 2.5 veto players with complete ab- sorption and 5 without. Unless Bolivia’s parties act in a programmatic and dis- ciplined way (thus suggesting that the lower veto-players count is more appropriate), the situation seems quite unstable.24 In sum, the veto-players framework yields important insights for Latin America in general and Brazil in particular.25 Excesses of veto players as well as greater ideological distances between veto players both decrease the chances of signiµcant legislation. Supermajority requirements, common for constitu- tional amendments, also reduce the chances of signiµcant legislation. Finally, control of the agenda matters. In presidential systems, the legislature controls the policy-making agenda. Because the legislature elaborates and modiµes leg- islation, and because the legislature can override presidential vetoes, it main- tains agenda control. In Brazil, where the parties cannot control their members and where individuals or groups trade cooperation for particularistic beneµts or 23. Stallings (1978, 33–34) also points out that the failure of the Ibañez administration led to a union of labor organizations as well as the coming together of the two socialist parties. In other words, the number of veto players, both societal and partisan, declined. The 1973 overthrow of Sal- vador Allende, the most obvious case of democratic breakdown, occurred when the number of veto players was unexceptional. However, the ideological distance between the parties was extremely large. 24. Ecuador had a large number of veto players between 1948 and 1961, and though a mil- itary coup occurred in 1961, the democratic regime survived three presidential administrations. I cannot explain the Ecuadorean case, though it is interesting that Martz (1990, 382) regards this pe- riod as an anomaly in Ecuador’s normally turbulent politics. 25. Although this discussion of veto players has centered on presidents and parties, the con- cept clearly has implications for other institutional actors. In systems of many veto players, courts and bureaucracies typically take larger legislative roles. Courts may act as veto players by inter- preting the constitution and by providing judicial review of legislation. Bureaucracies adopt the same kind of interpretive role. When the number of veto players is small, the legislature can allow bureaucrats wide discretion, since the legislature can easily restrict them in the future. When the number of veto players is large, however, legislatures whose veto players are close together may try to restrict bureaucracies. If the legislative veto players are far apart, then legislatures are more likely to allow bureaucrats more leeway. 18 The Deadlock of Democracy in Brazil concessions, agenda control means that most legislation includes a pork-barrel component. It is important to note, however, that the argument that Brazil suffers from an excess of veto players is not equivalent to claiming that Brazil has too many parties. Although Brazil has a great many parties and that number in itself may frustrate the adoption of policy innovations, an overall count of veto players based purely on parties represents a simpliµcation that results from aggregat- ing across issues. On any given issue, veto players may include state governors and their delegations, issue caucuses (such as bankers or rural property own- ers), or individual legislators. It follows, moreover, that these individual or col- lective veto players’ ideologies or motivations also affect the outcome of pol- icy struggles. The Origins of Brazil’s Institutional Problem An excess of veto players creates difµculties for democratic regimes, and by any standard Brazil suffers from such an excess. This book argues that the root causes of Brazil’s high number of veto players lie in the nation’s institutional framework, especially in the electoral system. Still, institutions do not descend from the sky. Political and economic elites create them. Two factors, federal- ism and the pervasiveness of patronage and pork, have been particularly im- portant in shaping the choice of institutions, and certain historical events have locked Brazil into particular institutional patterns. Over the course of recent Brazilian history, federalism and the extensiveness of patronage and pork vary in form and importance, but they always matter.26 I will discuss these two fac- tors from a rational choice perspective, but their origins lie so far back in Brazil- ian history that they can reasonably be taken as givens. Federalism States and municipalities, Brazil’s subnational units of government, have been important political actors since colonial times. Major governmental activities, including such crucial social services as primary and secondary education, are the responsibility of states and municipalities. These subnational units elect their own ofµcials and possess their own sources of tax revenue. They may is- 26. Of course, these background elements potentially raise the same “chicken and egg” cau- sation problem seen earlier. Is their existence a result of institutional or cultural conditions? It seems to me that in the time period analyzed here, the setting can be taken as given. Introduction 19 sue bonds subject only to the approval of the territorially based Senate. Resid- ual powers not enumerated in the constitution fall to the states. Overall, Brazil’s federal system meets Riker’s (1964) famous criteria: Two levels of government rule the same land and people, each level has a well-deµned scope of authority, and each possesses a guarantee of autonomy within its own sphere. At the national level, federal systems typically represent territories in one legislative chamber and population in another. With three senators per state, Brazil’s Senate gives Roraima, with fewer than 250,000 people, representation equal to São Paulo, with more than 30 million. One vote in Roraima thus has 144 times the weight of a vote in São Paulo, and senators representing 13 per- cent of Brazil’s population can block legislation supported by 87 percent.27 But the disadvantage to more populous regions also occurs in the Chamber of Deputies. No state is allowed fewer than eight or more than seventy deputies. Purely on the basis of population, Roraima’s 8 deputies ought to be 1; São Paulo’s 70 ought to be 115. Why did Brazil adopt federalism? Historically, the country had no alterna- tive. Portugal was simply too weak to maintain a bureaucracy capable of con- trolling the colony. Emperor João III (1521–57) divided the country into hered- itary captaincies and handed them over to landowners rich enough to defend and colonize them (José Murilo de Carvalho 1993). Though the Marquis de Pombal ended these captaincies in the eighteenth century, Portugal was really incapable of centralizing. It had to rely on political and administrative decen- tralization (including tax farmers) and on private power based on big land- holdings and on slavery. The constitutional reform of 1834, a few years after Brazil’s break from Portugal, produced a new centralism based on Brazil’s emperor. Coffee pro- ducers, concentrated in the province of Rio de Janeiro, paid most of the central government’s taxes and joined exporters and bureaucrats in support of a strong, monarchical central government.28 Nineteenth-century centralization was bol- stered by the in×uence of the monarchy in the rural population and by fear of upsetting slave society and fragmenting the country. The proponents of decen- tralization, by contrast, included liberal professionals and farmers producing for the domestic market. Though early coffee production had stimulated centralization, the spread 27. The worst ratio in the United States is Wyoming to California: One Wyoming vote equals 66 California votes (see Stepan 1999, 35). 28. By 1877 the central government employed 69 percent of all public employees and col- lected 77 percent of all public income, a µgure that fell to 37 percent in 1902 (José Murilo de Car- valho 1993, 65). 20 The Deadlock of Democracy in Brazil of coffee into São Paulo favored decentralization. The paulistas thought a cen- tralized system would transfer resources to backward provinces. But resistance to central power also had an oligarchical component: federalism supported pri- vate power, inequality, and hierarchy. When the monarchy ended in 1889, Brazilian states were much stronger than states in federalist Mexico and Argentina. Brazil’s states could write their own civil codes, negotiate foreign loans, and sell bonds outside the country (Love 1993, 187). In the 1920s and 1930s, coffee-producing states forced their exchange-rate preferences—favoring devaluation—on the rest of the country. The policy imposed losses on consumers, importers, and the central govern- ment itself, which had to repay its foreign loans in a weakened currency. Brazil- ian states also maintained serious military forces. In 1925–26, for example, São Paulo had a 14,000-man army, its own military academy, and a foreign military mission (Love 1993, 202). Even in the centralizing years of Getúlio Vargas and the Estado Novo (1937–45), states retained considerable authority. State taxes as a percentage of federal taxes reached 55.9 percent between 1931 and 1937 and 55.7 percent between 1938 and 1945. Comparable µgures from the Mexican federation dur- ing the same years were 22.7 percent and 17.3 percent, respectively (Love 1993, 218). Landowning was the traditional base of local power, and though twentieth- century economic development weakened landowners, they never lost their po- litical force, especially in rural areas. With large land areas, sparse populations, and few good roads, states and localities naturally developed independent bases of power. Federalism also grew out of Brazil’s strong regionalist tradition. Each region has quite different social, cultural, economic, and political conditions.29 Brazil’s Northeast contains about 30 percent of its population but produces only 14 percent of the gross domestic product. Sugar production once made the Northeast the center of the Brazilian economy, but sugar has been declining since 1800. Poverty, a recurring cycle of droughts, and the continuing in×uence of rural bosses all combine to make the Northeast dependent on central gov- ernment resource transfers. As a result, politics has often revolved around ex- changes of political support for pork-barrel beneµts from the legislature or the executive. Aid programs have produced countless dams and hydroelectric proj- 29. Brazil’s regions are deµned as follows. The Northeast includes Alagoas, Bahia, Ceará, Maranhão, Paraíba, Piauí, Pernambuco, Rio Grande do Norte, and Sergipe. The North and Center- West include Acre, Amapá, Amazonas, Goiás, Mato Grosso, Mato Grosso do Sul, Rondônia, Ro- raima, and Tocantins. The Southeast includes Espírito Santo, Minas Gerais, Rio de Janeiro, and São Paulo. The South includes Paraná, Rio Grande do Sul, and Santa Catarina. Introduction 21 ects, mostly beneµting big farmers and local bosses. Business interests from wealthier regions support Northeast aid, even though their taxes µnance the projects, because their µrms build the dams and roads. The Northeast, in essence, is Brazil’s Mezzogiorno. The North and Center-West are frontier regions, with small but rapidly growing populations. In the absence of traditional landholding families, pa- tronage-based political networks link key political actors. These networks in- clude members of federal and state congressional delegations, state government agencies, and local governments. States in these regions depend on the central government, so the alliances µght to control major local governments and, through these governments, federal and state patronage resources. The Southeast shelters the bulk of the nation’s industry. Coffee proµts µnanced São Paulo’s early industrial ventures, and today the state dominates advanced industry. Rio de Janeiro and Minas Gerais are also heavily industri- alized. With heavy migration from poor regions (especially from the Northeast), Rio de Janeiro and São Paulo have become megacities, plagued by grossly deµcient social services, debilitating pollution, and high levels of crime. The South is mainly agricultural. With a strong economic base and rela- tively even income distribution, politics is less likely to be the only lucrative economic activity. As a result, corruption is somewhat lower, bureaucratic com- petence is higher, and education and health conditions are better. Given regional income variations and the long tradition of federalism, it is no surprise that government-society relations vary across states. In some, poli- tics has traditionally been a proµtable business, monopolized by a small num- ber of families supported by major economic groups such as sugar planters or cattle ranchers. In other states, principally in the South and Southeast, economic interests are more diverse. Ties between citizens and representatives are more direct, and politicians are less a class unto themselves. In Bahia, for example, 40 percent of all deputies in the 1991–94 federal legislature had a close relative holding political ofµce. Among São Paulo deputies, only 5 percent came from such political families. Brazilian federalism has followed a pendular pattern, with greater central- ization under authoritarian rule. During the military regime, from 1964 to 1985, the central government sharply increased its power at the expense of the states. Not unexpectedly, the current democratic constitution strengthened states and municipalities by granting them new tax resources. Though the constitution failed to transfer responsibility for program implementation along with the new money, states have been gradually assuming responsibility for formerly federal programs. Accelerating this uncoordinated and slow shift in responsibilities has 22 The Deadlock of Democracy in Brazil been the federal government’s policy of holding back transfers to states and mu- nicipalities as part of the 1994 Plano Real economic stabilization program.30 Though Brazilian federalism meets Riker’s narrow deµnition, the country’s federal structure, like most nominally federal systems, is much more a marble cake, with overlapping and interlocking jurisdictions, than a layer cake. And Brazil’s federalism is far from the standards that Montinola, Qian, and Wein- gast (1995) impose for their “market-preserving federalism.” Brazil fails this more rigorous test on multiple counts: its subnational governments lack primary economic authority within their jurisdictions, they avoid hard budget con- straints by borrowing, they share revenue extensively, and the central govern- ment can unilaterally alter allocations of authority and responsibility among the levels of government. Weingast and his colleagues argue that their version of federalism pro- motes economic growth through a “Tiebout-like” decentralization mechanism (Tiebout 1956). In a truly decentralized polity, local leaders compete for mo- bile sources of revenue by avoiding debilitating regulations and the conµsca- tion of private wealth. If leaders fail to implement efµcient policies, labor and capital exit. At the national level, the central government must remain weak enough so that it cannot conµscate private wealth but strong enough to enforce contracts and provide public goods. In practice the political conditions supporting the theory of market-pre- serving federalism are very hard to realize.31 Rich but immobile interests exer- cise de facto mobility, increasing interjurisdictional inequality by walling them- selves off in their own municipalities. Cut off from revenue sources, poor and marginalized municipalities see no alternative to welfare and the extraction of pork from the central government. Local leaders, whether operating in demo- cratic or authoritarian environments, often answer to the demands of immobile capital rather than mobile capital or labor. Leaders in democracies respond not only to capital but also to electoral threats, and immobile interests may have greater incentives to organize electorally. As this brief historical review has demonstrated, Brazil had periods of fed- eralism close to the decentralization of market-preserving federalism, espe- 30. Debt µnancing played a large role in states’ assumption of new responsibilities. Between 1990 and 1993 (the last pre-Real year), state debts as a proportion of state internal product (the pro- duto interno bruto) climbed from 4.6 percent to 20.6 percent in the North, from 24.5 percent to 47.1 percent in the Northeast, from 10.6 percent to 24.7 percent in the Southeast, from 20 percent to 31 percent in the South, and from 25.7 percent to 45.8 percent in the Center-West (see C. Souza 1998, 584). 31. This section draws on Rodden and Rose-Ackerman 1997. Those authors also note that in the long run local boundaries themselves are ×exible. Introduction 23 cially after the fall of the emperor in 1889, as well as periods of central domi- nation. The record of those experiences is not positive. The advantages con- ferred by money and traditional patron-client ties enormously favored landowners. State-level politicians were able to devalue the currency, aiding im- mobile coffee producers but penalizing the vast majority of consumers. State leaders arranged debt-forgiveness deals beneµting another immobile group, the northeastern sugar producers. State and local leaders created a pervasive sys- tem of pork-barrel projects and political appointments. In sum, because Brazil- ian federalism grew out of central governments’ inability to dominate the na- tional territory, this federalism catered to the power of local interests, especially holders of µxed capital. Federalism provided guarantees to local oligarchs. Rather than facilitating economic progress through the adoption of efµcient policies, federalism advanced the interests of the most backward economic groups and increased regional economic inequality. How does Brazil compare with other federal systems? Federalism always constrains national majorities. In a landmark study, Alfred Stepan (1999) as- sesses the “majority-constraining” quality of twelve federal democracies along four dimensions: the degree of overrepresentation in the territorial chamber, the “policy scope” of that chamber, the degree to which policy-making is constitu- tionally allocated to subunits of the federation, and the degree to which the party system is polity-wide in its orientation and incentive systems.32 After Ar- gentina, Brazil has the greatest overrepresentation of the twelve nations. Its Senate has all the areas of policy-making competence enjoyed by the Chamber of Deputies plus some areas the Chamber lacks. Its constitution is extremely detailed, and in a wide variety of policy areas substantive change necessitates constitutional amendments that require three-µfths of the members of both houses. The electoral system, as part 1 of this book will demonstrate, power- fully hinders the development of polity-wide parties. The Pervasiveness of Pork and Patronage Since the early nineteenth century, Brazilian politics has centered on politi- cians’ attempts to µll bureaucratic jobs with their allies and to supply individu- alized or geographically speciµc political goods, what Americans call pork bar- 32. The twelve systems are Argentina, Australia, Austria, Belgium, Brazil, Canada, Ger- many, India, Russia, Spain, Switzerland, and the United States. Stepan (1999) uses the terms demos constraining and demos enhancing rather than majority constraining and enhancing to avoid im- plying that a majority always exists or is right. 24 The Deadlock of Democracy in Brazil rel, to supporters.33 Without question, political support is exchanged for gov- ernment jobs and public works in every society, but Brazil is unique in the per- vasiveness of these exchanges and in their tendency to substitute for broader, more ideologically or programmatically driven policy-making. Pork and pa- tronage particularize policy-making. Politicians sustain themselves not by pro- moting local prosperity and providing public goods but by supplying pork and services to individuals. My emphasis on pork and patronage is not just another label for Brazil’s political clientelism. For two reasons, I deliberately avoid the term clientelism. First, the corrupt, vote-buying behavior of politicians that is commonly labeled clientelistic is more often an outcome of Brazil’s institutional structure than a precondition for it. Second, discussions of clientelism frequently drag along in- tellectual baggage that is wrong or misleading. Anthropological uses of clien- telism typically refer to individual exchanges of private goods between actors of unequal power, actors called patrons and clients (Greenµeld 1977). The ori- gins of such relationships are said to lie in “traditional” rural society, in the ties between landowner and peasant. They are based on reciprocity, trust, and loy- alty. Implicitly or explicitly, “modern” society rejects such relationships in fa- vor of ideological or group-based links. As a recent essay by Geert Banck (1999) makes clear, the anthropologists’ concept of clientelism travels poorly to the political realm. Trust and loyalty are not central to contemporary exchanges of patronage and pork precisely because such exchanges are deals (negócios) between traders in a political marketplace. Buyers and sellers have to prove the quality and reliability of their goods and their commitments. Loyalty and trust are irrelevant. And since the state, as the source of resources, is intrinsic to the transaction, patronage and pork are as much urban as rural in origin. As the discussion of federalism illustrated, nineteenth-century property owners wanted a strong, centralized government. Only a strong central gov- ernment could guarantee order, and in a slave society of grossly unequal wealth, order was the primary concern.34 At the same time, property owners worked within a formal structure of politics in which elections, though restricted to a 33. To any student of developing-country politics, the pervasiveness in the U.S. media of pork-barrel accusations is laughably exaggerated. See Stein and Bickers 1995 for a demonstration of the unimportance of pork in the United States. As to patronage, a prominent U.S. congressman for whom I worked had at most two patronage positions (excluding his personal staff) to which he controlled appointments. 34. The following discussion of nineteenth-century politics owes a great deal to Graham 1990. For criticisms of Graham’s use of clientelism, see Bezerra 1999 and José Murilo de Carvalho 1997. Introduction 25 small number of propertied males, were important. The emperor and his cabi- net sat at the pinnacle of the system, but the indirectly elected Chamber of Deputies had to ratify the cabinet’s decisions. Thus the cabinet needed the sup- port of the deputies. Both the Conservatives and Liberals, the parties dominating the Chamber of Deputies throughout the nineteenth century, were fundamentally patronage vehicles, exchanging government posts for votes. Both in the Chamber and at the local level, party meant simply an afµliation, not a durable commitment to program or policy. Parliamentary parties formed, split, and reformed, taking ap- parently contradictory positions on major issues right up to the empire’s end in 1889. In the electorate, personal ties, not ideological considerations, deter- mined political divisions. As Graham (1990, 148–49) makes clear, the rivalry and violence of local politics often came not from two distinct parties but from two factions both claiming to belong to the party then in power. The cabinet de- sired electoral support from the local faction most likely to win; from its point of view, party label was irrelevant. The cabinet controlled deputies by granting patronage to or withholding patronage from local patrons. In Graham’s words (1990, 148), “power f×owed simultaneously ‘downwards’ from the Cabinet through the provincial president and ‘upward’ from local bigwigs to the president and Cabinet.” Seekers of bu- reaucratic positions pursued the preservation or improvement of their place. Job seekers justiµed their claims, in the rigid hierarchy of Brazilian society, on the basis of social place, deference, and constant loyalty. Graham (1990, 217) shows that deputies and senators, more than other ofµcials, acted as intermedi- aries in the search for jobs. Legislators’ requests went primarily to the prime minister and to the ministers of justice, agriculture and public works, and war. Over the long run, pervasive patronage affected the quality of public em- ployees, their behavior, and the content of policy itself. Government appoint- ments became political rather than merit-based not merely at the level of min- ister or secretary-general but also at µve, six, or even more levels down.35 Political parties expected to nominate party faithful to quite technical jobs, and major disputes developed over “fair division” of the spoils (Geddes 1994a). Turnover in technical positions has always been quite high, since each ad- ministration replaces its predecessor’s appointees. Because holders of high 35. For purposes of comparison, in Colombia the secretary of basic education, the third- highest post in the Education Ministry, directs all elementary and secondary schooling. A techni- cal person holds that job. In Brazil under President Collor, the equivalent position was held by an appointee with no experience who was a former teacher of the president’s son. All jobs for the next four to µve levels down were also political. 26 The Deadlock of Democracy in Brazil bureaucratic positions often expect to run for elected posts, they use their posi- tions to develop personal followings. Politicians are unlikely to devote much effort to making the bureaucracy less oppressive and remote, because they proµt from mediating between constituents and the distrusted civil servants. The fall of the empire in 1889 and its replacement with Republican gov- ernment had little effect on the centrality of patronage. With the twentieth cen- tury’s expansion of the scope of the state, however, politicians developed a new resource in the form of public works: dams, roads, infrastructure, and so on. Be- ginning in the 1930s and accelerating in the 1950s, Brazilian governments adopted a policy of import substitution industrialization (ISI). Characterized by inward-looking growth, ISI included tariffs, import quotas, denial of mineral rights to foreign capital, nationalization of foreign-owned utilities, and over- valuation of the currency (facilitating easy importation of capital goods). Gov- ernment invested in “strategic” areas: iron and steel, alkali processing, aircraft and truck engines, river valley development. Patronage-based politics adapts easily to state-led industrialization.36 Chubb (1981) makes this point effectively in her discussion of Italy’s south. In Brazil, faced with the challenge of accommodating the diverse regional and economic interests whose cooperation was critical, the state responded by politicizing its programs and by buying off regional and economic interests through growth and subsidies. In 1950 public administration absorbed 3 per- cent of the economically active population; by 1990 that percentage had reached 5 percent (Brazil 1990). Regional politicians, big farmers, industrial- ists—whoever had sufµcient clout commanded a subsidy.37 Inside the state apparatus, the expansionary drive of the state privileged bu- reaucrats themselves. State banks offered their employees higher rates of return on the investment opportunities available to the general public. University pro- 36. A system characterized by pork-barrel politics and patronage does not require that the government supply large quantities of resources in the form of pork-barrel beneµts. If resources are plentiful, in fact, brokers lose their monopoly and hence their control, so patronage can thrive in situations of scarcity and uncertainty. 37. Certain government programs were “privatized” long before neoliberal economics. In the heavy construction sector, for example, a few µrms dominate and have become deeply involved in campaign µnance. Government reciprocates by shaping major construction projects to beneµt the builders. Few observers were surprised by the 1993 revelation that construction companies had been paying legislators hundreds of millions in kickbacks. But construction companies are not alone in shaping public policy to beneµt themselves. For years, for example, the central government funded a program providing meals for schoolchildren. Rather than sending money to the schools to buy lo- cal products, the central government built warehouses and shipped processed food to the schools. Transportation costs were high, schools ran out of food, and local farmers lost potential markets. The program’s logic was political; its real beneµciaries were large food processors. Introduction 27 fessors retired in their forties with generous pensions. Though irrational privi- leges can doubtless be found in government agencies everywhere, in Brazil the scope and depth of the largesse became exceptional. The combination of state expansionism and patronage yielded not just a bit more in-house corruption or a bit more bias toward the private sector, but a hypertrophied monster.38 My stress on patronage and pork does not imply that Brazil lacks politi- cians devoted to programmatic objectives. The Congress contains a solid con- tingent of deputies and senators whose interests lie in legislation affecting the whole society. But as this book unfolds, it will become clear that such issue- oriented legislators are a minority. Politicians whose careers center on supply- ing public works and bureaucratic jobs dominated the conventions that created Brazil’s constitutions, and the same politicians have dominated the legislatures those constitutions created. Such politicians have a hard time seeing political beneµt in rules that strengthen parties or minimize incentives to distribute pork. Historical Continuities and Their Consequences Three historical continuities are important to understanding the linkage be- tween, on the one hand, federalism and pervasive patronage cum pork, and, on the other, the choice of institutions and long-run political outcomes. The µrst continuity concerns institutional traditions themselves. The framers of Brazil’s 1988 Constitution preserved the institutional framework under which they had lived between 1947 and 1964, before the military coup. The second continuity is one of personnel. A substantial number of congressional deputies and sena- tors began their political careers during the pluralist or military periods, and they brought a particular set of preferences to the choice of institutions. The third continuity is one of state-level political organization, a continuity result- ing from the legacy of the military regime itself. In what kind of setting was Brazil’s new constitution framed? The Con- 38. The private economic exchanges characteristic of patronage-based systems are also in- herent in populism, a concept crucial to post-1945 Latin American politics. Unlike its U.S. coun- terpart, Latin American populism is primarily urban. As cities swelled with the in×ux of millions of rural immigrants, political entrepreneurs searched for ways to compete for support with social- ists and communists, the traditional working-class parties. Populists emphasized social welfare pro- grams and immediate beneµts, including government jobs. They bypassed intermediate political or- ganizations such as class-based parties to forge direct links with followers. Economic nationalism helped build coalitions with domestic industrialists. Populist politicians like Vargas and Juan Perón maintained their popularity—and the support of domestic business—as long as their economies en- joyed vigorous growth. For a deµnitive treatment of the pre-Republican foundations of clientelism, see Graham 1990. 28 The Deadlock of Democracy in Brazil stituent Assembly of 1987–88 was simply a joint meeting, held every morning, of the Senate and the Chamber of Deputies. The deputies had been elected to the Chamber in 1986 under the same open-list proportional representation that Brazil had used since 1947, while senators continued under majoritarian rules. Considerable discussion about alternatives to open-list proportional represen- tation took place, and the delegates extensively debated proposals for a switch from presidentialism to parliamentarism. Not surprisingly, they could not bring themselves to change the electoral rules that had brought them victory.39 Perhaps the constitution framers retained the institutional framework of the 1947–64 democracy simply because it had elected them, but that explanation seems simplistic. I will now consider the problem through the optics of the veto- players framework, the role of ideas, and the effects of interests and preferences. The evidence for an excess of veto players is inconclusive but suggestive. The PMDB, the Party of the Brazilian Democratic Movement, had an absolute ma- jority in both chambers at the beginning of this period (in 1986); by the method of complete absorption, the number of veto players is one. But the party was undisciplined and factionalized. During the Constituent Assembly dozens of deputies from the PMDB’s left wing deserted to form the Brazilian Social De- mocratic Party, the PSDB, and even before the split the PMDB had ceased to en- joy any sort of programmatic coherence or legislative discipline. While the num- ber of effective veto players may never have climbed as high as it did in the next administration (reaching seven), it likely did contribute to the absence of signiµcant institutional change. Equally important, however, was the power of pork: President Sarney found many delegates indifferent to institutional issues but not at all indifferent to the friendly persuasion of the pork barrel. To understand the role of ideas in the Constituent Assembly, a picture of the delegates’ perceptions of the 1964 coup is needed. Was there a sense that institutional pathologies played a part in democracy’s breakdown?40 Though 39. I do not deny the existence of cases elsewhere in Latin America in which sitting legisla- tors have changed the rules that brought them victory. In these cases researchers might examine party discipline and, in the more disciplined cases, the goals of party leaders as well as the expected career trajectories of individual legislators. 40. Wanderley Guilherme dos Santos (1979) attributes the coup to legislative deadlock. Per- manent stalemate, he argues, resulted from presidential inability to command a majority, the con- servatism of a legislature elected in malapportioned districts and boss-controlled rural areas, and the weakness of party discipline. Youssef Cohen (1994) argues that moderates on both left and right found themselves in a classic “prisoner’s dilemma.” Moderates in the major parties could have agreed on reforms, but they had to disavow the extremists in their own camps. Neither group of moderates did so, because each feared the other would retain its links to its own extreme wing. The moderate right thought Goulart was seeking dictatorial powers. Goulart himself believed that the conservative moderates were plotting with right-wing soldiers and businessmen to block reforms Introduction 29 many PMDB leaders strongly supported parliamentarism, my interviews with delegates suggest that only a small fraction of the overall body of constitution writers blamed democracy’s failure on institutional factors, at least in a broad sense.41 The delegates were more likely to focus on short-term triggers: the ex- cesses of populist leaders, reactionary business interests (supported by the United States), the sheer incompetence of civilian President João Goulart, and the downturn in the business cycle. In the end, the Constituent Assembly nearly adopted parliamentarism, but the weight of the delegates’ interests led them to maintain the institutional sta- tus quo. A majority of deputies from the more developed states supported a switch to parliamentarism, as did most deputies on the Left, but the issue got caught up in the survival strategies of President José Sarney and other party leaders. As I demonstrate in chapter 5, Sarney’s immense pork-barrel power persuaded many deputies to support preserving presidentialism and lengthen- ing his term. Presidential hopefuls Leonel Brizola and Orestes Quércia (former governor and governor of Rio de Janeiro and São Paulo, respectively) also kept their supporters out of the parliamentarist camp. I will return to this issue in chapter 7; sufµce it to say here that without the executive’s ability to persuade deputies who were otherwise indifferent on the issue, and without the tactics of potential presidential candidates, parliamentarism would have triumphed. The question of the interests of the constitutional delegates leads to the sec- ond continuity, the continuity of personnel. As Francis Hagopian (1996, 246) demonstrates, Brazil’s traditional political families—those dominating politics from 1947 to 1964—survived and prospered through the military regime and into the current New Republic. Six governors elected in 1990, for example, had served in the mid-1980s as governors from the promilitary PDS party. The largest party in the Constituent Assembly, the PMDB, had historically opposed the military regime, but by 1986 one-fourth of its delegation were former mem- bers of the right-wing, promilitary ARENA party (Fleischer 1987, 2). Overall, more assembly members had begun their careers during the authoritarian regime with the promilitary party than with the opposition. and overthrow his administration. Thus moderates on both sides feared that if they jettisoned their radical allies, they would be overwhelmed by the other side—moderates plus extremists—and the outcome would be unacceptable. It was rational, therefore, for moderates to retain their ties to ex- tremists in their own camps. No one wanted the stalemate and the military coup, but both sides ended up with a worse outcome than if they had cooperated. 41. The constitution writers included one narrow institutional modiµcation: the president and vice president are no longer elected on separate ballots. Thus the situation of 1960, when a cen- ter-right president was elected alongside a left-wing vice president from the losing presidential ticket, is no longer possible. 30 The Deadlock of Democracy in Brazil Why stress the durability of the traditional political oligarchy? The mili- tary government repressed the Left but made center and right-wing politicians the incumbents in the political marketplace of Brazil’s new democracy. In ef- fect, these leaders got a head start, and political competition was waged on their terms. Most of the conservative delegates to the Constituent Assembly of 1987–88 had built personal political machines based on patronage politics; they knew that the electoral system would have a huge impact on their machines’ fu- ture.42 But the strategists behind the military regime did more than simply en- courage right-wing politicians. As the military withdrew from power, the gen- erals sought to safeguard their legacy by creating conditions maximizing postdeparture support. The regime tried to increase the in×uence of potential supporters and reduce the weight of likely opponents. Merging two industrial states, for example, eliminated three senators from an opposition stronghold. The formation of new states on the frontier created additional legislators likely to be conservative. Fomenting industrial growth away from Rio and São Paulo could increase jobs in the conservative Northeast. These strategic moves came on top of a long-standing malapportionment that cut São Paulo’s proportional representation in the Chamber of Deputies by about 40 percent, or about µfty seats.43 The result of all these forces was simple: the entire spectrum of political discourse moved to the right. In the Constituent Assembly and in the legisla- tures that followed, the delegates increased the length of President Sarney’s term and preserved presidentialism. They rejected agrarian reform, a German- style mixed-district electoral system, and party and administrative reform. Ob- viously, no one can say with certainty what would have happened if right-wing politicians had not been so favored, but (as chapter 1 will demonstrate) the closeness of votes on these issues suggests that many would have turned out differently. The third continuity is what I have referred to as the continuity of state- level political organizations. The Brazilian military regime (unlike its bureau- cratic-authoritarian counterpart in Chile) never implemented an overall bu- reaucratic reform. Instead, the junta imitated civilian President Juscelino Kubitschek (1956–61), creating, strengthening, and insulating agencies central to its economic project. Agencies delivering social services remained disor- 42. Even presidential elections re×ected the effects of local power: in the 1989 contest local political machines played a signiµcant part in an election supposedly dominated by national media messages (Ames 1994b). 43. This µgure is about 10 percent of the total Chamber membership (see table 1). Introduction 31 ganized and patronage ridden. Where state-level politicos had good connections with the junta and where they had the luck and skill to achieve momentary dom- inance, the availability of massive central-government resources enabled the construction of durable political machines that still exist today. These machines wielded a political force out of proportion to their numeric or economic strength, dominating their states’ electoral processes and controlling their del- egations in Congress. Only a few of these machines have survived until today, but, as chapter 4 will demonstrate, they remain a key part of the story. The Constitution of 1988 augmented the power of states and municipali- ties by increasing their share of overall tax revenues.44 To understand why fed- eral deputies and senators would weaken the central government, it is neces- sary to remember that career trajectories in Brazil differ from those of U.S. politicians. Turnover in the Chamber of Deputies is around 50 percent per term, but electoral defeat accounts for only about half of this µgure. Deputies simply do not seek indeµnite congressional careers (Samuels 2001). Legislators may run for state and local ofµces after a term or two in the federal chamber or re- turn to private business and then seek reelection. Federalism made states and municipalities into political arenas as desirable as the federal capital, and deputies acted to preserve those arenas’ prerogatives. At the same time, strengthening the states—in a context of weak parties—reinforced the power of state governors over their delegations in the Congress, especially in the Chamber of Deputies, and augmented their ability to nominate allies and veto enemies for cabinet positions and high bureaucratic posts (Abrucio 1998). A Note on Methods; or, Why I Did What I Did Scholarship on Latin American politics, including work by Latin Americans themselves, is certainly the most advanced of all developing regions. So many scholars have undertaken systematic research projects that there is far too much to cite here. Still relatively rare, however, is research that combines theory with empirical analysis and applies this perspective to a broad topic. This book tries to comprehend Brazil’s national-level political structure by marrying a consis- tent theoretical perspective—the perspective of rational choice theory—to 44. Between 1985 and 1993, the federal share of all public resources declined from 44.6 per- cent to 36.5 percent. Federal resources as a share of the gross domestic product (PIB) fell from 6.7 percent to 5.7 percent. State resources grew from 5.6 percent of the PIB to 6.3 percent, while the municipal share grew from 2.7 percent to 3.5 percent (see Rezende 1990, 161). 32 The Deadlock of Democracy in Brazil extensive empirical veriµcation. The book is framed as a series of puzzles. Not every puzzle is solved with equal precision, and the theoretical arguments go further in some cases than others, but every part follows the same pattern: the- oretical argument, hypotheses, empirical resolution. Rational choice arguments applied to political systems in Latin America make no sense unless accompanied by efforts at empirical veriµcation. Other- wise, these arguments cannot be falsiµed. In a sense, when forced to choose be- tween theoretical elegance and explanatory completeness, I opt for the latter. This commitment to empirical explanation may pose problems for readers un- familiar with Brazil. The nation has enormous cross-regional variation, and since states are important actors in national-level politics, this variation is of- ten an important factor in determining national political outcomes. In chapter 4, for example, I discuss patterns of political competition in groups of particu- lar states. At times, chance political events, random shocks no general theory would consider relevant, in×uence long-run outcomes in these states. The dan- ger, of course, is that explanations sacriµce breadth as they gain accuracy. In general, I include explanatory factors if they affect outcomes over the long term; I exclude factors whose effects are brief.45 Because my goal is to explain real political outcomes, I use institutional arguments probabilistically rather than deterministically. In other words, I pre- dict propensities rather than certainties. The fact that Brazil’s electoral system stimulates deputies to focus on pork does not mean that no deputy will focus on national issues. The fact that hundreds of deputies support macroeconomic stabilization policies only when paid off does not mean that stabilization can never be achieved. In fact, chapter 8 will show that legislators yielded a certain portion of their access to pork when they feared that the alternative was a vic- tory in the upcoming presidential election by the leftist candidate, Lula. A Lula victory, they understood, might lead to a permanent rather than temporary end to business as usual. Brazil as a Case Study If all studies focusing on a single country automatically deserve the dreaded la- bel of case study, I obviously plead guilty. More sensitive criticisms are based 45. For example, in a cross-national model explaining public expenditures, I once included a dummy variable for an earthquake in Ecuador. Under this criterion, I would include the earth- quake dummy if expenditures jumped to a higher level and remained there and exclude the dummy if expenditures soon returned to their pre-earthquake level. Introduction 33 on two criteria. Does the central puzzle interest only traditional country ex- perts? Do the explanatory concepts come from the case itself and remain in- variant within it? Hopefully, neither criterion applies here.46 The functioning of electoral and legislative political institutions interests scholars, politicians, and citizens in all competitive regimes. The core theoretical argument derives mainly from rational choice reasoning combined with historical institutional- ism, and there is sufµcient internal variation to enable hypothesis testing with standard empirical techniques. Criticisms of case studies often assume that each variable in a given case is only measured once. In assessing the scientiµc utility of research, however, the real issue is not the number of cases but the number of observations. As King, Keohane, and Verba (1994, 52) point out, only the number of observa- tions is important in judging “the amount of information a study brings to bear on a theoretical question.” Repeating Eckstein’s (1975, 85) famous example, “A study of six general elections in Britain may be, but need not be, an n = 1 study. It might also be an n = 6 study. It can also be an n = 120,000,000 study. It de- pends on whether the subject of study is electoral systems, elections or voters.” The data utilized in this book vary enormously. Here is a partial list: mu- nicipal-level electoral returns for thousands of congressional candidates over µve elections (between 1978 and 1994) in nineteen different states; 14,000 grants from the central government to individual municipalities; more than 10,000 budgetary amendments submitted by individual deputies to the con- gressional budget committee; more than 200 interviews with deputies, journal- ists, and academics; schedules of private meetings between ministers and deputies; and historical materials concerning particular states. In essence, the data provide measures across states, time periods, and levels of government. In other words, this book is about Brazil, but it is not a case study. Interviews as Data While the project ultimately employed a wide variety of information, I did not begin with all the data in hand or even in sight. At the start, the only hard data were electoral returns from ten states for the µrst two elections. More elections were scheduled, of course, but I expected interviews to become the central source of information complementing the electoral results. It turned out differently. 46. Ironically, this criticism applies much more strongly to those who make it most often— that is, scholars in American politics. I will return to this topic in the conclusion. 34 The Deadlock of Democracy in Brazil Interviewing congressional deputies is not fun. The µrst couple of times you make an appointment, the deputy cancels or simply fails to show up. The third time you wait an hour or two until the deputy arrives. When the interview µnally begins, the deputy assumes you need a lecture on Brazilian history, so the conversation starts with the 1930 revolution. When you get past the history lesson, lies begin. Interviewing is not µt work even for graduate students. Interviews really are not that bad. But since they are relatively easy to arrange, too many studies rely on interviews without considering their inherent problems. Interviews with key informants such as politicians almost never con- stitute samples in the scientiµc sense, not least because politicians willing to be interviewed tend to be the more high-minded, public-spirited types. And al- though we all know that information

The Deadlock of Democracy in Brazil (2001) PDF - Introduction

Document Details

Tags

Related

Summary

Full Transcript

Upgrade to continue