From Patronage to Program

FROM PATRONAGE TO PROGRAM: THE EMERGENCE OF PARTY-ORIENTED LEGISLATORS IN BRAZIL Frances Hagopian, Carlos Gervasoni, and Juan Andres Moraes Working Paper #344 – December 2007 Frances Hagopian is Michael P. Grace II Associate Professor of Latin American Studies in the Department of Political Science and a faculty fellow of the Kellogg Institute at the University of Notre Dame. Her recent works include The Third Wave of Democratization in Latin America (Cambridge University Press, 2005), (coedited with Scott Mainwaring) and Religious Pluralism, Democracy, and the Catholic Church in Latin America (Notre Dame University Press, forthcoming). Carlos Gervasoni (MA Political Science, Latin American Studies, Stanford University) is a PhD candidate in the Political Science Department at the University of Notre Dame and a professor of political science (on leave) at the Pontificia Universidad Católica Argentina and Universidad Torcuato Di Tella in Buenos Aires. He has published several articles on the politics of economic reform in Latin America, Argentine politics, and electoral behavior. Juan Andres Moraes is a PhD candidate in the Political Science Department at the University of Notre Dame. His dissertation focuses on constituency service in Latin America, with emphasis on Uruguay and Costa Rica. He is coauthor (with Daniel Buquet and Daniel Chasquetti) of Fragmentación Política y Gobierno en Uruguay: ¿Un enfermo imaginario? [Political Fragmentation and Government in Uruguay: An Imaginary Patient?] (Universidad de la República, 1998). Acknowledgements We thank Diego Miranda, Guilherme Canela, and students at the University of Brasília for help with administering the surveys on which this article is based; Fabiano Santos and Marcio Grijó Vilarouca for sharing data on legislative roll-call votes in the 51st legislature from the NECON/IUPERJ databank; Jon Bischof for superb research assistance; and David Nickerson for methodological advice. We also thank for comments on earlier versions of this work participants in the 2005 Kellogg Institute Regional Workshops, Michael Coppedge, Wendy Hunter, Scott Mainwaring, Carlos Pereira, and Ben Schneider, and at the David Rockefeller Center for Latin American Studies at Harvard University, John Carey, Scott Desposato, Jorge Domínguez, Gretchen Helmke, Mala Htun, and Steven Levitsky. 2 3 ABSTRACT This paper explains the unanticipated emergence of party-oriented legislators and rising party discipline in Brazil since the early 1990s. We contend that deputies in Brazil have become increasingly party-oriented because the utilities of party-programmatic and patronage-based electoral strategies shifted with market reforms, which created a programmatic cleavage in Brazilian politics and diminished the resource base for state patronage. Based on an original survey of the Brazilian Congress, we introduce new measures of partisan campaigns, party polarization, and the values legislators attach to party program and voter loyalty. Regression analysis confirms that deputies who believe voters value party program run partisan, programmatic campaigns, and those in polarized parties and those who believe voters are loyal to the party are willing to delegate authority to party leaders and do not switch parties. Party polarization and the proximity of deputies’ policy preferences to their party’s mean explain discipline on 236 roll-call votes in the 51st legislature (1999–2001). RESUMEN Este artículo explica el inesperado surgimiento de legisladores orientados hacia los partidos y la creciente disciplina partidaria en Brasil desde principios de los 90s. Sostenemos que los diputados en Brasil han devenido crecientemente orientados hacia los partidos porque las utilidades de las estrategias electorales partido-programáticas y basadas en el patronazgo cambiaron con las reformas de mercado, las que crearon un clivaje programático en la política brasileña y redujeron la base de recursos para el patronazgo estatal. Con base en una encuesta original del Congreso Brasileño, presentamos nuevas mediciones acerca de las campañas partidarias, la polarización de partidos y los valores que los legisladores asignan al programa partidario y a la lealtad de los votantes. El análisis de regresión confirma que los diputados que creen que los votantes valoran el programa del partido, desarrollan campañas partidarias y programáticas y que aquellos que se encuentran en partidos polarizados y quienes creen que los votantes son leales al partido, están dispuestos a delegar autoridad a los líderes de los partidos y no cambian de partido. La polarización de los partidos y la proximidad de las preferencias de política de los diputados respecto de la media de sus partidos explican la disciplina en 236 votaciones nominales en la 51º legislatura (1999–2001). 4 Hagopian, Gervasoni, and Moraes 1 Why, in emerging democracies, where stable bonds between parties and voters are often elusive and candidates compete for office more on the basis of the particularistic benefits they can deliver to constituents than on their party’s programs of government, do party parliamentary delegations become more disciplined and parties and politicians transform the basis of their linkages with voters from clientelism to program? The move to partyoriented representation is typically explained, if at all, in one of two ways: as a response to voters demanding—and rewarding—disciplined, programmatic challengers at the polls, usually because their political tastes change along with socioeconomic development; or as the result of change imposed by leaders who enjoy centralized control over their parties. There has been less theoretical and empirical attention paid to the alternative—incumbent politicians voluntarily surrender the autonomy they enjoy under a particularistic system to their party leaders and shift the basis on which they mobilize voters. This article focuses on such a possibility in the case of Brazil. Brazilian parties of the center and right, hobbled by shallow roots in the electorate and high rates of electoral volatility, have long been reputed to be among the weakest in Latin America. In a system marked by permissive party legislation, electoral rules that disjoin legislators from party leaders, and a brand of decentralized federalism that makes nonlegislative political careers attractive, the common wisdom is that opportunistic politicians hoping to build post-legislative careers in state and municipal executive office switch parties at will and defect from their parties in legislative committees and on the floor of Congress because leaders lack ballot control and other means with which to sanction them. Unreliable parliamentary majorities, in turn, supposedly make party governance nearly impossible or at minimum very expensive, as deputies must be bought. Even former President Fernando Henrique Cardoso (1996: 13) once lamented that Brazil’s fragmented, “ideologically fuzzy” political parties and autonomous politicians made the task of gathering a majority for a bill a “case-by-case exercise” requiring arduous negotiations. Of late, however, there are signs that Brazilian parties are growing stronger: electoral volatility is down and party unity in congressional voting has risen. Arguably, parties are being transformed from loose patronage machines to programmatically coherent and distinctive groupings. Key to this transformation is the emergence of 2 Hagopian, Gervasoni, and Moraes legislators that forego parochial campaigns, express willingness to delegate more authority to party leaders, and accept party discipline in floor votes. Such party-oriented behavior was once believed to be unique to members of the Workers’ Party (PT)—the 2005 corruption scandal that exposed illegal campaign contributions and monthly sidepayments to members of Congress in exchange for votes notwithstanding—but there is mounting evidence that legislators across the board have exhibited such support of parties. This article attempts to explain the emergence in the 1990s of party-oriented politicians, which we take to be a prelude to stronger parties. We contend that legislators invest in their parties when the value of their parties’ reputations rises relative to their personal ones, and that they look to parties to organize their competition in a climate of growing insecurity over their ability to compete independently. The reasons behind such a shift in the value of party and personal reputations can vary; in Brazil in the 1990s, the catalyst was sweeping state and market reform. Once party legislative delegations began to cohere on a range of contentious economic and administrative reforms and diverge from their opponents, a distinctive, programmatic cleavage emerged in legislative voting. At the same time that parties were becoming homogenized and polarized, creating value for the party brands, state reforms that shrank the scope and resource base for state patronage diminished the efficiency of personal, patronage-based electoral strategies. A favorable response by voters to the offer of cohesive and programmatic parties reinforced the shift from patronage and the personal vote to party and program. To support our claim that Brazilian politicians are in fact becoming more partyoriented in both the electoral and legislative arenas, we move beyond the familiar records of floor voting and party switching to examine also the basis of deputies’ campaign strategies—the first study that we know of to do so—as well as attitudes toward delegating resources and authority to party leaders. We also introduce new measures of individual and partisan policy preferences, progressive career ambition, contact with public officials, and politicians’ beliefs about the reasons for their party’s electoral success. These measures are built upon responses to a survey of the Brazilian Congress conducted in 1999-2001. While elite surveys have potential pitfalls—most notably, respondents may exaggerate the nobility of their work or conceal their true preferences Hagopian, Gervasoni, and Moraes 3 and behavior, especially if they engage in clientelism, patronage or corruption (Kitschelt, 2000: 869)—surveys allow us to observe directly individual attitudes, motivations, and behaviors, enabling a more powerful individual-level analysis of the orientation of politicians toward parties and constituents than would otherwise be possible. While our ability to draw conclusions about the degree and source of change is limited by the fact that our survey captures the attitudes and behavior of politicians from only one Congress, comparable legislative surveys conducted in 1988 (Mainwaring, 1999) and 1990 (Power, 2000) provide a baseline to chart at least some attitudinal change over time. The paper proceeds as follows. First, we situate the theoretical debates about the partisan and parochial behavior of legislators in the context of Brazil, and propose an alternative explanation that attributes the emergence of party-oriented legislators to the development of a programmatic dimension of party competition and the increase in the value of party labels commensurate with a decline in patronage. Electoral, legislative, and survey data confirm the rise of aggregate levels of party cohesion, programmatic competition, partisan campaigns, legislative discipline, and pro-party attitudes. Next, we test our hypotheses and several competing ones at the individual level. We show that 1) the more polarized their parties, the more legislators perceive voters are loyal to their party brand and value its program, and 2) the more they cohere around their party’s positions on key issues, the more loyal, disciplined, and party-oriented they are. The final part concludes. PARTY AND PATRONAGE IN BRAZIL Why do politicians privilege their personal reputations over party loyalty? One view holds that where voters come cheap—they are poor, relatively uneducated, live in rural isolation, cast ballots that can be easily monitored, and heavily discount the future— politicians will make personal and especially patronage-based appeals. Another maintains that politicians respond to their institutional incentives; in systems with candidatecentered electoral rules, they will cultivate a personal vote. Especially important are rules governing ballot control, vote pooling, vote types, and district magnitude (Carey and Shugart, 1995), but in federal systems or those with proscriptions on immediate 4 Hagopian, Gervasoni, and Moraes reelection, politicians are also accountable to whoever holds the key to their future career ambition. For many years, these theories appeared to explain why Brazilian parties were fragmented, had shallow roots in the electorate, and lacked legislative discipline.1 On the one hand, Brazilian politicians competed for the votes of the poor and barely literate in an underdeveloped society whose late industrialization had delayed the emergence of a working-class constituency, and parties had been weakened by a strong state and a neocorporatist, neopatrimonial order (Faoro, 1958; Campello de Souza, 1976: 31, 36; Mainwaring, 1999: 225-32).2 On the other, elected under a system of open-list proportional representation in large, statewide, multimember districts (ranging in magnitude from 8 to 70 and averaging 19.3) in which voters cast a single ballot for the candidate of their choice, legislators faced with internal party competition had cause to stake out autonomy from party leaders who did not control ballot access and to develop their personal reputations (Mainwaring, 1999: 243–62; Ames and Power, 2007). Also detracting from national party unity were federalism and fiscal decentralization that by making state- and even municipal-level careers attractive options to reelection shifted the focus of politicians to state and local parties and diminished their incentive to respond to national party leaders (Mainwaring, 1999: 264-66; Samuels, 2003). Enough (10 to 15 percent) consistently voted against the majority of their parties on key roll-call votes to force presidents to assembly fiscally costly multiparty coalitions (Lamounier, 2003: 275). Moreover, in response to the substantial distributive rewards for joining a governing coalition or even forming a new party (all leaders would hold seats on legislative governing and agenda-setting bodies) and in the absence of sanctions by parties or voters, party switching was endemic. From 1991 to 1995, 32 percent of deputies in the Chamber changed affiliation, many more than once and even to ideologically noncontiguous parties (Melo, 2004: 65-66, 102).3 But beginning in the mid-1990s, parties began to sink roots into the electorate and engender greater loyalty from politicians. Volatility rates in lower chamber elections fell from 45 percent in 1990 to 21 percent in 1994, and again to 12.5 percent in 1998, a rate comparable to those of the most institutionalized party systems in Latin America from 1980 to 1997—Argentina (13 percent), Chile (10 percent), and Mexico (15 percent) Hagopian, Gervasoni, and Moraes 5 (Roberts and Wibbels, 1999: 577). At the same time, the number of deputies that defected from their parties fell modestly to 26 percent (with fewer jumping to noncontiguous ideological camps) (Melo, 2004: 65, 102), and party discipline rose. The Rice Index score for the Congress as a whole climbed from 68 to 80 from 1986–90 to 1995–99, and it did so in the catchall parties with low barriers to entry: between the 48th Congress (1987– 1991) and the 51st (1999–2003) party unity rose in the Party of the Liberal Front (PFL) from 80 to 90, the Party of Brazilian Social Democracy (PSDB) from 67 to 91, and the Party of the Brazilian Democratic Movement (PMDB), from 62 to 72 (Roma, 2004: 77). Even in the PFL, known for its autonomous politicians, only 3 percent of legislators voted against the recommendations of party leaders (with 13 percent abstaining) in the 51st Congress (1998–2002) (Roma, 2004: 81). Why parties with weak programmatic identities and parochial patterns of representation would develop more disciplined legislative delegations is puzzling. Brazil has developed, but not fast enough to account for such impressive party discipline across delegations from undeveloped states, and institutional incentives did not markedly change. It is also doubtful that party leaders were able to impose programmatic change and party discipline, as their counterparts did in Eastern Europe and Argentina (Grzymala-Busse, 2002; Levitsky, 2003). Some scholars contend that legislative rule changes in the late 1980s endowed parliamentary party leaders with substantial agendasetting powers and the authority to represent their delegations in procedural questions, compensating them for candidate-centered electoral rules (Figueiredo and Limongi, 1999: 20–29), as well as the prerogative to make legislative committee assignments and recommend that individual budget amendments be funded. Other scholars (Amorim Neto et al., 2003: 558; Samuels 2004) are skeptical, contending that the appeal of committee appointments is a hollow one and that executives, not party leaders, hold the key to the pork barrel. Indeed, the impact of party loyalty on gaining seats on key congressional standing committees during this period was not significant (Santos, 2002). Moreover, party discipline rates did not begin to rise when the new rules were introduced in 1988 but only after 1994, when Fernando Henrique Cardoso formed a “parliamentary agenda cartel” (Amorim Neto et al., 2003). 6 Hagopian, Gervasoni, and Moraes The thesis attributing party transformation to the demand of voters for more cohesive and disciplined parties and politicians—because they become wealthier or better educated; they want sharper partisan differences (as in the second half of the twentieth century in the United States) (Rohde, 1991; Jacobson, 2000); or they want to influence national policy, as occurred in Britain in the nineteenth century, when institutional change deprived backbenchers of access to pork, patronage, and policy initiative and power shifted to the cabinet and by implication the party that controlled the executive (Cox, 1987: 136)—does not fare much better. Brazilian voters may have somehow overcome their own collective action problem to value collective over particularistic goods (Lyne, 2004), but the evidence that voters in fact demanded more programmatic parties and collective goods in advance of their offer is thin. Even given falling rates of electoral volatility, an identical proportion of the Brazilian electorate—45.6 percent— identified with parties in 1989–1994 and 1995–2002 (Carreirão and Kinzo, 2004: 141– 42). As late as 2002, the only party that was deemed to have true partisans was the PT (Samuels, 2006). Based on in-depth interviews conducted during the 1994 campaign, one researcher characterized the typical Brazilian voter as lacking in political information, moving away from voting on the basis of clientelism but not yet voting on the basis of program, ideology, or party loyalties, and guided by preferences that were “unstable, autonomous, and unpredictable” (Silveira, 1998). Respondents to the 2002 Brazilian National Election Study (ESEB 2002) cited as their most important considerations in voting for federal deputy the candidate’s record (33 percent) and issue positions (32.5 percent), followed by personal qualities (17 percent); only 7 percent mentioned the candidate’s party. One possible solution to our puzzle is that, as the skeptics contend, politicians are not becoming more party-oriented, but legislative delegations are maintaining discipline for strategic reasons (Ames, 2002). According to the “two-arena” thesis, autonomous members support party-backed presidential initiatives in exchange for side payments from the pork barrel critical for their reelection, and thus parties that are apparently disciplined in the legislature are in fact weak and fractious in the electoral arena (Pereira and Mueller 2004). Alternatively, as other leading Brazilian scholars contend, legislators may allow party leaders to coordinate their votes in order to strengthen their collective Hagopian, Gervasoni, and Moraes 7 bargaining position with, and credibly threaten the legislative agenda of, a powerful executive that concentrates budget-making initiative (Figueiredo and Limongi, 2002: 333; Santos, 2002). Both arguments could explain disciplined floor voting, but neither would anticipate a broader pattern of party-oriented behavior and a turn from patronagebased to programmatic linkages between politicians and voters. We propose that politicians are motivated to become party-oriented—to observe party discipline, remain in their parties, campaign on party program, and be willing to invest their leaders with real authority—not merely to bargain with the executive but also to develop their party brands and organize their electoral competition in a context of structural change. FROM PATRONAGE TO PARTY: A THEORY OF SHIFTING UTILITIES The scholarly consensus on the choices of linkage strategies made by parties and politicians generally and in Brazil in particular rests on the assumptions that the resource base for dispensing patronage to politicians and to voters is stable, and that given access to state resources demanded by citizens of the districts that hold the greatest electoral value for them, politicians will cultivate a personal reputation. Further, with most legislators invested in a personal vote, heterogeneous, catchall parties will be a semipermanent feature of the political landscape, unless economic development depresses the demand for patronage and raises the cost of providing it (Kitschelt and Wilkinson, 2007: 24–28). We begin from the premise that if economic reforms were to limit levels and access to federal and state government discretionary and patronage spending—the asset base upon which parties and deputies traditionally made patronage appeals—politicians’ incentives to preserve their autonomy from party leaders and their parties as heterogeneous might also change. When such reforms engender programmatic competition and parties become more homogeneous and polarized—the condition of “conditional party government” (Aldrich and Rohde, 1998: 5)—at the same time that access to patronage became restricted, the utility of personal reputations will diminish relative to the value of the party brand, making party-oriented, programmatic linkage strategies more attractive than personal, patronage-based ones. We can think about this shifting calculus theoretically on two levels—that of individual politicians and parties. Despite the collective incentives to strengthen the party 8 Hagopian, Gervasoni, and Moraes brand, individual legislators may yet succumb to their collective action dilemma, buck the party leadership (which lacks sanctioning power over them), and persist in cultivating a personal vote even with a shrinking patronage war chest. In the end, politicians in systems such as Brazil with open-list proportional representation and large district magnitudes ultimately must compete against one another, and deputies who can bring pork to their districts may be more likely to be reelected than those who cannot or do not (Pereira and Renno, 2003). For some deputies, even outright vote buying might still be an efficient electoral strategy in areas of high poverty or unemployment. But with pork in short supply and access to it restricted, most will not be able to buy votes. Whether they cultivate a personal vote on another basis, such as constituency service, personal experience, or competence; move to valence issues; or stand with their parties in supporting or opposing government policies on salient—and polarizing—issues of the day to gain election, they will now need the party label more than before, at minimum in addition to their own personal reputation (Cox and McCubbins, 1993; Rohde, 1991). The incentives now exist to work to create value for the party label by observing floor discipline, conducting programmatic campaigns, and surrendering autonomy to party leaders to coordinate their competition. In time, a virtuous circle will form: as some deputies invest in and become oriented toward their parties, more will follow suit. This argument necessarily assumes voter demand for the party brand. Even if voters do not yet seek programmatic cohesion specifically and are not the initiators of the move toward programmatic parties, as we have argued was the case in Brazil, voters most certainly would reinforce it by becoming more attuned to policy differences between government and opposition, more appreciative of ideological consistency between national party leaders and members of Congress, and more responsive to party-oriented appeals. An offer of party-oriented representation would have to resonate with voters, or else it would be meaningless. Party strength, of course, can be based on different things, and the emergence of party-oriented legislators provides no assurance that party competition and party-citizen linkages will become more programmatic. To explain the growing value of the party program, we must view the problem from the perspective of party leaders who seek to maximize their party’s electoral fortunes. Let us assume they may choose between Hagopian, Gervasoni, and Moraes 9 developing a program and coordinating the distribution of patronage resources, as Liberal Democratic Party leaders did in Japan before 1994 (McCubbins and Rosenbluth 1995). Even if competition might intensify and the ability to deliver pork might become an even more prized resource for incumbent parties and politicians when the overall pool of resources shrinks (Chubb, 1982), liberalizing economic reforms will reduce the number of candidates who can count on the effectiveness of such a strategy, making it inefficient from the party’s standpoint. If party leaders cannot distribute enough patronage to win elections, claiming credit for good public policy outputs or opposing unpopular ones becomes an attractive alternative. Now not only should deputies adopt more discipline in the legislature to preserve the party brand, but they should also become more attached to the party program. Once more disciplined parliamentary parties make partisan-based electoral competition imaginable, they also make it viable. This argument inverts the logic of the two-arena thesis: parties with program value—those that are polarized on policy—create incentives for autonomous politicians to abandon patronage-based, personal, and parochial campaign strategies in favor of flying the party banner. Theoretically, the emergence of conditional party government creates feedback loops to preserve party discipline in the legislature and partisan value in the electoral arena. We may visualize this argument as follows: Figure 1 THE CHANGING UTILITIES OF PROGRAM AND PATRONAGE Legislative arena Parties’ increasing polarization & homogenization Increased value of party and program Delegation to leaders Permanence in party Economic Reform Fewer resources for patronage Diminished value of patronage-based campaigns Legislators invest more in parties and less in their personal reputations Partisan campaigns Voters demand for programmatic parties Electoral arena Legislative discipline 10 Hagopian, Gervasoni, and Moraes This theory has several implications. In the aggregate, programmatic competition and a decline in patronage should create incentives for legislators to vote with party leaders and campaign on party program. If feedback loops indeed exist, as voters become loyal to programmatic parties we should expect to see a greater incidence of both. We would also expect the reduction in patronage and the emergence of programmatically distinctive parties to lead deputies to confer authority on leaders to apply the whip in floor voting and guard party ballot access. The implications are not as straightforward for party switching. Absent sanctions for party-switching, nonideological deputies who seek career advancement should have an incentive to defect from a party with a weak brand to one with a stronger one, or to a party that offers opportunity for individual career advancement in elective or appointed office. And although party leaders should want to stem the tide in order to provide a credible product to voters, they also have a clear incentive to welcome heavy hitters, as long as the new entrants are willing to observe discipline. At the aggregate level, then, we would not necessarily expect party switching to decline sharply. Our theory also makes several predictions about which parties and politicians should exemplify that behavior in the legislative and electoral arenas. 1. At the party level, politicians in parties whose legislative delegations cohere around a set of key issues on the programmatic divide of state-market reform, and whose mean position on those issues is polarized, that is, farther from the mean of the entire legislature, should exhibit more party-oriented behavior than their counterparts in heterogeneous, centrist parties.4 2. At the individual level, the more deputies a. believe that program (patronage) has electoral value, the more (less) partyoriented they will be in the electoral arena. b. believe that the party label has value, the more willing they will be to delegate authority to party leaders and remain in that party. c. cohere around their party’s positions on key issues, the more disciplined they will be. Hagopian, Gervasoni, and Moraes 11 SURVEYING PARTY CHANGE IN BRAZIL Brazil provides an apt case to apply this theory. The signature reforms of the Cardoso administration—fiscal discipline, privatization of state-owned enterprises, deregulation of markets, and state administration reforms—removed state discretion over economic activity, reduced and professionalized the ranks of federal and state-government employees, and diminished the scope and resource base for patronage and pork (Gaetani and Heredia, 2002). Legislators also imposed severe limits on the number of budget amendments that individual members could sponsor in order to recover their credibility with voters in the wake of the 1993 congressional budget scandal.5 Moreover, the Camata Law of 1995 (strengthened in 1999), which limited all state government payrolls to 60 percent of state revenue, and the Fiscal Responsibility Law (FRL) of 2000, which set strict spending limits for all levels of government and prohibited the central government from refinancing subnational debt, restricted the ability of governors to dispense patronage.6 At the same time that the utility of particularistic strategies of representation was declining, economic reforms requiring legislative supermajorities to amend the constitution to allow foreign investment into strategic sectors, privatize state monopolies, and reform the state administration sparked vibrant congressional and public debates in the 1990s and eventually a programmatic divide. To ascertain whether Brazilian politicians are oriented toward their parties or their constituents, and if and why they are abandoning patronage for program, we must move beyond the voting records, constituency characteristics, and ideological self-placement scores that have not resolved the debate on Brazilian parties. We draw upon our 1999– 2000 survey of the Brazilian Congress to ascertain whether or not partisan legislative delegations have become more cohesive and distinctive and to develop new measures of party-oriented behavior and attitudes. A sixteen-page questionnaire was mailed to every member of the Brazilian Congress in June 1999. Initial responses were supplemented with personal interviews to achieve a sample representative of the Congress by party, region, gender, and level of legislative activity7 (the chi2 statistics testing the null hypothesis that the observed and expected number of cases by party, region, gender, and legislative activity are the same yield p-values of 0.52, 0.91, 0.95, and 0.47 respectively) (See Appendix A). 12 Hagopian, Gervasoni, and Moraes To build a measure of party homogenization and cohesion, we use a simple standard deviation score to measure the dispersion of policy preferences for politicians on a cluster of twenty state and market reforms included in our survey that constitute a single dimension of competition.8 Partisan cohesion rates ranged from a fairly cohesive .40 for the PSDB and .51 for the PT to a moderate .68 for the rightist Brazilian Progressive Party (PPB) (now the PP), and a higher score of .85 for the PMDB and .87 for the PFL (now the Democratas). If not all Brazilian parties are homogeneous, they are comparatively quite programmatically polarized on policy. To gauge the spread of partisan preferences, we measured the distance between mean scores on this cluster of state and market reforms for the PSDB-PFL governing alliance on the one hand and its principal opposition, the PT and other left parties on the other, by calculating the absolute distance between party means, which range from -2 to 2, and dividing these by the maximum spread (4)—an adaptation of the Sani and Sartori (1983) measure of ideological polarization. This distance (.55) was wider than that between the Peronists and the Radical-Frepaso opposition in Argentina in 1997 (.50), and substantially wider than between the Concertación and its rightist opposition in Chile in 1998 (.31) and the PAN and the PRI-PRD opposition in Mexico in 2001 (.31) on similar clusters of statemarket reforms.9 Without a reliable baseline it is hard to make temporal comparisons of party cohesiveness and distinctiveness, but based on an analysis of the ideal points of legislators drawn from policy preferences expressed in 1995 and their voting records from 1986 through 2004, Roma (2004: 69) concluded that party polarization sharpened during the Cardoso presidency (1995–2002), especially in 1996 and 1999.10 To determine whether or not homogeneous and polarized parties also generated the conditions for the emergence of party-oriented legislators, we turn to other survey evidence, which confirms that in the aggregate attitudes also changed over time.11 The percentage of legislators who, when given a choice between favoring the interests of their district (their state), their party, or both would choose their state fell from 56 percent in Mainwaring’s 1988 survey to 40 percent in 1999–2000, and twice as many (22 percent) said they would take the interests of both party and state more or less equally into account in 1999–2000 than in 1988 (when 11 percent did) (Mainwaring, 1999: 160). Brazilian legislators also evinced a greater willingness to surrender autonomy to party leaders. 13 Hagopian, Gervasoni, and Moraes Deputies were asked by Power in 1990 and again in our survey in 1999–2000 to offer their views on whether party leaders should exercise the whip on votes and expel a member who voted against the party line, whether Brazil should adopt a closed-list proportional representation electoral system, and whether there should be sanctions for party switching. While support for application of the party whip was stable and support for expelling members who defied the party whip actually fell slightly, support for a change from open- to closed-list proportional representation increased by a third (from 27 to 36 percent) and support for expelling from the legislature representatives who switched out of the parties on whose ticket they had been elected rose by 15 percent (from 55 to 63 percent) (Mainwaring, 1999: 168). The value of the party label for Brazilian politicians also rose. By 2000, 37 percent of Brazilian politicians believed both their party label and personal reputation were important to their election, and 58 percent attributed their election “almost exclusively” (23 percent) or “largely” (35 percent) to their personal effort, a significant decline since 1990 when 82 percent did (Power, 2000: 126).12 The factors politicians believed most contributed to their party’s electoral success at the state level were the party’s image, the image of its state leaders, and the loyalty of its voters; among the perceived least important factors were the resources the party delivered to the state in the preceding electoral cycle and those anticipated by voters in the future.13 By 1998 deputies on the whole also campaigned to a greater extent on issues and on themes overlapping with what they perceived to be key aspects of their party’s platform than on the basis of patronage, parochial, and personal appeals. Only 9 deputies (out of 119 who answered this question) in the survey identified as one of their three most important campaign appeals in 1998 their promise to work for resources or projects for the municipality or the region. Of 357 (=119*3) total possible campaign messages, only 69 (19 percent) courted a personal vote (34 made parochial appeals, 24 stressed such personal qualities as honesty, integrity, experience, and dedication; and 11 promised patronage). Most appealed to voters on the valence issues that mattered most to them— health, education, corruption/ethics in government, and unemployment (35 percent), on more sharply polarized issues (38 percent), or on party-oriented appeals (4 percent). The remaining 4 percent corresponds to missing answers. 14 Hagopian, Gervasoni, and Moraes Accompanying this shift, voters also attached value to party brands and ideological consistency. In addition to declining electoral volatility, which could be explained by other factors, 14 Lyne (2004: 24–27) contends that “partisan tides” (which measure the change in party fortunes across elections and are generally taken to indicate that some component of voters’ choice is driven by partisan positions on national issues [cf. Cox and McCubbins, 1993]) were in evidence for all major parties except the PDT in the 1990s. She also finds evidence of a clear decline in the pattern of successful deputy candidates gathering most of their vote in municipalities in which they dominate—the archetypal pattern evident in clientelistic systems—in favor of more dispersed votes across their states over the period (Lyne 2004: 31). Voters rewarded ideological consistency in the 1998 legislative elections by punishing incumbents in the rightist coalition that had not voted with the government and their counterparts in the leftist opposition that had not voted against the government (Pereira and Renno, 2003: 442–43). Vote buying campaigns may also be less attractive than they once were; the Catholic Church has spearheaded grassroots campaigns exhorting citizens not to sell their votes but to cast them for parties with proposals for dealing with poverty. CHOOSING PARTY AND PROGRAM: VARIABLES AND MEASUREMENT To learn if Brazilian politicians are investing in strengthening their party labels and programs or voting in a disciplined fashion because they are being handsomely rewarded for their votes requires individual-level analysis. We draw from our survey to develop new measures of politicians’ campaign strategies, willingness to accept rules changes, cohesion on party program, and beliefs about the value voters place on patronage, program, and the party label. We also reproduce measures used by other scholars to test competing hypotheses about the behavior of Brazilian legislators. (See Appendix B for the definition, measurement, and summary statistics for all dependent and independent variables, including the wording of survey questions that generated them). Dependent Variables Our four dependent variables measure politicians’ behavior and attitudes. The first (PARTISAN CAMPAIGN) measures campaign strategies based on program, taking into Hagopian, Gervasoni, and Moraes 15 account both the content of a deputy’s campaign appeals and their overlap with her party’s positions. (See Appendix C for coding rules). The second (DELEGATION) represents an index of average agreement with four party-oriented rule changes: (1) the party whip should be applied to floor votes; (2) deputies should be expelled from their party for voting against the party whip; (3) closed-list proportional representation should be adopted; and (4) legislators should forfeit their mandates for party switching. While questions about floor voting are theoretically separable from those about ballot access and expulsion for party switching, they are empirically linked and do capture a single dimension of underlying orientation toward parties (factor analysis strongly confirmed the unidimensionality of the four items; Cronbach’s alpha=0.81). Our third dependent variable (NO SWITCH) is a measure of post-1998 party switching. Since we want to measure party loyalty, not a deputy’s degree of fickleness, we built this as a dichotomous measure in which deputies who switched parties at least once are coded as one. Our fourth and final dependent variable is voting discipline (DISCIPLINE), which represents the percentage of the total number of votes on which a deputy followed or defected from the recommendation of her party leaders on a set of 236 roll-call votes held during the first three years of the 51st legislature (1999–2001) in which 10 percent of the Chamber either voted contrary to the majority or absented themselves from the floor vote, and on which the leader of at least one party issued a recommendation contrary to her counterparts.15 Changing Utilities To test the effect of programmatic cohesion and polarization on the orientation of legislators toward parties, we create variables of cohesion (COHESION) and polarization (PARTY POLARIZATION) for each party, which we assign to deputies in those parties. Because these are highly correlated (r=0.71) and produce severe multicollinearity when both are used in our regressions, we include only the latter (results are broadly similar when the former is used). We expect this variable to be significant for each of our four dependent variables. At the individual level, our predictions are more fine-tuned. We expect the more politicians believe the party label has value, which we measure by deputies’ perceptions of the importance of voter loyalty to their party’s state-level performance in the last election (VOTER LOYALTY VALUE), the less likely to abandon the 16 Hagopian, Gervasoni, and Moraes party and more willing to delegate authority to party leaders they will be. But because party loyalty can be generated for a number of reasons (program, competence, patronage), we believe the impact of voter loyalty on the type of campaigns deputies run should be neutral. Here, we expect the more deputies credit the party program for their party’s electoral success (PROGRAM VALUE), the more they will campaign on a programmatic basis, and the more they credit that success to patronage (PATRONAGE VALUE), the less they will do so. Finally, to determine the independent impact of a deputy’s own policy preferences on legislative behavior, we calculate the distance between the average mean score of each deputy on the cluster of twenty salient state and market reforms and initiatives and the mean score of her party on those same policies (POLICY DISTANCE). We expect the higher this score, and thus the more distant a legislator’s views from her party’s position, the less likely she would be to delegate authority to party leaders and to follow their instructions on floor votes Rival Hypotheses and Control Variables The first group of variables tests the competing “electoral connection hypothesis”: (a) electoral security (ELECTORAL SECURITY), which we measure as the number of votes a deputy received minus the number of votes of the first alternate, over number of votes of first alternate (this number is easily interpretable as the factor by which a given deputy cleared the minimum number of votes needed to be elected). To take into account the electoral risk associated with a certain party or coalition not reaching the electoral threshold, we performed a similar calculation for the deputy’s party or coalition. When this number was smaller than 1 (i.e., the party or coalition obtained less than twice as many votes as the threshold), we used it to adjust the security of the deputy downward (by multiplying her individual electoral security score by the party/coalition score). The resulting number was logged to reflect the diminishing value of each additional vote over the minimum to be elected; (b) electoral dominance (DOMINANCE), first introduced by Ames (1995a, 1995b), which is an average of the percentage of the total votes a deputy receives in each municipality of the state, weighted by the contribution of that municipality to her overall total in the state (the district); and (c) the logarithm of district magnitude (DISTRICT MAGNITUDE) commonly used in studies of Brazilian legislative Hagopian, Gervasoni, and Moraes 17 politics. Deputies with higher vote totals relative to their copartisans and those who dominate districts—those with electoral bailiwicks—should rely on patronage and personal appeals rather than party reputation and program and be in a stronger position to switch parties, defy party leaders in floor voting, and maintain rules that preserve their autonomy. Moreover, if, as institutionalists contend, under open-list proportional representation the larger the district magnitude, the higher the level of intraparty competition and the greater the incentives of individual legislators to cultivate a personal reputation, then deputies in states with large district magnitudes should display less partisan loyalty in campaigning. A second set of variables tests a second major competing hypothesis for the orientation of individual legislators: their progressive career ambition. By drawing responses from our survey about which post the deputy sought next (reelection; senator; governor; mayor; federal administrative post; state administrative post; or a position outside politics), we can improve on the standard measure of career ambition—the post a deputy next contested and/or filled—which imputes a retrospective consciousness to the decisions guiding earlier actions and eliminates all ambitions that did not materialize, especially for administrative appointments. We create a dummy variable, STATE AMBITION (those with ambitions to be elected mayor or governor or to acquire an administrative post at the state level), and expect legislators focused on local and state political careers to exhibit a negative sign on all four party-oriented behaviors (most legislators who do not report state ambition report federal ambition; the resulting high collinearity does not allow us to include both variables in the same regression). We also consider the resources of individual politicians and the strength of their networks by tapping survey responses on the frequency of contact the deputy has with local politicians to create a LOCAL CONTACT variable; and with the state governor, cabinet secretaries, and directors and superintendents of state agencies to create a variable of STATE CONTACT (the three items are clearly unidimensional; Cronbach’s alpha = .91). We interact the latter with membership in the state governing coalitions (STATE COALITION), as the effect of state contacts are expected to be different for politicians in the government or the opposition. Finally, we control for levels of development—expected to depress voter demand for patronage-based campaigns, party switching, and politicians’ autonomy and 18 Hagopian, Gervasoni, and Moraes indiscipline—using the log of per capita state income (GDP PER CAPITA). To test whether rising rates of party unity on legislative votes are a product of a successful bargaining process in which deputies are satisfied with their individual payoffs (Ames, 2002: 193–194), or part of a broader phenomenon of growing party-oriented behavior among politicians also requires a measure of strategic incentives. Following other scholars (Ames, 2002; Pereira and Mueller, 2004), we identify the percentage of individual budget amendments funded by the executive in the two-year period from 2000 to 2001 (for amendments passed in 1999 and 2000) (AMENDMENTS FUNDED [%]), but we interact this variable with membership in the federal governing coalition (COALITION) to allow for different expectations for members of the governing and opposition parties. We also test whether these same incentives may have a bearing on party switching and proparty attitudes, but do not expect them to have retrospectively influenced the kinds of campaigns politicians ran. We test these hypotheses using OLS regression models for party campaigning, voting discipline, and delegation, and a logistic regression model for party switching. For each dependent variable we proceed in two steps: we begin with a baseline model that tests our hypotheses as well as existing competing ones, then drop all variables that are far from statistical significance (t-ratio<1) in order to obtain more precise estimates for the more significant coefficients. Table 1 summarizes our predictions and our results. Hagopian, Gervasoni, and Moraes 19 TABLE 1 EXPLAINING PARTY-ORIENTED BEHAVIOR Theoretical Approach and Partisan Campaigns Delegation Variables (predicted effects Model 1 Model 2 Model 3 Model 4 in parentheses) Electoral Connection (-) ELECTORAL SECURITY .28 .25 .01 -(.20) (.19) (.16) (-) DISTRICT MAGNITUDE .17 ---(LOG) (.25) (-) DOMINANCE -.10 -.60 -(1.37) (.95) Progressive Career Ambition (-) STATE AMBITION .12 --.09 -(.25) (.20) Level of Development (+) GDP PER CAPITA .50* .63*** .33* .19 (LOG) (.28) (.20) (.19) (.15) Networks (-) LOCAL CONTACT -22* -.23* .08 -(.13) (.12) (.11) (-) STATE CONTACT .04 --.04 -(.09) (.12) (*) STATE COALITION ---.34 -0.39** (.22) (.16) (+) STATE CONTACT * ---.04 -STATE COALITION (.16) (*) COALITION --.26 -(.35) Material Payoffs (-) AMENDMENTS --.004 -FUNDED (%) (.010) (+) COALITION* ---.006 -A MENDMENTS (.011) FUNDED (%) Changing Utilities (+) PARTY POLARIZATION (+) PROGRAM VALUE (-) PATRONAGE VALUE (+) VOTER LOYALTY VALUE POLICY DISTANCE -.25 (.21) .28** (.11) -.10 (.09) -- -26 (.20) .27*** (.10) -.10 (.08) -- .70*** (.19) --- .66*** (.14) --- No Party Switching Model 5 Model 6 -.04 (.61) -- -- .85 (2.66) -- -.57 (.62) -.46 (.56) -8.94*** (3.05) -7.12*** (2.61) .77 (.57) .83* (.50) 3.59 (2.75) 2.59 (2.25) -- -2.00 (1.72) -.82 (1.74) 4.55 (3.44) -.98 (2.50) 3.65 (5.07) -1.66 (1.37) -- .11 (.15) -.14 (.17) -- .22 (.35) -1.10*** (.38) .41 (.63) 1.38*** (.51) .01 (1.01) -- -1.11*** (.33) .34 (.57) 1.41*** (.45) -- -1.43 (2.56) -- -- 11.11 (14.42) -- -.03 (.02) .02 (.02) -- .92* (.52) -- .98* (.51) -- 9.21*** (2.77) -- -- -- -- .29*** .29*** .50* (.08) (.07) (.27) (-) --.38 .36* -(.22) (.19) Intercept -1.72 -2.30 -3.57** -1.97 -6.78 (2.01) (1.70) (1.84) (1.28) (5.18) N 107 110 102 113 110 R2 /PSEUDO R2 .20 .19 .36 .36 .26 2 ADJ R .11 .14 .26 .33 -Prob > F/ Prob. > chi2 .015 .001 .000 .000 .002 Standard errors in parentheses. *p<.10; **p < .05; ***p < .01. (*) No effect is predicted because this variable is part of an interaction effect Voting Discipline Model 7 Model 8 -- .67*** (.25) --7.22* (4.32) 120 .27 -.000 ---6.14* (3.42) 47.67* (26.23) 106 .32 .23 .000 -- 3.71 (2.47) --- -- 9.13*** (2.14) -----5.34* (2.98) 58.49*** (19.83) 116 .28 .24 .000 20 Hagopian, Gervasoni, and Moraes RESULTS AND DISCUSSION The empirical findings of this analysis are as significant for what they tell us is not true about Brazilian politics as for what is. First, contrary to expectations that under open-list proportional representation deputies would campaign most heartily against one another in larger districts, the impact of district magnitude on legislators’ campaigns was not significant. Deputies enjoying clear dominance in select municipalities (rather than drawing a scattered shot of votes from across the state) were no less likely to campaign on a partisan and programmatic basis than a patronage, personal, or parochial one. In fact, our measures of dominance and electoral security were not statistically significant in any model. More powerful is the impact of development, and our findings suggest we take seriously the thesis that voter taste changes along with socioeconomic development: the logged value of per capita state GDP exercised significant and powerful effects on promoting partisan campaigning. It had an effect on deterring party switching, as other scholars have found (Desposato, 2006), and it was close to significant on delegation and voting discipline. It is also the case that we have overreached in the power we attribute to progressive career ambition, at least in the case of Brazil. An ambition for a state level career does exercise a significant and negative effect on floor voting, but not on party switching, delegation, or the basis of campaign appeals. What this implies is that deputies who wish to remain in the legislature or join the federal administration have an incentive to follow parliamentary party leaders, either to preserve the value of the party brand or for the endorsement or campaign finance assistance of party leaders, relative to those who plan to shift their careers to local and state venues. At a minimum, these results suggest that our expectations about the influence of electoral rules and progressive career ambitions need finer theoretical specification. They are not equally weighty in every form of behavior. Most surprising was the apparently limited impact of material payoffs. Using an interaction term, we find that the percentage of individual budget amendments funded was not a significant incentive for members of the opposition to defect from their party leaders or for members of the governing coalition to observe discipline. These findings Hagopian, Gervasoni, and Moraes 21 contradict the results obtained by Ames (2002) based on pooled votes in the 49th and 50th congresses (1991–1998) and Pereira and Mueller (2004: 785) for the 50th, but are consistent with our theory that the value of patronage as an electoral strategy declined during the 1990s. By the 51st Congress, when deputies placed greater value on their party labels and the pork barrel was running dry, their incentives to cooperate with party leaders in Congress increased. If party switching is primarily driven by political opportunism—we know that members switch in the months after the election to better position themselves for party leadership and choice administrative positions—politicians in the state opposition who are well connected to state officials were significantly more disposed to switch out of their party, and their well-connected coalition counterparts to stay loyal. Those who meet frequently with local politicians are less likely to run partisan campaigns. This is intuitively obvious and should not be controversial. Connections to local and state politics, however, do not appear to impact voting records. A startling finding of our survey is that governors were cited as the twelfth and least important influence on congressional votes. Our models suggest that other long-standing stereotypes of Brazilian politics may need to be modified. The presumption that PT membership has a powerful, positive impact on party-oriented behavior relative to the catchall parties notwithstanding, the PT in the electorate is not as distinctive from other parties as the PT in the Congress: in unreported results, a PT dummy was not significant in any party campaign model, and controlling for party polarization and the value of voter loyalty, PT membership exercised no independent impact on delegation, party switching, or voting discipline. In another unreported result, politicians from the Northeast are slightly more likely to employ clientelist, localist, or personal vote campaigning strategies, but they are more, not less, likely to observe party discipline. That the Northeast is changing should not be surprising: a casual glance at the 2006 electoral map confirms that Lula has painted the Northeast red. By comparison, the theory of changing utilities finds some support. Membership in polarized parties is a significant predictor of sticking with one’s party, delegating authority to party leaders, and heeding the recommendations of party leaders on controversial floor votes. Legislators who believe the party program holds value in 22 Hagopian, Gervasoni, and Moraes elections and that patronage does not are embracing party-oriented campaigning. The perception that voter loyalty was critical to the party’s electoral success in their home state significantly influenced deputies not to switch out of their parties, and also served as a strong inducement to surrender autonomy to party leaders. Finally, deputies who share their party’s views on the state-market divide are significantly more likely to vote with their party leaders’ recommendations, though the sign on delegation was not in the predicted direction. While one might reasonably question whether issue preferences are endogenous to votes and hence not a good measure of party-enhancing discipline, in fact the universe of votes was considerably broader than the state-market divide from which we generated these scores. CONCLUSIONS Brazilian politicians, once believed to be among the least party-oriented in Latin America, are today more faithful to their parties in the legislature and in their constituencies than they have been at any time since Brazil returned to democratic rule in 1985. Falling rates of electoral volatility and rising legislative voting unity are part of a broader trend: party delegations are also programmatically coherent and distinctive, and Brazilian politicians who value their party’s reputation do not sell their votes to the highest bidder but voluntarily close ranks behind party leaders. While we cannot really know the extent of this transformation due to a blurry baseline, if we assume that previous scholarship was not completely erroneous, and if we are correct that there are party-oriented legislators in Brazil today in parties across the board where strategic incentives do not exist for them to become so, then we must address the sources and significance of that transformation. A series of new measures enables us to provide a more nuanced interpretation of the effects of the electoral connection, progressive career ambition, and the effects of development and partisanship than many previous studies. We find that the ambition to enter state and local politics does indeed exercise a negative effect on floor discipline, and opposition deputies who have more contact with state government officials are more likely to switch out of their parties. We also find that high levels of socioeconomic development spur partisan campaigns and depress party switching. Most surprisingly, a Hagopian, Gervasoni, and Moraes 23 pattern of electoral domination of municipalities, highly secure seats, and strategic payoffs does not predict politicians’ behavior in the electoral or legislative arenas during the 51st legislature (1999–2002). In light of these results, we need to rethink the determining influence of electoral institutions, and shift our focus away from using them to justify an apparent weakness of Brazilian political parties to identifying the causes of stronger, more disciplined, and more programmatically cohesive parties, absent those incentives. In other words, we need to identify the countervailing incentives that motivate legislators to strengthen parties. In this work, we explored two directions of change—the declining utility of constituent-based strategies that raised the value of the party label vis-à-vis personal reputation and the growing homogenization of parties along programmatic lines—which, we argue, increased the incentives for deputies to delegate authority to party leaders to coordinate their parties’ strategies to meet electoral competition. Party-oriented legislators may have emerged for the same reasons they did in nineteenth-century England—the loss of backbench access to patronage and an empty pork barrel (Cox, 1987)—but in the past decade in Brazil, this drama apparently unfolded in the opposite sequence. Following lively debates over the country’s economic course, parties first became more homogenous and polarized; indeed, we should not be using the term “catchall” any longer as synonymous with Brazilian parties of the center and right. Moreover, the more a deputy believed in his party program and that voters were loyal to his party, the more willing he was to defer to his party leaders; the more he believed that program had electoral value, the more likely he was to campaign on it; and the more he shared the policy views of his party, the more likely he was to observe voting discipline. These findings have important implications for Brazilian politics. As these programs create value for the party brand and voters respond favorably to programmatic coherence as well as governing competence, the incentive for parties and politicians to offer such distinctive choices is reinforced. Any future narrowing of party polarization could, of course, reverse the incentives to compete on program in favor of valence issues or competence. But while only time will tell, it is possible the trend toward legislators that are oriented toward their parties will outlive the structural and strategic changes that gave rise to it. This might occur because the logic of programmatic competition may 24 Hagopian, Gervasoni, and Moraes create the necessary incentives for politicians to strengthen their parties by penalizing party switching with a loss of mandate and by adopting a closed-list proportional representation electoral system. Indeed, rule changes have already begun. In 2002 the candidato nato (birthright candidate), which had guaranteed ballot access to incumbents, was eliminated and efforts to reduce party fragmentation via the “barrier clause,” which would have restricted the access of small parties to legislative leadership bodies and the media beginning with the 2006 election, have been stalled only by the courts. Even a mixed-member system to solve the dilemma of intraparty competition is now foreseeable. Along with rule changes has come the greater insulation of social services from clientelism. The loosening of political control over the distribution of the most basic social services to the neediest families implied by the conditional cash transfer program, Bolsa Familia (family stipend), and its precursor, Bolsa Escola (school stipend), was only possible in a context of a diminishing value of patronage. Not surprisingly, the two most cohesive parties—the PSDB and the PT—also substantially polarized on policy, were the ones to initiate such changes. Scholars of Brazil have long decried that autonomous politicians and the weak parties they insist on preserving erode party discipline, make Brazil ungovernable, and undermine political responsiveness—and representation. Rising party discipline has mitigated Brazil’s worst governability problems, breaking the “deadlock” of Brazilian democracy (Ames, 2001), and even the end of the scourge of party switching that so openly flouts the popular will is in sight. A measure making party switching more difficult was approved in the Chamber of Deputies in 2007 (as of this writing it is still awaiting Senate action), but in any case the Supreme Court also in 2007 upheld an Electoral Court interpretation of the 1990 Party Law that legislative seats belong to parties, not individuals. Taken together, the break in the patronage chain, rising programmatic cohesion and party loyalty, and rules changes that have already made party government possible may now or will soon allow constituents to hold Brazilian parties accountable on program and policy as well as for goods and services delivered. In time, these changes may shape a new, ultimately stronger, pattern of political representation in Brazil. Hagopian, Gervasoni, and Moraes 25 APPENDIX A Returned questionnaires, supplemented by interviews conducted by students from the University of Brasília, yielded 151 total responses (25.4 percent of the Congress), but to make our analysis comparable to that of other researchers we omitted senators from this study, resulting in a sample of 127. The sample overrepresents first-term members including alternates (which were half of our sample but only about 40 percent of the entire Congress). Although some nontrivial level of sample bias is inevitable in any random sample of human subjects in general and legislators in particular, the size of the bias in our sample is not large and the sample does reproduce the population in terms of four important variables—party, region, gender, and level of legislative activity, making it well within accepted standards for this type of population 26 Hagopian, Gervasoni, and Moraes Table A Representativeness of Survey: Sample and Population Compared Party PFL PSDB PMDB PPB PT PTB PDT PSB PL PCdoB PPS Other* Total Number of seats Population Sample 105 22 99 27 83 27 60 10 59 19 31 4 25 4 18 4 12 4 7 2 3 0 11 4 513 127 Percentage of seats Population Sample 20.5% 17.3% 19.3% 21.3% 16.2% 21.3% 11.7% 7.9% 11.5% 15.0% 6.0% 3.1% 4.9% 3.1% 3.5% 3.1% 2.3% 3.1% 1.4% 1.6% 0.6% 0.0% 2.1% 3.1% 100.0% 100.0% Chi-square P-value 0.52 Region Northeast North Southeast Center-West South Total Gender Women Men Total Population 149 65 179 41 77 511 Population 29 484 513 Sample 41 15 44 8 19 127 Sample 7 120 127 Population 29.2% 12.7% 35.0% 8.0% 15.1% 100.0% Population 5.7% 94.3% 100.0% Sample 32.3% 11.8% 34.6% 6.3% 15.0% 100.0% Sample 5.5% 94.5% 100.0% P-value Level of legislative activity Very active Active Average Weak Very weak Population 155 112 99 58 14 Sample 42 27 22 10 1 Population 35.4% 25.6% 22.6% 13.2% 3.2% Sample 41.2% 26.5% 21.6% 9.8% 1.0% P-value 0.91 P-value 0.95 0.47 Total** 438 102 100.0% 100.0% Note: Partisan delegations for the entire congress are based on their size resulting from the 1998 election. * PMN, PRONA, PSD, PSL, and PST. ** Source: Folha de Sao Paulo (2002). Caderno Olho no Voto (September 27). Available at http://www1.folha.uol.com.br/folha/brasil/ult96u38553.shtml. Only 438 (of 513) deputies were scored, 102 of whom were included in the sample. 27 Hagopian, Gervasoni, and Moraes APPENDIX B. VARIABLE MEASUREMENT AND SUMMARY DESCRIPTIVE STATISTICS Variable Name Dependent variables PART CAMPAIGN DELEGATION NO SWITCH DISCIPLINE Measurement Obs Mean St. Dev. See Appendix C for coding rules and scores Index of pro-party attitudes on four survey itemsa Dummy variable denoting no switches after 1998 from party on which elected Percent of roll-call votes cast in agreement with party leadership’s recommendation, 1999-2001 117 2.71 1.15 0 5 119 0.22 1.02 -2 2 126 0.75 .44 0 1 127 81.5 17.5 98.3 Measure of deputy’s electoral security adjusted by coalition security Logged value of district magnitude 127 0.50 0.55 -0.55 3.25 127 3.27 0.75 2.08 4.25 Measure of pattern of deputy’s electoral support; two Federal District cases, where DM = 1, dropped Next position sought is mayor, governor, or a post in the state administration Logged value of per capita state gross domestic productb Frequency of meetings with local politiciansc Frequency of meetings with high-level members of state administration c Member of state governing coalition (1) or opposition (all others) (0) Member of federal governing coalition (PSDB, PFL, PMDB, PPB) (1) or opposition (all others) (0) Percentage of individual budget amendments funded by executive during 2000 and 2001 Absolute distance of mean position of party’s legislators on cluster of statemarket issue from all legislators in sample Degree of importance deputy attached to party’s platform in party’s state election resultd Degree of importance deputy attached to resources delivered for party’s state election resultd Degree of importance deputy attached to loyalty of state voters to party for party’s state election resultd Absolute distance of mean position on cluster of 20 state-market issue items from party mean positione 125 0.09 0.11 .001 0.74 127 0.28 0.45 0 1 127 8.61 .55 7.48 9.65 123 1.63 .86 -2 2 124 -0.52 1.33 -2 2 127 .39 .49 0 1 127 0.65 .48 0 1 119 29.2 0 93.6 127 0.77 .60 119 -0.12 1.19 -2 2 121 -0.40 1.25 -2 2 122 0.16 1.17 -2 2 120 0.50 .42 0.005 2.11 14.8 Min Max Independent variables ELECTORAL SECURITY DISTRICT MAGNITUDE (LOG) DOMINANCE STATE AMBITION GDP PER CAPITA (LOG) LOCAL CONTACT STATE CONTACT STATE COALITION COALITION AMENDMENTS FUNDED (%) PARTY POLARIZATION PROGRAM VALUE PATRONAGE VALUE VOTER LOYALTY VALUE POLICY DISTANCE 27.8 0.02 1.99 28 a Hagopian, Gervasoni, and Moraes Average of the responses to the four following questions: What is your view of the following propositions (very negative [-2], negative [-1], neutral [0], positive [1], very positive [2])? (1) The party closes the question and applies the party whip; (2) The party determines the order of the candidates to the Chamber of Deputies on the list (instead of the open system that exists now); (3) A legislator loses his mandate if he switches party after the elections; and (4) The party expels a legislator who votes against the party whip. b Figures are from 2001, and are available at www.ibge.gov.br. c No contact (-2); every three months (-1); monthly (0); every two weeks (1); and every week (2). d Values generated from responses to questions, “How important would you say the following national (state) factors (your party’s electoral platform (the loyalty of state voters to your party) (the resources the federal government recently transferred to your state) to your party’s performance in your state in the 1998 election?” Very little (-2); little (1); average (0); a lot (1); enormous (2) e Twenty items drawn from three questions (A-C): (A) “Independently of the position of your party, what was your personal opinion with respect to the following federal government proposals (total opposition [scored as -2], partial opposition [-1], indifferent [0], partial support [1], total support [2]): (1) Fund for Fiscal Stabilization; (2) Relaxation of monopolies (petroleum, energy, telecommunications); (3) Change in the definition of a national capital enterprise; (4) Privatization of the Rio Doce Valley Mining Company; (5) Tax on financial transactions (CPMF); (6) End of job tenure for public servants; (7) Social security reform of 1998; (8) Temporary labor contract; (9) Sanctions against states that violate the Camata Law.” (B) “How would you classify the effect of the following federal government policies on the country? (10) Trade liberalization; (11) Privatization of federal enterprises; (12) Flexibilization of labor law; (13) Social security reform; (14) Public administration reform.” Very negative (-2); very positive (2) (C) “How would you classify the effect of the following policies of the federal government on your state? (15) Trade liberalization; (16) Privatization of federal enterprises; (17) Privatization of state government enterprises; (18) Flexibilization of labor law; (19) Social security reform; (20) Public administration reform.” Very negative (-2); very positive (2). Hagopian, Gervasoni, and Moraes 29 APPENDIX C. CODING INDIVIDUAL CAMPAIGN APPEALS In separate open-ended questions, deputies were asked to name the three most important items in their party’s platform, and their three most important campaign appeals. We coded the latter as personal, programmatic, or partisan, and awarded a score of 1 for each programmatic or party-oriented campaign appeal (minimum = 0, maximum = 3). For each campaign message in which there was overlap between the message offered by the candidate and one of the most important points of the party platform as identified by the respondent, the candidate received a score of 1, regardless of whether they were valence issues or programmatic ones. Thus the maximum value for a party-oriented campaign is 6. We scored partial responses only up to the number of answers provided and dropped nine deputies who did not provide an answer to either question. Criteria Personal Vote Patronage/Pork Parochial Personal Qualities Program Valence Issues – I Valence Issues – II Other Program – I Other Program – II Partisan Pro-party appeal Political appeal Endorsement of partystrengthening changes Explicitly mention “resources” or “projects” Appeals to regional interests, either generally or by name Personality attributes; record in office; type of commitment Issue items in authors’ survey generating the least amount of controversy (as measured by standard deviation score) Issues not asked, but not likely to generate conflict Issue items in authors’ survey generating the greatest amount of controversy; those on which all interests not likely to align Highly charged issues not asked in survey; ideological visions Showing commitment to party as representatives Political alliances Identified as one of most important points on party platform Sample Responses Defense of concrete interests (public works, resources); more resources Defense of my state; region needs representation; recovery of state of Pernambuco; economic development for the region; water resources, irrigation, and dam construction”(including São Francisco River region) Honesty (but not honesty in politics); seriousness (but not seriousness in politics); competence; experience; hard work; work already accomplished as representative of the region; loyalty to grassroots groups Unemployment; corruption/ethics; health; education (but not lack of funds for education) Tourism; violence; physical security (segurança) (but not seguridade, which also implies personal economic security) Privatizations; pension reform; workers’ rights; tax reform; monetary stability; role of state in the economy; sovereignty (definition of national capital); agrarian reform; agriculture Review of external debt; socialism; political reforms (parliamentarism, check on executive and judiciary); women’s rights; social justice; distribution of income; alleviation of regional disparities Voting with party platform; voting with party; strengthening parties Loyalty to president; opposition to governor Party fidelity (whip); internal party democracy; active participation in party; party spirit 30 Hagopian, Gervasoni, and Moraes ENDNOTES 1 On Carey’s (2002) “unity index,” which differs from the familiar Rice Index of party cohesion (the difference between the percentage of party members voting yes and no averaged over the votes in a particular Congress) by weighting votes by attendance and the closeness of the vote to capture the degree of controversy legislation generates, party unity in Brazil was .75 from 1989 to 1998, lower than in Chile (.81 from 1997 to 2000), and Argentina (.88 from 1989 to 1991). 2 Although voters in large cities exhibited fairly stable partisan identification from 1945 to 1964 (Lavareda, 1991; Soares, 1967), in the interior parties did not espouse coherent programs and formed opportunistic alliances in local elections with ideologically noncontiguous parties. 3 These figures differ from Desposato’s (2006: 69) and Mainwaring’s (1999: 145–46) because the base includes all deputies who have served in the life of a legislature, including alternates (N=620). 4 Morgenstern (2004: 127) also found the more ideologically cohesive the party, the higher voting unity scores were in Argentina, Brazil, Chile, and Mexico. 5 Such amendments, an important source of district pork, declined from 73,642 in 1992 to 13,915 in 1994 to only 8,334 in 2000 (Samuels, 2003: 138). 6 To enforce compliance, the FRL imposed heavy fines, loss of mandate, and even jail time for public officials in noncompliant states (18 of 27 at the time) (Bressan, 2002: 384). 7 Legislators are classified according to their “level of legislative activity” (an ordinal scale consisting of five categories ranging from “very active” to “very weak”) based on their attendance, committee membership and leadership, bill sponsorship, voting record, and participation in floor debates (Folha de São Paulo, 2002). 8 The standardized item alpha score was .97 (see Appendix B for question wording), and factor analysis confirmed the internal consistency of these categories. That this divide constituted the single issue dimension of party competition in Brazilian politics in the 1990s is supported by Leoni’s (2002) calculation of W-Nominate scores on roll-call votes in the 49th and 50th Brazilian congresses. 9 All coalition scores are weighted by the size of the party delegations in the sample. Data on Argentina, Chile, and Mexico are drawn from responses to comparable issue clusters from the authors’ surveys. 10 Contrary to our survey findings, Power’s 1990, 1993 and 1997 surveys suggest that ideological distance narrowed during the 1990s. But Roma (2004: 119–20) found the correlation between the self-positioning of Brazilian deputies on the ten-point left-right scale and their location on the same scale based on revealed ideal point positions (derived from stated opinions in 1995 and their W-nominate scores on roll-call voting across legislatures in the 1990s) was only .44 (with 0 representing perfect correspondence and 1 total dispersion between ideological self-placement and revealed positions), or “moderate” at best, and concluded based on their stated preferences on policy and their voting record that politicians on both the left and right understated their extremism later in the decade. 11 Power’s 1990 sample is also representative by party and region (Power 2000: 246). 12 Cross-temporal comparisons are complicated because respondents to our survey were offered the option of responding “equally because of party label and personal effort” (which Power’s survey unfortunately did not) to: “Some parliamentarians are elected because of their party label—that is, the power of the party’s organization or its profile in public opinion. Others are elected due to their individual capacity for organization or personal behavior in politics. Would you say you were elected …” (other options were “almost exclusively” [2] or “mostly” [1] because of party label). 13 Mean values of responses were: quality of candidates (.64); image of state leaders (.36); image of party (.33); voter loyalty (.16); image of governor (.15); economy in state (.15); resources Hagopian, Gervasoni, and Moraes 31 anticipated in the future (.06); role of party members in state governments (-.22); and resources delivered (-.40). 14 Such as a 1994 electoral calendar reform producing concurrent elections for president, Congress, and state assemblies and improving economic performance and greater price stability (Mainwaring, 1999: 221–22; Panizza, 2000: 507, 509, 520; Amorim Neto et al., 2003: 576). 15 We scored a vote coinciding with a party leader’s recommendation as 1, and a contrary vote, an unexcused absence (a convenient means of opposing party leaders without voting against the party whip), or an abstention as 0. We also included in our sample votes on which a quorum was lacking if the quorum failed for this reason. Since we were testing the rival thesis that strategic payoffs drove discipline, we dropped twenty-four votes in 2002 to match the “percentage of amendments funded” variable (for which data were available only through 2001) as closely as possible. 32 Hagopian, Gervasoni, and Moraes REFERENCES Aldrich, J. H. & D. W. Rohde (1998). Measuring Conditional Party Government, Paper presented at the annual meeting of the Midwest Political Science Association, Chicago. Ames, Barry (1995a). “Electoral Rules, Constituency Pressures, and Pork Barrel: Bases of Voting in the Brazilian Congress.” The Journal of Politics 57, 2 (May): 324– 43. Ames, Barry (1995b). “Electoral Strategy under Open-List Proportional Representation.” American Journal of Political Science 39, 2 (May): 406–33. Ames, B. (2001). The Deadlock of Democracy in Brazil. Ann Arbor: University of Michigan Press. Ames, B. (2002). A Theory of Legislative Parties in Brazil. In S. Morgenstern & B. Nacif (Eds.), Legislative Politics in Latin America (pp. 185–221). Cambridge: Cambridge University Press. Ames, B. and T. Power (2007). Parties and Governability in Brazil. In P. Webb & S. White (Eds.), Political Parties in Transitional Democracies (pp. 179–212). Oxford: Oxford University Press. Amorim Neto, O., G. W. Cox, & M. D. McCubbins (2003). Agenda Power in Brazil’s Câmara dos Deputados, 1989–98. World Politics 55, 4: 550–78. Bressan, S. (2002). Reforma Administrativa [Administrative Reform]. In B. Lamounier & R. Figueiredo (Eds.), A Era FHC: Um Balanço [The FHC Era: An Accounting] (pp. 369–94). São Paulo: Cultura. Campello de Souza, M. do C. (1976). Estado e Partidos Políticos no Brasil (1930 a 1964) [State and Political Parties in Brazil (1930 to 1964]. São Paulo: Alfa-Omega. Cardoso, F. H. (1996). In Praise of the Art of Politics. Journal of Democracy 7, 3: 8–19. Carey, J. (2002). Getting Their Way or Getting in the Way? Presidents and Party Unity in Legislative Voting. Paper presented at the annual meeting of the American Political Science Association, Boston, MA. Carey, J. & M. S. Shugart (1995). Incentives to Cultivate a Personal Vote: A Rank Ordering of Electoral Formulas. Electoral Studies 14, 4: 417–39. Carreirão, Y. de S. & M. D. A. G. Kinzo (2004). Partidos Políticos, Preferencia Partidária e Decisão Eleitoral no Brasil (1989/2002) [Political Parties, Party Preference, and Electoral Choice in Brazil (1989–2002)]. Dados 47, 1: 131–68. Hagopian, Gervasoni, and Moraes 33 Chubb, J. (1982). Patronage, Power, and Poverty in Southern Italy: A Tale of Two Cities. Cambridge: Cambridge University Press. Cox, G. W. (1987). The Efficient Secret: The Cabinet and the Development of Political Parties in Victorian England. Cambridge: Cambridge University Press. Cox, G. W. & M. D. McCubbins (1993). Legislative Leviathan: Party Government in the House. Berkeley: University of California Press. Desposato, S. (2006). Parties for Rent? Ambition, Ideology, and Party Switching in Brazil’s Chamber of Deputies. American Journal of Political Science 50, 1: 62–80. Estudo Eleitoral Brasileiro (ESEB). (2002). CESOP-FGV/Brasil02.Dez-1838. In Banco de Dados do Centro de Estudos de Opinião Pública – CESOP – UNICAMP. Available at: http://www.cesop.unicamp.br/busca/CESOP/pesquisa_usuario. Faoro, R. (1958). Os Donos do Poder: Formação do Patronato Político Brasileiro [The Power Barons: The Formation of the Brazilian Patrimonial Order]. Porto Alegre: Globo. Figueiredo, A. C. & F. Limongi (1999). Executivo e Legislativo na nova ordem constitucional [Executive and Legislature in the New Constitutional Order]. Rio de Janeiro: FGV. Figueiredo, A. C. & F. Limongi (2002). Incentivos Eleitorais, Partidos e Política Orçamentária [Electoral Incentives, Parties, and Budgetary Politics]. Dados 45, 2: 303–44. Folha de São Paulo (2002). Caderno Olho no Voto [Eye on the Vote]. (September 27). Available at: http://www1.folha.uol.com.br/folha/brasil/ult96u38553.shtml. Gaetani, F. & B. Heredia (2002). The Political Economy of Civil Service Reform in Brazil: The Cardoso Years. Prepared for the Inter-American Development Bank. Grzymala-Busse, A. M. (2002). Redeeming the Communist Past: The Regeneration of Communist Parties in East Central Europe. Cambridge: Cambridge University Press. Jacobson, G. C. (2000). Party Polarization in National Politics. In J. R. Bond & R. Fleischer (Eds.), Polarized Politics: Congress and the President in a Partisan Era (pp. 9–30). Washington, DC: Congressional Quarterly Press. Kitschelt, H. (2000). Linkages Between Citizens and Politicians in Democratic Polities. Comparative Political Studies 33, 6–7: 845–79. 34 Hagopian, Gervasoni, and Moraes Kitschelt, H. & S. I. Wilkinson. (2007). Citizen-politician Linkages: An Introduction. In H. Kitschelt & S. I. Wilkinson (Eds.), Patrons, Clients, and Policies: Patterns of Democratic Accountability and Political Competition (pp. 1–49). Cambridge: Cambridge University Press. Lamounier, B. (2003). Brazil: An Assessment of the Cardoso Administration. In J. I. Domínguez & M. Shifter (Eds.), Constructing Democratic Governance in Latin America (pp. 269–91). Second Edition. Baltimore, MD: Johns Hopkins University Press. Lavareda, A. (1991). A Democracia nas Urnas: O Processo Partidário-Eleitoral Brasileiro [Democracy at the Polls: The Brazilian Partisan-Electoral Process]. Rio de Janeiro: IUPERJ/Rio Fundo. Leoni, E. (2002). Ideologia, Democracia e Comportamento Parlamentar: A Câmara dos Deputados (1991–1998) [Ideology, Democracy, and Parliamentary Behavior: The Chamber of Deputies (1991–1998)]. Dados 45, 3: 361–86. Levitsky, S. (2003). Transforming Labor-based Parties in Latin America: Argentine Peronism in Comparative Perspective. Cambridge: Cambridge University Press. Lyne, M. M. (2004). “Endogenous Institutions: Electoral Law and Internal Party Dynamics in Brazil.” Paper presented at the annual meeting of the American Political Science Association, Chicago, IL. McCubbins, M. D. & F. M. Rosenbluth (1995). Party Provision for Personal Politics: Dividing the Vote in Japan. In P. F. Cowhey & M. D. McCubbins (Eds.), Structure and Policy in Japan and the United States (pp. 35–55). Cambridge: Cambridge University Press. Mainwaring, S. P. (1999). Rethinking Party Systems in the Third Wave of Democratization: The Case of Brazil. Stanford, CA: Stanford University Press. Melo, C. R. (2004). Retirando as Cadeiras do Lugar: Migração Partidaria na Câmara dos Deputados (1985–2002) [Pulling the Seats out of Place: Party Migration in the Chamber of Deputies (1985–2002)]. Belo Horizonte: UFMG. Morgenstern, S. (2004). Patterns of Legislative Politics: Roll-Call Voting in Latin America and the United States. Cambridge: Cambridge University Press. Panizza, F. (2000). Is Brazil becoming a ‘boring’ country? Bulletin of Latin American Research 19, 501–25. Pereira, C. & L. Renno (2003). Successful re-election strategies in Brazil: The electoral impact of distinct institutional incentives. Electoral Studies 22, 3: 425–48. Hagopian, Gervasoni, and Moraes 35 Pereira, C. & B. Mueller (2004). The Cost of Governing: Strategic Behavior of the President and Legislators in Brazil’s Budgetary Process. Comparative Political Studies 37, 7: 781–815. Power, T. (2000). The Political Right in Postauthoritarian Brazil: Elites, Institutions, and Democratization. University Park, PA: Pennsylvania State University Press. Roberts, K. M. & E. Wibbels (1999). Party Systems and Electoral Volatility in Latin America: A Test of Economic, Institutional, and Structural Explanations. American Political Science Review 93, 3: 575–90. Rohde, D. (1991). Parties and Leaders in the Postreform House. Chicago: University of Chicago Press. Roma, C. R. (2004). Atores, Preferências e Instituição na Câmara dos Deputados [Actors, Preferences, and Institutions in the Chamber of Deputies]. Unpublished PhD Diss., University of São Paulo. Samuels, D. (2003). Ambition, Federalism, and Legislative Politics in Brazil. Cambridge: Cambridge University Press. Samuels, D. (2004). Political Ambition, Candidate Recruitment, and Legislative Politics in Brazil. Paper prepared for the conference Pathways to Power: Political Recruitment and Candidate Selection in Latin America. Wake Forest University, Winston-Salem, NC, April 3–4. Samuels, D. (2006). Sources of Mass Partisanship in Brazil, Latin American Politics and Society, 48, 2: 1–27. Sani, G. & G. Sartori. (1983). Polarization, Fragmentation and Competition in Western Democracies. In H. Dalder & P. Mair (Eds.), Western European Party Systems (pp. 307–40). Beverly Hills: Sage. Santos, F. (2002). Partidos e Comissões no Presidencialismo de Coalizão[Parties and Committees in Coalition Presidentialism]. Dados 45, 2: 37–64. Silveira, F. E. (1998). A Decisão do Voto no Brasil [Deciding to Vote in Brazil]. Porto Alegre: EDIPUCRS. Soares, G. A. D. (1967). The Politics of Uneven Development: The Case of Brazil. In S. M. Lipset and S. Rokkan (Eds.), Party System and Voter Alignments: CrossNational Perspectives (pp. 467–96). New York: Free Press.