Lucio Rennó. University of Pittsburgh Department of Political Science 4L01 Posvar Hall, Pittsburgh, PA, USA (

Transcription

1 Fabiano Santos Instituto Universitário de Pesquisas do Rio de Janeiro -- IUPERJ Rua da Matriz, 82 Botafogo Rio de Janeiro, RJ Brazil ( Lucio Rennó University of Pittsburgh Department of Political Science 4L01 Posvar Hall, Pittsburgh, PA, USA (

2 ABSTRACT This article contributes to the study of legislative organization by contrasting the process of committee chair selection in two distinct institutional environments. The goal is to verify how party loyalty affects the distribution of committee chairmanship in the Brazilian Chamber of Deputies, focusing in two distinct historical periods. The main hypotheses are tested using a data set that contains information about Federal Deputies career characteristics, both inside and outside the Chamber, measured in an annually basis, in two distinct democratic periods 1946 to 1964 and 1989 to The article s main conclusions are: first, in a situation where parties are oriented by a factional logic and where power distribution between party leaders and committee chairs is more equilibrated, party loyalty is a crucial factor in the allocation of committee chairmanships. Second, in a broader perspective, the interaction between committees and parties cannot only be explained by incentives emanating from the electoral connection, for it also suffers the influence of executive-legislative relations. 2

3 The Selection of Committee Leadership in the Brazilian Chamber of Deputies Fabiano Santos Lucio Rennó I. Introduction This article intends to contribute to the study of legislative organization by contrasting the process of committee chair selection in two distinct institutional settings. The recent Brazilian history, with two democratic periods namely, and separated by an authoritarian interregnum offers the perfect opportunity to do so. Both periods are distinct from each other in some important institutional ways, and these differences, we argue, have a drastic impact in the factors that influence committee chair selection. Our main hypothesis is that in the period from 1951 to 1963, in face of higher intra party divisiveness and a more decentralized Chamber, it was fundamental for institutional leaders in the Chamber to have the support of a majority of party members. Consequently, party loyalty served as a crucial criterion for the selection of institutional leadership. The same does not hold in the post 1988 period. The current democratic period, especially during Cardoso s administration, has been marked by considerably higher levels of preference congruity between Congress and the president 1, less divisiveness than the previous period, and more centralized decisionmaking in the Chamber 2. Hence, committees are more limited legislative actors today than they were in the period and party discipline is higher. In such an environment, party leaders get more leeway to choose who will be committee 3

4 chair, and less stringent criteria are employed; party loyalty no longer plays the decisive role it did before. The balance of power in the Brazilian Chamber of Deputies shifted towards party leaders, and this lead to changes in the functioning of the Chamber. In order to test our main propositions, we rely on a data set that contains information about Federal Deputies career characteristics, both in and out of the Chamber, measured in an annually basis in the two periods. Data include positions occupied by representatives in the Chamber s hierarchical structure, previous personal and political career, votes received in previous elections, and, most importantly, support of the party s majority preferences in roll call votings. This unique data set allows us to compare the impact of the distinct factors in influencing the selection of committee chairs. The article is organized as follows: in the following section, we discuss the relation between parties and committees in organizing the legislature. In section three, we focus on the blossoming literature on legislative politics in Latin America. Then we advance our explanation of committee chair selection in the two periods. In the fifth section, we describe our variables. In the sixth, we test our propositions. In the final section, we conclude. II. Political Parties and Congressional Committees The comparative literature on legislatures identifies two fundamental agents in parliament s internal organization: parties and committees. In the U.S., for instance, the latter plays a key role in legislative organization, whereas in Europe parties are certainly the most relevant legislative agent 3. There seems to be a tradeoff between party power and committee power; when committees are strong legislative actors, parties are less influential. The case of the US is paradigmatic; 4

5 party leaders only gained more influence in the flow of legislative business after reforms that weakened committee chairs. In Latin America, Constitutions and assemblies standing orders usually endow committees with considerable parliamentary rights, albeit lower than the US case and with wide variation inside Latin America 4. For example, in Brazil bills must be referred to a committee before any floor action. Moreover, under certain conditions, committees can approve laws with no need of agreement by the floor. Consequently, committees are potentially powerful legislative actors. Pereira and Mueller 5 go as far as claiming that committees play an important informational and gate keeping role in the Brazilian chamber. But urgency requests from party leaders, which send legislation directly to the floor of the Chamber, can quite easily circumvent committees decision-making power. Similarly, the power of party leaders also varies by country. It has been consistently strong in Argentina and Chile 6. Party leaders were very strong in Venezuela up until the late 1990s 7, but their power decreased ever since 8. In Brazil, the power of party leaders has been increasing, but this is not a consensual conclusion 9. Hence, Latin American countries lay midway in a continuum between powerful committees in one extreme and powerful parties on the other. Furthermore, in Latin American presidential systems, presidents are extremely strong legislative actors, much more so than the US case 10. Therefore, in addition to parties and committees, studies of legislative organization in Latin America cannot ignore the central role of executive legislative relations. Such complex configuration of legislative actors offers interesting opportunities to test hypotheses about legislative organization and further advance theoretical claims that were initially based on the cases of the United States and 5

6 Western European countries. A comparison between the two democratic periods in Brazil is especially interesting because it allows tests of how changes in the standing orders of the Chamber of Deputies and the party system, keeping constant the electoral system and the type of government, affect the balance of power between parties and committees. Moreover, it helps place the current levels of party loyalty in perspective, indicating that such levels are in fact high in comparison to other historical periods in Brazil. III. Legislative Politics in Latin America The study of legislative politics in Latin American has seen tremendous development in the past 10 years. Cox and Morgenstern have pointed out that central topics in this discussion are Congresspersons career motivations, executive legislative relations, and the power of party leaders 11. In some countries, like Argentina, Chile, and Mexico there is a consensus that party leaders are strong legislative actors. Leaders directly affect legislators career paths and electoral success, by not just controlling access to ballots, but also resources that are valuable to incumbents, such as power positions inside Congress 12. In other cases, Brazil being the prime example, there is an open debate about the role of party leaders. On one side, Figueiredo and Limongi have pointed out that party members do vote together consistently in the current democratic period and argue that the role of party leaders in coordinating legislative business is grossly underestimated 13. It is important to highlight that such finding is path breaking in the study of Brazilian politics. Before Figueiredo and Limongi, it was common sense that political parties played only a marginal role inside the Chamber. Figueiredo and Limongi challenged the common sense and their view is now a cornerstone of the debate about the Brazilian legislature. 6

7 But some authors have disputed Figueiredo and Limongi s claims. Ames 14, for example, argues that the microfoundation of partisan cooperation is based on selective incentives, like distribution of public goods and patronage, and on preference congruence between backbenchers and leaders. Another approach emphasizes that the central role of governors and local level pressures in orienting legislators careers mine the power of party leaders 15. A core question of this dispute is if party leaders control resources that might enforce backbenchers cooperation. For Ames, party leaders have the capacity of distributing positions inside the chamber, like committee chairmanships, to loyal members as a way of rewarding them, but such distribution does not necessarily lead to increases in cooperation of Deputies who do not benefit from those rewards 16. In Ames view, what affects ordinary Deputies cooperation is the distribution of public funds to their constituencies and party leaders have little influence on this process. Our view differs in important ways from that of Figueiredo and Limongi in one hand, and Ames and Samuels on another. First, Figueiredo and Limongi correctly emphasize the greater significance of political parties in the current democratic period, a factor that Ames and Samuels incorrectly dismiss as secondary. On the other hand, Figueiredo and Limongi ignore the powerful role that the distribution of extra-policy incentives, like patronage and what Ames and Samuels refer to as pork barrel, have on enforcing more effective partisan collective action. Figueiredo and Limongi seem to refuse to accept that legislators individual interests play an important role in the modus operandi of the Chamber. In our view, party leaders of the governing coalition craft a compromise between individual and partisan, collective interests, aiding the executive branch in 7

8 getting its agenda through Congress. We argue that party leaders are key actors in the connection between local and national political issues in Brazil, by mediating the disbursement of federal largess to specific localities and by convincing backbenchers of the importance of their support to specific national level policies. Because of their intermediary role, party leaders affect the electoral connection between voters and representatives. In their struggle to construct majorities that support the government, party leaders from the governing coalition are backed not only by a consistent policy agenda but also by the budgetary disbursement power of the executive branch and by the allocation of power positions inside the chamber. Ultimately, party leaders have a great influence in the distribution of resources that affect Federal Deputies careers. This enforces individual cooperation and is reflected by the higher levels of party discipline in the current period. In other words, we agree with Ames' and Samuels claim that extra-policy incentives are an important inducement for cooperation, but reject their view that party leaders do not affect the distribution of such goods. 17 In addition to the distribution of parochial incentives, intra-legislative rewards, such as access to committee chairmanship, clearly is part of the roster of resources leaders have to reward allies. As Jones points out for the Argentine case, influence over committee selection is one of the tools party leaders have to punish and reward followers, in spite of career paths that lead politicians away from reelection attempts 18. Notice that political career patterns in Argentina and Brazil are similar in that local level positions, and not just re-election ambition, are very enticing for congresspersons 19. In fact, re-election rates are actually lower in 8

9 Argentina than Brazil. Hence, the same disclaimer made by Jones to the Argentine case that regardless of the legislator s career ambition, these resources (intrachamber resources) are vital to fulfillment of his/her career goals 20, also applies to Brazil. In fact, research about the 1998 election in Brazil indicates that being Chair directly affects Federal Deputies career choices and, indirectly their re-election chances 21. Being chair offers some advantages for Federal Deputies: 1) it increases Federal Deputies access to the media, increasing incumbents visibility and name recognition; 2) being chair is an indication of the Federal Deputies higher status in the Chamber compared to his/her counterparts. So if Deputies were given the choice of holding a position of chair or not, most would choose to be chair. The difference between the two periods in Brazil is that both access to committee positions and party loyalties are not as scarce resources today as they were from 1951 to Our argument is that in Brazil, distribution of committee chairmanship is used as a tool to reward party loyalists; no matter which period is analyzed. The difference is that when access to committee chairmanship is less restricted and party loyalty levels are overall higher, leaders have more leeway to choose which Deputy will be chair and less stringent criteria can be applied to committee chair selection. When access to committee chairmanship is more restricted and party loyalty is lower and more erratic, leaders have less freedom to choose and more stringent criteria is employed in the selection of committee chairs. The next section expands our arguments about how institutional changes increased party leaders discretion. Committee Chair Selection under Different Institutional Contexts 9

10 Any explanation about the organization of the Brazilian legislature, and of presidential systems in general for that matter, must take into account the impact of the Executive branch actions upon legislators behavior. In this sense, it is our contention that deputies behavior vis-à-vis their parties is a function of presidents efforts towards the implementation of policy programs. In this regard, there is a fundamental difference between the two periods under analysis in this article during the first democratic period, , parties were split among various factions, the same not being true for the post 1988 period. Since 1946, with the exception of the authoritarian period, executive legislative relationship has been marked by what has become customarily known as coalition presidential system. Given the separation of powers and party pluralism in Congress, the president organizes his basis of support through the distribution of ministerial offices and budget resources among members of the largest parties, and these members secure the votes necessary for the implementation of his government program. 22 In this setting, one of the fundamental functions of legislative parties is to protect politicians from president s monopolist behavior. When there is no room for the distribution of public goods that benefit specific localities inside the legislature and the exclusive source for the flow of patronage is the executive, party discipline becomes an efficient strategy against president s exploitation of his privileged bargaining position. Seen as an individual action, any threat of rebellion against government s proposals is not credible. If, otherwise, representatives who are members of a disciplined group gather around common objectives, their demands bear higher leverage upon the executive. As a corollary, in a situation of cohesive parties, the president also knows that he can rely on the full support of that party, 10

11 represented by the complete or almost so correspondence between the number of seats under control of such party, and the amount of votes its members cast favoring the president s proposal. The situation differs when the executive is not the sole provider of distributive policies. In this juncture, congressmen do not rely massively on the support of their party leadership in order to deliver particularistic goods for their personal followers, meaning that they are relatively free to follow their own mind on votes taken in the floor. Consequently, it is only natural to observe parties splitting into several factions. This last scenario better represents the political context inside the Chamber of Deputies during the period, whereas the former is a more accurate portrait of the present, post-1988, situation 23. In both settings, legislators overcome their collective action problems delegating the task of coordination to a party leader. More precisely, the party leader, in the Brazilian setting, functions as an intermediary between the executive branch and legislators. The former needs congressional support for the approval of its policy proposals and the latter need monies that benefit their constituencies. The leader informs his caucus about the executive s policy priority and conveys information to the president about possible congressional reactions towards these proposals as well as legislators desires of patronage benefits in exchange for supporting the executive program. Again, the difference is that during the period, leaders had control only over a faction of their caucus, while nowadays parties tend to behave more cohesively. The factional logic that oriented the functioning of political parties in the pre- 64 period is also in the root of the differences between the two democratic periods. For example, lets consider the selection of party leaders: according to the Chamber 11

12 standing orders, leaders were elected every two years by the majority of his/her caucus. Therefore, if the party was split into factions and if a leader is supposed to obtain the support of a majority of his/her peers, hence he/she must also be a member of the majority faction. The selection of committee chairs followed similar logic. Committee chairs were formally elected by a vote taken inside the committee. However, the distribution of these posts resulted from a negotiation among party leaders an informal rule dictated that chairmanships of the most important committees should be bestowed in favor of the largest parties. Again, if parties were divided into factions and if party leaders were members of the majority faction and if they were also responsible for the selection of committee chairs, then chairmanship positions would likewise be distributed among majority faction members. In this respect, an important question emerges concerning leaders comparative ability to control caucus behavior: what is the relationship between institutional/organizational changes and the increase of party discipline observed on the floor? It is common sense that under an open-list PR with open endorsement and single vote, party leaders have fewer resources to impose electoral sanctions on members (Ames 1995). However, in Brazil there are some things party leaders can do to induce members cooperation. 24 In the Chamber of Deputies, party leadership helps the Board of Directors formulate the Chamber s schedule and stipulate the size of committees, as well as refer bills to committees. Furthermore, leaders are directly responsible for choosing party caucus in committees and authorizing members to 12

13 make speeches in the Chamber s floor. In this particular, there is no difference between the two periods under consideration. Nonetheless, there are two important changes that were implemented in the 1994 revision of the Internal Rules of the Chamber of Deputies that might have impacted the balance of power between parties and committees. These are 25 : 1) Obligatory rotation of committee chairs; 2) Standing committees were more powerful in the period before 1964 than they are now, both in their capacity of allocating funds as well as of influencing legislation. During the period, there was no rule dictating rotation of committee chairs, while nowadays committees chairpersons are prohibited from being reelected after one term. The frequency with which committee chairs are rotated hinges on the issue of alternative sources of legislative leadership emergence. The rotation of committee chairs, which is also an institutional characteristic of other legislatures, like the Chilean one for example 26, democratizes the access to power positions in the Chamber by assuring that a greater number of Federal Deputies are given the chance of being chairs. Even though such rule appears to be a reminiscent of the military regime s efforts of weakening congressional leadership 27 and unquestionably it does impose specialization costs to Congress, it was probably maintained in the revision of the Regimento Interno because it assured a more equitable access to power positions in the Chamber. In fact, a probably unanticipated result of the rule was to render access to chairmanship less of a scarce resource in the hands of party leaders and minimize a potential source of conflict, which could spur from disputes for strongholds over committees. 13

14 As for the second issue, three organizational differences are worth mentioning: the control that party leaders have over committees timetable; the power leaders have to reallocate bill proposals to ad hoc committees; and the budgetary function of committees. During the period, leaders were not able to interfere with a committee timetable, as they are nowadays. After the 1988 Constitution, a board, called College of Leaders, was created in order to support the Chamber s Mesa Diretora in matters related to the assembly s voting schedule. Its decisive influence consists on the ability to require urgency for specific proposals, meaning that the proposal has to be voted within 45 days. The urgency request can be sent to the Board by the college, being the signature of each leader weighted by the size of the caucus under his/her leadership. After this, the request is sent directly to the floor for a vote, skipping the committee stage altogether 28. During the period, an institution like the College of Leaders did not exist, meaning that party leaders did not have much say in the organization of committees schedule. The second organizational difference refers to the ability of leaders to reallocate bills submitted to specific committees. This task is facilitated nowadays through the creation of ad hoc committees. This kind of committee has the objective of considering specific subjects, and it is dissolved when it has accomplished the purpose for which it was created. Differently from what occurred during the pre-64 period, the Chamber s internal rules dictate that special committees shall be created in cases of constitution amendment or when there is jurisdiction conflict around proposed legislation. When legislation is submitted for the consideration of more than three permanent committees, a special committee shall be created. The relevant point here is that members of special committees are chosen in ad hoc basis, which 14

15 allows the leaders increased latitude to influence committee members preferences. Hence, whenever there is a conflict of interests between a party leader and its caucus in a permanent committee around a specific bill, the former can try to reallocate this bill through the creation of special committees, thus, circumventing obstacles posed by the committee the bill was originally referred to. The third organizational difference relates to the powers permanent committees have to allocate funds. First, during the period, as nowadays, the executive had the constitutional power to initiate the budgetary law. Nevertheless, differently from what occurs in the present regime, legislators were able, through various procedural instruments, to avoid the manipulation of the timetable by the president in order to impose his preference upon the legislature. Chamber s standing orders, concerning the definition of permanent committees responsibilities, established that the Finances Committee could send a budget proposal to the floor in occasions in which the executive failed to send the budget in the stated period determined by the Constitution. This means that the reversal point, in the cases in which the budget is not voted in due time, was not as favorable to the president as it is under the 1988 Constitution. If the president tried to use his power of initiative before for instance, by delaying the sending of the budgetary proposal, and thus forcing a decision away from the median legislator s ideal point - legislators could always resort to the Finances Committee, which, in turn, could propose an alternative budgetary package potentially far from the president s ideal point. In summary, the pendulum of power in the new institutional environment favors party leaders, whereas in the previous period Committee Chairs were much more influential than today. Finally, as we argue, these changes correspond to the 15

16 interests of the rank and file members in the sense of protecting politicians from president s monopolist behavior. How can we derive empirically testable propositions from the preceding discussion? In the period, given the internal divisions of political parties and the greater power of Committee Chairs, party leaders strove to maintain their influence over the decisions of committees by using their political muscle to push forward the election of allies to committee chair positions. What can be said about the post-88 period? As it has been discussed before, we argue that the fractionalized logic that predominated previously is not as widespread nowadays. In this setting, the propensity to follow the majority faction of one s party does not distinguish between legislators behaviors. Loyalty levels are overall higher and party leaders have a large pool of allies to choose from. Consequently, even though only allies will receive committee chairmanships, the difference in loyalty level between committee chairs and other members is not statistically significant. In other words, because it is so widespread, party loyalty no longer is a decisive criterion for leadership selection. Our basic model can be described in the following way. First, we observe the relationship between presidential allocation powers and legislative organization based on parties. Second, we note the impact of party discipline upon the criteria for the distribution of power inside the legislature. The first nexus of the model says that the emergence of incentives for party discipline inside the Chamber is a positive and linear function of presidential prerogatives in budgetary matters. The second one shows that the criteria for assigning power within the assembly varies with a party s degree of cohesion. As the cohesion rate increases, the relevance of party loyalty, as an individual attribute, decreases. Conversely, when loyalty is a scarce resource for 16

17 party leaders, its importance naturally grows as a means of allocating power among the caucus. Thus, our basic proposition can be stated as follows: During the first period, that is, from 1951 to 1963, besides seniority, expertise and electoral performance, party loyalty rates were a crucial criterion for the selection of legislative leadership. Nowadays, only the three former criteria are relevant. Before testing the proposition, though, it is important to present our main variables and how we measured them. We do this on the next section. Variables and Measures Some attributes of a legislator s career must necessarily characterize any politician intending to bear leadership posts. Obvious and very intuitive examples are the attributes of seniority, electoral performance and expertise. The first one is an important feature because the exercise of leadership presupposes a good knowledge of an assembly s standing orders and a trustful relationship with other legislators and executive clerks, which only time can provide. The second one is also a crucial attribute because political vulnerability weakens the exercise of authority. The more insecure a politician feels towards his/her future, the less will be his/hers bargaining power vis-à-vis those with whom he/she must deal with, such as, caucus members, leaders of opposing parties, members of the Board of Directors, committee chairs, and the executive. Beyond these criteria, it is important to consider informational constraints. For a deputy to be selected committee chair, he/she must have a minimal degree of expertise on a committee s subject matter. In Brazil, the specialization requirement restricts the profile of a potential candidate to such legislative post in face of the instability of committee assignments that characterizes the Chamber s committee 17

18 system. Therefore, given that it is not realistic to suppose that members will become experts on policy matters during her/his term, parties will rely almost exclusively on members previous specialization when deciding which members allocate to committee panels. In other words, parties will permanently look for members who can specialize at a low cost. These three variables are operationalized in the following way. Seniority: This variable measures the number of years a deputy has been a Chamber member before being elected party leader. District Competition: This variable shows the number of seats available in the district divided by the number of candidates. It is an indicator of how competitive the elections were in that district. In other words, what the variable expresses is how electorally impervious the Deputy is. 29 Committee Experience. This variable refers to members professional background, or main economic activity before becoming a deputy. Data for this component was collected from the Repertório Biográfico dos Deputados Brasileiros (Biographical Data on Brazilian Deputies), which are volumes published usually at the beginning of each legislature, a few months before the first session 30. The variable assumes value 1 if a deputy s professional background or previous economic activity is related to the subject matter of a committee of which he/she is member and 0 otherwise. Since previous career experience can indicate experience with the topic of various committees, we computed an additive term of experience in all the possible issue areas a deputy could be considered an expert in. Hence there are Federal Deputies who can be seen as experts in distinct areas, and these will have a higher chance of being committee chairs. 18

19 Our fundamental explanatory variable, an index called party loyalty rate is defined as the percentage of relevant roll calls on which a deputy votes with the majority of her/his party caucuses weighted by this deputy s turn out rate on these same roll calls. Therefore, we control for Deputies absences from important voting sessions. The criterion of roll call relevance was taken from Carey 31 (1998), who points out two features: participation and closeness. Participation is the percentage of legislators voting on a roll call and closeness is the absolute difference between the percentage of legislators who vote yes and the percentage of legislators who vote no. We can thus write the relevance index as: Carey s relevance index = % of legislators voting x (1 [ %aye - %nay] ) (6) In this article, only roll calls whose score on the index were above the 0.4 threshold were considered relevant. Notice that a 0.4 threshold denotes that in each roll call taken under consideration at least 60% of the legislators were present and a 2/3 majority won over a 1/3 minority. Using this threshold criterion, 213 roll calls were considered in the first period and 259 in the second one. The basic role of this variable is to show the rate at which a deputy supports the majoritarian position of her/his party on the floor, thus, high degrees of loyalty indicate that a specific deputy is more reliable, from the point of view of party leaders, than deputies that display relatively low rates of this same variable. Data for the roll calls were collected from the Diário do Congresso and from Nicolau 32. In addition to the variables discussed above, we also controlled for other factors that may influence the selection of chairs. First we control for serial correlation and for the impact of inertia in committee chair selection by including a 19

20 lagged term of the dependent variable (lcommittee president). The inertial component of holding a committee chair position should be statistically significant only in the first democratic period. This variable should not be significant in the current period due to the forced rotation rule. In fact, such variable could be ignored in the current period, but in order to increase the comparability of the results, we keep model specification constant in both samples. Second, we control for being a member of the directing table of the Chamber (similar to House Speaker in the US House of Representatives) (directing table). Third, we control for having previous legislative experience in state and municipal level, which indicates familiarity with the functioning of legislative organizations (legislative experience). Fourth, we include in the equation if the Federal Deputy has occupied a position in the federal, state or municipal bureaucracy also to control for previous experience with executive legislative relations (bureaucrat). Fifth, we control for the status of the Federal Deputy, if a main office holder or if a substitute (main office holder). Substitutes take office when a main office holder takes a leave of absence either for personal reasons or to occupy an appointed position in national, state or municipal level bureaucracies. Once substitutes take office, they have identical rights as main office holders and keep the position until the holder of the mandate returns to the Chamber. Since substitutes tend to stay in office for shorter periods than main office holders, they probably have less access to being committee chairs. Finally, we also test if being a member of the governing coalition influences access to chairmanship positions (government). Given the legislative power of presidents in the Brazilian system, being from the governing coalition should affect 20

21 access to committee chairmanship. This variable was created based on party membership of the deputy and the partisan configuration of the presidents cabinet. Amorim Neto 33 has advanced an interesting form of measuring coalition management in coalition presidential systems by relying in the cabinet partisan composition as an indicator of the parties that compose the governing coalition. We use his classification of cabinet compositions to define if a Federal Deputy is a member of a party in the president s supporting coalition. We created a dummy variable where the Federal Deputy receives value one if she/he is a member of a party that is part of the president s cabinet. 34 Data Analysis Our theoretical claim is that because party loyalty and access to committee chairmanship were scarcer in the period, party loyalty was a decisive element in the selection of committee chairs. Nowadays, since neither of the above are as scarce as before, party loyalty is not as decisive as it was. Notice that we are not claiming that party loyalty is irrelevant today. In fact, our argument is just the opposite. Party loyalty is the predominant behavior nowadays, not the exceptional one as it was before. Party leaders still favor their allies when choosing committee chairs, but since party leaders can choose from a larger pool of allies, this criteria is not as vital as it was before. Hence, in order to show that our claims resonate with reality, we must test 1) if levels of party loyalty are higher in the current period; 2) if committee chairs display higher levels of party loyalty than other Federal Deputies in both periods, and 3) if party loyalty is a statistically significant criteria for committee chair selection in a multivariate model in the previous period and isn t in the current one. 21

22 Party loyalty and Committee Chairmanship The histograms of party loyalty presented in Graphs 1 and 2 indicate that the period before 1964 was one of lower levels of party loyalty than the period after This is true for both the aggregate, floor measure of party loyalty shown here as well as when disaggregated by parties. 35 But one point must be clear, even though the levels of party loyalty are higher today; the standard deviations around such measures are not any smaller than those of the previous period, which indicates that there still is quite a bit of defection today and that not all deputies of the same parties display similar patterns of loyalty. This supports our claim that an important aspect of the functioning of the Chamber of Deputies, in both periods, is the negotiation with individual Deputies. However, levels of loyalty are overall higher today, due mostly to the power leaders currently have to coordinate the disbursement of resources that affect Deputies careers. (Graphs 1 and 2 about here.) Table 1 indicates that committee chairs in both periods display higher levels of party loyalty than other party members. The difference between both periods, as we predicted, is that in the current period, the difference in means is not statistically significant. This indicates that party loyalty is a relevant criterion of selection in both periods; only allies get access to power positions. However, in the current period, leaders levels of party loyalty are as discrepant from other members as they were in the past. (Table 1 about here.) Correlates of Committee Chairmanship In order to more appropriately test our main hypothesis we ran logistic regressions with identical models for the two different periods. The dependent 22

23 variable is a simple dummy variable where the value one indicates that a given Federal Deputy was a chair of a committee in a specific year. We expect the indicator of party loyalty to be positive and significant in the period from 1951 to 1963 and positive but insignificant in the current period. Party loyalty is a more powerful currency for gaining prominence in Congress when it is a scarcer resource. When holding a committee chairmanship presents a higher stake for parties, as it seems to be the case in the 1951 to 1963 period than in the current democratic period, party leaders will be sure to have allies in such positions. Table 2 contains the odds ratio, robust standard errors, and the statistical significance of the variables that operationalize our different hypotheses for both periods. We highlight in the table the result of our main hypothesis; the impact of party loyalty. In accordance with our theoretical argument, party loyalty was a defining factor in influencing access to a committee chairmanship in the pre-1964 period. From 1951 to 1963, the combination of institutional rules with intra and inter party politics lead to a greater prominence of party loyalty as an influential factor in the distribution of power inside the Chamber of Deputies. Party loyalty increased by 5 times the likelihood of being a committee chair in the previous period, whereas in the current period it is positive but not statistically significant. 36. (Table 2 about here) In addition to party loyalty, several other variables we controlled for in the period are also important in affecting the allocation of chairmanships. A first one is the decisive impact of inertia. That is, the lagged term of having occupied the position of chair on the previous year is highly correlated with occupying it again in the subsequent year. This indicates that there was continuity and stability in the patterns of chairmanship position allocation. On the other hand, in the current 23

24 period, the lagged term of being a chair decreases the chances of staying as chair, however it is not statistically significant. The forced rotation of committee chairs is what explains this finding. It is no surprise that having occupied a position of chair in the previous year does not lead to occupying it again in the following year if rotation as committee chair is forced. A factor with decisive influence in the allocation of committee chairs in the previous democratic period, as well as in the current one, is having professional experience in the matters concerning the jurisdiction of the committee. Federal Deputies whose professional carrier or academic training made them specialists in the issue area of the committee had a greater chance of holding a chairmanship. This variable drops off in the period because it perfectly predicts holding a committee chair. That is, all committee presidents were specialists in the topic of jurisdiction of that committee. Therefore, not only intra-legislative experience is important in the choice of committee chairs, but also their previous career trajectories. 37 Another variable that positively influences the allotment of chairs in both periods is seniority. Hence, similar to what occurs in the US, and even in a context where there is forced rotation of committee chairs, being an experienced legislator, knowledgeable of the rules and functioning of the Chamber of Deputies is a relevant factor in occupying a committee chair. Representatives who occupied more terms have a greater chance of being committee chairs, independent of differences in the rules of chair allocation. One factor negatively influences the likelihood of being a committee chair in the previous democratic period; being a member of the Directing Table of the Chair. This is further evidence that there wasn t much circulation of Federal Deputies 24

25 among the main posts in the hierarchical structure of the Chamber of Deputies in that period. Holding positions in the Directing Table decreased the odds of also being a chair, indicating that Congresspersons specialized in distinct roles inside the hierarchy of the Chamber. The difference in the explanatory power of the identical models in the two periods 44% in the first period and only 1% on the current one is evidence that the combination of institutional changes plus parties higher capacity of influencing Chamber decisions increased the randomness of chair selection process. Chairs forced rotation, lower budgetary allocation power of Congress, higher centralization of decision-making power on the hands of party leaders, and overall higher levels of party loyalty in the present period gave leaders more leeway to choose committee chairs from a larger pool of allies. Consequently, loyalty levels of committee chairs are not much higher than that of backbenchers in the current period. CONCLUSION We provided evidence in this article of the impact of distinct institutional settings in federal deputies legislative behavior. In a situation where parties are oriented by a factional logic and where power is more equally distributed between institutional actors, cooperative behavior is a central element in defining power positions in the Chamber. This line of reasoning is confirmed by the influence of party loyalty in affecting committee chair selection. In a broader perspective, we may say that the differences between the two periods under analysis shows that changes in the interaction between committees and parties is not solely due to variation in the incentives originated from the electoral connection, for it also suffers 25

26 the influence of executive legislative relations, what Amorim Neto and Santos (2001) called the presidential connection 38. We also find similarities between the transfer of power from committee chairs to party leaders in Brazil with Rohde s 39 and Sinclair s 40 description of trends in U.S. House of Representatives. As these authors claimed, as the incentives for party cohesion increased and as House Speakers became more influential, committees relevance in the functioning of the House decreased. The same logic applies to the Brazilian case: the higher fluidity of party dominance over specific committees is evidence of the diminishing importance of committees in a environment of higher levels of party loyalty and lower degree of committee s influence in policy decision inside the Chamber. Finally, our results indicate that in the current period, the forced rotation of committee chairmanship increases the randomness of Deputies occupying leadership positions in the Chamber. Hence, the change in rules inside the Chamber from the previous period and the increase in levels of party loyalty made the allocation of supporters to committees less of a central concern for party leaders. What our results indicate is that in the period from 1951 to 1964 the criteria for the allocation of committee chairmanship in the Chamber were more rigid than nowadays. Today, since most deputies defect less from the party s majority opinion, loyalty of individual Federal Deputies becomes less of a decisive element in the distribution of power positions inside the Chamber. This is so not because Federal Deputies and party leaders are not interested in committees nor because party leaders don t use committee positions to reward supporters, it simply is this way because both access to committees and party loyalty are not as scarce resources in the current period as they were in the previous one. 26

27 Notes 1 B. Ames, The Deadlock of Democracy in Brazil (Ann Arbor: Michigan University Press, 2001). 2 A. Figueiredo and F. Limongi, Executivo e Legislativo na Nova Ordem Constitucional. (Rio de Janeiro: Editora da Fundação Getulio Vargas, 1999). 3 There is, however, an enormous controversy around the topic. For a good view of the U.S. case, see C. J. Deering and S. S. Smith, Committees in Congress (Third Edition) (United State of America: Congressional Quarterly, 1997). For an introduction to committees in the western European parliaments, see K. Strom, Parliamentary Committees in European Democracies, in L. D. Longley and R. H. Davidson, eds., The New Roles of Parliamentary Committees (London: Frank Cass, 1998). 4 G. Tsebelis and E. Alemán, Agenda Control in Latin American Presidential Democracies. Paper presented at the Annual Meeting of the American Political Science Association, August 29-September 1, C. Pereira and B. Mueller, Uma Teoria da Preponderância do Poder Executivo: O Sistema de Comissões no Legislativo Brasileiro, Revista Brasileira de Ciências Sociais, 15 (2000), pp M. Jones, Explaining the High Level of Party Discipline in the Argentine Congress, in S. Morgenstern and B. Nacif, eds., Legislative Politics in Latin America (Cambridge: Cambridge University Press, 2001); P. Siavelis, Exagerated Presidentialism and Moderate Presidents: Executive-Legislative Relations in Chile, in S. Morgenstern and B. Nacif, eds., Legislative Politics in Latin America (Cambridge: Cambridge University Press, 2001). 7 B. Crisp, Presidential Behavior in a System with Strong Parties: Venezuela, in S. Mainwaring and M. S. Shugart, eds., Presidentialism and Democracy in Latin America (Cambridge: Cambridge University Press, 1997). 8 Ames, The Deadlock of Democracy in Brazil. 9 Figueiredo and Limongi, Executivo e Legislativo na Nova Ordem Constitucional; Ames, The Deadlock of Democracy in Brazil. 27

28 10 G. W. Cox and S. Morgenstern, Epilogue: Latin America s Reactive Assemblies and Proactive Presidents in S. Morgenstern and B. Nacif, eds., Legislative Politics in Latin America (Cambridge: Cambridge University Press, 2001); S. Mainwaring and M. S. Shugart, Presidentialism and Democracy in Latin America (Cambridge: Cambridge University Press, 1997). 11 Cox and Morgenstern, Epilogue: Latin America s Reactive Assemblies and Proactive Presidents, pp. 12 Jones, Explaining the High Level of Party Discipline in the Argentine Congress, pp. ; Siavelis, Exagerated Presidentialism and Moderate Presidents: Executive-Legislative Relations in Chile, pp. and B. Nacif (2001), Understanding Party Discipline in the Mexican Chamber of Deputies: The Centralized Model in S. Morgenstern and B. Nacif, eds., Legislative Politics in Latin America (Cambridge: Cambridge University Press, 2001). 13 Figueiredo and Limongi, Executivo e Legislativo na Nova Ordem Constitucional. 14 Ames, The Deadlock of Democracy in Brazil. 15 D. Samuels, Progressive Ambition, Federalism, and Pork-Barreling in Brazil, in S. Morgenstern and B. Nacif, eds., Legislative Politics in Latin America (Cambridge: Cambridge University Press, 2001). 16 Ames, The Deadlock of Democracy in Brazil, p Federal Deputies in Brazil have the right to present individual amendments to the budgetary law transferring federal funds to whichever municipality and state they so wish. We agree with Samuels (2001) that local level pressures orient Federal Deputies choices of which municipalities and states will be favored by their amendments and that Governors are especially influential in such moment. However, the approval of the amendments does not stipulate when and if the executive branch will in fact disburse the money. Usually, the disbursement of such funds is affected by Federal Deputies cooperation in votings inside the chamber (Pereira 2000, Ames 2001). Our point is that party leaders play a central role in informing the executive branch about which Deputies deserves to receive the money and when. 18 Jones, Explaining the High Level of Party Discipline in the Argentine Congress. 19 Samuels, Progressive Ambition, Federalism, and Pork-Barreling in Brazil. 28