Over the course of a few decades, political scientists and economists developed a rich collection of formal models of political coalition formation and survival in parliamentary regimes (see summaries in Laver 1998, 2003; Martin and Stevenson 2001). This first wave of work in coalitional politics tended to think in general terms about the coalition characteristics that facilitated survival. Yet, as Druckman noted more recently, "Extant work offers scant insight into the processes of governance" (2008, 482). A newer branch of literature has begun exploring the idea of ongoing administration or "coalition management," particularly in the context of multiparty presidential regimes (e.g., Chaisty, Cheeseman, and Power 2014; Hiroi and Renno 2014; Pereira and Melo 2012; Pereira, Power, and Raile 2011; Pereira, Power, and Renno 2005; Praca, Freitas, and Hoepers 2011; Raile, Pereira, and Power 2011). The current project builds on such work by considering the various decisions and tools involved in coalition management and by considering the ways a multiparty president can minimize governing costs with coalition parties.
Multiparty presidential regimes provide fertile ground for examining coalition management. Importantly, only a president in a multiparty system must simultaneously manage: (1) the entire executive branch, (2) his own political party, (3) relations between constitutionally separate branches, and (4) a true potential multiparty cabinet. The president is clearly the head of any coalition even prior to formation and is more difficult to replace than a prime minister due to the lack of confidence votes and the constitutional separation of origin and survival from the legislature. The president can often construct new cabinets in the middle of a term in a fairly unilateral manner, as well. A multiparty president is required to make a range of important management decisions due to these diverse, interrelated responsibilities of a political and administrative nature.
Multiparty presidents often operate in difficult environments but are also often given considerable institutional tools and resources to overcome this fragmentation and to facilitate governance (Shugart and Carey 1992). Some presidents fall into the trap of treating resource distribution as the end itself, which is a situation that lends itself to political corruption (Power and Taylor 2011). Other executives use their tools as a means to the end of achieving policy goals. Such executives may be concerned about politics and the spoils of office, but they also have sought the highest office in the land in order to implement certain policies. For these presidents, coalition management is an ongoing and complicated process.
This project asks how a president in a multiparty system might optimize outcomes in a complex and fragmented management environment. The use of resources in managing the coalition not only constitutes a cost in itself but also often means bearing the cost of policies drifting away from the president's preferences. The argument developed here is an important step in understanding governance and policymaking in multiparty presidential regimes and in being able to evaluate the performance of presidents in such regimes. This theoretical development and the derivation of testable implications are crucial for advancing the literature on coalition management.
This project examines these complex issues using Brazil as a case study. Although the last few Brazilian presidents have governed practically under the same institutional setting, they have faced different constraints and have made different decisions about managing their coalitions. (1) Once elected, the president faces at least three interconnected constraints: (1) the level of party fragmentation in the legislature, (2) the size of the president's party relative to the sizes of other parties, and (3) the ideological distances between the president's party and the other political parties in the legislature. Working within such constraints, the executive makes decisions about the size and ideological heterogeneity of his coalition and about proportionality in rewarding cabinet seats. The latter refers to the proportionality with which a coalition party's number of legislative members translates to seats in the cabinet. We argue that these coalition management decisions then affect the need for other expenditures. In particular, we propose that the number of cabinet posts, the allocation of funds to the ministries, and pork for legislators are among the costs affected by these initial decisions.
The section that follows discusses presidential decisions about coalition management and governing costs, states our primary hypotheses, and considers other external and internal influences. The section after describes our data and methods. The two subsequent sections provide a more case-based analysis of data and then the results of seemingly unrelated regression (SUR) analyses. These analyses examine the effects of coalition management strategies on governing costs, finding that choices about the size and ideological spread of the coalition and cabinet proportionality do indeed influence governing costs even when controlling for other constraints and factors. A final section discusses these results and their implications.
Strategic Coalition Management
Presidents elected in multiparty, hyperrepresentative systems with significant party fragmentation in the legislature, as in Brazil, often face a minority condition. Therefore, presidents must build postelectoral coalitions if they wish to enjoy majority status. In so doing, presidents must make decisions about: (1) the number of parties that will take part in the coalition, (2) the degree of ideological difference between coalition partners and the president's party, and (3) and how power will be shared within the cabinet among coalition allies. Though their focus is ultimately on the generation of legislative support, Raile, Pereira, and Power (2011) argue that such managerial choices engender different trade-offs and costs for the executive.
Coalition Management Choices and Governing Costs
Government formation theories have long focused on the relative sizes and ideologies of political parties that serve as potential partners in a coalition. The concept of the "minimal winning coalition" and its equilibrium status in cooperative games emerged in early models of government formation (e.g., see von Neumann and Morgenstern 1953). Each member in a minimal winning coalition was essential to the winning status. Amendments to this concept incorporated the minimal winning coalition with the smallest weight (Riker 1962) or the minimal winning coalition with the smallest number of members (i.e., the "bargaining proposition" of Leiserson 1968). The addition of ideology and policy-oriented parties resulted in the "minimal connected winning coalition" (Axel-rod 1970). Parties in such a coalition were adjacent to one another on the left-right dimension. An amendment here suggested that a minimal winning coalition would include the smallest possible ideological range (de Swaan 1973). Martin and Stevenson (2001) have also introduced more recent theories in which the largest party is the centripetal actor in coalition negotiations or the party containing the median legislator enters the government and becomes a policy dictator.
Other scholars have argued that the value of the pie being divided among coalition partners also matters (see Diermeier, Eraslan, and Merlo 2003). This observation is key for Strom (1990) and Muller and Strom (1999) in understanding that government membership has costs that affect the occurrence of minority governments. In other words, a government has to be valuable in order to attract the participation of coalition members. The value of taking part in a coalition directly relates to the richness of the spoils of office, the policy-making opportunities that holding office entails, and the expected electoral gains (Strom 1990). The influence of Strom's work has gone beyond the recognition of minority government by also investigating government stability and duration.
Yet other authors have argued that parties that form the government do not have complete information about the true preferences of their rivals in the real world. This lack of complete information reduces the ability of parties to make credible commitments and increases incentives for opportunistic behavior and blackmail. Accordingly, we should expect an increase in information asymmetries between players to result in an increased coalition size (Dodd 1976). Carrubba and Volden (2000) predict that a "minimal necessary coalition" results as parties attempt to create a more stable trading environment less subject to the high costs of defection; however, they also predict that oversized coalitions make the budget more difficult to approve. The occasional disconnect between the legislative seat share of a party and the ability of that party to influence behavior at the polls (Ansolabehere et al. 2005) is yet another reason a formateur may choose to construct something other than a minimal winning coalition.
Researchers who argue that supermajorities can be cheaper also detract from arguments in favor of minimal winning coalitions. Supermajorities may occur so frequently in empirical terms because they prevent small parties from acting as pivots, thus decreasing overall costs (see Groseclose and Snyder 1996). In some presidential systems, lack of party loyalty and lack of discipline mean that a minimal winning coalition may not be enough to win consistently over time. Larger coalitions can also occur when two buyers (e.g., interest groups, political parties) with opposite preferences have significant resources and bargaining power (Groseclose and Snyder 1996; Wiseman 2004). On the other hand, empirical research has shown that minority (i.e., smaller than...