This dissertation examines how the structure of party systems affects the stability of presidential coalitions in Latin America. Scholars believe that the incoherent and unstable party systems in several Latin American countries contribute to poor governance, economic instability, and the fragility of democracy in the region. This dissertation seeks to explain how regional interest differences within countries affect the dimensions of conflict and the stability of coalitions in their party systems. It will show that although conditions conducive to the aggregation of local and regional interests into stable parties or multiparty coalitions exist in some Latin American countries, they are absent in others. Although we have a theoretical expectation that in seeking to control government, politicians and the interests they represent will form long-lived, majority parties or coalitions, this dissertation will argue that for some types of regional interests aggregated into regional parties and for presidents, short coalitions often make political sense. Regional parties that care most about pork for their districts and policies orthogonal to the main left-right policy dimension can only meet their constituents. needs by becoming part of the president's coalition. Consequently, they compete fiercely for places in coalitions of any ideological coloration. This competition tends to drive down the price presidents have to pay to attract coalition partners and to lead to coalition instability as different parties jockey for places. In the empirical investigation of this argument, the researchers expect to find that the presence of parties seeking region-specific and distributive policies is associated with shorter presidential coalitions, even after taking into account different institutional and historical factors that might influence coalition building. The research will involve both in-depth examinations of social and electoral data at the sub-national level in some countries and cross-national empirical analyses for most of Latin America in the current democratic period. The investigation will measure the extent of variation in the vote shares parties receive in different regions of each country, and identify the types of interests represented by regionally-based parties. It will use census and other socio- economic data to map local interests onto vote shares for different kinds of parties; data on the ideological positioning and policy goals of parties to show that some regional parties do not fit onto the standard left-right policy dimension; and data on the length of presidential cabinets, legislative coalitions, and alliances across second-round presidential elections to show that regionalized party systems are associated with shorter coalitions. Broader Impact: The results of this dissertation project will give political scientists a better understanding of the electoral landscape in Latin America and the conditions under which narrow, regionally-based parties survive to play an important role in national politics in some of these countries. These findings will address the relationship, or trade-off, between broad or diverse representation and stable government in the context of highly unequal and territorially heterogeneous societies.
