Coalition Formation and Survival under Multiparty Presidential Democracies in Latin America: Between the Tyranny of the Electoral Calendar, the Irony of Ideological Polarization, and Inertial Effects

This paper attempts to achieve two goals: first to find under which circumstances coalitions are more likely to form, and second how do they survive once formation is achieved. Because this paper wants to deal with these two major questions, it has two sets of working hypotheses and two different databases. The first part of the analysis, is centered on executive coalition formation; the second, is about who are the actors likely to join an executive coalition and if there is certain pattern of coalition evolution. For the first question a datab ase composed of 129 observation (countryyears) was built. For the second question I utilized a database composed of 1326 cases. Each observation corresponded to a party per year per country and a dummy indicated whether a party belonged to the executive coalition that respective year. Although the paper uses GLS and OLS, the methods were essentially logistic regressions with dummy variables. The findings suggest that governing coalitions are subject to the tyranny of the electoral calendar, and that the higher the ideological dispersion of a party system, the less likely a coalition is formed. They also show that the presidential share in the legislature has not effect on the prospects of coalition formation per-se but rather coalitions are contingent on the degree of effective opposition. The research finds that the inertial effect of having being member at the executive coalition (a variable constantly omitted by the literature) is highly significant in order to foresee which parties are more likely to remain in a coalition government. It suggests that the ideological proximity between a given party and the president play an important effect on entering to the governing coalition. Nonetheless, whether a party remains in the executive coalition is also subject to the tyranny of the electoral calendar.