Pre-electoral coalitions have been largely ignored in Presidential systems. In this dissertation I study the case of Mexico where since 1991 a pre-electoral coalition has been formed by two non-contiguous parties; the far-right party (PAN) and the far-left party (PRD). Despite dramatic differences in policy agendas, they have won important elections at the state and municipal level. The creation of this coalition creates a puzzle that is not addressed by existing theories such as spatial models, party alignment, incumbency advantage, federalism, electoral budget cycles and ideological expenditures. The first chapter of this dissertation answers the following questions: Will voters support their preferred party despite the non-contiguous coalition? Will voters punish coalition members if it breaks while in office? Will coalitions stay together through multiple electoral cycles? This chapter develops a Game Theoretical Bayesian model to analyze voters' electoral behavior. The main finding is a separating equilibrium in the multiple-period game. Reputation effects will lead to a long-term collaboration while in the short-term the coalition will break. Voters support these coalitions if they stay together once in office and implement a policy platform that maximizes their payoffs. The second chapter introduces the concept of "Partial Alignment'', created by a pre-electoral coalition in power, and its effects on resource allocation. Do municipalities where pre-electoral coalition governments win an election receive significantly more or fewer resources from the federal and state level governments than those where the coalition loses? The theoretical answer is ambiguous: Partial alignment could lead to more resources in an effort to keep the party in power - even as part of a coalition - or fewer resource due to the desire to stop sharing power in order to govern alone. Analysis uses a Regression Discontinuity Design with a matched dataset that combines data on municipality level income with municipal election results between 1989 and 2016. Results show that municipalities where the pre-electoral coalition barely won received fewer federal resources in the year of the election and the year before. The third chapter looks at how the pre-electoral coalition spend their resources. Do pre-electoral coalitions spend differently than single party governments? Will pre-electoral coalition governments increase expenditure after the election or wait until the next electoral cycle? Will they strategically pick between short and long-term expenditures to signal voters their competence? Analysis uses a Regression Discontinuity Design and a Fixed Effects Panel Model with a matched dataset that combines data on municipality level expenditures with municipal election results between 1989 and 2016. Results show that municipalities where the pre-electoral coalition barely won create a new electoral budget cycle by increasing expenditures during their first year in office. It is also shown that these governments are not strategic when selecting between short and long-term expenditures to signal voters their competence in office.