党派政治、辖区间合作与执法：来自墨西哥的证据

We study how partisan politics affect inter-jurisdictional cooperation and public good provision in federal systems. We focus on law enforcement in Mexico, a country marked by high violent crime and strong partisan cleavages. Using a Regression Discontinuity Design in close municipal elections, we show that municipalities where the party in power in most neighbors barely won are more likely to cooperate with their neighbors on law enforcement than those where it barely lost, especially during periods of widespread violence. Such cooperation, in turn, improves crime prevention and significantly reduces homicide rates, with effects increasing the more neighbors are governed by the same party. These results hold regardless of party identity, incumbency, or cooperation with federal or state authorities. Overall, our findings indicate that while horizontal cooperation can improve local public goods in the presence of geographical spillovers, partisan divisions may hinder this process.