Divide to Rule: Deconcentration As Coalition Manipulation

Abstract: Why do countries create new sub-national units of government? Recent studies have argued that deconcentration is the product of local and national elites jointly pursuing new districts for patronage and electoral gain. Yet this perspective leaves unresolved why patterns of contestation over deconcentration fail to map onto dominant cleavages such as partisanship or ethnicity, and how such conflicts get resolved. I argue that rulers instead pursue deconcentration to manipulate political coalitions such as factions or parties. I study a model of non-cooperative coalition formation, finding that elite preferences over deconcentration can be induced by expectations about coalitional alignments under alternative player sets. A central implication of the theory is that changes to existing sets of elites can precipitate deconcentration by making alternative coalition structures more attractive. I test this prediction using new data on deconcentration worldwide from 1960 through 2010. I find that exogenous variation in player sets --- sudden leadership deaths --- leads to a significant increase in the probability of deconcentration.