How do multiethnic coalitions affect the risk of violent conflict? We argue that government leaders form larger ruling coalitions as the uncertainty over violent threats from radical sub-groups increases. We formalize this argument in a novel multi-actor model. Our game-theoretic approach illustrates how power-maximizing government leaders reduce the risk of rebellion by sharing power with radical subgroup leaders, while only slightly increasing the risk of coups. Empirically, we rely on a novel selection estimator and data on ethnic coalitions to investigate the predictions from our theoretical model. Our findings show that the conflict-reducing effect of power sharing increases once we consider the endogeneity of coalition formation and conflict. Our results also demonstrate that power-sharing coalitions are most effective when including radical sub-groups. Our study has important implications for research on coalition formation, authoritarian politics, and armed conflict.
Read full abstract