Divide and Diverge

Political polarization can be beneficial to competing political parties. I study how electoral competition itself generates incentives to polarize voters, even when parties are ex ante identical and motivated purely by political power, interpreted as office rents or influence. I develop a probabilistic voting model with aggregate popularity shocks in which parties have decreasing marginal utility from political power. Equilibrium policy convergence fails. Platform differentiation provides insurance against electoral volatility by securing loyal voter bases and stabilizing political power. In a unidimensional policy space, parties' equilibrium payoffs rise as voters on opposite sides of the median become more extreme, including when polarization is driven by changes in the opponent's supporters. In a multidimensional setting, parties benefit from ideological coherence, the alignment of disagreements across issues. The results have implications for polarizing political communication, party identity, and electoral institutions.