Top-Down Divergence

Although the traditional Hotelling—Downs—Black (Hotelling, 1929; Downs, 1957; Black, 1958) spatial model of voting predicts candidate convergence, several empirical studies show that convergence in actual elections is rare. In response, researchers have designed models that produce the more empirically tenable candidate divergence outcome. While most of these models rely on assumptions about the election or the electorate to derive divergence predictions, I show that divergence is possible based on assumptions about institutional power seeking among legislators. More specifically, I assume that ideological proximity to political parties within the legislature determines how much power over policy outcomes an individual legislator receives. Given this assumption, I find that candidates competing for seats in legislatures with two parties virtually never converge, because their parties pull them in different directions and away from their districts’ median voter ideal point. Party divergence within the legislature, then, creates candidate divergence at the electoral level.