Quality-Based Explanations of Incumbency Effects.

Empirical studies of incumbency effects continue to accumulate, but progress in explaining these findings is modest. I introduce a simple framework that clarifies how differences in candidate characteristics (i.e., quality) could account for empirical findings of incumbency effects based on regression discontinuity designs (RDD). The key observation is that although RDD ensures that marginal winners and losers of close elections are comparable, the candidates who compete in the next election may differ for many reasons, of which "scare-off" (the tendency of incumbents to deter strong challengers) is just one. Using a simple model, I show that quality differences between marginal incumbent-party candidates and their opponents may persist even when incumbency is irrelevant to voters, incumbent retirement is random, and all new candidates are drawn fromthe same candidate pool. I conclude by showing how quality differences can help explain puzzling findings in prior empirical work. [ABSTRACT FROM AUTHOR]