Using Objective Local Economic Conditions to Estimate the Effect of Subjective National Economic Evaluations on Vote Choice.

One of the most replicated findings in political science is links individuals' perceptions of past economic performance with the vote. Yet there are a number of reasons to expect that self-reported evaluations of the economy are endogenous to vote choice. Indeed, if endogeneity is present, then existing models cannot be viewed as providing a valid test of the causal claim derived from retrospective theories of voting. In this paper, we show that objective local economic conditions help to frame individuals' perceptions of the national economic situation and are a useful instrument for retrospective sociotropic evaluations. We use an instrumental variables probit model to assess the causal effect of these evaluations on the decision to vote for the incumbent president or his party's presidential candidate in the 1980 to 1996 U.S. presidential elections. We find that, while there is much greater uncertainty regarding the IV estimate for sociotropic evaluations, it is quite similar to the estimate obtained by a traditional probit model in which any potential endogeneity is ignored. Our results suggest that existing models of retrospective sociotropic voting may be robust against claims of endogeneity. [ABSTRACT FROM AUTHOR]