Presidential campaign expenditures: Evidence on allocations and effects

This paper analyzes the impact of presidential campaign spending on election results. Analyses of expenditures and voting are often plagued by simultaneity between campaign spending and expected vote share. However, game-theoretic models of resource-allocation decisions made by a central actor (i.e., a presidential campaign) suggest that candidates will spend more in close races and in races likely to be pivotal. We provide empirical support for this theory; using Federal Communications Commission data from the 1972 presidential election, we find that expenditures were higher in states where the election was expected to be closer and in states likely to be pivotal. We use these two factors as instruments in a two-stage least squares model to estimate the effect of spending on votes. We find that, contrary to previous theory and research, presidential campaign spending significantly increases a candidate's vote share.