Registration costs and voter turnout

Abstract

We exploit a natural experiment in Massachusetts in 2012 to estimate the causal effect of lowering voter information and registration costs on: voter registration, turnout and voting behavior in presidential elections. Both a within-Massachusetts specification and a cross-state specification (utilizing Vermont, Maine and New Hampshire data) show a statistically significant effect on voter registration and turnout that is of a material magnitude. However, conditional on registration we find no material difference in turnout. Finally, we find a large treatment effect on Democrat voteshare. Our results highlight the importance of voter registration and information costs for electoral participation, especially for citizens from lower socioeconomic backgrounds.