Abstract

Electoral rules are assumed to influence turnout. However, assessing this empirically is challenging because they rarely change and, when they do, counterfactuals are hard to come by. In 2012, Chile moved from voluntary and permanent registration and mandatory voting to automatic registration and voluntary voting. We study how electoral rules influence turnout by analyzing variations in turnout for presidential elections before and after this reform, with an original approach and using novel data. We estimate changes attributable to voluntary voting among the registered population as the increase in abstention rates among them and changes attributable to both automatic registration and voluntary voting among the non-registered population as the increase in turnout among them. We estimate counterfactual abstention and registration rates based on past behavior and use bounds to account for uncertainty. Our estimates suggest that while automatic registration and voluntary voting brought 7.1% of eligible population who were unregistered to the polls, voluntary voting pulled away 12% who were previously registered. The explicit purposes of this reform were to increase turnout and reduce the age bias in voting. We estimate a reduction in turnout of almost 5% of eligible population and a 39% reduction in the age bias of voters.