Why votes have value: Instrumental voting with overconfidence and overestimation of others' errors

We perform an experiment in which subjects bid for participating in a vote. The setting precludes conflicts of interests or direct benefits from voting. The theoretical value of participating in the vote is therefore zero if subjects have only instrumental reasons to vote and form correct beliefs. Yet, we find that experimental subjects are willing to pay for the vote and that they do so for instrumental reasons. The observed voting premium in the main treatment is high and can only be accounted for if some subjects either overestimate their pivotality or do not pay attention to pivotality at all. A model of instrumental voting, which assumes that individuals are overconfident and that they overestimate the errors of others, is consistent with results from treatments that make the issue of pivotality salient to experimental subjects.