The Swing Voters' Blessing

Can democracy work well if the electorate is neither fully informed nor fully rational? I provide an affirmative answer in a model with quality differences between two ideological candidates running for office. The candidates commit to policy platforms before the elections take place. All voters care about the quality of the candidates as well as the policies they offer; however, the quality differences are only observable to a limited number of informed voters. I consider two different scenarios. In the first one, all uninformed voters are fully rational and follow an optimal strategy of making their voting decisions dependent only on the position of their own policy bliss point relative to the median policy bliss point. As in the standard case with only informed voters, the candidate who is preferred by the median voter wins. In equilibrium, this is the higher quality candidate and the policy implemented is the same as if all voters had been fully informed. In the second scenario, I show that the existence of boundedly rational uninformed 'swing voters' increases the welfare of the majority of voters. These swing voters do not consider the fact that their vote influences their utility only when their vote is pivotal. Consequently, they always support the candidate whose policy platform they prefer. In this scenario, the winning high quality candidate's policy is closer to the median voter's bliss point than in the first scenario with only fully rational voters. This is the "Swing Voters' Blessing" that makes the majority of voters better off.