Risk aversion and bandwagon effect in the pivotal voter model

Abstract

The empirical literature on the effects of opinion polls on election outcomes has recently found substantial evidence of a bandwagon effect, defined as the phenomenon according to which the publication of opinion polls is advantageous to the candidate with the greatest support. This result is driven, in the lab experiments, by a higher turnout rate among the majority than among the minority. Such evidence is however in stark contrast with the main theoretical model of electoral participation in public choice, the pivotal voter model, which predicts that the supporters of an underdog candidate participate at a higher rate, given the higher probability of casting a pivotal vote. This paper tries to reconcile this discrepancy by showing that a bandwagon effect can be generated within the pivotal voter model by concavity in the voters’ utility function, which makes electoral participation more costly for the expected loser supporters. Given the strict relationship between concavity and risk aversion, the paper also establishes the role of risk aversion as a determinant of bandwagon.