A Referendum Experiment with Participation Quorums

This paper conducts a laboratory experiment focused on a yes/no referendum in which the outcome is valid only if voter turnout is greater than a predetermined level. Such a participation quorum is argued to induce the minority group of voters to abstain strategically in particular when the quorum is large. In our experiment, subjects are randomly divided into two groups yes and no in advance of voting, and the level of quorum changes round by round. We observe that (i) if the quorum is small, all subjects go to the polls, and (ii) if the quorum is large, subjects in the ex‐ante majority group go to the polls, whereas subjects in the ex‐ante minority group tend to abstain. As a result, if the quorum is large, it frequently happens that the outcome is rendered invalid because of low voter turnout. Furthermore, when the number of subjects in each referendum increases, subjects of the ex‐ante minority abstain more aggressively, and there arises room for the ex‐post minority to win if the expected number of members is similar between the two groups. Therefore, if policy makers would like to avoid the possibility of strategic abstention, then our results suggest that they should impose either no quorum or only a relatively small quorum which most voters expect will be dwarfed by the actual turnout. If policy makers also care about the possible determination of either yes or no with extremely low voter turnout, as well as strategic abstention, then a relatively small quorum is recommended.