Tiebout (1956) suggests that individuals with similar preferences for different bundles of local public goods sort themselves into communities governed by different institutions in expectation of interacting with others who have chosen the same institution. He argues that if communities are sufficiently heterogeneous and consumer-voters are fully mobile, voting with feet generates considerable efficiency gains in public goods provision.
An attempt to study this suggested beneficial role of voting with feet in public goods provision was made by Gürerk, Irlenbusch, and Rockenbach (2006) who experimentally show that a reward-and-punishment-community efficiently provides public goods in a voting with feet setting. Despite an initial reluctance to join the sanctioning institution over time almost all subjects vote with their feet for that institution while the sanction-free institution becomes depopulated. Contributions in the sanctioning institution reach almost full levels and high efficiency is attained. Cooperation is surprisingly stable and even continues when the community grows large. These results document a clear competitive advantage of the sanctioning institution. They, however, leave open important questions concerning the determinants of the success. In this paper, we report a series of five new experiments, designed to shed light on these questions.
Our first research question concerns the nature of the institutions at choice. Specifically, we ask whether voting with feet would be equally successful if subjects would not have the choice between the combined reward/punishment and a non-sanctioning institution. In the new experiments we study voting with feet between a pure punishment and a non-sanctioning community and additionally between a pure reward and a non-sanctioning community. We show that in community choice with pure punishment possibilities cooperation is significantly higher than in settings with pure reward possibilities.
Our second research question aims at disentangling the determinants of success inherent in the dynamic pattern of the punishment institution. The experimental results of Gürerk, Irlenbusch, and Rockenbach (2006) leave room for two non-exclusive explanations: the self-selection of subjects and the slow growth of the punishment community. We show that the initial endogenous self-selection of subjects is an important key for the establishment and efficient maintenance of cooperation, while slow community growth is less decisive.
Our findings highlight a so far undervalued feature of the voting with feet mechanism: In our voting with feet setting consumers do not choose between communities with different public goods but they choose between communities with different institutional rules that are intended to govern the provision of the public goods. Thus, in addition to the efficiency improvement as suggested by Tiebout (1956), the voting with feet mechanism improves efficiency by facilitating the right initial match between consumers and different institutional rules.