This paper provides laboratory evidence on the efficiency-enhancing properties of the Tiebout model as a decentralized system of public goods provision. Tiebout (1956) shows that if a sufficient number of local communities exist to accommodate different types of preferences, individuals sort themselves in a way that provides an efficient allocation of public goods and taxes. Our experiment aims to disentangle the effect of voting participation and is composed of two treatments. In the non-participation treatment, local public good provision is chosen by only one subject, while the other members of the community can only stay in or move to another community. In the participation treatment, all the community members have the right to vote as well as to move to another community and collective decisions are taken by majority rule. Our findings show that social welfare is greater in the participation than in the non-participation treatment. We conclude that voting with one’s feet increases efficiency if all the community members vote and that the influence of voting participation on the allocation of local public goods should be taken into account to assess the viability of the Tiebout model.
