Comparing Public-Good Provision Mechanisms in Large Groups: Theory and Experiment

This paper investigates threshold-public-goods provision with budget-balancing condition. The condition is achieved through rebate policies. In addition to the proportional rebate (PR) and winner-take-all (WTA) rebate policies from the literature, we introduce two novel rules: the uniform price cap (UPC) and the uniform price auction (UPA). Both feature endogenously determined prices and discriminatory payment rules. We conduct two large-group experiments to examine the efficiency property of these rebate policies. Our results show that, over-contribution exists and efficient provision rates are higher under the UPC and UPA mechanisms, while over-contribution does not appear under the PR or WTA rebate rules. The latter result differs from the findings of Ronbeau et.al.(1999) and Spencer et.al.(2009), in which the provision threshold is unknown to subjects. Our belief elicitation results show that, subjects expect the provision threshold to be much higher than the actual, and their contributions are significantly higher than the optimal levels calculated from reported beliefs of others' contributions. Moreover, group contribution declines once subjects are informed of the exact provision threshold and when decisions are repeated. It suggests that the over-contribution phenomenon from the literature might have arisen from the uncertainty in the information environment.