[Untitled]

In this paper we use a laboratory experiment toexamine the Pivotal mechanism when applied to a binarydecision on a public project of a fixed size. Wefirst point out that the well-known incentivecompatibility of the Pivotal mechanism is true only in aweak sense; There are always strategies otherthan truth-telling that do no worse for a subject thantruth-telling, in any particular set of circumstances.This weakness of the incentive compatibility ofthe Pivotal mechanism makes it difficult for subjects tounderstand that truth-telling is the unique dominantstrategy for the mechanism, unless they havecomprehensive understanding of the payoff structure,with the result that subjects often do not play thedominant strategy. We suggest that this tendency todepart from the dominant strategy can be overcome byproviding subjects with more information about the payoffstructure. We controlled the level of informationprovision in the laboratory and verified that thestrategies used are closer to the dominant strategywhen more information is provided. Under noinformation provision conditions were outcomes poor,but under ``wide'' provision conditions, in which eachsubject experienced a variety of true personalvaluations of the project, departure from thedominant strategy was smaller in magnitude, and under``deep'' provision conditions, in which detailed payofftables were available to each subject, the rate of useof the dominant strategy increased significantly.