The Groves-Ledyard mechanism: An experimental study of institutional design

Abstract

The Groves-Ledyard mechanism theoretically can solve the ‘free-rider’ problem in public good provision in certain environments. Two questions are of overriding importance in implementing the mechanism. The first is related to the actual performance of the mechanism in general. The second is the choice of a ‘punishment parameter’, γ, which is the only parameter that is available for those who may want to actually use the mechanism. Thus the determination of the role of this variable on mechanism performance is fundamental for any advances along the lines of actual implementation. In studying the Groves-Ledyard mechanism, we show that the punishment parameter, γ, plays a crucial role in the performance of the mechanism. By using γ = 1 and 100, we show that under the higher punishment parameter, the Groves-Ledyard equilibrium is chosen much more frequently; a higher level of the public good is provided and efficiency is higher. By examining two behavioral models, we show that a higher γ leads to an increase in the probability of an individual choosing a best response predicted by the model. The parameter γ alone explains nearly 70% of the data in both the Cournot and the Carlson-Auster behavioral model. We also found that convergence to Cournot behavior is faster and more stable under a high γ than under a low γ.