GROUP-SIZE AND THE VOLUNTARY PROVISION OF PUBLIC-GOODS - EXPERIMENTAL-EVIDENCE UTILIZING LARGE GROUPS

New experimental evidence extending the investigation of free-riding behavior in public goods provision is presented. Procedures are developed to deal with the logistical problems inherent in experiments involving many subjects. Data from Voluntary Contribution Mechanism experiments are reported utilizing group sizes of 4, 10, 40 and 100. These experiments provide replicable results that contradict the widely held view that a group's ability to provide the optimal level of a pure public good is inversely related to group size. On the contrary, groups of size 40 and 100 provided the public good more efficiently than groups of size 4 and 10. Several possible alternative explanations are discussed.