Cooperation, Sanctions, and Strategy in On-Shot and Finitely Repeated Prisoners' Dilemmas: Experiments with the Voluntary Contribution Mechanism with Punishment

Brown University, Providence RI

Investigators

Abstract

The prisoners' dilemma, the most studied game in economics and the social sciences, captures key features of the problem of cooperation and collective action. Early work in economics used the solution concept of eliminating dominated strategies to make clear-cut predictions of no cooperation by self-interested, rational agents in finitely repeated-prisoners' dilemmas. But theory and experiments with the voluntary contributions mechanism showed higher levels of cooperation and provided explanations for it. The explanations focused on the presence of non-standard preferences, and strategic incentives to mimic such preferences by signaling. Current research has begun to document the presence of heterogeneous agents with non-standard preferences. This project further develop this line of research by using experimental methods to study agent heterogeneity in order to: 1. Distinguish between departures from standard predictions attributable to nonstandard preferences and departures attributable to strategic incentives to signaling, and 2. Learn how institutional environments can be modified to increase efficiency in collective action. Specifically, a set of experiments are conducted using the general framework of the voluntary contribution mechanism (VCM) with the option to sanction individual group members by reducing their earnings at a cost to oneself after contribution decisions have been taken and announced. In the VCM, subjects are assigned to a group, given an endowment of funds, and asked to divide the endowment between a group project and an individual account. Combined earnings are highest when all contribute their full endowment to the group project, but there is an incentive to "free ride" by contributing less and sharing the proceeds from the contributions of their group members. The option to punish non-contributors helps to produce more optimal results. The analysis focuses both upon the effects of varying the treatment conditions, for instance number of repetitions, group size, and punishment cost, and upon differences in the outcomes of different groups facing the same treatment. By identifying differences in the behaviors of individuals which may derive from differences in their preferences or types, the researchers study how differences in the types of individuals composing different groups can account for different group outcomes.