Cooperative Networks: Theory and Experimental Evidence

We consider a modi¯ed pure public good game characterized by a pre-play negotiation stage, onwhich pairs of players can form binding cooperation commitments. As the introduced mecha-nism only supports pairwise rather than more inclusive commitments, it does not implement thee±cient outcome. We theoretically derive the incentive compatible and e±cient cooperative net-works and evaluate the behavioral e±cacy of the suggested mechanism to promote and stabilizecooperation. We present the results of two experiments. The ¯rst implemented environmentfollows the standard procedures for the voluntary contributions mechanism and establishes thatneither repetition with an unknown end nor voluntary costly monitoring are behaviorally suf-¯cient to induce cooperative outcomes. In the second experiment we introduce the pairwisecommitment mechanism and we observe aggregate cooperation rates beyond those observedin the ¯rst experiment whatever the monitoring cost but also beyond the rates supported bythe formation of incentive compatible networks when individual monitoring is relatively cheap.Groups di®er largely in their ability to make use of the pairwise commitment mechanism: whilesome groups converge to full cooperation by managing to coordinate on the formation of e±cientnetworks over time, both networks and cooperation rates unravel in other groups. An extendedversion of our theoretical setting with inequity averse players in the form suggested by Fehr andSchmidt (1999) captures the stylized facts of both experiments.