Approval mechanism to solve prisoner’s dilemma: comparison with Varian’s compensation mechanism

After having played a prisoner’s dilemma, players can approve or reject the other’s choice of cooperation or defection. If both players approve the other’s choice, the outcome is the result of the chosen strategies in the prisoner’s dilemma; however, if either rejects the other’s choice, the outcome is the same as if they had mutually defected from the prisoner’s dilemma. In theory, such an approval mechanism implements cooperation in backward elimination of weakly dominated strategies, although this is not the case in the subgame perfect Nash equilibrium. By contrast, the compensation mechanism proposed by Varian (Am Econ Rev 84(5):1278–1293, 1994) implements cooperation in the latter but not in the former. This result motivates the present experimental study of the two mechanisms. The approval mechanism sessions yield a cooperation rate of 90% in the first period and 93.2% across periods, while the compensation mechanism sessions yield a cooperation rate of 63.3% in the first period and 75.2% across periods. In addition, the backward elimination of weakly dominated strategies better predicts subjects’ behavior than does the subgame perfect Nash equilibrium in both mechanism sessions.