Co-operation in Infinitely Repeated Games: Extending Theory and Experimental Evidence

preliminary, comments welcome A proper understanding of the determinants of co-operation is crucial for the social sciences. In this paper we propose a novel theoretical tool that enhances this understanding. Accordingly we present experimental evidence designed to differentiate clearly among the prevailing theories based mostly on repeated games non-deviation constraints and our theoretical tool. Our theory intuitively takes into account the often neglected "sucker’s payoff " in infinitely repeated prisoner’s dilemma games. This yields a new cutoff level for the discount factor, δ∗, compared to the more traditional cutoff level δ below which co-operation is not supportable as an equilibrium. We simulate infinitely repeated prisoner’s dilemma games in the laboratory with random continuation and matching rules and many experimental subjects, and find that changes in the frequency of co-operation are predicted much better by our novel indicator δ ∗ as by δ. We finally estimate experimental subjects ’ risk aversion and demonstrate that our results, which square well with evidence in the literature, do not change when taking into account risk attitudes