A Theory of Agreements in Infinitely Repeated Games

In environments where players interact repeatedly, it is natural to assume that players can communicate and hence coordinate their actions. This paper offers a theory of self-enforcing agreements in infinitely repeated games as an alternative to the notion of renegotiation-proofness. The notion of stable (selfenforcing) agreements is defined and its central (distinct) feature is that in contemplating a deviation, a coalition does not, ex ante, count on renegotiating with nonmembers, even though renegotiation might occur ex post ; a coalition deviates only if by solely coordinating the actions of its own members, it can guarantee a higher payoff for each member; moreover, our theory allows every coalition to deviate. ∗ I am grateful to Professor Joseph Greenberg for many valuable suggestions, and to Professor Geir B. Asheim for helpful comments. Financial support from Social Sciences and Humanities Research Council of Canada (SSHRC) is gratefully acknowledged.