Cheap-talk and cooperation in a society (Q1176679): Difference between revisions

Traditional noncooperative game theory is much oriented toward the study of strategic stability. This approach has failed in demonstrating the cooperative effects of costless communication (often called ``cheap talk''). The author adopts a different point of view and investigates social stability. He models society as infinitely many anonymous individuals, who may be of two types. He analyzes the following repeated situation: at every stage, two individuals of different type are randomly matched and simultaneously choose one of two possible actions. More precisely, they play a two by two noncooperative game with common interest (i.e. with a unique Pareto optimal outcome). The author uses a set-valued solution concept called `` cyclically stable set'', which he has developed in a more general context, with \textit{I. Gilboa} [Econometrica 59, No. 3, 859-867 (1991; Zbl 0728.90099)]. In the present paper, the author proves that if cheap talk is allowed before every stage, the Pareto optimal outcome is the only cyclically stable set. The statement does not necessarily hold in games without cheap-talk. The result can be extended to two-person pure coordination games with cheap-talk where the players have finitely many actions.