ABSTRACT

As is well known, players involved in a prisoners' dilemma mayAs is well known, players involved in a prisoners' dilemma maysolve it through its repetition over an infinite horizon, provided theirdiscount factors meet the requirement arising from the version of thefolk theorem being adopted. We prove that if the dilemma is impure,players may indeed neglect this route to achieve the Pareto-optimaloutcome of the constituent stage game through a correlation devicewhich does not identify a correlated equilibrium strategy in the stagegame but does so in the repeated game and, in presence of a Nashbargaining solution, may indeed make time preferences immaterial.Then, we show that the same conclusion holds in continuous strategygames as well, by means of an example based upon an oligopoly modelwith rational expectations and network externalities, with and withoutproduct di¤erentiation, and either quantity or price competition. Thisexample additionally shows that firms may activate a form of collusionwhich is not only independent of their time preferences but also verydifficult to interpret from the standpoint of an antitrust authority.

KEYWORDS

Impure prisoners' dilemma; Correlation; RepeatedImpure prisoner’s dilemma; Correlation; Repeatedgames; Network externalities; Expectations.