Equilibrium selection in signaling games with teams: Forward induction or faster adaptive learning?

Abstract

Teams are shown to violate the most basic of equilibrium refinements in signaling games: single-round deletion of dominated strategies (part of the Cho–Kreps intuitive criteria). This is important because, to the extent that teams can be easily induced to violate the most basic of equilibrium refinements even under a “best case” scenario (teams that rapidly develop strategic play in games of this sort), it implies that one must rely on learning models, and past empirical research with these models, when predicting equilibrium outcomes.