Abstract:
Saez-Marti and Weibull [4] investigate the consequences of letting some agents play a myopic best reply to the myopic best reply in Young's [8] bargaining model. This is how they introduce cleverness of players. We analyze such clever agents in general finite two-player games. We show that Young's [9] prediction is robust: adaptive learning with clever agents does select the same minimal curb set as in the absence of clever agents, if their population share is less than one. However, the long-run strategies distribution in such a curb set may vary with the share of clever agents.