Learning, Experimentation and Equilibrium Selection in Games

OF THE DISSERTATION LEARNING, EXPERIMENTATION, AND EQUILIBRIUM SELECTION IN GAMES by Youngse Kim Doctor of Philosophy in Economics University of California, Los Angeles, 1992 Professor David K. Levine, Chair This dissertation addresses and resolves problem of selection among multiple equilibria in games, by perturbing the original system and then characterizing the outcome of resulting perturbed system. Chapter 2 examines reputational sequential equilibrium of what we call participation games, that have many economic applicability, such as entry deterrence and product quality control. By perturbing the original game with types, we show that the lower bound of the single long run player’s payoff is almost his Stackelberg commitment payoff in the limit as the finite horizon grows. Discontinuity exists between the infinite horizon and the limiting finite horizon solution. Our result is robust to a model modification in which the long run player announces the payoff structure just before the whole game begins so that the rational type of long run player has to mimic not only the strategy but also the initial payoff announcement of the Stackelberg commitment type. Chapter 3 and chapter 4 analyze a game played by randomly and anonymously matched players from a large population. The game of interest is a multiperson