| The first chapter, joint with Salvador Navarro, develops a semiparametric test of misspecification of agent's information for games of incomplete information. The test uses the intuition that, for any proposed information structure, the opponent's choices should not predict a player's choice, conditional on the proposed information available to the player. In particular, the test is designed to check against some of the null hypotheses commonly used in the literature. We show that our tests have satisfactory power against common alternatives, even in small samples. In a simple empirical application using entry data in the US airline industry, we find that the assumption of independent private shocks is not supported by the data.;The second chapter analyzes the effect of competition and demand decline (due to TV diffusion) on the exit process of movie theaters in the US from 1950--1965. I explain background information on the industry. I argue that the industry in this time period resembles a war of attrition (a dynamic game of exit), in that the benefits of staying in the market were declining in an arguably exogenous way and that competition was local. Then, I provide data analyses. In particular, using panel data and an IV regression, I show that theaters' exits were both positively correlated with TV diffusion in the local market and with competition.;The third chapter develops and estimates a dynamic game of exit using data on movie theaters in the US from 1950--1965. I modify Fudenberg and Tirole (1986)'s model of exit in a duopoly with incomplete information to an oligopoly. I estimate this model with panel data, using variations in TV diffusion across households and other market characteristics to identify the parameters in the theater's payoff function and the distribution of unobservable exit values. Using the estimated model, I show that theaters who are making negative profits choose to remain in the market, in hopes that they outlast their competitors, because at that point their profits would increase. This creates a significant delay in the exit process. |
