A game theoretic approach to economic policy analysis

In the first chapter, Legislative Bargaining and Coalition Formation, I analyze a widely used model of endogenous policy determination. The main question is if the stationary equilibrium, a popular equilibrium selection in applied work, can be rationalized as approximating the unique backwards induction equilibrium with a long finite horizon. The answer is negative: I show that there is a continuum subgame perfect equilibrium outcomes in the finite horizon model. With sufficiently patient players a "folk theorem" applies. In contrast, I obtain generic uniqueness of equilibria in a generalized model where differences in time preferences are allowed. However, a unique equilibrium is always non-stationary and even if differences in discount rates are arbitrarily small, the non-stationarity is non-negligible. In the second chapter, Affirmative Action in a Competitive Economy (with Andrea Moro), the objective is to investigate how affirmative action affects incentives for human capital acquisition in a model of statistical discrimination with endogenous human capital. Affirmative action may "fail" in the sense that there may still be discrimination in equilibrium. However, the incentives to invest for the disadvantaged group are better in any equilibrium under affirmative action than in the most discriminatory equilibrium without the policy. Welfare effects are ambiguous: the policy may even hurt the intended beneficiaries, also when the initial equilibrium is the most discriminatory equilibrium. The third chapter, Statistical Discrimination and Efficiency, investigates if there is an efficiency rationale for policies aimed at statistical discrimination. In the same framework as the second chapter I show that there is always an efficiency rationale for intervention. However, the inefficiencies arise because of a "free-riding" problem, which occurs independently of whether there is discrimination or not. A planner may want to discriminate between two identical groups of workers. The reason is that discrimination makes it possible for certain workers to specialize as qualified workers, which reduces the problem to match workers with jobs efficiently. This positive effect must be weighted against inefficiencies in investment behavior created by discrimination, but examples can be constructed where the surplus maximizing allocation involves discrimination.