The first chapter of this dissertation examines the incidence of tax credits for hybrid vehicles in the United States. A variety of state and federal tax incentives have been used to subsidize gas-electric hybrid vehicles. I estimate the incidence of these tax incentives using microdata on sales of the Toyota Prius. In order to account for heterogeneity in strategic timing of sales around tax changes, I develop a partial identification methodology which provides sharp bounds on the effect of tax changes. I find that consumers capture nearly all of the benefits. This finding is difficult to explain in a standard rational choice model. I propose an alternative framework based on the possibility that current prices shift future demand in a particular way, which rationalizes the empirical findings.The second chapter, written with Rebecca Blank and Kerwin Charles, investigates the response of young people in the United States to state laws dictating the minimum age at which individuals could marry. Our analysis focuses on how individuals avoided binding laws, which is revealed by broad discrepancies between marriages reported in survey and administrative data. We provide evidence that individuals both traveled across state boundaries and lied about their age in order to avoid binding laws. Our results have important implications about the quality of administrative data when it is poorly monitored; about the effect of laws when agents can avoid them; and about the validity of estimates using cross-state variation in laws as an instrumental variable.The third chapter, written with Alexandra Resch and Paul Courant, studies how a social planner should allocate students and resources throughout a higher education system. We posit a distribution of student ability and a limited pool of resources and demonstrate the optimal rules for establishing, populating and funding universities. If student ability and school resources are complements, and if there is a fixed cost to establishing a school, then the optimal allocation will involve a tiered system of higher education that sorts students by ability. In contrast to previous research, we show this tiered system is optimal even in the absence of peer effects.