Date of Award

Spring 2021

Document Type


Degree Name

Doctor of Philosophy (PhD)



First Advisor

Berry, Steven


This dissertation includes three chapters on industrial organization and the economics of education, centered around New York City's public high school choice procedure. In the first chapter, I document evidence of informational frictions in the usage of the public high school choice in New York City and the patterns of racial disparities. The second chapter develops a model of application behavior and considers its identification. The third chapter estimates a model of high school applications and uses it to analyze the impacts of New York City's high school choice procedure. Centralized school choice procedures are gaining wider adoption, based in part on theoretical results promising desirable properties such as matching stability and strategy-proofness. However, these theoretical results are limited in two dimensions. First, they do not directly address distributional outcomes, such as racial integration and equity, that parents and policymakers are often interested in. Second, although they have been argued to indirectly assist in achieving distributional goals, they rely on idealized assumptions regarding the matching mechanism and information. In the first chapter, I use the administrative data provided by the NYC DOE for the 2016–2017 academic year to provide evidence that suggests the applicants in New York City are not aware of all the available schools. The evidence also suggests that applicants take the admission probabilities into account when they apply and that they may not understand certain properties of the mechanism. I then document evidence of racial disparities with respect to the schools that the students live close to, apply to, and are matched to. In the second chapter, I develop a model of application behavior that allows applicants to consider only a limited set of school options and to have incorrect beliefs about admission chances. Rich information in applicants' rank-ordered lists of schools, coupled with certain restrictions, allows identification of the model. For instance, while a lack of consideration may affect which schools are listed, it cannot affect how the listed schools are ranked. Furthermore, while strategic behavior to shift admission chances may affect the rankings, it cannot affect which schools are listed unless the list length constraint binds. Identification is also aided by an assumption that certain observables, such as the page on which a school is listed in the school directory, can affect the consideration set but not preferences. I formalize these intuitive ideas with sufficient conditions for nonparametric identification. In the third chapter, I estimate a model of students’ application behavior, which is a parametric version of the model considered in the second chapter. I also analyze the impact of New York City's 2016-2017 school choice on integration and equity of welfare across different demographic groups. I simulate the estimated model under different counterfactual scenarios to measure the contributions of different factors: students’ residential locations, preferences, limited consideration sets, admission priority groups, and the screening policies of the schools. The model also allows me to quantify matching stability. Results show that, while school choice slightly integrates race and improves welfare across all races, these gains and the stability of the school assignments are compromised by deviations from fully informed behavior. Schools’ admission priorities and screening policies contribute to racial segregation and tend to place Asian and White students in their preferred schools.