Essays in Economics of Education

Abstract

This thesis is composed of three essays on the economics of education. The first essay is about the heterogeneity of gains from selective school admission. The question of who benefits from selective school enrollment remains controversial. I show that Boston exam schools have heterogeneous effects on achievement. Impact differences are driven primarily by the quality of an applicant's non-exam-school alternative rather than by student demographic characteristics like race. Admission policies prioritizing students with the weakest schooling alternatives have the potential to increase the impact of exam schools on academic achievement. In particular, simulations of alternative admissions criteria suggests schemes that reserve seats for students with lower-quality neighborhood schools are likely to yield the largest gains.

The second essay is about understanding the impact of selective school admission screens on segregation in New York City schools. 70 years after \textit{Brown v. Board of Education}, US school districts are still economically and racially segregated. School segregation is especially apparent in NYC, the largest US school district. I analyze the impact of two integration plans which reduced the role of screens in admission in two local NYC school districts. I show that abolishing selective admissions reduced both economic and racial segregation. Amending selective admission criteria also elicits substantial behavioral response from applicants. I find evidence that reducing the role of admission screens leads to White and high-income enrollment losses, which decreases the effect of the plans. On the other hand, applicants' changes in application behavior in response to the reforms increased the plans' impact on segregation.

The final essay is about predicting the effect of changes in school admission on students' enrollment. Such predictions are based on estimated student preferences, which in turn are obtained from the ranked order lists they submit. A concern is that an applicant with fixed preferences might submit different lists when faced with different admission criteria. For instance, an applicant could strategically take into account their probability of admission at each school, therefore violating the truthfulness assumption. A solution is to estimate preferences allowing students to strategically choose over all possible lists, but this runs into the curse of dimensionality as the choice space is large. This paper provides a model of applicants' list formation which presumes applicants use a simple heuristic in selecting their lists. In the model, applicants fill their list sequentially, without fully internalizing the dynamic consequences of each choice. Using this simplification, I estimate applicants’ preferences, circumventing the dimensionality problem. I leverage an admission reform in NYC to estimate the model. Allowing applicants to deviate from truthfulness affects substantially their estimated preferences.