Job Market Paper
Racial segregation is an enduring feature of U.S. K-12 education. Up to half of it originates within schools and is due to how classrooms are formed. This paper develops an empirical framework to understand the implications of discretionary classroom formation in competitive education markets. I leverage a school competition reform to document via an event-study that in anticipation of the competitive shock, public schools change students’ assignments to classrooms along the racial dimension. On average, this response increases classroom segregation. I then estimate an empirical model of school choice and competition to understand whether schools choose their level of classroom segregation so as to differentiate horizontally, thereby relaxing vertical competition on costly academic quality. The novelty of the model is that it embeds classroom segregation both on the demand side, as a dimension that parents value, and on the supply side, as a margin of differentiation that schools choose directly alongside academic quality. I estimate preferences for classroom segregation so as to rationalize the reduced-form effects of competition identified through the event-study. I estimate substantial heterogeneity in household preferences for classroom segregation, both across and within racial groups. On the supply side, schools face incentives to enroll a larger share of white students. I use the model to evaluate a policy that requires schools to form racially integrated classrooms, given the composition of the student body. I find that the policy raises aggregate academic quality and the average test score in equilibrium. Magnitude-wise, present value lifetime earnings increase by up to $1,620 per student. Because the schools that increase academic quality the most are located in non-white areas, the learning gains accrue mostly to non-white students, and the racial test score gap decreases by 2%.
R&R AEJ: Economic Policy
Co-winner of 2021 NYU Best Third Year Paper Award
I study the effect of charter openings on racial segregation across classrooms at traditional public schools. Exploiting almost 100 charter entries in North Carolina from 1997 to 2015, I compare segregation across classrooms in nearby public schools to those further away. I find that the announcement of a charter opening increases classroom segregation significantly. Charter entry also raises ability tracking and the fraction of white students classified as gifted. Overall, test score inequality increases upon charter entry, driven by a reduction in the performance of low-achieving students.
Joint with M. Daniele Paserman and Liang Zhong
In many contexts, a decision-maker must screen applicants using only imperfect information about their quality. The decision-maker may use information about the applicant’s race or gender to guide their decision, resulting in discrimination. Using data on the group differences in the output of applicants that pass the screening process, is assessing the extent and nature of discrimination possible? In this paper, we tackle this question in the context of the motion picture industry. Using human raters supplemented by a machine learning algorithm, we construct a new dataset with racial identifiers of the cast of more than 7,000 motion pictures released in the US between 1997 and 2017. We use this dataset to test the predictions of a model of discrimination. Empirically, we document the following facts: (a) Average box-office revenue of non-white movies is substantially higher than white movies; (b) this difference is driven primarily by the left tail of the distribution, suggesting non-white movies with low box-office potential are never produced; and (c) relative to white movies, non-white movies substantially overperform relative to expectations. The pattern of results cannot be rationalized by customer discrimination or by a model of statistical discrimination where the signal embodied in non-white movies is less precise. Instead, the findings are consistent with our theoretical model if producers hold non-white movies to a higher standard or if they systematically underestimate the revenue potential of non-white movies.
Joint with Brendan Yap
We use novel high frequency employment data to study the effect of federal transfers toward states on states’ employment over the COVID-19 pandemic. We find that $150 of state aid per person decrease employment by 23.4 jobs per 1000 people. The drop is concentrated in the private sector and driven by non-Democratic states. Our findings are consistent with large-aid states deterring job search, especially among the longstanding unemployed, as well as vacancy creation.
Peer Effects in College Applications
We study classroom peer effects in college applications. We use novel, rich student-level administrative data from a large school district in Minnesota alongside an instrumental variable approach to deal with endogenous classroom formation. We find that students whose peers apply to better colleges apply themselves to better colleges. The effect is statistically and economically significant. Peer influence is strongest among high-achieving students and operates within achievement groups.
The Effect of Job Loss on Pharmaceutical Prescriptions Joint with Lorenzo Rocco, Lorenzo Simonato and Laura Cestari, Social Science and Medicine 217: 73-83.
Work in Progress
Skimming Through Advertising: Evidence from Charter Schools
How Do Schools Allocate Spending? Competitive Incentives and Implications for Value-Added
The Determinants and Consequences of School District Secessions and Consolidations
Tax Incentives and Voters’ Support for Spending on Education Joint with Andrea Menini
In: Axel Börsch-Supan, Johanna Bristle, Karen Andersen-Ranberg, Agar Brugiavini, Florence Jusot, Howard Litwin, and Guglielmo Weber (Eds.), Health and socio-economic status over the life course: First results from SHARE Waves 6 and 7. 2019. Berlin, Boston: De Gruyter.