Association for Public Policy Analysis & Management

Many school districts in the United States have transitioned from a neighborhood assignment system where students are assigned a seat in their neighborhood school, to a school choice assignment mechanism. School choice mechanisms typically involve parents submitting a list of preferred choices to school districts who assign seats trying to guarantee that as many students get their first choices while making sure that parents have incentives to report their true preferences. By implementing school choice, school districts aim to reduce school segregation and create new opportunities for students outside of their neighborhood schools.

Using data from Boston Public Schools pre-kindergarten assignments for the years 2010 to 2015, I document that despite BPS’s choice system is built to guarantee equal access to quality, African-American students are less likely to get assigned to their first-choice school than White and Hispanic students are. This paper tests two potential non-exclusive mechanisms behind this fact. Both mechanisms share the same underlying rationale: that African-American students face more competition for seats in the school choice system, and in consequence, a higher fraction of them end up assigned to lower ranked options. This excess competition can be a product of (i) a reduced supply of seats associated to the residential location of African-American students, and (ii) a reduced supply of seats with the characteristics that African-American students demand.

Preliminary evidence suggests that differences in supply associated with the residential location may account for around 1/3 of the effect. To quantify the effect of racial specific preferences in explaining the fact, I propose a novel strategy to estimate preferences using a discrete choice model augmented with the structure of the underlying assignment algorithm. Using the model, I will derive a set of moment restrictions that I will estimate via GMM to obtain racial-specific preference parameters and each school admission threshold.

I will use racial-specific preference parameters to estimate counterfactual assignments to identify the contribution of preferences in explaining the gap in first-choice assignments. Also, counterfactual exercises will be used to estimate the impact of different policy changes on the resulting assignments and welfare of different subgroups of students.