Simulating the Desegregation Potential of Controlled Choice in New York City

• In Event: The Gap Between Promise and Practice: Three Logics of School Choice Under Scrutiny

Abstract

Despite persistent school segregation across the United States, policymakers face a fundamental tradeoff: mandated desegregation efforts often provoke resistance from advantaged families, while unfettered school choice tends to reinforce existing inequities. While proponents tout school choice's potential to improve educational opportunity for historically underserved students, the extent to which it can meaningfully address longstanding segregation remains unclear. Controlled choice systems, where families rank school preferences and a centralized algorithm allocates seats using equity-focused guardrails, have been proposed as a means of balancing family autonomy with integration. Yet the scope of permissible approaches is constrained by Parents Involved (2007), which limits the use of individual student race in assignment decisions.

This paper examines how much racial segregation can be reduced through district choice-based school assignment. We focus on New York City Public Schools—home to nearly one million students and by some measures the most segregated urban school system in the nation—where newly elected Mayor Zohran Mamdani's identification of school desegregation as a top priority makes this study especially policy-relevant. NYC already uses a district-wide choice system but not one designed to reduce segregation. Relying on administrative kindergarten application data from 2016–2019, encompassing over 150,000 students, we simulate four distinct assignment approaches We evaluate each approach using Theil's H, a measure of how evenly racial groups are distributed across schools, and compare results against a baseline that replicates NYC's existing algorithm.

The findings reveal a stark gap between what is theoretically achievable and legally feasible. The first of our four simulations is purely theoretical, using individual student race to determine which students receive seats at oversubscribed schools, which is impermissible under current legal interpretation. We find that such an approach could reduce overall segregation by more than 15 percent while assigning over 95% of families one of their top three school choices. However, our remaining three simulations, which instead use various permissible proxies for student race, find no legally viable pathway to comparable outcomes. Among permissible alternatives, the neighborhood racial composition proxy is most effective, but even this simulation reduces segregation by only roughly five percent. Results vary meaningfully across NYC's 32 Community School Districts (CSDs), however, revealing greater potential in some local contexts. In roughly a quarter of CSDs, segregation is reduced by ten percentage points or more even relying only on legal approaches. Two factors drive this variation: districts with greater residential racial diversity tend to see larger integration gains, while districts where available proxy measures are weakly correlated with race see smaller reductions.

Taken together, these findings point to real limits on school choice as an equity tool under current legal constraints. Without the ability to consider individual student race, no choice-based system can come close to what is theoretically achievable. That said, even small reductions in segregation may carry meaningful consequences for students. The district-level findings further suggest that education and housing policy must be considered in tandem: where residential segregation is entrenched, the gains from even well-designed choice systems will remain constrained.

Author

• Sierra McCormick, Teachers College, Columbia University
  Presenting Author