School choice policies — from charter schools and vouchers to open enrollment and centralized assignment systems — emerged in large part as a response to the default assignment mechanism that governs most American public schooling: the attendance zone. By tying school enrollment to residential address, attendance boundaries embed neighborhood racial and economic composition directly into school composition, concentrating disadvantage in predictable geographic patterns. Yet despite widespread recognition that attendance zones are central to how segregation is produced and reproduced, surprisingly little quantitative work had established how much of observed school segregation is actually attributable to the way boundaries are drawn, as opposed to the residential patterns that precede them. A related but distinct question concerns school siting — the decision about where to physically locate school buildings within a district — since placing schools in racially homogeneous neighborhoods can generate segregated enrollment even under a neutral assignment rule that simply sends students to the nearest school. Without decomposing these forces, it was difficult to evaluate whether reforming attendance zones, repositioning schools, or replacing the whole system with choice mechanisms offered a meaningful path toward integration.
Tomás Monarrez addresses these questions by developing a decomposition framework applied to census block data and attendance boundary maps for nearly 1,600 school districts, covering roughly 52 percent of national enrollment. His approach constructs a counterfactual set of boundaries based on minimizing the distance between students and schools — a neutral benchmark that replicates residential geography without any integrative or segregative intent — and compares segregation under this counterfactual to segregation under actual boundaries. He separately isolates the contribution of school siting by simulating what segregation would look like if school locations were randomized within the same residential environment, holding everything else constant.
The results are striking in their clarity. Residential segregation alone explains more than 100 percent of school segregation in the average district, meaning that if attendance zones simply replicated neighborhood geography in the most geographically efficient way, schools would actually be no more segregated than they are in practice. Actual attendance boundaries produce roughly 5 percent more integration than the distance-minimizing baseline, suggesting that on average local governments use their discretion to draw boundaries that are modestly more integrative than pure geography would dictate. School siting, by contrast, explains almost none of the variation in school segregation — the physical placement of schools turns out not to be a meaningful independent driver once residential patterns are accounted for. A validation exercise using Charlotte's 2002 redistricting reform — which ended decades of desegregation busing and reverted to neighborhood schools — confirms that the decomposition behaves as expected, with the boundary component of segregation jumping from negative to zero precisely at the reform. Districts with racially desegregated boundaries show an 86 percent pass-through to actual school segregation, and evidence of smaller racial gaps in educational inputs such as experienced teachers and gifted program access. The findings suggest that while attendance zone reform can reduce segregation at the margin, the primary driver of school segregation in the United States is residential sorting — a force that choice mechanisms operating within a single district are unlikely to overcome on their own.
- Journal Article: https://www.aeaweb.org/article...
- Working Paper: https://www.urban.org/research...