Reza Moradi

This paper studies the distributional consequences of school consolidation when families differ in access to effective alternatives. I develop and estimate a structural model of school choice and educational production using Puerto Rico's 2013–17 public-school closures. Linked student–teacher records and school enrollment choices allow me to separately estimate school and teacher contributions to achievement and families' responses to distance and the removal of public-school options. The results reveal substantial heterogeneity in the consequences of closure. The nearest surviving school stood about 3.4 km from rural children's closed schools, against about 2 km for urban children. Rural-poor children's change in school effects from the school closed to the school entered was about 0.08 test-score standard deviations less favorable than that of all other displaced children, mainly reflecting the schools closed. Their model-implied access loss per displaced child is 84 percent larger than urban non-poor children's, measured in km-equivalents; this conversion reflects rural families' lower estimated costs per km. At unchanged closure counts, proximity-based lists lower rural-poor distances to surviving schools, but their school-effect changes are not statistically distinguishable from the realized closures'. Keeping open 47 closures farthest from a surviving alternative avoids about 30 percent of rural-poor modeled access loss. The comparisons measure access and school contributions, not causal effects of closure on children's own achievement. These findings show why enrollment alone provides an incomplete basis for evaluating consolidation: distributional comparisons depend on both the effectiveness and accessibility of the alternatives available to displaced students.