Neighborhood sorting and the value of public school quality

This paper develops a method to estimate the parental value of public school quality with two novel features. First, it estimates the value of public school quality in the same unit in which public schools' costs are measured and private school tuition is charged: per year, per child at each grade level. Second, it develops a novel approach to control for unobservables correlated to school quality, including those generated by sorting. People without school-age children enjoy neighborhood-level amenities but do not enjoy school-level amenities, so data about their residential choice can be used to control for neighborhood unobservables, isolating the value of school quality per se. I embed this idea into a dynamic model of neighborhood choice, building on previously unconnected literatures. Using the 2000 U.S. Census data, I find that parents tend to value school quality more in elementary and high school grades relative to middle school grades. However, improving public school quality currently costs more than is worth to parents even at the most valued grades, so externalities in education are necessary to justify such investments. These findings highlight the importance of improving the efficiency with which school resources are spent.