Using data from both the National Longitudinal Survey of Youth (NLSY) and High School and Beyond (HSB), we investigate if public high schools differ in the production of earnings and if rates of return to future education vary with public high school attended. Given evidence of such variation, we seek to explain why schools differ by proposing that standard measures of school quality as well as proxies for community characteristics can explain the observed parameter variation across high schools. Since analysis of widely-used data sets such as the NLSY and HSB necessarily involves observing only a few students per high school, we employ an exact finite sample estimation approach. We find evidence that schools differ and that most proxies for high school quality play modest roles in explaining the variation in outcomes across public high schools. We do find evidence that the education of the teachers in the high school as well as the average family income associated with students in the school play a small part in explaining variation at the school-level.