9511343 Hoxby This project empirically examines three topics in the economics of education. The first part of the project employs natural variation in cohort size within school districts to analyze the effects of class size on student performance in this way avoiding many criticisms of the existing literature on the productivity of school inputs. The second part of the project extends a substantial literature on the effects of school finance centralizations on average public school per-pupil spending in a state in a new direction. This project examines the effect of centralization on school productivity, or student performance and per-pupil inputs. This new line of research is timely and important because from 1972-92, 36 states centralized school finance to some degree, twelve by court order. The third and final part of the project studies how the emergence of a national college market, with schools drawing students from across the nation or within a several-state region, affects the characteristics of the college attended by a given student and the value of the student s college degree. School-based results on the effects of alass size typically suffer from an omitted variables problem: the student-teacher ratio is endogenous to unobserved student characteristics. Results based on aggregate (state-wide) data have untenable identifying restrictions and suffer from aggregation bias. The first part of this project makes use of the fact that school districts with populations under 25,000 have natural year-to-year variation in cohort size. Because teachers are relatively indivisible and grade-specific, schools cannot perfectly re- allocate resources so as to equalize class sizes between different sized cohorts. The strategy uses panel data on Connecticut school districts, and instruments for the change in class size between succeeding cohorts with the change in cohort size. The second part of the project analyzes sixteen recent (1977-87) reforms th at increase school finance centralization at the state level. For each district, Census of Governments data are used to categorize the change in the tax-price of marginal local school spending as a result of the reform. These vary from no change to extremely large increases in the case of districts that have reached maximum spending levels in their states. The project develops and tests empirically a model in which a high tax-price of local school spending decreases school productivity (student performance for inputs) because it weakens mechanisms by which less productive schools are made more productive.
