Abstract:
While test-score value-added models are increasingly popular measures of school quality, concerns about their limitations remain. Schools may reallocate resources toward tested subjects or spend excess time on exam preparation, and standardized exams may not accurately measure content knowledge. I create a theoretical framework outlining how schools may allocate limited resources under high-stakes accountability, and follow it by examining the empirical relationship between schools' test-score value added and the grades their graduates earn at public colleges in Michigan. In preliminary results, I find that schools with high value added in reading contribute to better grades in college, both in English courses and in other subjects, while math value added has only a marginally-significant relationship with college math grades and no relationship with any other outcomes. The effects of attending a school with high reading value added are particularly large for black students and economically disadvantaged students.