School value-added and longterm student outcomes

Abstract

Several recent studies find that interventions in schools can have important lasting consequences for students, and that schools differ in their contribution to students' learning. However, there is less research investigating how these differences between schools influence longer-term outcomes, especially outside the US. In this paper I study school value-added (VA) in Norwegian compulsory school, where between-school differences are smaller than in the US. I find that VA indicators are able to predict in-school performance without bias. Furthermore, VA is strongly related to long-term outcomes, and differences between schools in VA correspond to meaningful differences in long-term outcomes. For example, a one standard deviation higher VA correspond to 1.5 percent higher earnings around age 32. Three quasi-experiments using variation from student mobility and changes in neighborhood school assignments indicate that the differences captured by the VA indicators do indeed reflect differences in school quality, rather than unobserved student characteristics. Analysis of teacher grades and exam scores suggest that the former is heavily influenced by relative grading, and that the effect of exam score VA on long-term outcomes reflects the effects of skills acquired in school. In addition to shedding lights on the differences in and mechanisms of school quality, the findings help connect learning outcomes with later labor market outcomes, e.g. for cost-benefit analysis of interventions in schools.