Title: A New Method of Estimating Teacher Value-Added

Paper link: GGM_TVA

Abstract: This paper proposes a new methodology for estimating teacher value-added. Rather than imposing a normality assumption on unobserved teacher quality (as in the standard empirical Bayes approach), our nonparametric estimator permits the underlying distribution to be estimated directly and in a computationally feasible way. We implement the approach using two separate large-scale administrative data sets, finding that our estimated teacher value-added distributions depart from normality and differ from each other. We then draw out the implications of our method for policies that are based on teacher value-added. First, considering a widely discussed policy that releases teachers in the bottom five percent of the value-added distribution, we compare predicted test score gains using our approach with those using parametric empirical Bayes. The parametric method predicts similar policy gains in one data set while overestimating the gains in the other by around 30 percent. More generally, we simulate the aggregate test score effects of policies that release any given percentage of teachers from the bottom of the value- added distributions under the two approaches, and also policies that reassign teachers at the top of the distribution. The results highlight the benefit of using a nonparametric approach, given that the underlying (unobserved) distribution of value-added is likely to be context-specific.