Determinants of teacher value-added in public primary schools : evidence from administrative panel data

This study estimates teacher value-added (TVA) for language arts and mathematics test scores of students in public primary schools to investigate the empirical relationship between testscore TVA and observable traits and promotions of teachers. Our empirical strategy employs Chetty, Friedman, and Rockoff (2014a) with school-year fixed effects as an additional control for potential sorting of students across schools. Using unique administrative panel data of students in public primary schools of a large municipality of Japan, we find TVA distribution to have variance comparable to ones observed in the U.S. schools. Using TVA estimates, we examine their associations with gender, teaching experience, age, and promotions of teachers. We find that these observable characteristics of teachers are statistically significantly associated with TVA estimates. Additionally, we find that TVA estimates are positively associated with teacher promotions.