Do Teachers Boost Test Scores in Public High Schools in Delhi?

We measure teacher effects by their ability to raise test scores in high‐stakes examinations. We present estimates for value‐added teacher effects from public high schools , while prior work in developing countries has focused on primary and middle schools or private high schools. Our data come from 22 schools and represent the cohort of public‐school students in Delhi who took their end of grade 12 examinations in 2016. We estimate teacher effects using within‐student and across‐subject variation in test scores, and additionally account for contemporaneous student inputs. We find substantial variation in teacher quality that contradicts the perception that most public‐school teachers are ineffective: over the course of two years, namely grades 11 and 12, being assigned a one standard deviation better than average teacher from the same school moves a student at the median to the 63rd percentile of their cohort's test score distribution. Furthermore, being female is the only teacher attribute that positively predicts teacher effectiveness; training, experience, and personality traits are uncorrelated with teacher effectiveness. Lastly, we find low correlation between value‐added teacher effects and principals' ratings of teachers implying that principals' ratings cannot function as a proxy for value‐added estimates when deciding teacher compensation.