This paper presents a model where teacher effects on long-run outcomes reflect effects on both cognitive skills (measured by test-scores) and non-cognitive skills (measured by non-test-score outcomes). In administrative data, teachers have causal effects on test-scores and student absences, suspensions, grades, and on-time grade progression. Teacher effects on a weighted average of these non-test score outcomes (a proxy for non-cognitive skills) predict teacher effects on dropout, high-school completion, and college-entrance-exam taking above and beyond their effects on test scores. Accordingly, test-score effects alone fail to identify excellent teachers and may understate the importance of teachers for longer-run outcomes
↧