We revisit a Swedish comprehensive school reform first evaluated by Meghir and Palme (2005). This reform increased years of schooling and abolished tracking. We extend the original analysis to the full population and introduce an improved education measure. Our results confirm the original overall finding of small average earnings effects. However, we find considerably larger increases in educational attainment and no evidence of decreased labor earnings for students with high-educated fathers. Our analysis provides two new important insights: First, we find no evidence that de-tracking had differential earnings effects across socioeconomic groups. Second, previous instrumental variable (IV) estimates using similar administrative education data are substantially upward biased.
