Do Students Learn More When Studying with Similar Students? Evidence from Korean High Schools

This paper examines the effects of high school de-tracking on Korean SAT scores. After accounting for the sample selection bias using an event study approach with matching, we identify three distinct channels of de-tracking effects: classroom effects from peer quality and teaching effectiveness, campus effects from school facilities and resources, and supplement effects from parental inputs. We find that de-tracking reduces the test scores for each subject by 0.05-0.09 of one standard deviation, and that classroom effects are the most important. The negative classroom effect is primarily driven by teaching effectiveness (i.e., teacher-student match quality), and the effect heterogeneity comes mostly from peer quality changes. We also find that the negative effect of de-tracking is the strongest for students at the top percentiles of the score distribution and the part of classroom effects is mitigated by supplementary effects for the parents of students at the top percentiles of the score distribution.