UC Berkeley

This dissertation addresses three questions in the economics of education. Chapter 1 analyzes whether segregating students by gender is beneficial for students' academic achievement. Students or parents often choose peer groups by selecting school types, assuming that peers are important determinants of one's academic achievement. Among the various types of peer effects, this study addresses whether segregating students from the opposite sex is beneficial for one's academic performance by making use of the variation created by randomly assigning students to either same- or mixed-sex high schools. By using seven years of administrative data on scores in college entrance exams, I find that both male and female students benefit by being in same-sex schools. Moreover, the quantile regression analysis reveals that the effect is greater for students located at the middle quantile of the distribution of test scores. I conducted a sensitivity analysis by using a different type of test that students take, and the results are robust.

In Chapter 2, unlike estimating the effect of conventional incentive mechanisms in which good schools are rewarded and bad schools are punished, I estimate the impact of "rewarding" poor-performing schools on students' academic achievement. Because of the simple discontinuous eligibility that determines the provision of categorical school funding to underachieving schools, I use regression discontinuity designs to causally estimate the treatment effect. The results of the analysis reveal that students' academic performance in poor-performing schools improved significantly (7 to 10 percentile points) after the treatment. Moreover, the ratio of underachieving students decreased in schools that received funding (5 to 10 percentage points), relative to those that did not receive funding.

Finally, in Chapter 3, I explore whether grouping students by ability benefits students. Local education agencies often engage in educational reforms with limited resources aimed at improving the academic achievement of students. One of the low cost methods that the agency frequently employs is the use of ability tracking. In this chapter, by making use of the randomized social experiment conducted in Seoul, I provide causal estimates of the effect of ability tracking on students' achievement, using administrative data on students' test scores. Based on the results, I find that, on average, tracking promotes achievement of not only high-achieving students, but also of low-achieving students. Moreover, the magnitude of the treatment effect is similar across various quantiles of the distribution of students' performance. Therefore, contrary to the view that tracking may be detrimental to the learning of low-achieving students, tracking may not worsen inequality in students' achievement.