Abstract:

We study the effect of single-sex school on student academic performance using 10 year panel data on 202 randomized Korean high schools in Seoul, Korea. The sample includes all randomized schools during 2000-2010 in Seoul.

Students and teachers are randomly assigned to one of two types of school: single-sex school or co-ed school. Some important features of random assignment are worth mentioning. First, random assignment applies to general high schools, not special-purpose high schools such as science, foreign language, art etc. Middle school graduates who want to attend special-purpose high schools are admitted based on an entrance exam or other screening processes which are not random. Second, the Committee on Assignment of High School Entrance is established every year to guarantee the assignment process to be random. In particular, it makes sure that each high school's average score is equalized within 11 districts of Seoul. Last, a student can be allocated a high school within a district where her residence is located. Out of the 202 randomized high schools in Seoul, there are 59 female schools, 69 male schools and 74 co-ed schools in 2010.

We find evidence that the average Korean Scholastic Ability Test (hereafter KSAT) scores for single-sex schools are higher than those of co-ed schools for both all-boy and all-girl schools. Furthermore, the effect is much more significant for all-boy schools than all-girl schools. It is also revealed that the positive effects of single-sex school is much bigger for lower quantile of KSAT score. In particular, when it comes to girls, the effect becomes statistically insignificant for KSAT score higher than in 60 percentile and above.

Keywords:

Single-sex school, Education production function, Random experiment, Quantile regression.