Effects of Class-Size Reduction on Cognitive and Non-Cognitive Skills

Abstract

We estimate the effects of class-size reduction by exploiting exogenous variation caused by Maimonides’ rule, which requires that the maximum class size is 40 students and that classes be split into two when 41 students are enrolled. Our data cover all fourth to ninth graders in 1,064 public schools in an anonymous prefecture of Japan for three years. We find that the effects of class-size reduction on academic test scores are statistically and/or economically insignificant when school fixed effects are controlled. We find no evidence that small class size improves non-cognitive skills.