↵Notes: Estimates for the baseline sample of students who graduated between 1978 and 2001 from a combined school that does not take up students from other primary schools. The schooling outcome is defined as years of educational attainment and measured when the individuals are 27–42 years old. Column 1 shows the results from a specification that uses a flexible function that is linear in enrollment, while Column 2 uses a quadratic function of enrollment. Column 3 and 4 use linear and quadratic functions of enrollment that are interacted with the segment around the maximum-class-size thresholds, while Columns 5 and 6 interact the segment-specific linear and quadratic functions with the class-size thresholds. Column 5 is our baseline specification that is used throughout the paper. Column 7 shows the estimates of the baseline specification without excluding observations close to the thresholds (donut strategy). Columns 8 uses the baseline specification but excludes enrollment observations further than five pupils from the threshold, while Column 9 excludes enrollment observations further than ten pupils from the threshold. All regressions include year of graduation fixed effects and segment dummies and interact the flexible primary school enrollment function with the three maximum-class-size regimes. Except for Column 10 all regressions include individual background controls for gender, parental education, immigrant background, month of birth, and age at graduation. Standard errors clustered at the school level are reported in parentheses. *p < 0.10, **p < 0.05.