Elite schools, magnet classes, and academic performances : Regression-discontinuity evidence from China

Employing a unique data from a county in rural China, we use the regression-discontinuity design to study the causal effects of elite school and magnet class enrollment. Our data contains two admissive processes, allowing us to separately examine elite school and magnet class effects on two groups of students with different abilities. Results show that enrollment in elite schools has small and insignificant effect on borderline student scores in the college entrance examination, whereas studying at a magnet class can significantly boast borderline student CEE score by 0.435 standard deviations. We provide suggestive evidence that teacher effect is roughly 40% of the magnitude of the peer effects associated with the score gain in magnet classes. We also find magnet class experience can improve the probability of entering high-quality academic universities. By comparing the effects of two cutoffs with different student abilities, our findings support the claim that the effect of elite school/magnet class enrollment depends on student abilities.