The Best In The Class

I estimate the effect of being the best in the class in primary school on performance in secondary school. I implement a novel methodology that exploits that some students are the best in the class because better students in the same school are assigned to other classes. If students were randomly assigned, the probability of being the best in the class would be a well-known function of students ranking in the school and the number of classes. I use this exogenous probability as an instrument for actually being the best in the class. I find a positive impact of being the best in the class on future performance: being the best in primary school increases test scores by 0.13 standard deviations in secondary school. My instrument is suitable to account for the sorting of units into groups in other contexts.