The reason parents try get their children in good schools is not because of the better teachers, it is because of the better class mates. If they are among stable, studious and ambitious students, they are expected to perform better in expected terms. (Of course there is also the tactic to send your kid to an inner-city school, where she outperforms the others and lands easy acceptances in colleges, with scholarships, if all goes right). But how much do these peer effects matter?
Eleonora Patacchini, Edoardo Rainone and Yves Zenou study the social networks of children in the AddHealth longitudinal survey and find that it is the very last years of high school that matter, but those years matter a lot as their impact is very persistent. Specifically, they observe that if a child's friends have the equivalent of two more completed high school educations, that child will get 3.5 months more education. But it would be good to know whether this dominates any effect from neighborhoods, which is taken into account but not reported.