School Choice: The Case for the Boston Mechanism

I study school assignment problems, focusing on two popular mechanisms: the Boston Mechanism (BM) and Deferred Acceptance (DA). The former has been criticized regarding both efficiency and fairness, particularly its treatment of naive (non-strategic) students. The latter has been suggested in its place, and has already replaced the former in several cities. The formal critique of BM and support of DA were founded on the assumption of strict priorities, i.e., schools rank every child so that there are as many priority classes as there are students. In almost all cities where these mechanisms are applied, however, the actual number of priority classes (e.g., walking-distance and sibling in school) that may be used is orders of magnitude smaller than the number of students, and tie-breaking lotteries are needed. Approximating this case by assuming only one priority class, I show that BM outperforms DA according to several ex ante e¢ ciency criteria. DA performs very poorly if all students share identical ordinal preferences over schools. Simulations show that these analytical results extend to more realistic cases. Finally, I suggest a simple modification to BM, which, according to simulations, protects naive students while largely preserving its efficiency properties.