DP16544 Do Peer Preferences Matter in School Choice Market Design? Theory and Evidence
Can a centralized school choice clearinghouse generate a stable matching if it does not allow
students to express their preferences over both programs and peers? Theoretically, we show
that a stable matching exists with peer preferences under mild conditions, but finding one via
canonical mechanisms is unlikely. We show that increasing transparency about the previous
cohort of students enrolling at each program induces a tâtonnement process wherein the distributions
of former student types play the role of prices. We theoretically model this process
and develop a test for match stability. We implement this test empirically in the college admissions
market in New SouthWales (NSW), Australia, where we find evidence of preferences
over relative peer ability. We show that the NSW market fails to converge to stability over time.
We propose a new mechanism that improves upon the current design, and we show that this
mechanism generates a stable matching in the NSW market.