The Iterative Deferred Acceptance Mechanism
52 Pages Posted: 7 Dec 2016 Last revised: 1 Dec 2018
Date Written: August 1, 2018
We introduce a new mechanism for one-sided matching markets, inspired by procedures currently being used to match millions of students to public universities in Brazil and China. Unlike most options available in the literature, where students are asked for a full preference ranking over all colleges, they are instead sequentially asked to make choices among sets of colleges. These choices are used to produce, in each step, a tentative allocation. If at some point it is determined that a student cannot be accepted into a college, then she is asked to make another choice among those which would tentatively accept her. Participants following the simple strategy of choosing the most preferred college in each period is a robust equilibrium which yields the Student Optimal Stable Matching. We analyze the mechanism currently being used in Brazil and, based on data from 2016, we provide evidence that it suffers from shortcomings: it fails to provide reliable information to the students about the universities that they are able to attend; and it is subject to manipulation via cutoffs, a new type of strategic behavior. We also provide an extension in which after running the sequential mechanism for a number of steps students are asked to submit a ranking over the colleges that are still within reach. This constitutes a novel approach to matching mechanisms. We show that the initial sequential stage clears a substantial part of the market before the rankings submission. This finding, together with empirical and simulation results, makes our proposal an attractive alternative to sequential mechanisms currently being used and the standard Gale-Shapley Deferred Acceptance mechanism for practical applications.
Keywords: Market Design, Matching, Iterative Mechanisms, College Admissions
JEL Classification: C78, C92, D63, D78, D82
Suggested Citation: Suggested Citation