Abstract

We build a school choice model where students can learn about their cardinal fits to schools before submitting rank-order lists to a clearinghouse. Learning is flexible; students can choose how much to learn and what to learn. The model allows us to revisit the Deferred-Acceptance mechanism's utilitarian inefficiency relative to the Boston mechanism. Our model suggests that students acquire less information under the DA, which is driven by and reinforces DA's greater reliance on tie-breaking due to homogeneous rank submissions. Hence, the inefficiency of DA can be more substantial than previously known.