Ex-ante fairness under constrained school choice: An experimental approach

Abstract

In a college admission mechanism, students are often matched with colleges by using a noisy signal of their true abilities (e.g., their exam scores). The matching outcome thus may be imperfect in terms of ex-ante fairness, which suggests matching students with higher ability to better colleges. To achieve ex-ante fairness, we consider constraining student choice over colleges, by designing treatments with different constraint levels under the Boston and Serial Dictatorship mechanisms, with preference submission before or after the exam. Constraining student choice increases the probability of achieving ex-ante fairness under the Boston and Serial Dictatorship mechanisms with preference submission before the exam, compared with unconstrained mechanisms and mechanisms with preference submission after the exam. However, the probability of achieving highly unfair matching is also increased, resulting in a riskier matching outcome. Learning the game or providing students with recommended equilibrium strategies can decrease this risk and further increase ex-ante fairness.