The Content of Data: Backing out Preferences in Non-strategy Proof Mechanisms
summary
Presents a structural estimator that backs out preferences from submitted rank lists by modeling a mixed naive versus sophisticated population and solving optimal lists via backward induction, then validates on 2006 Barcelona data with 2007 priority reform validation. Finds many households behave rationally, estimates type probabilities from observables, and shows neighborhood priorities and a few bad schools push safe play so that 44% of schools fill in first round and switching to Gale Shapley reduces aggregate welfare while TTC gives modest gains.
AI-written summary · about