Revealed Preferences of One-Sided Matching
summary
Builds an agent-level directed graph with an arc from each agent to every agent who holds the object that agent receives, then partitions the graph into strongly connected components; the allocation is rationalizable (in the core / competitive equilibrium) iff within each SCC every agent of the same observable type gets the same object — a condition checkable in linear time. Extends to transfers by adding net payments to arc weights so rationalizability ⇔ no positive cycle, and gives moment-inequality estimators for utilities from repeated aggregate matchings.
AI-written summary · about