Information-Credible Stability in Matching with Incomplete Information
Abstract
In this paper, I develop a refinement of stability for matching markets with incomplete information. I introduce Information-Credible Pairwise Stability (ICPS), a solution concept in which deviating pairs can use credible, costly tests to reveal match-relevant information before deciding whether to block. By leveraging the option value of information, ICPS strictly refines Bayesian stability, rules out fear-driven matchings, and connects belief-based and information-based notions of stability. ICPS collapses to Bayesian stability when testing is uninformative or infeasible and coincides with complete-information stability when testing is perfect and free. I show that any ICPS-blocking deviation strictly increases total expected surplus, ensuring welfare improvement. I also prove that ICPS-stable allocations always exist, promote positive assortative matching, and are unique when the test power is sufficiently strong. The framework extends to settings with non-transferable utility, correlated types, and endogenous or sequential testing.