Bayesian Stable States

This paper extends the Bayesian stability notion of Liu (2020) to define the Bayesian stability of a market state, which consists of a matching outcome and an information structure. The information structure can be arbitrarily heterogeneous and can accommodate learning among agents. We first establish that a Bayesian stable matching function of Liu (2020) can be recast as Bayesian stable market states with homogeneous information. We then illustrate the usefulness of such an extension by (i) refining Liu's Bayesian efficiency notion to define the Bayesian efficiency of a market state and (ii) generalizing his result—that Bayesian stable matching functions are Bayesian efficient—to an analogous one for market states. More importantly, we show that (iii) a decentralized matching process converges to a Bayesian stable market state and thereby offer a decentralized foundation for Liu's Bayesian stable matching function.