Matching With Noise and the Acceptance Curse

Abstract

Matching in environments with search and information frictions requires agents to evaluate potential partners based on noisy signals of their hidden attributes. In this context, rational decision-making incorporates not only the observed signal but also the informational content of a partner’s acceptance decision. This phenomenon, termed the acceptance curse, implies that being accepted provides a negative signal regarding a partner’s type, as it reveals they were willing to match given their own private information and selectivity. Despite this endogenous adverse selection, an equilibrium exists characterized by a stochastic form of positive assortative mating. As an agent’s type increases, the severity of the acceptance curse diminishes, while the option value of continued search rises. The interplay of these forces ensures that higher-type agents remain more selective, resulting in a positive correlation between the types of matched pairs. By reinterpreting the dynamic matching problem as a Bayesian game with a continuum of types and actions, the existence of a nontrivial equilibrium in increasing strategies is established. Furthermore, matching equilibria in decreasing strategies are shown to be non-existent, confirming that the incentive to match with higher-quality partners remains dominant even under significant informational noise. – AI-generated abstract.