Learning through Transient Matching in Congested Markets

A Google TechTalk, presented by Andrew Ferdowsian, 2024-05-23 Google Algorithms Seminar. ABSTRACT: I introduce a framework for studying transient matching in decentralized markets where workers learn about their preferences through their experiences. Limits on the number of available positions force workers to compete over matches. Each capacity-constrained firm employs workers whose match value exceeds a threshold. Since employment offers both payoff and information benefits, workers effectively face a multi-armed bandit problem. To them each firm acts as a bandit where the probability of "success'' at the firm is driven by market competition. In such markets, aggregate demand for firms satisfies the gross substitutes condition which ensures equilibrium existence. The resulting search patterns match a variety of stylized facts from labor market data. High-quality workers search less and tenure increases with age. In general, equilibria are inefficient because competition depresses the level of search. Natural interventions designed to improve efficiency are effective in uncongested markets, but can fail when congestion is severe. From a market design perspective, the utilization of headhunters has differential effects depending on workers’ quality, conclusively improving both outcomes for low-quality workers and overall efficiency. Reducing congestion through unemployment benefits, can depress search and may ultimately reduce match efficiency. About the Speaker: Andrew Ferdowsian completed his Ph.D. in economics at Princeton University in May 2023. While at Princeton he earned the Goldfeld Fellowship, alongside multiple Dietrich Economic Theory Center Grants. This past year he served as a tenure track lecturer (assistant professor) at the University of Exeter, and this fall he will be joining the University of Notre Dame as a tenure track assistant professor. His research focuses on improving market places, such as the marketplace for public housing in Singapore and the labor market for medical residents. His research can be found at https://ferdowsian.net/pages/research/.