How to Use Prices for Efficient Online Matching [link]
Many matching markets feature unknown, dynamic arrivals of agents that must match immediately. A caseworker must match an abused child to a foster home, a hospital must assign a patient in critical condition to a room, or a city must place a homeless individual into a shelter. We design an online matching algorithm---the Sequential Equilibrium Mechanism (SEM)---that approximates large market equilibria to match arriving agents to objects. SEM is asymptotically efficient, fair, and strategy-proof with probability one. Our application plans to deploy a lab-in-the-field experiment where real caseworkers match vulnerable children to host homes, and we currently provide simulation evidence that SEM can substantially improve welfare.
Presentations: Conference on Mechanism and Institution Design (2026)
A Dynamic Matching Framework for Faster Child Adoptions [link]
Caseworkers in foster care systems match waiting children to adoptive homes. We use dynamic matching market design to characterize a class of mechanisms that incentivize expedient matches that homes can accept or decline. We design mechanisms satisfying fairness and limited strategy-proofness. They also avoid costly patience. Our empirically-based simulations suggest the mechanisms could increase adoptions by at least 25% versus the status quo. A naive dynamic extension of Deferred Acceptance does not attain these benefits. Our mechanisms sidestep direct preference elicitation by predicting preferences, and they are robust to prediction error.
Do Preferences Matter in Fair Task Allocation? (draft available upon request)
I model fair task allocation where tasks stochastically arrive and must be matched to a fixed set of agents; the novel constraint is that agents must receive allocations that require the same level of average effort. Social work supervisors, call center managers, and courts all rotate case allocation across employees to satisfy this balance constraint, but the Rotation mechanism is not Pareto efficient. I design the Dynamic Pseudo-Market (DPM) mechanism, and it satisfies Pareto efficiency and asymptotic balance. It has strong finite-sample balance bounds, and a simple modification is asymptotically Pareto efficient and ex-post balanced. Because Pareto inefficiency suggests that employees might be persistently dissatisfied and turnover as a result, a common problem in all of these applications, I plan to test DPM in an online task experiment to analyze its effects on retention and task productivity as well as its robustness to manipulability.
Presentations: POMS (2026)
How Are Good Matches Made in Foster Care? (pre-analysis draft available upon request)
In the United States, caseworkers match neglected and abused children to temporary foster homes in decentralized markets with uncertainty over match quality. I theoretically and experimentally study the impact of supply and demand in foster care. There are not enough family foster homes for children (supply-side constraint), and caseworkers representing children often have limited time to deliberate matches (demand-side constraint). I build a theoretical framework that shows how supply and demand could impact match quality. Alleviating the demand and supply constraints could either increase or worsen child welfare, suggesting that policies aimed at recruiting foster homes must be paired with effective interventions to aid caseworker decision-making. In a novel experiment design, I recruit real caseworkers to participate in foster care matching markets while varying these constraints.