An Experimental Evaluation of a Matching Market Mechanism

Project Outcomes Statement

Overview
This project tested whether matching Army offices to units using the deferred acceptance algorithm improves job matching compared to the U.S. Army's traditional system of officer placement. Each year, thousands of Army officers are matched to units, and until recently this process relied heavily on manual assignments by human resource staff. Deferred acceptance is a transparent algorithm that allows both officers and units to rank preferences, producing stable matches that reduce inefficiencies.

The research team conducted the first randomized controlled trial (RCT) of such a mechanism in a large-scale labor market, assigning over 100 Army officer labor markets to use either the new algorithm or the traditional process. Outcomes were tracked using administrative data and officer surveys.

Intellectual Merit
The study provides the first experimental evidence on how strategyproof matching algorithms affect outcomes in practice. We found that deferred acceptance significantly improved satisfaction and reduced inefficiencies in assignments. Officers matched with the new system were more likely to remain in the Army in the short term, though longer-term retention effects were smaller. Performance outcomes such as promotions did not differ across systems.

These findings advance economic theory by testing predictions about stability, truthful reporting, and efficiency in a real-world labor market. They also demonstrate how randomized experiments can be used to study institutional design in high-stakes settings. The research has been accepted for publication in Econometrica, one of the leading journals in economics, ensuring its visibility to scholars and policymakers alike.

Broader Impacts
Beyond the Army, this project informs the design of matching systems used in medicine, education, and public policy. The results highlight both the benefits and the limits of algorithmic assignment, helping decision-makers weigh transparency, fairness, and efficiency in their own settings.

The project also contributed to workforce development by providing undergraduate and graduate students with training in research design, data analysis, and project management. Several student researchers have since gone on to graduate study and research careers.

Importantly, the U.S. Army has already incorporated deferred acceptance into its broader assignment processes, showing how evidence-based research can directly improve institutions. More broadly, the study illustrates how carefully designed experiments can yield insights that make systems fairer, more transparent, and more effective.

The replication package for "An Experimental Evaluation of the Deferred Acceptance: Evidence from Over 100 Army Officer Labor Markets" (Econometrica, 2025) can be found here: https://zenodo.org/records/15776949