By Martin B. Hackmann

Choosing a school, accepting a job, or even starting a relationship all involve the same challenge: finding the right fit. Yet in many important markets, people make these decisions with incomplete information—and, crucially, the gaps run in both directions.

In new research with Benjamin Friedrich (Northwestern University), Adam Kapor and Sofia Moroni (Princeton University), and Anne Brink Nandrup (VIVE), UCLA Professor Hackmann studies how this two-sided information problem shapes match quality in one of society’s most important labor pipelines: medical education. Schools rarely know how applicants rank them, or what competing schools have learned about the same applicant. Applicants, in turn, do not know how schools view them relative to others. Standard models of matching markets assume these frictions away; this paper takes them seriously, in a setting where match quality is unusually consequential. More than 15 percent of admitted students in Denmark’s medical school programs drop out before graduating—a rate high in international comparisons, and a costly outcome in a country already facing shortages of doctors.

The study finds that the information schools collect plays a central role in determining student success. Applicants who voluntarily complete supplemental essays, interviews, or knowledge tests drop out at significantly lower rates than otherwise comparable applicants admitted on grades alone—the act of submitting itself reveals commitment that grades cannot. Schools’ own rankings of applicants through these channels are similarly predictive of who graduates: when programs can observe more than grades, they identify likely persisters with markedly greater accuracy.

A natural experiment at the University of Southern Denmark (Odense), which in 2002 introduced a knowledge test and a personal interview to better screen applicants, reinforces the point: dropout rates fell sharply at Odense, but rose at its closest rival, Aarhus, which absorbed a fair share of the applicants Odense had screened out. Matching markets are deeply interconnected, and changes at one institution reshape outcomes across the system.

To explore broader reforms, the researchers build and estimate a structural model of the Danish admissions market. In simulations with full information on both sides, dropout rates fall sharply—there is real room for improvement. Yet intuitive interventions fall flat. Revealing each applicant’s first-choice school to programs, for example, produces almost no benefit, because applicants begin to misreport their top choice strategically, turning a would-be signal into noise.
The findings offer a broader lesson for policymakers and market designers. Whether in education, labor markets, or healthcare, improving outcomes often requires more than simply increasing transparency. Because participants adapt strategically, successful policy design must account not only for what information is revealed, but for how individuals and institutions respond once it is.

The paper, “Interdependent Values in Matching Markets: Evidence from Medical Programs in Denmark” is available here.