Amsterdam replaced its gameable school lottery with a strategy-proof match

Amsterdam's popular secondary schools were chronically oversubscribed, and the old lottery-based system gave students little reason to state their true favorite: if you listed the popular school first and lost the draw, you might get nothing, so roughly 8% of students applied to a school that was not their most preferred, research by De Haan, Gautier, Oosterbeek and Van der Klaauw found.

In 2015 the city's schools, organized under OSVO, switched to a centralized matching system using deferred acceptance with a separate lottery per school. Students submit five ranked preferences and the algorithm assigns and re-assigns seats so no student can be displaced by a lower-priority applicant; the design came from the economists' own simulation study.

The first year was turbulent — about a hundred students were placed outside their top five, and parents sued to swap seats, but the court upheld the no-swap rule because allowing swaps would invite strategic gaming. The economists' evaluation found the mechanism worked: deferred acceptance placed more students in their top choices than the old system and raised average welfare, with the largest gains for disadvantaged families.

A strategy-proof match makes honest preferences safe

Deferred acceptance is strategy-proof, so students can always rank their real favorite first.

Separate lotteries per school keep children near oversubscribed schools from losing everywhere at once.

Banning after-the-fact swaps preserved the anti-gaming property of the mechanism.

Empirical evaluation showed welfare gains, concentrated among disadvantaged students.

If an allocation rule punishes honesty, people will lie and the sophisticated will win; make truth-telling a dominant strategy and the system stops rewarding gaming.

The system was refined after 2015, and Amsterdam became Europe's most studied school-choice experiment; the evaluation was published in the Journal of Political Economy in 2023.