We have conducted laboratory experiments to study why the dynamic mechanism to assign college seats to applicants implemented in 2015 in Vietnam failed. Note that similar mechanisms were introduced in Brazil (2010) and in Inner Mongolia, China (2007) as well.

 

With the help of four different experimental treatments, we sought to understand which defining characteristics of the mechanism contributed the most to its failure. In particular, we measured the impact of time constraints and the impact of the lack of commitment on the final matching outcomes (whether they are stable and whether they maximize aggregate welfare) and the strategic incentives (whether applicants have incentives to report their preferences truthfully) created by the mechanism.

 

We have collected data on decisions made by 180 participants in November-December, 2025, in the experimental laboratory at the School of Political Science and Economics (Waseda University). All our sessions were computerized (using zTree) and monetarily incentivized.

 

The statistical analysis of the data supports the theoretical predictions that we derived for the mechanism implemented in Vietnam, suggesting that the mechanism is manipulable by applicants submitting truncated preference lists of college seats (instead of telling the truth) and that it would not deliver a stable outcome even based on the applicants' true preferences.

We find that, when compared to the well-known static benchmark with desirable theoretical properties (i.e., the Gale-Shapley deferred-acceptance mechanism), a dynamic mechanism based on the same rules leads to substantial improvements in terms of inducing truthful behavior, stability, and efficiency. Regarding our main research question, we conclude that the introduction of a time constraint slightly reduces stability, while the lack of commitment in applications significantly reduces stability and efficiency.