Timing is Everything? An Experimental Investigation of Incomplete Information in Centralized Matching Mechanisms

In this paper, we experimentally investigate in a large game four matching mechanisms related to China’s college admission system: Boston or serial dictatorship with students submitting their preferences either before (incomplete information) or after (complete information) the score realization. We find significant differences in the proportion of truth-telling across the mechanisms, but there is no significant difference in non-truth-telling reporting strategies. Students' reporting strategies affect their matching outcome: down-reporting on the first choice school leads to worse outcome on average, while up-reporting is associated with higher risk. The Boston mechanism with incomplete information is not superior to the other three mechanisms in ex-ante efficiency and ex-ante/post fairness when the score uncertainty is small, but may be superior to others in ex-ante fairness when the score uncertainty is large.