An Improved Bound to Manipulation in Large Stable Matches

This paper provides an upper bound to the expected proportion of agents that have incentives to misreport their true preferences or vacancies in many-to-one stable matching mechanisms, given that others report their preferences and vacancies truthfully. The paper shows that the upper bound converges to zero as the number of participants in the market goes to infinity at a faster rate as compared to previous results from the literature. Moreover, the paper relaxes some of the assumptions that are commonly made for getting this convergence result. In particular, it does not require that the stable mechanism generates an allocation with many listed positions unfilled. So this paper adds evidence to the fact that, though stable matching mechanisms are not completely strategy-proof, in practice, when the number of participants in the market is sufficiently large, they are effectively strategy-proof