[UTMD-094] Robust Exchange (by Yuichiro Kamada, Yosuke Yasuda)

Yuichiro Kamada, Yosuke Yasuda

This paper offers the first rigorous economic analysis of the dropout problem— a practical challenge in designing exchange markets. We introduce a model incorporating situations where participants may drop out of the market before transactions occur, which results in the failure of all transactions within the same exchange cycle. We show that the celebrated Top Trading Cycles (TTC) mechanism is disadvantaged due to its potential reliance on large exchange cycles, which implies a high likelihood of transaction defaults. Furthermore, agents may not have an incentive to truthfully report their preferences under the TTC mechanism when dropouts can happen. To improve efficiency, we propose k-greedy mechanisms designed to manage the risks associated with exchange cycle sizes while maintaining reasonable efficiency, with a constraint on the maximum possible cycle size of k. We show, both theoretically and through simulations, that the k-greedy mechanisms (with small values of k) outperform the TTC mechanism.