The manipulability of matching rules via segmentation (Q868616)

The paper analyses the manipulation of matching rules via several versions of segmentation. The considered matching problem feature agents with endowments that face certain division rules. In particular, a coalition of agents may get matched according to some matching rule under a given division rule, then rejoin the complementary coalition with their new endowments and get re-matched in the grand coalition according to the same matching rule and under the same division rule. If each member of such a coalition is at least as well off and some members are better off than if they were to be matched in the original problem, it is said by the authors that the matching rule is manipulable by this set of agents via active segmentation. Moreover, if each member of such a coalition is better off than if they were to be matched in the original problem they say that the matching rule is strongly manipulable by this set of agents via active segmentation. The authors show that all stable matching rules are strongly manipulable under a fairly wide class of division rules. They also provide a class of division rules under which the men-optimal matching rule is not manipulable via active segmentation, and analyse whether all agents in a society can be made at least as well off and some agents can be made better off by separating into coalitions, one of which participates in segmentation actively and the other one actively or passively. Finally, the authors show the possibility of cycles of matching pairs under iterations of problems through the men-optimal matching rules, and that such cycles can repeat themselves many times, and for some sequences of problems even infinitely in the limit.