Publications 

Unequal Peace (joint with Charles Zheng), June 2024, Forthcoming, IER

A mediator proposes a settlement between two contestants to avoid a conflict where the cost each contestant bears is inversely related to the contestant's privately known strength. Their strength levels are identically distributed, and their welfares  weigh equally in the mediator's objective. However, the optimal proposal offers one contestant much more than it does the other so that the former accepts it always, while the latter only occasionally. This unequal treatment improves the prospect of peace by making one contestant willing to settle without fearing that the action signals his weakness that his opponent can exploit should conflict occur. 

Work in Progress 

Favoritism in Manipulative Conflict Mediation  

In the literature on dynamic conflict management, a received insight is that full participation in the negotiation mechanism is not guaranteed by the revelation principle. This paper studies a conflict model where a mediator, who can offer economic incentives and for reputation and practical motivations would like to guarantee full participation in her negotiation mechanism, proposes a split of peace surplus to two players to minimize the probability of conflict. Players have private valuations for the prize. Participation is voluntary. Players decide whether to participate in the mediation and, if participating, whether to accept or reject the split. If mediation fails, conflict ensues, in which the continuation play depends on posterior beliefs that are indirectly determined by the proposal. Despite ex-ante identical rivals, a certain kind of biased proposals, called lopsided, are optimal if peace surplus is lower than a threshold and the equal proposal is optimal if otherwise. Lopsided proposals incentivize the favored player to always accept, thereby revealing no further information. The equal proposal, due to inducing a symmetric posterior information structure, violates full participation when the high type has optimistic forecasts of nonparticipation. Given such beliefs, randomized lopsided proposals, due to asymmetric revelation of information, are conflict minimizing splits subject to the full participation condition. When the prior is extended to a continuum of types, the equal proposal violates full participation while randomized lopsided proposals, which the favored player always accepts, satisfy it. 

Peace Settlements with Possibilities to Renege

This paper studies a conflict model where adversaries lack commitment and can renege on an accepted agreement. The primitive is such that no negotiation mechanism exists that fully preempts conflict. A mediator whose objective is to maximize welfare subject to renege-proof constraint proposes a peaceful split of the contested prize to two ex-ante identical rivals. By observing each other's decisions, after a successful or failed mediation, adversaries learn about each other and update their forecast of conflict. Despite ex-ante identical players, the admissible optimal proposal is a biased proposal where the favored player always accepts and reveals no further information. This biased proposal is even more extreme than the biased proposal that maximizes welfare in a renege-banning model where players have full commitment. The same results hold when the objective of the mediator is minimizing the probability of conflict.