讲座题目：Stable Allocations with Network-Based Comparisons

　　报告人：程琛

　　报告时间：2018年12月21日(周五)下午15点30分

　　报告地点：经管院A208

　　主办单位：数理经济与数理金融系

　　主持人：韩立宁

　　摘要:We consider a model in which an agent’s payoff is based on her (local) ranking, i.e., the ranking of her allocation among her neighbors’ in the network. An allocation is stable if it is not revoked under α-majority voting; that is, there exists no alternative allocation, such that a fraction of at least α of the population have their rankings strictly improved under the alternative. We find a sufficient and necessary condition for a network to permit any stable allocation: the network has an independent set of size at least (1 − α) of the population. A network is more permissive if it permits stable allocation for a larger set of α. We then characterize the size of the largest independent set for Erdős–Rényi random networks, which reflects how permissive a network is: for large enough population, the level of permissibility solely depends on the expected degree. We provide several interesting comparative statics results: more connected networks, more populated networks (with a fixed link probability), or more homophilous networks are less permissive. Given expected degree, whether a network is segregated or integrated would not affect the level of permissibility. We generalize our model to arbitrary blocking coalitions and provide a sufficient and necessary condition for this case. We also extend the model to directed networks.

　　简介：Chen Cheng is currently an Assistant Professor at the Carey Business School, Johns Hopkins University. She received Ph.D. from Managerial Economics and Strategy Department (MECS), Kellogg School of Management, Northwestern University in 2016. She got M.A. from China Center for Economic Research (CCER) at Peking University in 2010 and B.A. in economics from Huazhong University of Science and Technology in 2008.