【主讲】Zhentao Shi (The Chinese University of Hong Kong)
【主题】A Structural Network Pairwise Regression Model with Individual Heterogeneity
【时间】2016年4月22日 (周五) 15：30-17：00
【地点】上海财经大学经济学院楼701室
【语言】英文
【摘要】Pairwise binary regressions are used in labor economics and development economics to predict links in moderate-scale networks by observed characteristics of each pair of agents. With no micro foundation, their usefulness is limited in causal inference about economic phenomena.We propose a new structural pairwise model based on individual utility maximization that maybe useful in such contexts. In this model, the individual utility function governs whether anagent is willing to create a link with each peer. The econometrician, as an outsider, witnessesa pairwise link if and only if it is preferred by both agents. Besides observed personal characteristics, unobservable individual heterogeneity enters the utility function as an additive xed effect.The fixed effects enhance the applicability of the model, but they impose technical challengesas they are buried in nonlinear functions. To identify and estimate the model, we exploit thespecial panel-data structure. In a network of n individuals, we can view one's linking outcomesto all his n1 peers as a within-group observation, so that the network provides n overlappinggroups. We treat each xed effect term as an unknown nuisance parameter. Under a distri-butional assumption about the link-speci c idiosyncratic shock, we establish consistency andasymptotic distribution for the maximum likelihood estimator (MLE) of the common param-eter. We apply the new estimator to study the determinants of gift exchange in a network ofvillage households in China. We apply the new estimator to study the gift exchange behaviorin a network of rural households in a Chinese village.
