Statistical Theory for Economic Models of Network Formation

长期以来，社会科学家一直在研究网络对行为的影响，以及网络如何维持合作、传递信息和促进经济交易。随着各种社会背景下网络数据的日益可用性，科学家们现在有机会对理论进行严格的实证检验，并发现新的事实，以提高对社会行为的理解，更好地设计经济政策，以促进国家福利。为此，本研究开发了新的计量经济学工具来分析网络模型，特别是网络形成模型。需要新的工具来解释网络数据的独特结构和网络模型的复杂性。研究者研究了具有战略相互作用的网络形成的计量经济模型。这些是潜在指数模型，允许其他人形成的链接直接影响自己形成链接的倾向。本研究的一个主要目标是证明一个中心极限定理，该定理既适用于仅使用大型网络的单个快照作为数据的静态模型，也适用于利用大型网络的短时间序列的动态模型。研究者还开发了实用的推理方法。在研究大型网络的依赖结构时，本研究也适用于内源性社会互动模型的推断。该奖项反映了美国国家科学基金会的法定使命，并通过使用基金会的知识价值和更广泛的影响审查标准进行评估，被认为值得支持。

Social scientists have long theorized about the influence of networks on behavior, how they sustain cooperation, transmit information, and facilitate economic transactions. With the increasing availability of network data in a variety of social contexts, scientists now have the opportunity to subject theory to rigorous empirical testing and uncover new facts to improve understanding of social behavior and better design economic policies to advance the national welfare. Toward this end, this research develops new econometric tools for the analysis of network models, especially models of network formation. New tools are needed to account for the unique structure of network data and complexity of network models. The investigator studies econometric models of network formation with strategic interactions. These are latent-index models that allow links formed by other individuals to directly influence one's own propensity to form links. A main objective of this research is to prove a central limit theorem relevant for inference both in static models which use as data only a single snapshot of a large network, and dynamic models which exploit a short time series of a large network. The investigator also develops practical methods for inference. In studying the dependence structure of large networks, this research is also relevant for inference in models of social interactions when the network is endogenous.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.