Jordan Norris, NYU Abu Dhabi

"Bounding Network Comparative Statics"

Abstract

Many fields of economics can be framed as the study of networks between agents, with comparative static analyses constituting a core purpose of the associated economic modeling. Yet, except in highly stylized, empirically unrealistic models, the theoretically implied relation between an endogenous outcome and exogenous shock is extremely complex. In this paper, we derive new properties for a wide class of network models, and show they imply non-trivial yet simple bounds on network comparative statics — the effect of a shock to a node on the outcome of any node in the network. This result provides a new tool for establishing sufficiency for qualitative questions (such as when is a treatment effect positive), and identification in settings of incomplete network data.

Contact person: Egor Starkov