Matthew Thirkettle - Cornell University

Identification and Estimation of Network Statistics with Missing Link Data

Abstract:  A wide array of economic outcomes result from individuals interacting in social networks.  Individuals' behavior and decisions are affected by their position in a network e.g., their centrality.  Network statistics are used to measure these outcomes, but they are not point identified when the network is partially observed.  Partially observed networks are commonplace due to, for example, subsampling or incomplete responses in surveys. Using cross-sectional data and an endogenous social network-formation model, I obtain informative bounds on network statistics and economic outcomes of interest when a network is partially observed.  In contrast, previous studies require panel data or exogenous network formation to obtain point identification. I provide a computationally tractable characterization of an outer identified region for both the preferences determining network formation processes and for the network statistics.  In a simulation study on peer effort, I find that worst-case bounds that do not use the network formation model are 11 times wider than the bounds I obtain from my procedure.