Title: Identification and estimation of a partially linear regression model using network data
Authors: Eric Auerbach - Northwestern University (United States) [presenting]
Abstract: A regression model is studied in which one covariate is an unknown function of a latent driver of link formation in a network. Rather than specify and fit a parametric network formation model, a new method is introduced based on matching pairs of agents with similar columns of the squared adjacency matrix, the $ij$-th entry of which contains the number of other agents linked to both agents $i$ and $j$. The intuition behind this approach is that for a large class of network formation models the columns of this matrix characterize all the identifiable information about individual linking behavior. We first describe the model and formalize this intuition. We then introduce estimators for the parameters of the regression model and characterize their large sample properties.