A1849
Title: Networks as proxy controls: Nonparametric identification and estimation of partial effects
Authors: Gaoqian Xu - University of Washington (United States) [presenting]
Abstract: A nonparametric model is studied where a latent variable creates endogeneity by affecting both network formation and an outcome of interest. The existing network control function approach is generalized to nonparametric outcome models, using individuals' link functions to account for unobserved heterogeneity. Identification is achieved through a form of matching on unobservables: individuals are conceptually matched based on their latent link functions. Implementation involves two steps: first, the distances or dissimilarities between the latent link functions are estimated using network data; second, functional kernel smoothing is applied over these distances to estimate the structural parameter. Asymptotic analysis reveals a fundamental trade-off: the robustness gained from this approach comes at the unavoidable cost of a slow convergence rate, driven by the difficulty of matching on latent objects. This statistical cost is characterized by deriving a minimax lower bound.
