Title: Spillovers of Program Benefits with Missing Network Links

Abstract: The issue of missing network links in partially observed networks is frequently neglected in empirical studies. This paper aims to address this issue when investigating spillovers of program benefits in the presence of network interactions. We focus on a nonparametric outcome model that allows for flexible forms of heterogeneity, and we propose two methods to point identify the treatment and spillover effects in the case of bounded network degree. When the network degree is unbounded, these methods can be used to mitigate the bias caused by missing network links. The first method utilizes two easily-constructed network measures based on the incoming and outgoing links. The second method relies on a single network measure with a known rate of missing links. We demonstrate the effective bias reduction provided by our methods using Monte Carlo experiments and a naturalistic simulation on real-world school network data. Finally, we re-examine the spillovers of home computer use on children's self-empowered learning.