The Local Approach to Causal Inference under Network Interference
This addresses the problem of estimating treatment effects in social or economic networks for researchers in fields like economics and sociology, though it appears incremental as it builds on existing network interference literature.
The paper tackles causal inference under network interference by proposing a nonparametric framework that pools outcome data across similarly configured agents, deriving finite-sample error bounds for a k-nearest-neighbor estimator and an asymptotically valid test for policy irrelevance.
We propose a new nonparametric modeling framework for causal inference when outcomes depend on how agents are linked in a social or economic network. Such network interference describes a large literature on treatment spillovers, social interactions, social learning, information diffusion, disease and financial contagion, social capital formation, and more. Our approach works by first characterizing how an agent is linked in the network using the configuration of other agents and connections nearby as measured by path distance. The impact of a policy or treatment assignment is then learned by pooling outcome data across similarly configured agents. We demonstrate the approach by deriving finite-sample bounds on the mean-squared error of a k-nearest-neighbor estimator for the average treatment response as well as proposing an asymptotically valid test for the hypothesis of policy irrelevance.