Unobserved Heterogeneous Spillover Effects in Instrumental Variable Models

2025-11-27

AI Summary

This paper addresses the causal identification challenge in instrumental variable (IV) models arising from unobserved peer spillover effects. We propose generalized Local Average Controlled Spillover and Direct Effects (LACSEs/LACDEs), and further define Marginal Controlled Spillover and Direct Effects (MCSEs/MCDEs) to enable policy-relevant heterogeneous analysis. To our knowledge, this is the first extension of the marginal treatment effect framework to IV settings with spillovers. We derive novel point-identification conditions that do not require discretization of the instrument, thereby clarifying the necessity of conventional restrictions under binary IVs. Our approach supports nonparametric identification and semiparametric/parametric estimation, accommodating both continuous instruments and networked group interactions. An empirical application using Add Health data—focusing on best-friend networks—demonstrates statistically significant and individually heterogeneous educational spillovers, validating both the methodological soundness and real-world applicability of our framework.

Technology Category

• Machine Learning: Causal Learning
• Reasoning under Uncertainty: Causality
• Application Domains: Humanities & Computational Social Science

Application Category

• Graph Algorithms and Modeling for the Web: Algorithms and analysis for incomplete, noisy, or partially observed Web-related graphs
• User Modeling, Personalization and Recommendation: Social recommender systems and personalization for the social Web
• Security and Privacy: Large-scale security measurements

Abstract

This paper develops a general framework for identifying causal effects in settings with spillovers, where both outcomes and endogenous treatment decisions are influenced by peers within a known group. It introduces the generalized local average controlled spillover and direct effects (LACSEs and LACDEs), which extend the local average treatment effect framework to settings with spillovers and establish sufficient conditions for their point identification without restricting the cardinality of the support of instrumental variables. These conditions clarify the necessity of commonly imposed restrictions to achieve point identification with binary instruments in related studies. The paper then defines the marginal controlled spillover and direct effects (MCSEs and MCDEs), which naturally extend the marginal treatment effect framework to settings with spillovers and are nonparametrically point identified from continuous variation in instruments. These marginal effects serve as building blocks for a broad class of policy-relevant treatment effects, including some causal spillover parameters in the related literature. Semiparametric and parametric estimators are developed, and an application using Add Health data reveals heterogeneity in education spillovers within best-friend networks.

Problem

• Identifies causal effects with peer spillovers in groups
• Extends treatment effect frameworks to include spillover conditions
• Develops estimators for policy-relevant spillover effects in networks

Innovation

• Extends LATE framework to spillover settings
• Defines marginal effects for continuous instrument variation
• Develops semiparametric estimators for peer effects