Difference-in-Differences in the Presence of Unknown Interference

2025-12-24

AI Summary

This paper addresses the identification problem of difference-in-differences (DiD) estimation when the Stable Unit Treatment Value Assumption (SUTVA) is violated—particularly under unobserved interference and spillovers. We formally reinterpret the DiD estimand within the potential outcomes framework, revealing it as a weighted contrast of causal effects rather than a homogeneous treatment effect. Building on causal inference theory, we rigorously characterize the semantic interpretation of the DiD estimand under unobserved interference—the first such formal treatment. We propose three empirically verifiable, weakened identification assumptions that respectively ensure identifiability of the Group Average Treatment Effect (GATE), Local Average Treatment Effect (LATE), and overall Average Treatment Effect (ATE). Furthermore, we develop sensitivity analysis and empirical re-evaluation procedures, applying them to the seminal Card and Krueger (1994) minimum-wage study to refine its original causal interpretation. Our results provide both theoretical foundations and practical guidance for valid DiD inference in settings where spillovers are pervasive.

Technology Category

• Reasoning under Uncertainty: Causality
• Machine Learning: Causal Learning
• Knowledge Representation and Reasoning: Diagnosis and Abductive Reasoning

Application Category

• Security and Privacy: Large-scale security measurements
• Responsible Web: Measurement, analysis, and circumvention of Web censorship
• Web Mining and Content Analysis: Content-based information diffusion

Abstract

The stable unit treatment value (SUTVA) is a crucial assumption in the Difference-in-Differences (DiD) research design. It rules out hidden versions of treatment and any sort of interference and spillover effects across units. Even if this is a strong assumption, it has not received much attention from DiD practitioners and, in many cases, it is not even explicitly stated as an assumption, especially the no-interference assumption. In this technical note, we investigate what the DiD estimand identifies in the presence of unknown interference. We show that the DiD estimand identifies a contrast of causal effects, but it is not informative on any of these causal effects separately, without invoking further assumptions. Then, we explore different sets of assumptions under which the DiD estimand becomes informative about specific causal effects. We illustrate these results by revisiting the seminal paper on minimum wages and employment by Card and Krueger (1994).

Problem

• The paper examines DiD estimand identification under unknown interference.
• It explores assumptions needed to interpret DiD estimates causally.
• The study applies findings to Card and Krueger's minimum wage analysis.

Innovation

• DiD estimand identifies causal effect contrast under interference
• Assumptions needed to separate causal effects in DiD
• Illustrated via Card and Krueger minimum wage study