Sensitivity Analysis for Treatment Effects in Difference-in-Differences Models using Riesz Representation

2025-10-10

AI Summary

This paper addresses the violation of the conditional parallel trends assumption in difference-in-differences (DiD) due to unobserved confounding. We propose a novel sensitivity analysis framework grounded in the Riesz representation theorem, the first to extend this approach to group-time causal parameter estimation under staggered treatment adoption. Our method integrates empirical information—including pre-treatment tests and covariate benchmarking—to quantify the impact of assumption violations. Coupled with double machine learning, it delivers asymptotically efficient sensitivity-corrected point estimators and valid confidence intervals, accompanied by diagnostic simulation tools. Simulation studies confirm statistical validity, while two empirical applications demonstrate substantial improvements in transparency and robustness of causal inference. The core contribution lies in the original theoretical extension of the Riesz representation framework and its practical adaptability to complex, staggered DiD settings—enabling principled, assumption-robust estimation without requiring strong parametric or functional-form restrictions.

Technology Category

• Reasoning under Uncertainty: Causality
• Machine Learning: Causal Learning
• Intelligent Robots: State Estimation

Application Category

• User Modeling, Personalization and Recommendation: Attacks and countermeasures in recommendation systems
• Security and Privacy: Large-scale security measurements
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking

Abstract

Difference-in-differences (DiD) is one of the most popular approaches for empirical research in economics, political science, and beyond. Identification in these models is based on the conditional parallel trends assumption: In the absence of treatment, the average outcome of the treated and untreated group are assumed to evolve in parallel over time, conditional on pre-treatment covariates. We introduce a novel approach to sensitivity analysis for DiD models that assesses the robustness of DiD estimates to violations of this assumption due to unobservable confounders, allowing researchers to transparently assess and communicate the credibility of their causal estimation results. Our method focuses on estimation by Double Machine Learning and extends previous work on sensitivity analysis based on Riesz Representation in cross-sectional settings. We establish asymptotic bounds for point estimates and confidence intervals in the canonical $2 imes2$ setting and group-time causal parameters in settings with staggered treatment adoption. Our approach makes it possible to relate the formulation of parallel trends violation to empirical evidence from (1) pre-testing, (2) covariate benchmarking and (3) standard reporting statistics and visualizations. We provide extensive simulation experiments demonstrating the validity of our sensitivity approach and diagnostics and apply our approach to two empirical applications.

Problem

• Assesses DiD robustness to unobservable confounder violations
• Extends sensitivity analysis using Riesz Representation to DiD
• Relates parallel trends violations to empirical diagnostic evidence

Innovation

• Uses Riesz Representation for sensitivity analysis
• Extends Double Machine Learning for causal estimation
• Relates parallel trends violation to empirical diagnostics