Doubly Robust Difference-in-Differences Estimators

2018/11/29 by Pedro H. C. Sant'Anna, Pedro H.C. Sant’Anna, Pedro H. C. Sant’Anna +4 · 1 voice · 64 citations
Economics, Econometrics and Finance · Mathematics · #Advanced Causal Inference Techniques #Statistical Methods and Bayesian Inference #Statistical Methods and Inference #econ.EM

paper · pdf · doi:10.1016/j.jeconom.2020.06.003

arxiv created 2020/05/05 · arxiv updated 2020/05/07 · openalex publication_date 2020/07/07 · openalex created_date 2025/10/10 · openalex updated_date 2026/07/30

Abstract

This article proposes doubly robust estimators for the average treatment effect on the treated (ATT) in difference-in-differences (DID) research designs. In contrast to alternative DID estimators, the proposed estimators are consistent if either (but not necessarily both) a propensity score or outcome regression working models are correctly specified. We also derive the semiparametric efficiency bound for the ATT in DID designs when either panel or repeated cross-section data are available, and show that our proposed estimators attain the semiparametric efficiency bound when the working models are correctly specified. Furthermore, we quantify the potential efficiency gains of having access to panel data instead of repeated cross-section data. Finally, by paying articular attention to the estimation method used to estimate the nuisance parameters, we show that one can sometimes construct doubly robust DID estimators for the ATT that are also doubly robust for inference. Simulation studies and an empirical application illustrate the desirable finite-sample performance of the proposed estimators. Open-source software for implementing the proposed policy evaluation tools is available.

Citations

• Semiparametric Difference-in-Differences Estimators
• Doubly Robust Estimation in Missing Data and Causal Inference Models
• Recent Developments in the Econometrics of Program Evaluation
• A Generalization of Sampling Without Replacement from a Finite Universe
• Difference-in-Differences with Multiple Time Periods

Cited by

• Estimating Causal Effects With Observational Data: Guidelines for Agricultural and Applied Economists
• Difference-in-Differences with Interval Data
• Difference-in-differences with stochastic policy shifts of continuous treatments
• What’s trending in difference-in-differences? A synthesis of the recent econometrics literature
• Causal Inference About the Effects of Interventions From Observational Studies in Medical Journals
• AI Diffusion in Low Resource Language Countries
• Young women in cities: Urbanization and gender-biased migration
• Political Symbols and Social Order: Confederate Monuments and Performative Violence in the Post-Reconstruction U.S. South
• Revisiting Event-Study Designs: Robust and Efficient Estimation
• Effect of Cigarette Price and Tax Increases on Smoking in Europe: A Difference-in-Differences Study with Double Machine Learning
• Preliminary Independent Evaluation of Free2B: A Targeted Intervention to Promote the Mental Wellbeing of LGBTQIA+ Youth
• Hiring mental health professionals: Evidence from a large-scale policy in Brazil
• School Closures and Rural Population Decline*
• The Socio-Economic Impact of Coal Mining in Mozambique
• How one gesture curbed ethnic discrimination
• Pricing Congestion to Increase Traffic: The Case of Bogotá
• Efficient Difference-in-Differences Estimation when Outcomes are Missing at Random
• Behavioral Charity: Third‐Party Ratings of Nonprofits as Salience and Heuristics
• The Null Impact of Parenthood on Technical Risk Preferences
• Funding and productivity: Does winning grants affect the scientific productivity of recipients? Evidence from the social sciences and economics
• Resale Royalty in Non-Fungible Token Marketplaces: Blessing or Burden for Creators and Platforms?
• Public Services Under Private Management
• From Meat to Plant‐Based Products? The Enduring Impact of BSE on Beef Consumption
• Difference-in-differences with variation in treatment timing
• Institutional Innovation and the Adoption of New Technologies: The Case of Steam
• Socioeconomic impacts of concessions for visitor support services in protected areas: matching and difference-in-differences analysis
• A Fad or the New Norm for Student Access Today? Evaluating Enrollment Outcomes of Holistic Admissions in South Korea
• Breaking the Language Barrier? Exam Fee Subsidies and Spanish Proficiency
• When the money stops: fiscal and political reactions to changing grant eligibility
• How the Metaverse is Reshaping Multichannel Retail Through Balancing Complementarity and Substitution
• Modeling User Redemption Behavior in Complex Incentive Digital Environment: An Empirical Study Using Large-Scale Transactional Data
• Pretest with Caution: Event-Study Estimates after Testing for Parallel Trends
• Gender Differences in Healthcare Utilisation -- Evidence from Unexpected Adverse Health Shocks
• Dynamic Synthetic Controls vs. Panel-Aware Double Machine Learning for Geo-Level Marketing Impact Estimation
• The impact of super apps on the nutrition transition in low- and middle-income countries: Evidence from Indonesia
• Bayesian Sensitivity Analyses for Policy Evaluation with Difference-in-Differences under Violations of Parallel Trends
• A difference-in-differences estimator by covariate balancing propensity score
• Estimating dynamic treatment effects in event studies with heterogeneous treatment effects
• How much should we trust staggered difference-in-differences estimates?
• Broad‐Based Categorical Eligibility Policy and SNAP Participation
• Rural healthcare access and supply constraints: A causal analysis
• The impact of city-county amalgamations on small administrative units: evidence from land transfer in China
• Quasi-experimental evidence on Spanish VAT cuts during economic shocks
• Difference-in-Differences with Multiple Time Periods
• Good for the goose, bad for the gander? Corruption and income inequality
• Bridging Structural Causal Inference and Machine Learning The S-DIDML Estimator for Heterogeneous Treatment Effects
• Debiased Semiparametric Efficient Changes-in-Changes Estimation
• Estimating controlled direct effects with panel data: an application to reducing support for discriminatory policies
• Causal Panel Analysis under Parallel Trends: Lessons from a Large Reanalysis Study
• Mandated Sick Pay: Coverage, Utilization, and Crowding-in
• Veto and versatile: understanding the local implementation of environmental policies in China
• What do Firms Gain from Patenting? The Case of the Global ICT Industry
• Manufacturing Revolutions: Industrial Policy and Industrialization in South Korea
• Small Samples and Short Panels: Evaluating Policy Evaluation Methods with Realistic Data
• Evaluating the impact of Canadian cannabis legalization on cannabis use outcomes in emerging adults: Comparisons to a US control sample via a natural experiment
• Teen parent trap? The education and labor implications of motherhood and fatherhood during the transition from adolescence to adulthood in Cebu, the Philippines
• Ripe for contracts? Estimating the impact of an avocado producer organization contract farming intervention
• Institutional Effects of Adding Football: A Difference-in-Difference Analysis
• Certificates without Electrons? Theory and Evidence on Impacts from AI-Driven Power Demand
• Television and American consumerism
• Causal models for longitudinal and panel data: a survey
• Difference-in-differences with "bad controls"
• Skills Beget Skills: Evidence from Historical School Reforms Targeting Health and Further Education
• Leverage, liquidity, and investment: The cyclical effects of corporate debt
• Two-way fixed effects, the two-way mundlak regression, and difference-in-differences estimators

Discussions

• I'm a Sant'anna & Zhao (2020) doubly-robust boi: doi.org/10.1016/j.je... [bsky, 3 points, 0 comments]

Related

• A difference-in-differences estimator by covariate balancing propensity score
• Doubly Robust Instrumented Difference-in-Differences
• Efficient Difference-in-Differences Estimation when Outcomes are Missing at Random
• Robust Difference-in-differences Models
• Doubly robust estimation of average treatment effect revisited