mlsanityData to 14 Oct 2026

Treatment Effect Risk: Bounds and Inference

arXiv 2201.05893 ↗First posted 15 Jan 2022stat.ME econ.EM math.OCcited by 12

Not on the map. Why?

Abstract (arXiv)

Since the average treatment effect (ATE) measures the change in social welfare, even if positive, there is a risk of negative effect on, say, some 10% of the population. Assessing such risk is difficult, however, because any one individual treatment effect (ITE) is never observed, so the 10% worst-affected cannot be identified, while distributional treatment effects only compare the first deciles within each treatment group, which does not correspond to any 10%-subpopulation. In this paper we consider how to nonetheless assess this important risk measure, formalized as the conditional value at risk (CVaR) of the ITE-distribution. We leverage the availability of pre-treatment covariates and characterize the tightest-possible upper and lower bounds on ITE-CVaR given by the covariate-conditional average treatment effect (CATE) function. We then proceed to study how to estimate these bounds efficiently from data and construct confidence intervals. This is challenging even in randomized experiments as it requires understanding the distribution of the unknown CATE function, which can be very complex if we use rich covariates so as to best control for heterogeneity. We develop a debiasing method that overcomes this and prove it enjoys favorable statistical properties even when CATE and other nuisances are estimated by black-box machine learning or even inconsistently. Studying a hypothetical change to French job-search counseling services, our bounds and inference demonstrate a small social benefit entails a negative impact on a substantial subpopulation.

Key facts. Cited by 12 papers in the map (data through 14 Oct 2026). Builds on 25 of its 67 references that are in the map. First posted on arXiv 15 Jan 2022.

Builds on

25 of 67 references.

References · 25

  1. 2016"Why Should I Trust You?": Explaining the Predictions of Any Classifiercited by 5,070
  2. 2015Estimation and Inference of Heterogeneous Treatment Effects using Random Forestscited by 600
  3. 2017Meta-learners for Estimating Heterogeneous Treatment Effects using Machine Learningcited by 478
  4. 2017Preventing Fairness Gerrymandering: Auditing and Learning for Subgroup Fairnesscited by 384
  5. 2017Quasi-Oracle Estimation of Heterogeneous Treatment Effectscited by 332
  6. 2015Data-driven Distributionally Robust Optimization Using the Wasserstein Metric: Performance Guarantees and Tractable Reformulationscited by 291
  7. 2015Recursive Partitioning for Heterogeneous Causal Effectscited by 255
  8. 2020Fairness without Demographics through Adversarially Reweighted Learningcited by 251
  9. 2018Fairness Under Unawareness: Assessing Disparity When Protected Class Is Unobservedcited by 94
  10. 2019Assessing Algorithmic Fairness with Unobserved Protected Class Using Data Combinationcited by 79
  11. 2017Balanced Policy Evaluation and Learningcited by 52
  12. 2018Residual Unfairness in Fair Machine Learning from Prejudiced Datacited by 49
13 more
  1. 2020Experimental Design in Two-Sided Platforms: An Analysis of Biascited by 39
  2. 2019The Sensitivity of Counterfactual Fairness to Unmeasured Confoundingcited by 35
  3. 2013Program Evaluation and Causal Inference with High-Dimensional Datacited by 29
  4. 2016Recursive Partitioning for Personalization using Observational Datacited by 28
  5. 2021Doubly-Valid/Doubly-Sharp Sensitivity Analysis for Causal Inference with Unmeasured Confoundingcited by 28
  6. 2014Robust Sample Average Approximationcited by 25
  7. 2019Localized Debiased Machine Learning: Efficient Inference on Quantile Treatment Effects and Beyondcited by 23
  8. 2017Policy Learning with Observational Datacited by 19
  9. 2014Randomization Inference for Treatment Effect Variationcited by 9
  10. 2016Decomposing Treatment Effect Variationcited by 8
  11. 2016Evaluating the Impact of Treating the Optimal Subgroupcited by 5
  12. 2019Assessing Disparate Impacts of Personalized Interventions: Identifiability and Boundscited by 3
  13. 2020Rejoinder: New Objectives for Policy Learningcited by 1

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2022: 3; 2023: 5; 2024: 3; 2025: 0; 2026: 1.

12 papers.

  1. 2024Applied Causal Inference Powered by ML and AIcited by 30
  2. 2023B-Learner: Quasi-Oracle Bounds on Heterogeneous Causal Effects Under Hidden Confoundingcited by 28
  3. 2023Bounds on Representation-Induced Confounding Bias for Treatment Effect Estimationcited by 20
  4. 2022Robust and Agnostic Learning of Conditional Distributional Treatment Effectscited by 18
  5. 2022What's the Harm? Sharp Bounds on the Fraction Negatively Affected by Treatmentcited by 15
  6. 2022Partial Identification of Dose Responses with Hidden Confounderscited by 7
  7. 2023Ensembled Prediction Intervals for Causal Outcomes Under Hidden Confoundingcited by 7
  8. 2024Policy Learning for Balancing Short-Term and Long-Term Rewardscited by 7
  9. 2024Distributionally Robust Policy Learning under Concept Driftscited by 6
  10. 2023Policy Learning with Distributional Welfarecited by 5
  11. 2023Treatment Effects in Extreme Regimescited by 1
  12. 2026GAAVI: Global Asymptotic Anytime Valid Inference for the Conditional Mean Functioncited by 0