treatment effect estimation

This paper addresses counterfactual estimation under unobserved confounding in data-rich settings. Methodologically, it proposes a unified framework integrating structural causal models (SCMs) with latent factor models (LFMs), formally bridging graphical models and the potential outcomes paradigm. It establishes identifiability conditions for the average treatment effect (ATE), average treatment effect on the treated (ATT), and average treatment effect on the untreated (ATU), and derives general consistency conditions for estimation via principal component regression (PCR), latent factor modeling, and nonparametric smoothness analysis. The key contribution is a theoretical proof—under mild smoothness assumptions—that PCR consistently estimates all three average treatment effects, substantially relaxing conventional requirements of linearity, low dimensionality, and strong functional-form restrictions. This yields a robust, scalable solution for causal inference in high-dimensional observational data with large sample sizes.

Estimating causal effects from observational data requires selecting appropriate control and instrumental variables that satisfy causal identification conditions—a challenging task often reliant on strong domain knowledge or ad hoc assumptions. Method: This paper proposes the first end-to-end joint learning framework that automatically identifies valid combinations of control and instrumental variables. Grounded in conditional independence testing, the method integrates nonparametric dependence measures with structural search optimization, ensuring statistical consistency in variable selection under mild regularity conditions. Contribution/Results: Unlike conventional approaches requiring prespecified variable sets or strong prior assumptions, our framework is fully data-driven. In simulations, it achieves significantly higher variable identification accuracy. Empirically, applied to the Job Corps study, its estimated treatment effect closely aligns with results from the randomized controlled trial—demonstrating both validity and robustness in real-world causal inference.

Estimating causal effects in observational studies is challenging when unmeasured confounding violates the backdoor criterion. Method: This paper develops an average causal effect (ACE) estimation framework based on the front-door criterion, leveraging mediators to circumvent unobserved confounding paths. It systematically integrates targeted minimum loss estimation (TMLE) theory with front-door identification, accommodating binary, continuous, and multivariate mediators. The resulting estimator is √n-consistent, doubly robust, and semiparametrically efficient. The method combines data-adaptive machine learning (e.g., gradient boosting, random forests) with functional modeling of the front-door formula. Contribution/Results: Empirical evaluation on simulated data and real Finnish education–income data demonstrates that the proposed approach significantly improves finite-sample estimation accuracy and stability over existing methods, enabling reliable quantification of the causal impact of early academic performance on adult income.

This paper addresses the unreliability of causal and predictive parameter estimation under covariate shift. We propose a fully automated debiasing machine learning framework that eliminates regularization bias solely through parameter definition—without requiring explicit bias modeling. Our approach innovatively integrates training and target data within a unified debiasing mechanism, combining data fusion, high-dimensional statistical inference, doubly robust estimation, and the difference-in-differences (DID) principle—all under an unconfoundedness assumption. We establish theoretical guarantees of consistency and asymptotic normality. In simulation studies and an empirical analysis of minimum wage effects on teenage employment, our method reduces estimation bias by over 40% on average compared to benchmark approaches, while substantially improving estimation accuracy and robustness.

This study addresses selection bias in estimating long-term causal effects—such as graduation rates—from observational studies. We propose a novel control function approach that leverages experimental estimates of treatment effects on short-term outcomes (e.g., eighth-grade test scores) to correct for unobserved confounding in large-scale administrative observational data. Our method integrates insights from difference-in-differences estimation, covariate balancing, and cross-sample effect calibration, enabling the first systematic correction based on heterogeneity in short-term treatment effects. By bridging randomized experiments and observational datasets, the framework jointly preserves internal validity from experiments and external representativeness from administrative records, overcoming inferential limitations inherent to single-data-source designs. Empirical validation using the STAR randomized experiment and New York State school administrative data demonstrates substantial improvements in both accuracy and external validity of estimated causal effects of class size on academic performance.

This study addresses the limitations of traditional control-based causal inference methods—such as matching and difference-in-differences—in settings characterized by pervasive or structurally ambiguous spillover effects, where reliance on uncontaminated control units impedes accurate identification of both average direct and spillover effects. Within the potential outcomes framework, this work provides the first systematic comparison between control-based and prediction-based counterfactual approaches—including interrupted time series and machine learning control—in terms of their identification capabilities. Through simulation and empirical analyses, the authors demonstrate that in environments with widespread interference, prediction-based methods can more reliably estimate certain causal parameters over short horizons, circumventing the stringent assumption of unperturbed units and thereby offering a promising alternative for causal inference under complex interference.

This study addresses the core challenge in causal inference of accurately identifying true causal effects in settings characterized by high-dimensional observational data and endogenous selection. Leveraging both experimental data from a large technology company’s new feature rollout and observational data from users’ self-selection into the feature, this work provides the first joint validation of causal machine learning methods in a real-world product environment. By integrating propensity score modeling, doubly robust estimation, and high-dimensional covariate adjustment, the research demonstrates that careful modeling substantially improves the accuracy of causal effect estimates. The findings not only confirm the practical feasibility of modern causal inference techniques but also distill a set of best practices for enhancing estimation credibility, offering an empirical benchmark and actionable guidance for high-dimensional causal inference.

This study addresses the challenge in causal inference of accurately identifying treatment effects when using machine learning to predict outcome variables, compounded by the absence of effective criteria for model selection. The authors decompose prediction into three components: between-unit variation, within-unit temporal variation, and counterfactual treatment effects. They demonstrate that only the first two components are estimable from observed data and, for the first time, formally establish that the counterfactual component governs the accuracy of causal identification. To address this, they propose using within-unit temporal prediction accuracy as a structural proxy for this unobservable component, enabling model diagnostics and selection. Within a panel data framework that integrates causal theory with machine learning evaluation techniques, the proposed metric is validated on synthetic data and shown—under plausible assumptions—to yield approximately unbiased estimates of treatment effects.

This study addresses the challenge of causal inference when the outcome variable is latent and can only be indirectly measured through multiple imperfect proxies. Conventional methods are vulnerable to measurement incomparability across studies and model misspecification. To overcome these limitations, the authors propose a design-oriented nonparametric framework that identifies and estimates the average treatment effect on the latent outcome under randomized experiments. The key innovation lies in constructing an identifiable nonparametric bridge function that flexibly accommodates differences in measurement systems across studies and nonlinear relationships among proxies, without imposing strong parametric assumptions on the measurement model. Coupled with a debiased estimation procedure, the proposed method substantially outperforms benchmarks such as principal component analysis and inverse covariance weighting in simulations, accurately recovering comparable and consistent causal effects on the latent variable while eliminating spurious cross-study heterogeneity.

This study addresses the challenge of identifying and estimating causal effects under network interference, where an individual’s treatment may spill over and affect others’ outcomes. The authors propose a solution based on a linear outcome model that yields unbiased and consistent estimates of both binary and continuous treatment effects when the interference structure is known or partially known. The approach accommodates both fixed and random interference network specifications and innovatively eliminates interference-induced bias while remaining compatible with standard linear regression software. It also conveniently allows for the incorporation of random effects and heteroskedasticity- and autocorrelation-consistent (HAC) standard errors. Numerical simulations and empirical analyses demonstrate the method’s effectiveness in bias correction and practical applicability.