Juan Carlos Escanciano, Alejandro Puerta-Cuartas
arXiv 15 Oct 2026 · Econometrics
arXiv:2610.21756 · PDF · Extracted main text
Many parameters of interest in economics depend on outcomes generated by machine learning (ML).These generated outcomes are subsequently used in standard econometric estimation and typically treated as observed, a practice that threatens the validity of downstream inference. Empirically relevant cases include dynamic treatment effects under sequential exogeneity, long-term effects with surrogates, the partially linear model with an ML-predicted outcome, productivity persistence with an unrestricted production function and constructed instruments, and outcomes residualized of confounding variation. This paper develops a three-step locally robust GMM framework for valid inference when the dependent variable is generated by ML. We construct a debiased moment condition that augments the identifying moment with first- and second-step influence functions, making inference on the parameter of interest locally insensitive to regularization and model selection biases at both steps. We establish consistency and asymptotic normality of the proposed estimator and illustrate its finite-sample performance through Monte Carlo simulations. We also apply the method to U.S. mortgage data to study conditional Black--White disparities in a race-blind ML default-risk score. The score predicts subsequent default well overall, but the conditional disparity embedded in the score is much smaller than the corresponding disparity in realized serious delinquency; accounting for first-step estimation also raises the naive out-of-sample estimate of the score disparity.
appendix boundary found by appendix_command · 80% of the source is main text. Read the extracted text to check this.
The works this paper leans on most, across its whole bibliography — not restricted to papers in our corpus. Ranked by composite intensity, which combines how often a work is mentioned, how many sections mention it, and how much of that falls in the main text rather than the appendix.
| Reference | Intensity | Mentions | Sections | Main text | |
|---|---|---|---|---|---|
| 1 | Escanciano, Juan Carlos and Pérez-Izquierdo, Telmo (2023) Automatic Locally Robust Estimation with Generated Regressors self | 1.000 | 6 | 3 | 100% |
| 2 | Chernozhukov, Victor and Escanciano, Juan Carlos and Ichimura, Hideh… (2022) Locally robust semiparametric estimation self | 1.000 | 5 | 4 | 100% |
| 3 | Chernozhukov, Victor and Newey, Whitney and Singh, Rahul and Syrgkan… (2023) Automatic Debiased Machine Learning for Dynamic Treatment Effects and General Nested Functionals | 0.874 | 7 | 2 | 100% |
| 4 | Athey, Susan and Chetty, Raj and Imbens, Guido W. and Kang, Hyunseung (2025) The Surrogate Index: Combining Short-Term Proxies to Estimate Long-Term Treatment Effects More Rapidly and Precisely | 0.737 | 3 | 2 | 100% |
| 5 | Blundell, Richard and Pistaferri, Luigi and Preston, Ian (2008) Consumption inequality and partial insurance | 0.644 | 2 | 2 | 100% |
| 6 | Brunori, Paolo and Ferreira, Francisco H. G. and Neidhöfer, Guido (2025) Inequality of opportunity and intergenerational persistence in Latin America | 0.644 | 2 | 2 | 100% |
| 7 | Chernozhukov, Victor and Chetverikov, Denis and Demirer, Mert and Du… (2018) Double/debiased machine learning for treatment and structural parameters | 0.644 | 2 | 2 | 100% |
| 8 | Ackerberg, Daniel A. and Caves, Kevin and Frazer, Garth (2015) Identification Properties of Recent Production Function Estimators | 0.405 | 1 | 1 | 100% |
| 9 | Angelopoulos, Anastasios N. and Bates, Stephen and Fannjiang, Clara… (2023) Prediction-Powered Inference | 0.405 | 1 | 1 | 100% |
| 10 | Bodory, Hugo and Huber, Martin and Lafférs, Luká s (2022) Evaluating (weighted) dynamic treatment effects by double machine learning | 0.405 | 1 | 1 | 100% |
Showing the top 10 of 26 scored citations.