Abstract:Doubly robust estimators with cross-fitting have gained popularity in causal inference due to their favorable structure-agnostic error guarantees. However, when additional structure, such as Hölder smoothness, is available then more accurate "double cross-fit doubly robust" (DCDR) estimators can be constructed by splitting the training data and undersmoothing nuisance function estimators on independent samples. We study a DCDR estimator of the Expected Conditional Covariance, a functional of interest in causal inference and conditional independence testing, and derive a series of increasingly powerful results with progressively stronger assumptions. We first provide a structure-agnostic error analysis for the DCDR estimator with no assumptions on the nuisance functions or their estimators. Then, assuming the nuisance functions are Hölder smooth, but without assuming knowledge of the true smoothness level or the covariate density, we establish that DCDR estimators with several linear smoothers are semiparametric efficient under minimal conditions and achieve fast convergence rates in the non-$\sqrt{n}$ regime. When the covariate density and smoothnesses are known, we propose a minimax rate-optimal DCDR estimator based on undersmoothed kernel regression. Moreover, we show an undersmoothed DCDR estimator satisfies a slower-than-$\sqrt{n}$ central limit theorem, and that inference is possible even in the non-$\sqrt{n}$ regime. Finally, we support our theoretical results with simulations, providing intuition for double cross-fitting and undersmoothing, demonstrating where our estimator achieves semiparametric efficiency while the usual "single cross-fit" estimator fails, and illustrating asymptotic normality for the undersmoothed DCDR estimator.

DecoR: Deconfounding Time Series with Robust Regression

Sequential Deconfounding for Causal Inference with Unobserved Confounders

The DeCAMFounder: Non-Linear Causal Discovery in the Presence of Hidden Variables

Spectral Deconfounding for High-Dimensional Sparse Additive Models

Deconfounding Time Series Forecasting

The Decaying Missing-at-Random Framework: Doubly Robust Causal Inference with Partially Labeled Data

Recovering Latent Confounders from High-dimensional Proxy Variables

Spectral Representation for Causal Estimation with Hidden Confounders

Nonlinear causal discovery with confounders

Modeling Interference Via Symmetric Treatment Decomposition

Estimating Treatment Effects from Irregular Time Series Observations with Hidden Confounders

On Robust Inference in Time Series Regression

Conditional Instrumental Variable Regression with Representation Learning for Causal Inference

DOUBLY DEBIASED LASSO: HIGH-DIMENSIONAL INFERENCE UNDER HIDDEN CONFOUNDING.

Causal Discovery of Linear Non-Gaussian Causal Models with Unobserved Confounding

DARING: Differentiable Causal Discovery with Residual Independence

A Decorrelating and Debiasing Approach to Simultaneous Inference for High-Dimensional Confounded Models

Disentangled Counterfactual Recurrent Networks for Treatment Effect Inference over Time

Doubly Robust Inference in Causal Latent Factor Models

Two-way Deconfounder for Off-policy Evaluation in Causal Reinforcement Learning

Double Cross-fit Doubly Robust Estimators: Beyond Series Regression