Abstract:Many recent developments in causal inference, and functional estimation problems more generally, have been motivated by the fact that classical one-step (first-order) debiasing methods, or their more recent sample-split double machine-learning avatars, can outperform plugin estimators under surprisingly weak conditions. These first-order corrections improve on plugin estimators in a black-box fashion, and consequently are often used in conjunction with powerful off-the-shelf estimation methods. These first-order methods are however provably suboptimal in a minimax sense for functional estimation when the nuisance functions live in Holder-type function spaces. This suboptimality of first-order debiasing has motivated the development of "higher-order" debiasing methods. The resulting estimators are, in some cases, provably optimal over Holder-type spaces, but both the estimators which are minimax-optimal and their analyses are crucially tied to properties of the underlying function space. In this paper we investigate the fundamental limits of structure-agnostic functional estimation, where relatively weak conditions are placed on the underlying nuisance functions. We show that there is a strong sense in which existing first-order methods are optimal. We achieve this goal by providing a formalization of the problem of functional estimation with black-box nuisance function estimates, and deriving minimax lower bounds for this problem. Our results highlight some clear tradeoffs in functional estimation -- if we wish to remain agnostic to the underlying nuisance function spaces, impose only high-level rate conditions, and maintain compatibility with black-box nuisance estimators then first-order methods are optimal. When we have an understanding of the structure of the underlying nuisance functions then carefully constructed higher-order estimators can outperform first-order estimators.

Inference in models with partially identified control functions

Inference on Strongly Identified Functionals of Weakly Identified Functions

Causal Inference under Unmeasured Confounding with Negative Controls: A Minimax Learning Approach.

A UNIFIED STUDY OF NONPARAMETRIC INFERENCE FOR MONOTONE FUNCTIONS

Quasi-Bayesian Estimation and Inference with Control Functions

Inference In Partially Identified Panel Data Models With Interactive Fixed Effects

Inference in Semiparametric Conditional Moment Models with Partial Identification

Model-Agnostic Covariate-Assisted Inference on Partially Identified Causal Effects

Subvector Inference in Partially Identified Models with Many Moment Inequalities

Set-Valued Control Functions

Contrasting Identifying Assumptions of Average Causal Effects: Robustness and Semiparametric Efficiency

On extending Powell, Stock, and Stoker (1989) to indexes with functionally dependent covariates

Synthetic Control Method: Inference, Sensitivity Analysis and Confidence Sets

Causality for Complex Continuous-time Functional Longitudinal Studies with Dynamic Treatment Regimes

Inference for Synthetic Controls via Refined Placebo Tests

Assumption-Lean Post-Integrated Inference with Negative Control Outcomes

The Fundamental Limits of Structure-Agnostic Functional Estimation

Multiply robust causal inference with double-negative control adjustment for categorical unmeasured confounding

Control Function Instrumental Variable Estimation of Nonlinear Causal Effect Models

Bias Correction and Robust Inference in Semiparametric Models

Program Evaluation and Causal Inference with High-Dimensional Data