Kernel Instrumental Variable Regression

TL;DR

This paper introduces Kernel Instrumental Variable Regression (KIV), a nonparametric method extending 2SLS to model nonlinear relationships in causal inference, with proven consistency and superior empirical performance.

Contribution

It proposes KIV, a novel nonparametric IV regression method using RKHSs, with theoretical guarantees and improved accuracy over existing methods.

Findings

KIV outperforms state-of-the-art nonparametric IV methods in experiments.

Theoretical proof of KIV's consistency under mild conditions.

Derivation of optimal convergence rates for KIV.

Abstract

Instrumental variable (IV) regression is a strategy for learning causal relationships in observational data. If measurements of input X and output Y are confounded, the causal relationship can nonetheless be identified if an instrumental variable Z is available that influences X directly, but is conditionally independent of Y given X and the unmeasured confounder. The classic two-stage least squares algorithm (2SLS) simplifies the estimation problem by modeling all relationships as linear functions. We propose kernel instrumental variable regression (KIV), a nonparametric generalization of 2SLS, modeling relations among X, Y, and Z as nonlinear functions in reproducing kernel Hilbert spaces (RKHSs). We prove the consistency of KIV under mild assumptions, and derive conditions under which convergence occurs at the minimax optimal rate for unconfounded, single-stage RKHS regression. In doing so, we obtain an efficient ratio between training sample sizes used in the algorithm's first and second stages. In experiments, KIV outperforms state of the art alternatives for nonparametric IV regression.