Dual Instrumental Variable Regression

Abstract

We present a novel algorithm for non-linear instrumental variable (IV) regression, DualIV, which simplifies traditional two-stage methods via a dual formulation. Inspired by problems in stochastic programming, we show that two-stage procedures for non-linear IV regression can be reformulated as a convex-concave saddle-point problem. Our formulation enables us to circumvent the first-stage regression which is a potential bottleneck in real-world applications. We develop a simple kernel-based algorithm with an analytic solution based on this formulation. Empirical results show that we are competitive to existing, more complicated algorithms for non-linear instrumental variable regression.

arXiv commentAdvances in Neural Information Processing Systems 33 (NeurIPS 2020)

References in corpus (3)

• Nonparametric methods for inference in the presence of instrumental variables
• Deep Generalized Method of Moments for Instrumental Variable Analysis
• Bayesian Deconditional Kernel Mean Embeddings

Cited by in corpus (9)

• Auto IV: Counterfactual Prediction via Automatic Instrumental Variable Decomposition
• Minimax Estimation of Conditional Moment Models
• Learning Deep Features in Instrumental Variable Regression
• Biased Stochastic First-Order Methods for Conditional Stochastic Optimization and Applications in Meta Learning
• Kernel Conditional Moment Test via Maximum Moment Restriction
• Provably Efficient Neural Estimation of Structural Equation Model: An Adversarial Approach
• On Instrumental Variable Regression for Deep Offline Policy Evaluation
• Instrumental Variable Estimation for Compositional Treatments
• Quasi-Bayesian Dual Instrumental Variable Regression