Data-Driven Covariate Selection for Nonparametric and Cycle-Agnostic Causal Effect Estimation

2026-05-07

AI Summary

This study addresses the limitations of existing causal effect estimation methods, which often rely on acyclicity assumptions and thus struggle with observational data involving latent confounding and feedback loops. The authors propose a local, data-driven covariate selection approach that identifies valid adjustment sets using only conditional independence information, without requiring the learning of a global causal graph or assuming an acyclic structure. They establish, for the first time, the correctness and completeness of this method in cyclic causal models by leveraging the invariance of conditional independences under σ-acyclification, thereby constructing a unified cycle-agnostic framework for causal inference. Empirical results demonstrate that the method remains robust and effective in both cyclic and acyclic settings, enabling nonparametric estimation of causal effects.

Abstract

Estimating causal effects from observational data requires identifying valid adjustment sets. This task is especially challenging in realistic settings where latent confounding and feedback loops are present. Existing approaches typically assume acyclicity or rely on global causal structure learning, limiting applicability and computational efficiency. In this work, we study a local, data-driven method for covariate selection based on conditional independence information. While this method is known to be sound and complete in acyclic causal models, its validity in the presence of cycles has remained unclear. Our main contribution is to show that these guarantees extend to cyclic causal models. In particular, our result relies on the invariance of conditional independence assertions under $σ$-acyclification. These findings establish a unified, cycle-agnostic perspective on covariate selection and causal effect estimation, showing that the method applies across cyclic and acyclic settings without modification. Empirically, we validate this on extensive synthetic data, showing reliable performance in cyclic causal models.

Problem

• causal effect estimation
• covariate selection
• cyclic causal models
• latent confounding
• conditional independence

Innovation

• cycle-agnostic
• covariate selection
• causal effect estimation
• conditional independence
• σ-acyclification