Decomposing Discrimination: Causal Mediation Analysis for AI-Driven Credit Decisions

TL;DR

This paper introduces a causal mediation analysis framework to distinguish between direct discrimination and structural inequality in AI credit decisions, providing new identification strategies and an open-source tool.

Contribution

It formalizes causal effects in credit fairness, proposes a novel estimator under treatment-induced confounding, and offers an open-source implementation for practical use.

Findings

77% of racial denial disparity is mediated by financial features influenced by structural inequality

Proposed estimator has semiparametric efficiency and robustness properties

Empirical analysis on mortgage data demonstrates the method's practical utility

Abstract

Statistical fairness metrics in AI-driven credit decisions conflate two causally distinct mechanisms: discrimination operating directly from a protected attribute to a credit outcome, and structural inequality propagating through legitimate financial features. We formalise this distinction using Pearl's framework of natural direct and indirect effects applied to the credit decision setting. Our primary theoretical contribution is an identification strategy for natural direct and indirect effects under treatment-induced confounding -- the prevalent setting in which protected attributes causally affect both financial mediators and the final decision, violating standard sequential ignorability. We show that interventional direct and indirect effects (IDE/IIE) are identified under the weaker Modified Sequential Ignorability assumption, and prove that IDE/IIE provide conservative bounds on the unidentified natural effects under monotone indirect treatment response. We propose a doubly-robust augmented inverse probability weighted (AIPW) estimator for IDE/IIE with semiparametric efficiency properties, implemented via cross-fitting. An E-value sensitivity analysis addresses residual confounding on the direct pathway. Empirical evaluation on 89,465 real HMDA conventional purchase mortgage applications from New York State (2022) demonstrates that approximately 77% of the observed 7.9 percentage-point racial denial disparity operates through financial mediators shaped by structural inequality, while the remaining 23% constitutes a conservative lower bound on direct discrimination. The open-source CausalFair Python package implements the full pipeline for deployment at resource-constrained financial institutions.