Structural Gender Bias in Credit Scoring: Proxy Leakage

TL;DR

This study scrutinizes structural gender bias in credit scoring using the Taiwan Credit Default dataset and challenges the doctrine of fairness through blindness. It combines SHAP-based explainability with adversarial inverse modeling to reveal that non-sensitive features leakage of gender proxies sustains discrimination despite removing explicit gender attributes, achieving a gender-prediction ROC-AUC of $0.65$. While standard fairness metrics appear near-ideal, the analysis demonstrates that demographic features and even purely financial proxies such as LIMIT_BAL and BILL_AMT_1 carry gendered information that shapes model decisions. The work argues for causal-aware modeling and fairness-aware training as essential steps toward genuine accountability in financial AI, and outlines directions for formal causal frameworks and cross-dataset validation.

Abstract

As financial institutions increasingly adopt machine learning for credit risk assessment, the persistence of algorithmic bias remains a critical barrier to equitable financial inclusion. This study provides a comprehensive audit of structural gender bias within the Taiwan Credit Default dataset, specifically challenging the prevailing doctrine of "fairness through blindness." Despite the removal of explicit protected attributes and the application of industry standard fairness interventions, our results demonstrate that gendered predictive signals remain deeply embedded within non-sensitive features. Utilizing SHAP (SHapley Additive exPlanations), we identify that variables such as Marital Status, Age, and Credit Limit function as potent proxies for gender, allowing models to maintain discriminatory pathways while appearing statistically fair. To mathematically quantify this leakage, we employ an adversarial inverse modeling framework. Our findings reveal that the protected gender attribute can be reconstructed from purely non-sensitive financial features with an ROC AUC score of 0.65, demonstrating that traditional fairness audits are insufficient for detecting implicit structural bias. These results advocate for a shift from surface-level statistical parity toward causal-aware modeling and structural accountability in financial AI.

Figures (4)

• Figure 1: Methodology Outline
• Figure 2: Fairness and accuracy metrics across twelve experimental configurations
• Figure 3: Normalized SHAP values for demographic features in the primary credit default model
• Figure 4: Normalized SHAP values for gender prediction using demographic proxies