Credit Scores: Performance and Equity

ArXiv ID: 2409.00296 “View on arXiv”

Authors: Unknown

Abstract

Credit scores are critical for allocating consumer debt in the United States, yet little evidence is available on their performance. We benchmark a widely used credit score against a machine learning model of consumer default and find significant misclassification of borrowers, especially those with low scores. Our model improves predictive accuracy for young, low-income, and minority groups due to its superior performance with low quality data, resulting in a gain in standing for these populations. Our findings suggest that improving credit scoring performance could lead to more equitable access to credit.

Keywords: Credit scoring, Machine learning, Consumer default, Predictive modeling, Algorithmic fairness

Complexity vs Empirical Score

• Math Complexity: 6.5/10
• Empirical Rigor: 7.5/10
• Quadrant: Holy Grail
• Why: The paper employs advanced machine learning (e.g., AUC metrics, interpretability techniques) and statistical modeling, qualifying as high math complexity, while its empirical rigor is high due to the use of a large, proprietary dataset (1M households), backtesting on historical data (2004-2015), and clear performance benchmarking against a real-world credit score.

  flowchart TD
    A["Research Goal: Benchmark Credit Score<br/>Performance & Equity"] --> B["Data: Consumer Credit Records<br/>(e.g., TransUnion)"]
    B --> C["Method: Compare vs.<br/>Machine Learning Model"]
    C --> D["Process: Evaluate Misclassification<br/>& Predictive Accuracy"]
    D --> E{"Key Findings"}
    E --> F["Performance Gap:<br/>Significant misclassification<br/>in low-score borrowers"]
    E --> G["Equity Gain:<br/>ML model improves standing<br/>for young/low-income/minority groups"]
    E --> H["Outcome:<br/>Better scoring leads to<br/>more equitable credit access"]