Analysis of 44.7 million mortgages shows differences in calibration, stress-period behavior, and pricing alignment within the current credit scoring system

‍

WASHINGTON, D.C. – Prosperity Now today released, From Scores to Signals: Understanding Risk Classification in a Multi-Score Mortgage Market, a new analysis of approximately 44.7 million mortgages examining how credit scoring models function within the U.S. housing finance system and what their differences may mean as the market moves toward a multi-score framework.

‍

The report finds that while leading credit scoring models consistently rank borrowers from higher to lower risk, they differ in how that risk is segmented across score ranges and how borrowers perform under stress conditions. These differences are most visible near key decision thresholds and during periods of economic disruption, where how risk is defined and distributed can shape how borrowers are evaluated within the system.

‍

Drawing on loan-level data from Fannie Mae and Freddie Mac from 2013 through 2023, the analysis compares VantageScore 4.0 and FICO Classic Score across both stable and stress environments, including the COVID-19 period. Across the full dataset, both models demonstrate strong and consistent rank-ordering of risk. The more meaningful differences emerge in calibration, segmentation, and how borrower outcomes diverge under stress.

‍

During the COVID-19 period, when forbearance programs temporarily suppressed observed defaults, differences in how VantageScore 4.0 and FICO Classic Score segment risk became more visible. While both models maintained consistent risk ranking, borrowers in the lowest-risk segments under VantageScore 4.0 exhibited lower post-forbearance default rates than comparable segments under FICO Classic Score, highlighting differences in how each model defines and separates lower-risk borrowers.  

‍

The analysis also finds that pricing within the current system does not move proportionally with observed risk. Variation in default rates across score segments is substantially larger than variation in interest rates, reflecting structural features of a system built around a single primary score.

‍

“This analysis shows that the question is not whether risk can be identified, it can,” said Marisa Calderon, President and CEO of Prosperity Now. “The difference is in how that risk is defined and segmented, particularly under stress, and how those differences are reflected across the system. As the market moves toward a multi-score framework, understanding how these signals are interpreted and applied will be increasingly important for lenders, investors, and ultimately for borrowers.”

‍

The report underscores that credit scoring functions as a system-level input within the mortgage market, influencing how borrowers are evaluated and how risk is distributed. As the market moves toward incorporating multiple scoring approaches, the findings highlight the importance of how different risk signals are interpreted and applied in practice.

‍

Read the full report here.  

‍

###

‍