Skip to main content
WHITE PAPER

Fannie Mae and Freddie Mac: A comparison of internal mortgage risk score models

18 November 2026

How do Freddie Mac and Fannie Mae’s GSE scores compare with traditional credit scores?

Credit score modernization efforts have provided multiple datasets for analysis related to mortgage underwriting. Freddie Mac and Fannie Mae have released back-filled data allowing the industry to compare FICO 10T, VantageScore, and internal scores from both Freddie Mac and Fannie Mae (referred to as “the GSE score” in this analysis) relative to Classic FICO. Since the early 1990s, Classic FICO was the only credit score accepted by Fannie Mae and Freddie Mac (the government-sponsored enterprises, or GSEs) for the mortgages they purchase.

The GSE scores are proprietary scores developed by Freddie Mac and Fannie Mae to evaluate mortgage default risk. In September 2026, both GSEs released datasets that include these internal risk scores, making it possible to analyze borrower distributions by risk score and default rate performance. Lenders do not use and did not have access to the internal scores for mortgage underwriting. However, the scores offer insight into the scoring engines the GSEs have developed and allow for a comparison of the two enterprises’ approaches to assessing mortgage credit risk relative to commercial credit scoring models. (Commercial mortgage scoring models are estimated and calibrated to all types of mortgage credit, while the GSE scores are specific to mortgage trade lines.)

This paper summarizes Milliman’s analysis on the GSE scores. Specifically, it presents the distributions of the Fannie Mae and Freddie Mac risk scores, examines how each score varies with loan characteristics, compares default rates by score range between the two GSEs, and evaluates the relationship between the GSE scores and Classic FICO. Together, these results summarize the internal scores and show how their performance compares with that of other traditional credit scores.

We explore Fannie Mae and Freddie Mac’s internal scores across score distribution, relationship to loan characteristics, and default rate performance. We conclude that the GSEs’ internal scoring engines perform well at identifying high-risk mortgages and have stronger fit statistics relative to commercial credit-scoring models.

Our key findings include the following:

  • Fannie Mae (400 to 950) and Freddie Mac (240 to 1,000) internal risk scores span different scales, and the same raw score carries different risk at each GSE
  • Neither GSE uses the full official scoring range in practice; actual observed scores ranged from 516 to 898 for Fannie Mae and 372 to 960 for Freddie Mac
  • Both GSE score distributions are concentrated in the upper score bands and exhibit negative skewness, indicating a predominance of lower-risk loans while maintaining sensitivity to higher-risk segments through an extended lower-score tail
  • The divergence between Fannie Mae and Freddie Mac scoring approaches is most evident among higher-risk loan profiles, particularly for mortgages with elevated loan-to-value ratios, where Freddie Mac assigns relatively more favorable risk assessments
  • Both GSE scores effectively rank risk, with default rates declining consistently as scores increase; the lowest score deciles account for roughly two-thirds of all defaults
  • Freddie Mac's model shows slightly stronger risk separation overall, especially outside the COVID-19 period
  • Both GSE internal risk scores outperform commercial credit models in predicting mortgage defaults

Download the full paper (PDF).


About the Author(s)

Ricardo Nunez Magana

We’re here to help