Important questions have emerged about a transition to a multi-score framework for gauging mortgage credit risk. Recent developments include the approval of VantageScore 4.0 and FICO Score 10T for mortgages delivered to government-sponsored enterprises (GSEs). In addition, the Federal Housing Administration announced that both models could be used for mortgages it insures. The ability of lenders to choose which score to use for each individual loan, known as "vendor choice," has become a central discussion topic in the mortgage investment landscape.
As of June 2026, mortgages can be underwritten using Classic FICO and VantageScore 4.0 (Vantage). Milliman previously conducted research showing that, without any changes to the current pricing system, lender choice would lead to an adverse selection bias in default rates across credit score cohorts, which would require higher loan-level price adjustments to produce revenue-neutral pricing for the GSEs and higher mortgage rates for borrowers. This observation has been independently recreated in other industry research. Prior Milliman analysis on lender choice has focused on a two-score environment (Classic FICO and Vantage) and was performed using data published by Freddie Mac and Fannie Mae. When FICO 10T is accepted for loan delivery to the GSEs, lender choice will increase from a two-score to a three-score environment.
On July 1, 2026, the GSEs published updated loan-level datasets including historical VantageScore 4.0 and FICO 10T, which could be appended to their historical mortgage performance data. Using this new information, we discuss:
- An evaluation of the potential impact of adverse selection on mortgage default rates under a three-score environment.
- A demonstration of how the new environment might change if mortgages are exclusively underwritten using Vantage and FICO 10T in the future.
- An evaluation of the dispersion between scores for individual borrowers.
Download the full paper (PDF).