A quiet regulatory change in July has unleashed a chaotic transition in the U.S. mortgage market. When the Federal Housing Finance Agency allowed lenders to choose between traditional FICO scores and VantageScore 4.0, it looked technical. But the shift has rapidly evolved into a pricing war that’s forcing lenders to navigate competing models, new historical data, and growing legal exposure.
The stakes are higher than they appear. FICO responded to the regulatory opening by selling scores directly to lenders at roughly $10 per score, bypassing credit bureaus. Equifax slashed VantageScore 4.0 mortgage scores to $4.50 through 2027 and bundled some for free. Experian went further, offering VantageScore 4.0 at no cost for many mortgage clients.
Now, an even bigger shift is underway. In early December, FICO and FHFA reached an agreement to release historical FICO Score 10T data tied to single-family loan-level datasets. That release gives regulators, investors, and attorneys the ability to replay years of mortgage decisions under alternative scoring models—and ask why lenders didn’t act differently.
“The biggest change isn’t which score is ‘best,'” said an executive at Policy Edge AI, a regulatory intelligence platform advising lenders and investors on credit-model transitions. “It’s that everyone—regulators, investors, litigators—can now replay your book of business under alternative scores and ask why you didn’t act on what the new data shows.”
When Models Disagree, Risk Teams Scramble
FICO 10T and VantageScore 4.0 both aim to improve on the decades-old Classic FICO model, but they take different approaches. FICO 10T uses trended data—typically 24 months of balances and credit limits—to distinguish borrowers who are adding debt from those paying it down. VantageScore 4.0 uses machine learning and alternative data such as rent, utility, and cellphone payments, claiming it can score tens of millions more U.S. adults and expand access in Black and Hispanic communities.
Independent research sponsored by the American Enterprise Institute’s Housing Center found that FICO 10T flagged roughly 18 percent more defaulters than Classic FICO in a critical score band, while suggesting only marginal gains from VantageScore 4.0 over Classic FICO on several key measures.
Behind the scenes, risk teams are running scenario analyses that read more like litigation memos than pricing tables. Small movements in default probability or approval rates can translate into tens of millions of dollars in expected credit losses, capital consumption, or litigation exposure for a top-50 lender.
A New Kind of Oversight
The Consumer Financial Protection Bureau has warned that “black-box” algorithms do not relieve creditors of their duty to provide specific reasons for denials and manage algorithmic bias. A lender that clings to Classic FICO, or pivots heavily to a new model without documented analysis, could face questions about whether a less discriminatory alternative was available.
Policy Edge AI’s analysis suggests that institutions using new historical datasets early can sharpen risk segmentation and capture spread mispricings during the transition window. Lenders that treat model transition as a governed, scenario-tested program will have stronger defenses if repurchase demands or class-action suits attempt to retroactively apply new models to old decisions.
“In the coming years, the biggest risk in credit scoring may not be choosing the ‘wrong’ model,” the Policy Edge AI executive said. “It may be failing to anticipate how others will use the new data to rewrite your past decisions.”
