The Federal Housing Finance Agency (FHFA) is back to exploring the use of third-party credit score models to be used by Fannie Mae and Freddie Mac in the underwriting of mortgage loans.
The agency – which regulates government sponsored enterprises (GSEs) Fannie Mae and Freddie Mac – has been exploring the use alternative credit score models for several years now.
In July, however, the FHFA announced that it was suspending its plans for using alternative credit score models due to the passage of the Economic Growth, Regulatory Relief, and Consumer Protection Act (S.2155) in May.
Section 310 of the Act requires the FHFA to define, through rulemaking, the standards and criteria the GSEs will use to validate credit score models.
So, implementation of the Act essentially hit a reset button on the FHFA’s plans.
Now, the agency is back to seeking public comment on a proposed rule that would establish standards and criteria for the validation and approval of third-party credit score models used by the GSEs.
The proposed rule would establish a four-phase process for the companies to validate and approve credit score models, including solicitation of applications from credit score model developers; review of submitted applications; credit score assessment; and enterprise business assessment.
In December 2017, the FHFA announced that it was seeking feedback on the “operational and competition considerations” of changing Fannie Mae and Freddie Mac’s current credit score requirements.
At the time, the agency said the GSEs were evaluating three different credit score models, including the Classic FICO model, which they currently use, as well as the newer FICO 9 and VantageScore 3.0 models.
The FHFA had planned to issue a final decision this year but in February announced it had extended the deadline for feedback to March 30 from Feb. 20.
Mel Watt, outgoing director of the FHFA, last year said whether or not the GSEs should even use alternate credit scoring models is “a tough call.”
Importantly, the Economic Growth, Regulatory Relief, and Consumer Protection Act does not require the GSEs to use alternate credit score models; rather, it simply opens the door for them to do so if they choose.
The GSEs are free to develop and set their own credit score requirements under the Act, as well.
Meanwhile, Fannie Mae and Freddie Mac have each already upgraded their underwriting platforms to support alternative credit sources.
“Each Enterprise has updated its respective AUS in recent years to process loans for borrowers who lack a credit score,” the FHFA says in its proposal. “In September 2016, Fannie Mae upgraded Desktop Underwriter (DU) with the capability to underwrite loan applications where both the borrower and co-borrower lack a credit score. In June 2017, Freddie Mac updated Loan Product Advisor (LPA) with the same capability to underwrite both borrower and co-borrowers who lack a credit score.”
“Development of the ‘no score AUS’ reduces the significance of third party credit scores within each Enterprise’s AUS,” the FHFA adds. “The ability of an Enterprise AUS to assess borrowers who lack a credit score is an additional consideration in assessing the impact of the use of any credit score model on access to credit.”
In its proposal, the FHFA notes that the use of alternate credit score models by the GSEs could have wide ranging implications for the mortgage origination and secondary markets.
It cites feedback that it received during the last comment period in December 2017, when one respondent said, “Changes to Enterprise credit score requirements could have widely-felt implications for borrower access to credit, origination costs in the primary mortgage market, the ability to fully analyze and properly price mortgage credit risk, and liquidity in the secondary mortgage market.”
However, the proposal also states that these newer credit score models should prove to be more accurate, as they bring in more payment history data than current models – such as rent history and utilities payments – to support whether a borrower is creditworthy.
“Newer credit score models should statistically outperform legacy credit score models for several reasons,” the proposal states. “First, newer credit score models incorporate borrower information that was not available when the legacy credit score models were designed and estimated. Second, newer credit score models are estimated (or ‘trained’) on more recent borrower credit histories. More recent historical borrower behaviors better represent current borrower behaviors than older credit histories.”