Score modernization taking shape but ROI tough to quantify

Fannie Mae and Freddie Mac are set to accept newer credit scores that incorporate nontraditional data, a long-anticipated move that could expand access to home loans — but not without raising new questions and costs.

That was the conclusion some experts like James Bennison, former executive vice president and head of alternative markets at Arch Mortgage Insurance, reached at a recent Urban Institute's conference on the new commitment to adding Vantagescore 4.0 as an option.

"It's hard to argue that it wouldn't have some benefit," he said. "But in the short run (depending on how it's implemented), if it results in less certainty as to the predictive value of the credit scores that investors are using to evaluate the risks that they're taking, there's going to be an increase in cost."

The extent to which there are short-term expenses and operational change, and what the payoff will be, will guide policy and help mortgage firms determine when and whether to use alternative scores as two influential government-related mortgage buyers make them available.

What's known and yet to be determined about the score change

Some pundits at the meeting appreciated that Bill Pulte, director of Fannie and Freddie's oversight agency the Federal Housing Finance Agency, is prioritizing the legislative score modernization mandate and clarified some details with an FAQ that states it will be updated regularly. However, they'd also like details that'd help with fiscal calculations.

The FAQ clarifies lenders won't initially be able to submit multiple scores for a loan, and must pick the traditional classic FICO or the newer Vantagescore 4.0. Formal incorporation into the selling guidelines that makes the choice of 4.0 possible in practice was pending at deadline.

(Pulte has said he uses his social media account to get news to the market as quickly as possible and has noted that, in doing so, his announcements may come out before formal guidance.)

Lenders will "determine which credit score model to use on each loan they deliver," and the previously planned addition of FICO's advanced score, 10T is still in play, according to the FAQ, which notes "similar implementation efforts are underway" for it.

That appears to be in line with the distribution of past data that the GSEs and their regulator have directed lenders to use in modeling how new scores perform vs. FICO Classic for risk assessments. They'd provided past Vantagescore 4.0 data but were still obtaining FICO 10T's.

As a coalition of trade groups recently noted, getting stakeholders beyond the GSEs  comfortable with those risk assessments plays a key role in whether the market opts for modernized score use.

However, if lenders can vet 4.0 or 10T to their satisfaction the operational change could be as simple as establishing a contract with a provider. This is something that already has been done in the primary private-mortgage market. However, the new scores still face challenges to secondary market acceptance there, according to NMN columnist Chris Whalen, who has a background as an analyst and investor and has worked with certain mortgage companies. 

(Generally, reported use of advanced scores in private mortgage securitizations to date has involved situations where these were provided in addition to traditional Classic FICO metrics.)

Another important detail in Pulte's FAQ is that it adds a nuance to an earlier post suggesting Pulte would be retaining the traditional tri-merge for credit reports, noting that this is something that will not change "initially" in order to "minimize the cost and complexity of this transition."

That suggests flexibility involving the tri-merge is something that could still possibly surface down the road.

Whether the additional details around possible future flexibility in the tri-merge and eventual adoption of 10T will ease the minds of those who've shown concern that announcing Vantagescore's approval alone hands too much power to the credit bureaus remains to be seen.

Another concern lenders have had with a choice of score that could be determined by at the point of origination has been whether it would convince some mortgage companies to be motivated more by production volume than performance, although the latter could still hurt them.

"If I'm a loan originator, I might have an incentive or a view toward the score that's not the best predictor of delinquency, but the one that's going to let me sell most loans to GSEs," said Kenneth Brevoort, principal economist, Federal Reserve Board, during the conference.

The National Taxpayers Union and some other groups it's allied with also are concerned about this, President Pete Sepp told attendees at the meeting.

Factors that will go into determining the return on investment

A mortgage company's return on using one alternative credit metric the GSEs are in the process of adopting also is something that requires estimates.

To get a sense of the kind of differences between the advanced scores, consider that 10T requires a little bit of a longer timeline to establish credit history than 4.0, and emphasizes that it allows for the scoring of millions more borrowers than traditional credit risk models.

Both 10T and 4.0 also are said to differ from Classic in incorporating alternative data like rent. They also incorporate trended data. FICO indicates that 1.7% mortgages can obtain a high score of 740 or more with 10T compared to Classic, potentially obtaining more favorable pricing.

Certain alternative data is limited now but does appear to be on a path where it will be more widely available in the future, Jung Hyung Choi, principal research associate at the IU's Housing Finance Policy Center, told conference attendees.

"We do see the potential of adding alternative data into the credit scoring. Then the mortgage system could potentially expand the credit box safely and also lower the cost of mortgages for some borrowers," she said. "However, there are some additional existing barriers."

While there's been a lot of growing momentum behind efforts to increase the collection and use of rent data in credit reporting, less than 5% of rental payment history has been in the credit bureau files, possibly limiting the effectiveness to immediately adopting a score to that end.

"We still need to do a lot of work on reporting the rent payment data into the credit files," she said.

That said, getting scores with rental data accepted could accelerate progress in this area. There've been use cases for rent data by the GSEs that include underwriting before on an ad hoc basis but advanced score use would be more influential.

Pulte's reaction

Pulte in a post on X reacted to the UI meeting by commenting on a quote from Housing Finance Policy Center Laurie Goodman. 

Goodman's comment, like some others at the meeting, indicated she was "not quite certain what the overarching policy is" in terms of advanced scores.

He said Goodman instead should have focused on the uncertainties the recent announcement resolved, and "thanking us for allowing rent to be used to help people get a mortgage."