For years, the government-sponsored enterprises (GSEs) Fannie Mae and Freddie Mac have faced significant pressure by regulators and consumer advocates to adopt a new credit-scoring model when assessing homebuyers. The goal is to better predict consumer risk while giving more access to credit for under-served consumers, or those known as “credit invisibles.”
Since 2004, there has been one standard credit-scoring model accepted on loans eligible for purchase by Fannie Mae and Freddie Mac. This scoring model is known by various names but is most commonly called Classic FICO.
The mortgage business is the only industry to use this model as most other industries have adopted some variant of a newer model. So, why are Fannie and Freddie holding out? Originators will want to explore the dynamics of what is occurring as they help clients with problematic credit while understanding how credit-risk assessments may change in the future.
Competing models
Since Classic’s release more than two decades ago, FICO and VantageScore Solutions (the largest competing developer of scoring models) have released a suite of newer products. FICO 8 and 9, and the recently released FICO 10 and FICO 10T, are the more recent variants of FICO models. VantageScore, which is owned by the three major credit-reporting agencies (TransUnion, Equifax and Experian) also have variants of their models, including VantageScore 3.0 and 4.0.
These newer models all claim to be better predictors of risk. One example of a new feature is trended data. This technology tracks how much is paid each month on such things as credit cards regardless of the posted balance. This process helps identify whether a borrower is a “transactor” who pays their cards off each month, or a “revolver” who pays the minimum each month and carries a balance. It is obvious that a transactor has less risk, so they may have a higher credit score under a newer model.
Conversely, these newer models have some features that are more “consumer friendly” rather than being better predictors of risk. A key example of this is the treatment of defaulted debt that is paid off. Under some newer models, if a bad debt is reduced to a zero balance, it is bypassed from the scoring model as if it never existed. This could cause a consumer’s credit score to jump quickly.
Under Classic FICO, once a bad debt is paid, it may temporarily lower a credit score. As a consumer, this can be frustrating, but it is invaluable for lenders. For example, consumers that have bad debts often ignore them until they want to buy something big like a home. Paying off these debts does not automatically make them a better borrower.
Earned privilege
The logic of Classic FICO identifies that the consumer is taking action on negative accounts, which is good. Instead of bypassing this account as if it never existed, Classic looks for the consumer to show responsible use of credit in the subsequent months to demonstrate that they are now a lower risk. The reality is that homeownership is not a rite of passage — it is an earned privilege and Classic FICO functions as such.
The complexity of integrating a new credit-scoring model goes beyond how these models function. Upgrading to FICO 10 is not like updating to Windows 10. Fannie Mae’s Desktop Underwriter and Freddie Mac’s Loan Product Advisor are programmed with logic that considers Classic FICO algorithms. Changing models will likely require an overhaul of the guides and engines that determine loan pricing.
Next, it is important to understand that many banks have custom scoring models that work in concert with Classic FICO. These models also will have to be modified and retrofitted. Last and maybe most importantly, Wall Street investors buy mortgage-backed securities based on this scoring method. After the crash of 2008, it will likely take some serious convincing to get them to buy mortgages under a more “consumer friendly” model.
The complexity of integrating a new credit-scoring model goes beyond how these models function. Upgrading to FICO 10 is not like updating to Windows 10.
Path ahead
So, where does this leave the mortgage industry today? The answer is as clear as mud, but here is what is known. In 2017, Congress proposed the Credit Score Competition Act. This was designed to require the GSEs to establish procedures for considering new credit-scoring options. In 2018, this became a reality through the Economic Growth, Regulatory Relief and Consumer Protection Act.
The law required competing models to be tested and validated for use by Fannie and Freddie. An open invitation was announced for all developers to apply, and approximately two years later, the Federal Housing Finance Agency (FHFA) announced the validation of none other than FICO Classic, keeping it as the GSEs’ exclusive scoring model. FHFA is continuing to examine newer models, however, and may make an announcement of a new validation before the end of this year.
Francis Creighton, president of the Consumer Data Industry Association (CDIA), said he thinks it is foolish for Fannie and Freddie to keep using such an outdated model. The CDIA is the trade association for Equifax, TransUnion and Experian.
At CreditCon 2021, the largest consumer credit event in the U.S., Creighton was asked the following question: If you had to wager on FICO 10 being validated or Classic FICO remaining as the model for mortgages, where would you put your money? He did not have any definitive information, but he did say he’d place a bet on Classic. ●
