This post is the fifth in our Housing America Series, aimed at elevating the need for housing policy reforms—from housing finance to affordability—as we move through a pivotal federal election and transition.
Credit scores are a numeric metric that predicts borrower loan performance. These scores are derived from models that assess borrower information provided by the nationwide credit-reporting agencies. Widely used by lenders of all types for making a range of credit decisions, mortgage lenders use them to set qualifying requirements, price mortgages, and establish “overlays,” the line below which they will not lend because the likelihood of default exceeds their risk tolerance. This is why the technical debate over alternative credit-scoring models in housing finance is both important and politically fraught. This post reviews recent developments in the debate over these tools and offers five recommendations.
Fannie Mae and Freddie Mac, the two government-sponsored enterprises (GSEs), require all institutions that either sell them loans or service the loans they purchase to use credit scores developed by Fair, Isaac and Company—more commonly known as FICO. The Federal Housing Administration also requires lenders that seek their mortgage insurance to use FICO scores, effectively creating a credit-scoring monopoly in the mortgage space.
Yet according to the Consumer Financial Protection Bureau, 26 million consumers are “credit invisible” under the current system. Because they have no credit records recognized by the credit-reporting agencies, it is difficult if not impossible for FICO to score them. Another 19 million consumers have credit records deemed “unscorable” instead because of an insufficient credit history or lack of a recent history. Thus, about 45 million consumers—some of whom are no doubt creditworthy—are essentially shut out of the opportunity to access mortgages through the GSE channel. And unfortunately, this population is made up by a disproportionate number of minority consumers and those who live in low-income neighborhoods.
In recent years, alternative credit-scoring models have been developed by FICO, including FICO 9 and FICO XD. A company created by the three national credit bureaus, VantageScore, has also created new models, the latest of which is called VantageScore 3.0. Unlike the FICO model still in use (commonly referred to as Classic Fico), both of these newer models score more consumers—28 to 35 million credit-invisible consumers according to the two companies—by, among other things, taking account of more granular credit file data. These newer models include rental housing, telco, and utility payment data resulting in more credit information and, for many consumers, higher scores. In addition, while the Classic Fico approach downgrades scores for approximately 64 million people with medical collections on their credit reports, VantageScore 3.0 and FICO 9 exclude all paid collections and reduce the weight of unpaid medical debt, recognizing its weak correlation to ability or willingness to pay other types of debt.
Because mortgage credit remains tighter than it was pre-crisis, the credit-scoring monopoly has drawn the attention of lawmakers. In November 2014, Senators Mark Warner (D-VA) and Elizabeth Warren (D-MA) wrote to Federal Housing Finance Agency (FHFA) Director Mel Watt noting their support for continuing strong underwriting standards at the GSEs, while expressing interest in “the opportunities that additional competition in the credit scoring market could create.” And, last year, bipartisan legislation was introduced in the House of Representatives authorizing the GSEs to consider alternative credit-scoring models beyond just Classic Fico. When introducing H.R. 4211, called the Credit Score Competition Act of 2015, the bill’s sponsor, Representative Ed Royce (R-CA), argued, “….Breaking up the credit score monopoly at Fannie and Freddie introduces competition into the credit scoring industry and ultimately decreases the potential for another taxpayer bailout.”
Because mortgage credit remains tighter than it was pre-crisis, the credit-scoring monopoly has drawn the attention of lawmakers.
While some critics have concern that the adoption of more inclusive scoring models relaxes GSE standards, the effort has support from some conservative Republicans. As recently as September 2016, Rep. Scott Tipton (R-CO) endorsed the bill, noting that “as well as supporting many Americans’ dream of owning a home, the Credit Score Competition Act will increase innovation, alleviate portfolio risk, and lower systemic risk in the housing market.”
The FHFA has been working with the GSEs for the last two years to assess whether and how to leverage “alternate or updated credit scores for underwriting, pricing, and investor disclosures,” and a decision on the matter is expected soon. To help assess the policy implications of alternative credit-scoring models, we reached out to a cross-section of market participants. Here’s what we learned:
- The alternative credit-scoring systems under evaluation incorporate more data on individual consumer behavior than does Classic Fico. This raises the stakes for credit bureaus to keep errors to a minimum and to make it easier for consumers to review their files, challenge errors, and be confident that mistakes will be promptly corrected.
- FHFA must manage expectations around the conversion process to a new system. Any increase in borrowers approved for GSE-supported lending as a consequence of adopting either or both new scoring models will be tempered by the fact that both GSEs regularly improve and update their risk assessment analytics, partly by incorporating some of the same granular data used in the new models.1
- It is not clear how these systems will perform. While the GSEs have delivered to FHFA detailed analyses of how FICO 9 and VantageScore 3.0 would perform relative to Classic Fico and to each other, and the FHFA has conducted its own independent analysis, none of this information has been made public. Whatever decision FHFA reaches should be accompanied by a comprehensive narrative and supporting data that can be made available to the public.
- There will be an extended timeline for implementation. All participants in the securitization chain will have to adapt to the new system, including small and large lenders, banks and non-banks, mortgage brokers, private mortgage insurance companies, and investors. Some participants would also have to adapt their systems to accepting credit scores generated by both FICO 9 and VantageScore 3.0.
Based on these conversations, and to help expand access to credit for underserved borrowers, we recommend that FHFA:
- Direct the GSEs to modify their loan purchasing and assessment processes, upfront pricing grids, and securities disclosures to enable them to accept applications containing FICO 9 and VantageScore 3.0 credit scores. At a time to be specified by FHFA, the agency should direct the GSEs to no longer accept Classic Fico.
- Develop an initial equivalency score matrix for the market that illustrates what Classic Fico scores translate to in FICO 9 and VantageScore 3.0 in terms of default probabilities and release it to the public.
- Direct the GSEs to release historical loan data containing FICO 9 and VantageScore 3.0 loan-level credit scores, allowing market participants to recalibrate their prepayment/default models. Failure to do so in a timely manner could seriously disrupt the to be announced (TBA) and credit risk transfer markets.
- Require lenders and seller/servicers to adopt either FICO 9 or VantageScore Score 3.0, but not allow them to use more than one credit scoring system to evaluate the same consumer application. Lenders would be free to switch from one scoring system to the other on a periodic basis. Such a measure would limit “venue shopping” for whichever system will generate a higher score on a particular application and the potential for adversely selecting the GSEs relative to lenders’ choices as to which loans to retain in portfolio and which loans to sell into the secondary market.
- Because of the analytical challenges that conversion presents to investors and rating agencies, it may be operationally easier for the effective date of the change to follow the transition to the new single GSE security. This is now scheduled for implementation some time in 2018. While this may seem a long way off, the conversion to a lenders’ choice credit scoring system has many moving parts and will take time to get it right.
1 For example, the most recent update of Fannie Mae’s Desktop Underwriter enables lenders to submit applications for consumers who have no traditional sources of credit, although such loans would require at least a 10 percent down payment.