credit scores play a critical role in pricing too

Credit scores play a critical role in pricing too.

Date: June 23, 2020

Well, that went by quickly! With just a few more events to go, the fall blitz of conferences and speaking engagements is coming to a close. I want to thank the members of the VantageScore team and all of our partners and colleagues for the hard work that goes into each and every one of these opportunities.

Last month, my column challenged the notion that competition in the mortgage market among credit score model developers would lead to a “race to the bottom.” It triggered a lot of feedback endorsing the continuing education of how the credit scoring process works. Please keep the feedback loop going.

In fact, I believe credit scoring competition is a race to the top. That is because in our business lenders provide the ultimate quality control. And that is because before they implement a new scoring model they perform their own extensive testing: How predictive is the new model on their own portfolio or within subsets of it? How did it perform over time? Had they used it, how many more good loans could have been made or bad ones avoided? If short cuts were taken in developing new models, that independent testing would discover such shortcomings quickly.

This month I thought I’d touch on another topic that also relates to the argument in favor of breaking FICO’s exclusive franchise in the government-sponsored mortgage market: that loan approvals and getting through the proverbial door are only part of the way credit scores are used.

Credit scores also play a critical role in pricing. The price that a consumer pays for a loan that will be purchased by one of the GSEs is determined by a matrix. In Fannie Mae’s case, this is called the Loan-Level Pricing Adjustment (LLPA), which can be viewed here.

On one side of the matrix is the applicant’s credit score. The model which is required for use in this matrix is an outdated version of the FICO Score built prior to the recession using data samples from 1995 to 2000. On the other side of the matrix is the potential borrower’s “loan-to-value” (LTV), which represents the size of the borrower’s down payment relative to the total amount of the loan.

It’s often stated that credit scores are hardly, if at all, used in mortgage underwriting. While underwriting takes into account the entirety of an application, however, the credit score is one of only two factors that determine pricing. For borrowers without a credit score, the price they pay assumes the worst (which could be one factor explaining the paucity of loans to borrowers without credit scores).

We’ve written before about how newer models, like VantageScore 3.0 and VantageScore 4.0, could help improve access at the margins. This argument is based on the opportunity to expand the pool of applicants who pass that first bar of having a 620 score or better. Equally important, however, is the opportunity to price based on the latest, most predictive tools. This could have the dual benefit of pricing in some borrowers on the bubble and extending a fairer price—that is, a price that most accurately reflects their riskiness—to every borrower who takes out a mortgage.

We look forward to continuing our work with the various stakeholders to solve this challenge for the benefit of consumers and lenders.

Speaking of looking forward, I’d like to wish you all a happy and healthy holiday season.

Regards,

Barrett Burns