Credit & Risk

Moving Beyond Old Credit Scores

Ken Garcia

November 4, 2020

“Credit scores are currently being propped up by things like government stimulus payments and delayed mortgage payment plans for homeowners,” said Matt Schulz, chief industry analyst at CompareCards.com by LendingTree.

“Because of all the deferment we’re seeing and some stimulus, I’m not sure how predictive these credit scores will be of people’s financial situation in the future,” said Schulz.

Financial Institutions can’t Tell who is Creditworthy or Who is at Risk of Defaulting

The inability to rely on traditional credit scores is scrambling lender’s underwriting models, prompting a need for change. Lenders are trying to figure out how to avoid approving loan applicants who are unemployed and on the verge of running out of government assistance. And some are also looking to identify existing customers at higher risk of default.

To adapt, there is momentum to go “beyond the old credit score” and find new ways to evaluate risk. A recent article indicated Bank of America, JP Morgan Chase, Wells Fargo, and Citi Group, are seeking to augment credit reports and scores with real-time income or cash-flow data. As consensus builds to use more data, what’s the best approach to make sense of it to make smarter and faster decisions?

Machine Learning: Swapping out Risky Borrowers with Good Borrowers

As the industry looks for faster and more effective ways to assess credit and loan eligibility, machine learning models are getting a second look. Machine learning allows lenders to model using more variables, creating a more holistic, clearer picture of an applicant. Here is an example from one of our clients using an AI-powered model:

Traditional Model: The applicant that looks good on paper, gets approved. However, is John the more creditworthy applicant?

Zest AI machine learning model uses more variables and uncovered red flags that reveal John is a risky applicant.

The Zest AI machine learning model also reveals that Jane has a thinner file, but more high-quality accounts (i.e. mortgage)

Zest AI machine learning model swaps-out John (riskier applicant) and swaps-in Jane (less risky candidate).

With machine learning models, lenders can identify risky borrowers who may have looked good on paper and swap them out for better, creditworthy borrowers who—due to their lower score— were traditionally overlooked, ultimately delivering growth, increased productivity, and more inclusivity.

Considering the current economic environment, the ability to use more data and better math in a meaningful way to look “beyond the credit score” and more accurately rank risk is fast becoming table stakes. With a clearer picture of borrower risk, machine learning models enable you to make faster, more accurate decisions.

In our “Leaders in Lending” series, Mihaela Kobjerwoski, Chief Consumer Credit Officer, First National Bank of Omaha shared her thoughts on mainstreaming AI and machine learning in underwriting:

“I’m convinced that AI and machine learning can provide very impactful business results, and what I’m hoping happens with this new technology and new types of analytics is we consider the customer experience and the customer in this journey much more so than we did with previous types of analytics. I think we need to be very mindful of the experience we want to create and the human on the other side as we continue on this path.”

For lenders, there is an urgent business need to “move beyond the old score” and an opportunity to help more traditionally overlooked borrowers get access to credit, ultimately delivering a fairer and improved customer experience. If you looked at the average credit score, you wouldn’t be able to tell the U.S economy is struggling. In fact, amidst rising unemployment and increasing economic hardship, credit scores have reached record-breaking highs. But do these scores accurately reflect what’s really going on with consumers?

Thank you for subscribing!
Something went wrong while submitting the form.