Author | Jim Kasch
The more I talk about AI-driven underwriting, the more questions I entertain from credit union leaders. Among the most common questions I receive is “I understand that better decisions should lead to better loan performance, but how does that equate to a significant impact to my earnings?”
This question is naturally rooted in how better underwriting decisions will reduce delinquencies, which will in turn reduce charge-offs. Executives understand this natural consequence of better anticipating borrowers who are less likely to fulfill their borrowing obligation.
However, the challenge that lurks in the backs of these executives’ minds is that credit unions already have low loss rates. Our conservative nature, thoughtful underwriting approach, and committed collections efforts regularly result in loss rates far below 1.00% of total loans. In fact, according to the NCUA, the rolling 12-month charge off rate for the industry in March 2021 was 0.48. And this was after more than a year into the pandemic response that drastically affected household incomes.
So, if the typical credit union has loss rates of less than half of one percent, how much income could be recovered with smarter decisions driven by AI?
The answer lies in that last question. We shouldn’t be concerned with the amount of income we can recover, but how much additional income can we generate with smarter decisions driven by AI. The answer to that question is: Considerably more! Here’s what I mean:
As we have discussed, AI-driven underwriting uses substantially more information when making loan decisions. Ostensibly, this helps lenders in two scenarios: The first is what we outlined above. That is, declining loan applications of borrowers that are less likely to fulfill their borrowing obligation. This will reduce losses. The second is approving loan applications that would have otherwise been declined by our traditional underwriting methods. Many of these borrowers tend to have lower FICO scores, an important factor that discourages lenders from granting the loan request. However, as any seasoned underwriter knows, FICO does not always tell the entire story, especially for borrowers with scores below 700.
Instead of focusing on the limited components that go into establishing the FICO score, AI-driven underwriting engines will consider thousands of more pieces of data to make more informed decisions. Frequently, borrowers that would be traditionally turned down should and can be approved for the loan. Not only do these additional approvals result in higher loan balances, they also tend to carry with them higher interest rates, which are still tied to the FICO scoring model.
In other words, AI-driven underwriting engines will help your credit union uncover hidden gems in your borrowing pool. These are members who have lower FICOs but who are anticipated to perform like those with higher FICOs. Or put another way, your membership base has hundreds or thousands of borrowers with scores less than 650, but who will pay you back like a borrower with a 750.
The impact to earnings could be substantial. The easiest way to project the potential impact is to focus on your loan portfolio’s net loan yield, which is calculated by simply subtracting your net charge off rate from your gross loan yield. (Gross Loan Yield – Net Charge-Off Rate = Net Loan Yield). Obviously, there are two ways to affect the net loan yield: reducing charge-offs or increasing the gross loan yield. We’ve already established that most credit unions don’t have much room to improve their charge-off rates, but most can increase their gross loan yield.
We do this by granting more loans to members with lower FICO scores who we are confident will fulfill their loan obligations. This is more effectively accomplished through AI-driven loan underwriting decisions.
Consider the scenario below:
In this scenario, the credit union experiences only a 0.25% increase to their net loan yield on this relatively small loan portfolio, yet the impact is $125,000 in extra profit! (By the way, the cost to implement AI-driven underwriting is less than half of this windfall.)
Use AI-driven underwriting to find the hidden gems in your membership, and immediately improve your earnings!
Comments