In lending, in determining whether credit is likely to be repaid by a borrower, the devil is in the details.
What if the details are the wrong ones?
In an interview with Karen Webster, Steve Allocca, president of LendingClub, said that traditional credit scoring — such as through the FICO credit score — misrepresent creditworthiness of individuals.
The conversation came against the backdrop where Fair Isaac announced it would launch the latest model of its FICO model, FICO 10. The model, which changes some factors considered, and puts more weight on recent delinquencies and high credit utilization, may cause millions of consumers to see shifts of as much as 20 points in their credit scores — up and down.
The timing of it all, of Wednesday announcement of FICO 10, may not be all that surprising. As Allocca noted to Webster, it’s not entirely coincidental that Fair Isaac’s news comes in the same month Visa bought Plaid for $5.3 billion, heralding a shift in how financial data is collected and used.
Said Allocca of the credit bureaus: “It becomes obvious that if they don’t move dramatically, the world is going to quickly pass them by.”
In Allocca’s view, the FICO model, and using credit bureau data and historical loan performance data as primary ways that credit is underwritten and provisioned is increasingly antiquated in a world where financial life is digitized.
The digitalization of money and payments helps paint a much more accurate picture of a person’s ability to pay, said Allocca.
The Two C’s
And, contrary to looking solely at historical data — the way it’s always been done, in other words — Allocca said that “two C’s” of credit for unsecured lending are capacity and capital are relatively easy to measure today and provide a better picture of creditworthiness.
In an illustration of the ability to repay, Allocca defined capacity as the individual’s income, less non-discretionary expenses, and how the consumer handles discretionary expenses. This translates into recurring net savings, which translates into an accumulation of capital, which can be used for repayment.
There’s a seismic shift, too, in the workforce that demands a shift away from the traditional FICO models. Fewer workers have consistent W-2 jobs, and thus less stable income, have less stable expenses, so the age-old inputs to the risk models don’t work.
Moving toward a system that embraces the two C’s, he said, moves from “black box” models toward a transparent system. But the incumbents have a lot to lose by shifting to that system, he told Webster.
No Sure Bet
And perhaps it’s not a sure bet that the periodic revisions of the algorithms to make FICO any more relevant moving forward once FICO 10 debuts later in the year.
The FICO updates tend to happen every five years — the last two were in 2014 and 2009 — and yet there’s a long tail of adoption. Allocca said that many lenders are still using FICO 8 (we’re officially in the age of FICO 9), including LendingClub.
That would stand the logic that FICO 9, which used some additional data types and eliminated dings like civil judgments, was directly and solely responsible for a credit surge.
“That part of the story just doesn’t ring true,” said Allocca.
Also, companies such as Lending Club have been using custom models, where he said, “we get much more value from other data sources that go way beyond FICO and have nothing to do with a person’s historical performance on their loans … no one who is sophisticated is just taking the FICO score. They’re taking the score as an input to a custom model that has a lot of other attributes.”
As for personal loans, Webster noted that the FICO news in effect “calls out” such activity, and there are new tenets that ding consumers for taking out such loans. As noted in this space on Wednesday, unsecured personal loans have enjoyed a renaissance, having surged roughly 10 percent year over year in the second quarter, for example, to more than $300 billion.
Asked why the FICO scores would zero in on platforms such as Lending Club’s, where users can consolidate debt. Allocca countered that his firm’s customers are coming to the platform because they have “intent not to use credit to spend more money than their income justifies … they intend to use these credit products to ultimately increase their savings, which increases their ability to pay. And what credit scoring ultimately should do is be reflective of somebody’s ability to pay.”
Therein lies an inherent flaw in the traditional credit bureau driven system, said Allocca, where the scores were never really intended to anticipate future changes to that very ability to repay.
“In an age where your whole financial life is verifiable digitally, instantly, with a customer’s permission and credentials” — a nod to the FinTech driven realm, of say, Visa/Plaid — “why would you think that just historical loan performance would be a better indicator? ”
As Allocca said: “The credit score ‘black box’ is a dinosaur.”