FICO Banking Analytics
In this age of increasing reliance on analytics and rich alternative data sources, the firm grip on conventional credit scoring and risk assessments seems by FICO to be weakening. As non-traditional lenders flood the market, alternative business models make the need to watch the historical signifiers of credit risk — usage, debt levels, repayments, etc. — less pressing. Other data sources are being included to paint a broader picture of borrowers.
“Nontraditional borrowers have drawn the attention of lenders.”
A Credit Story
This has led to a new breed of borrower, a borrower who is credit-worthy but falls outside the guidelines established by scoring models used by FICO. With either low or nonexistent FICO scores, these borrowers have drawn the attention of lenders in both the traditional and non-traditional world. Lenders are now tasked to find a way understand what Tom Burnside, CEO of LendingPoint, calls the borrower’s “credit story.” And divining this story takes looking beyond the established models by FICO and going deep into the data.
“Traditionally, lenders look at credit through a series of questions like DTI (Debt to Income), PTI (Payment to Income), availability (how much credit do they have left?), the number of established creditors, and the consumer’s FICO score,” says Burnside in the LendIt 2016 conference blog. “What these questions do not address are broader, more complex credit considerations.”
Falling Through the Cracks
This is where examining other data points makes a significant difference — and where valuable insight on a consumer’s actual worthiness may fall through the cracks of the FICO scoring model. A consumer may only have a single credit card, resulting in low credit availability as well as low DTI. In the binary “approve or decline” scoring model, this may reflect negatively on the borrower’s potential credit worthiness, when in fact it could also tell the story of a borrower with plenty of cash reserves on hand to make payments.
“One argument for delving into nontraditional data is that 40 million Americans don’t have a FICO score due to no or ‘thin’ credit history,” writes John Birge, Chief Credit Officer of Credibility Capital, in another LendIt blog. “And perhaps not surprisingly Millennials represent a disproportionate percentage of this group. For this population new and promising data sources include positive rental payments and positive payment of utility bills.”
This has led to skepticism about the efficacy of the FICO model as a measure of credit-worthiness, particularly amongst new lending tech startups already outside the sphere of traditional lending. Many of these startups are relying more heavily on sophisticated credit analytics to get a sense not just of the historical risk posed by a borrower, but also projected risk as well. This involves factoring not just events but behavior and putting together a more comprehensive profile of a borrower — something that FICO has yet to fully integrate into their models.
This creates credit populations that are overly homogeneous and insulated against growth. Lenders only lend to borrowers who have already borrowed.
“In statistical parlance,” comments Birge, “this population is so super-prime and homogenous that FICO doesn’t provide any ‘orthogonal’ or incremental predictive value.”
This effectively locks potential borrowers out of the system, ensuring that they are never granted access to credit — credit needed to give them a score. But by looking more at behaviors, whether it be purchasing, repayment or social, lenders are able to look into credit invisible populations and recognize who might be a good candidate for credit. This data may even be able to help put together more accurate methodology to help borrowers with poor credit better use the credit they have.