Lenders are in the business of assessing risk, as it influences interest rates, principal, and ultimately whether or not a loan application is accepted or rejected. How accurately a lender is able to measure that risk has a direct impact on their bottom line.
Despite the importance of risk assessment, popularly used metrics like FICO scores provide a limited understanding of a borrower’s creditworthiness, which can lead to a lack of credit options for the underbanked. A new trend in risk modelling is emerging, however, as innovative lenders begin to take a broader view, looking at transaction histories and savings/spending patterns. By infusing this “alternative data” into their credit risk models, lenders are making them more holistic, robust, and predictive. Let’s take a look.
An Incomplete Picture
The limitations of FICO scores stem from the fact that they only consider a handful of credit-related criteria – credit bill payment history, amount owed, length of accounts, credit mix, and new credit. In other words, the score only paints a partial picture of a person’s financial health and behavior, and, therefore, their likelihood to make payments in a timely fashion throughout the course of the loan. Lenders can better assess risk, and set interest rates and other terms accordingly, by taking more data points into consideration. Years of transaction history, for example, can demonstrate whether or not a person consistently paid their rent and utility bills on time, providing an important glimpse into their financial behavior.
Not to mention the estimated 45 million Americans who are either “credit invisible” (26m) or “unscorable” (19m). The first classification refers to those who have no credit history whatsoever. The second points to those who have such little credit history that a credit score such as FICO cannot be generated. Clearly a more robust credit risk model must be developed to better tap into this pool, much of which is likely to be deemed creditworthy if viewed through a lens that takes into account more than just credit history.
Widening the Lens with Alternative Data
A number of forward-thinking lenders are exploring ways in which they can improve their lending models by moving beyond the traditional FICO score with the use of alternative data. SoFi, an online personal finance company that provides student loan refinancing, mortgages, personal loans, and wealth management services, no longer uses FICO scores to make credit decisions.
“Instead of relying on a three-digit number to tell us who’s qualified, we look for applicants who have historically paid their bills on time and make more money than they spend. It’s that simple,” said SoFi CEO and co-founder Mike Cagney in a press release.
Earnest, a startup that issues low-interest personal loans, uses alternative data such as savings history, investments, and career trajectory to model risk and offer competitive interest rates in the lending space as well. For example, Earnest examines whether or not an applicant has increasing bank account balances. Consistently growing bank balances means two things: 1) the person has some cash in order to make loan payments, and, more importantly, 2) that they have the financial discipline to save on a regular basis by spending less than they earn. This enables underwriters to go beyond the standard metrics and identify behavioral patterns that are indicative of good borrowers.
FICO seemingly recognizes the benefits of alternative data as well, having recently launched an “alternative credit score” of its own. During pilot trials of FICO XD, which includes cable, utility and cell phone payment history, 35-50% of people without enough credit history for a traditional credit score received an equivalent score of 620+, a big improvement from being “scoreless.”
A Brave New World
We are still in the early days of applying alternative data to the underwriting process. However, it is clear that credit risk models can be improved by taking a more holistic view than what traditional metrics of creditworthiness, such as the FICO score, can provide. If leveraged properly, more predictive credit risk models, informed by alternative data such as transaction history and savings patterns, can be deployed by lenders for more efficient and effective underwriting.
Interested in learning more about how you can enrich your credit risk model with Quovo? Sign up to start a conversation with one of our data experts!