More and more banks are using alternative data for many purposes: extending credit to people with no credit history, opening accounts, expanding relationships with small businesses, targeting marketing, detecting fraud, etc.
The First National Bank of Omaha in Nebraska, with assets of $27.2 billion, uses cellphone information, utility payment records and other non-traditional data “that can help establish positive payment behaviors” when underwriting loans, says Marc Butterfield, senior vice president, innovation and disruption.
“Machine learning and AI help create nonlinear models to maximize positive credit decisions,” says Butterfield. “They don’t necessarily conflict with traditional data sources. Instead, alternative data is used to augment and improve the decision to maximize the desired outcome.”
Butterfield notes that banks need to be careful when going down this path. Inconsistent use of alternative data and use of poor quality data sources can lead to suboptimal results and a lack of explainability and transparency about why someone was denied credit .
“Having a good data governance process is important for all types of data used,” he says. “The same practices should be applied to alternative data.”
Spring Bank of New York, with $293 million in assets, offers affordable consumer loans of up to $3,500 to people with limited, no or poor credit history, says Melanie Stern, director of consumer loans.
“We’ve learned over time that the best way to deliver consumer loans effectively is to partner with local employers, offer loans to their workers, and use their employment verification and income as a substitute for a credit score,” Stern says. “Currently we have 38 employer partners and we’ve never had a problem – delinquency rates are low.”
Upon loan approval, borrowers are required to open a savings account and set up loan payments as direct deposits from their employer’s payroll, she says. Borrowers can also use the account to build up savings, especially after repaying their loans. “Employers tell us that we have done them a great favor because they have been able to stop giving salary advances or lending money to their workers. And workers no longer borrow from their retirement savings.
Alternative data that banks can leverage to extend credit to small businesses includes their inventory and accounts receivable information, as well as tax filings and employment status, “not to mention examining macroeconomic factors that could impact their business,” says Isio Nelson, head of client engagement in BAI’s research division.
This data can be useful for targeted marketing, including household income, location data, hobbies and other personal interests. “There are many ways to develop hyper-personalized messages that will resonate better with that consumer,” Nelson says. On the fraud side, banks now use many different signals to detect if a fraudster is applying for credit or interacting with an existing account.
According to a report by Burnmark and CUBE, alternative data enables the analysis of digital behavior patterns for a wide range of banking functions. Examples include geolocation data for consumer buying behavior; drones for virtual site visits to approve and monitor commercial loans; Web scraping for lead generation and market analysis; and data from social media sites, super apps, and mobile phone usage for credit scoring.
A group of leaders from banking, business, technology and national civil rights organizations are exploring credit scoring models that leverage alternative data. The group, which is part of the REACh (Roundtable for Economic Access and Change) project, was convened by the Office of the Comptroller of the Currency of the US Treasury.
Several financial institutions within the group have started sharing data as part of a pilot program testing the predictability of alternative scoring models in underwriting low-value loan products and unsecured credit cards, Andrew said. Moss, OCC director for minority outreach.
“What we’ve heard so far from them is that it’s something that could be really beneficial with attractive opportunities,” Moss says. “The OCC’s role as steward of this group is to provide safeguards so that there are no issues that could trigger fair access or fair lending issues. We do not endorse any particular model – our advice is always aimed at ensuring that any type of modeling is backed by verifiable data and is reliable.
Katie Kuehner-Hebert is a BAI Contributing Writer.
Learn how financial services organizations can use data to build strong relationships and improve other business opportunities in BAI’s executive report, “The Power of Data: How Banks and Credit Unions Can Put It to Work” .