Plaid deploys AI to spot unconventional consumer cash-flow sources

Plaid website logo
Plaid is using AI to capture more under-the-radar income and expense activity to flesh out consumers' full cash-flow profiles from shared bank account data.
Postmodern Studio/Adobe Stock

Credit scores anchor most lenders' routine underwriting decisions, but certain intermittent-use financial products like buy now/pay later loans, earned wage access and gig-worker wages can fly under the radar without the proper tools to spot them, according to bank data aggregator Plaid.

Plaid aims to help close that gap with Consumer Report, a new product it announced Tuesday that gives lenders a more complete real-time overview of a prospective borrower's full cash flow picture. It provides deeper insights into creditworthiness alongside credit scores and other underwriting data. 

The move comes weeks after credit-scoring firm Experian introduced a new tool called Cashflow Attributes to expand consumers' access to credit, and credit scorer Vantage rolled out its own new model to augment thin consumers' financial profiles. 

San Francisco-based Plaid's newest tool goes beyond its core service of providing businesses with basic transaction data by providing additional detail, categorization and Plaid's own insights into that traffic, including which recurring payments are likely associated with buy now/pay later loans, rent, utilities and disbursements. Using traditional AI and machine learning to crunch Plaid's consumer-permissioned data, Consumer Report provides a more accurate estimate of a prospective borrower's gross income, net income and next likely paycheck, said Jonathan Gurwitz, credit lead at Plaid.

"When consumers give a lender permission for Plaid to access their bank data, it produces basic transaction strings and basic categorizations, and it's not always possible to see the big picture of a consumer's real-time cash flow, with every income stream and expense," Gurwitz said, noting that Consumer Report provides a more accurate estimate of gross and net income.

The feature is particularly useful in "edge cases," where a prospective borrower's income sources are diverse, he said. For example, Plaid's data indicates that on average, consumers who are likely to request a loan may have up to five different bank accounts, they may be using three or four different digital financial apps and many people are earning money on the side through gig work, according to Gurwitz.

"In these cases, lenders want to take a closer look of the borrower's actual cash flow, which usually requires manual checks of a borrower's most recent pay stubs, other bank accounts and proof of other income, and Consumer Report automatically pulls all of that data together in real time," he said.

Plaid designed Consumer Report to work in collaboration with its proprietary consumer reporting platform launched last fall that considers data from 500 million U.S. consumer accounts that have connected to Plaid.

Organizations access Consumer Report by embedding it into their underwriting processes via an application programming interface, and as usage builds, Plaid expects to see further insights emerge from consumers' cash flow patterns, including credit risk signals when borrowers are accessing an abnormal number of lending apps or "stacking" buy now/pay later loans, Gurwitz said.

H&R Block is one of several enterprises that have been among the first users of Plaid's expanding consumer cash-flow insights, according to Gurwitz. 

"We're seeing a movement where consumers are ready to share more information about their cash flow as their financial lives become more complicated, and Plaid is in a position to help lenders take a deeper look, as one in three Americans by now have linked to Plaid at some point," he said. 

For reprint and licensing requests for this article, click here.
Artificial intelligence Fintech Consumer lending
MORE FROM AMERICAN BANKER