Site icon Accelerating Commerce Excellence

Community bank, loan marketplace pilot transaction-based underwriting

Community bank, loan marketplace pilot transaction-based underwriting

Texas National Bank has been strategizing how to tackle social issues in the Rio Grande Valley. 

One example is its no-fee small-dollar loan, built with bank and credit union technology provider Velocity Solutions and meant to counter the abundance of payday lenders operating in the bank’s communities. The interest rate will hover just under 18% and the product is expected to launch within days.

Another initiative is a pilot project with Lendio to deploy transaction-based underwriting for small-business loans. Although Lendio is better known for its business loan marketplace, where it sources borrowers and solicits offers from its network of bank and nonbank lenders, it is testing alternative underwriting capabilities with Texas National, which is based in Mercedes.

“We realized through years of experience that fintech lenders have gotten really sophisticated at quickly analyzing borrower applications, generating decisions and often using transaction data as a way of evaluating the borrower rather than relying so heavily on income statements and credit scores,” said Philip Taliaferro, general manager and senior vice president of software as a service at Lendio.

The goals are reflective of other small banks and minority depository institutions: to learn how to lower costs, increase efficiency and improve the customer experience using technology, in this case with small-business loans. Taliaferro and Rey Garcia, executive vice president of Texas National, will discuss their pilot on June 12 at American Banker’s Digital Banking Conference.

“When we see the fintech world taking over the banking world, it keeps me up at night,” said Joe Quiroga, president of the $679 million-asset Texas National. “This was our small attempt to say, we’ve got to innovate and keep up with it; we might not develop the next best product but we will learn a lot,” such as which algorithms or automated processes can take over more efficiently from human analysis.

“We don’t have huge expectations for growth, but we have huge expectations for learning,” he continued.

Such experiments with alternative data may also trigger the question of how viable alternative transaction-based underwriting is compared to traditional credit scores.

Texas National’s largely immigrant customer base frequently operate in cash, and do not borrow often, which means creditworthiness may not be reflected in their credit scores.

“What got us excited about this product is how we can be socially mindful and create this responsible solution to allow small-business owners that primarily operate in cash to get credit,” said Garcia. “This tool that analyzes alternative data lets us improve the speed and accuracy with which we make decisions.”

Lendio’s technology will mine customer deposit data from the bank’s core system, run it through algorithms and classify the transactions. It will tabulate the results in its model, detect factors such as revenue over the last 30 days, or number of days where funds dropped below a certain threshold, and pre-qualify customers for a loan.

Taliaferro cites two reports: one from the Bank for International Settlements in 2022 that studied two lenders that used alternative data and found they predicted future loan performance more accurately than the traditional approach to credit scoring, particularly in areas with high unemployment; another co-written by New York University associate professor of finance Sabrina Howell, which found that process automation boosts lending to Black-owned businesses.

“If I rely exclusively on traditional credit scoring and traditional underwriting models I am more likely to exclude the type of borrower I should be more focused on supporting,” said Taliaferro.

Texas National is starting slow. Currently it is restricting the pilot to existing customers, because the bank has a rich history of their transactions and wants to test the solution with well-known customers of the bank before opening it to a larger market.

“We put on tighter risk limits so we can monitor performance, measure progress, and evaluate potential issues that come up, then iterate from that,” said Garcia. “We’re trying to be mindful about how much we are lending until we’re more comfortable.”

Although Texas National is its first pilot customer, Lendio is also piloting transaction-based underwriting with a large regional bank and another community development financial institution.

Transaction data has not reached the mainstream with underwriting among traditional banks.

“I’m willing to argue that statistically speaking, it won’t result in better loan repayment because historical transaction data is very difficult to utilize to extrapolate future transaction data,” said Mitch Wein, head of community banking and credit unions at Aite-Novarica. There are some businesses where transaction data may be more predictive than other types, he believes, for instance a business with fixed and steady contracts underpinning its revenue.

However, nontraditional data can augment credit score underwriting to potentially give users better outcomes in certain segments of lending, said Wein. “If you can collect that data in a usable repository, in a consistent way, and tune the algorithms effectively, you can get that outcome, which is a win for the banks.”

Texas National is currently using credit scores in addition to transaction-based data. The bank says it will continue to assess the underwriting evaluation process as the product evolves.

“Over time we will see the importance of alternative data continue to increase,” said Wein, “as there is more data being generated and because of capabilities like artificial intelligence and more advanced machine learning, [which means we] will be able to more deeply evaluate the quality of that data for algorithmic purposes.”

link

Exit mobile version