A community bank in South Carolina has come up with an ingenious way to keep track of insider loans and their rates using an artificial intellegence program.
Summit National Bank, Greenville, says the program helps it comply with Regulation O, which requires pricing for insider loans to reflect rates applied to other lending.
"Our exercise with Regulation O leads us to believe that, ultimately, our approach would help document our adherence to fair lending rules as well," says Bill Squadroni, head of deposit banking and compliance at the $60 million-asset bank.
Issue Raised by Auditor
Mr. Squadroni says he thought of applying an artificial intelligence program - Braincell, by Promised Land Technologies, New Haven, Conn. - after reading about such programs.
But the issue already had been raised by an external auditor, who advised the bank to attach explanations and rate comparisons to each insider loan file.
"We fed a data base of noninsider loans into the program, and showed what rates of interest were applied. Then we took 31 insider loans and fed in the inputs but not rates. We then asked the program to predict the insider rates, and its prediction fell within 25 basis points of the rates we actually applied."
Compliments from Examiners
"The OCC [Office of the Comptroller of the Currency] folks complimented us for taking the extra step," Mr. Squadroni said. "They said the numbers [in the prediction spreadsheets] looked fine."
Ed Furash, president of Furash & Co., Washington, calls Summit's approach ingenious, but adds: "AI is just a process of putting in an orderly fashion what is normally done intuitively. The real utility here is affording one more piece of documentation for the examiners to hang their hat on when they try to determine if a decision meets a set of standards."
Mr. Squadroni believes that determining this is no mean feat because "after enough [loan] comparisons, the argument [with an examiner] becomes very subjective. Each side thinks the other cherry-picked the loans in the comparisons."
