How banks can fine-tune regtech (and still reach the underserved)

Register now

Artificial intelligence can go far in helping banks keep bad actors out of the financial system, but one of the challenges remains a "black box" factor that can obscure decisions that are unintentionally shutting out underserved consumers as well.

Several industry experts at American Banker's RegTech conference in New York stressed two issues: creating a complete audit trail that can provide examiners an explanation of how they used AI, and ensuring that the evolving technology is not used to discriminate against those who struggle to have access to the banking system.

“Like other powerful tools in finance, it has the opportunity for suboptimal outcomes if it’s not accompanied by management judgment,” said Eugene Ludwig, chief executive of the IBM-owned Promontory Financial Group, who recommended having a designated officer in charge of reviewing AI applications.

For example, Ludwig said his team used machine learning models to identify subsets of banking customers that were falsely identified as high-risk — nonresident immigrants from high-risk countries who were being flagged incorrectly by anti-money-laundering tools. They amended risk policies to give banks a better command over their portfolio.

“In the process of finding the bad guys, you don’t want to root out a population of people who really ought to get financial services, but are put in the wrong category,” Ludwig said.

Some groups are also looking into alternative data like cash flow in underwriting to act as a conduit between industry firms and regulators. Melissa Koide, CEO of FinRegLab, a new research organization, is taking data shared by certain lenders that are underwriting borrowers based on cash flow data and comparing how those borrowers perform against others whose loans were underwritten by traditional means.

The end goal of the study is to determine whether the lives of minorities and low- and moderate-income consumers can be improved by underwriting with alternative data, said Koide, formerly the U.S. Treasury Department’s deputy assistant secretary for consumer policy.

“We have these really thorny questions around cash-flow data and where does the cash-flow data sit, who owns it, and what if it’s not accurate,” Koide said. “If it’s not accurate, we have much bigger financial problems.”

The lab is also trying to determine the responsibility and liability of data aggregators that sit between banks and fintechs and are trying to offer consumers new financial wellness solutions.

Yet with more availability of data and the technology to analyze that data, every organization in the public and private sector is now expected to make evidence-based decisions, said Jeremy Balkin, head of innovation at HSBC.

While data governance is a concern among consumers and regulators in the wake of data breaches at Equifax and Facebook, data also has the ability to help improve the relationship that banks have with the public, he said.

“In the last 10 years, there has been a breakdown of trust particularly with financial institutions in the U.S.,” Balkin said. “Part of what you’re seeing is a technology data evolution in the last decade that can lead to better decision-making and better trust in the economy and financial services.”

For reprint and licensing requests for this article, click here.
Financial inclusion Artificial intelligence Machine learning Regtech Digital banking HSBC IBM RegTech Conference