WASHINGTON — Acting Comptroller Michael Hsu Monday urged banks to reframe their view of fairness as a catalyst to innovation rather than an inhibitor, saying proactive efforts can provide banks a stronger foundation to experiment with emerging technology or new ideas.
In remarks — delivered to a crowd at the Community Bankers Association's For the People conference — the bank regulator attempted to dispel what he saw as a common misconception in banking: that safety and fairness must necessarily come at the expense of innovation.
"By elevating fairness, banks can improve their ability to anticipate and adapt to emerging compliance risk issues," he said. "The stronger a bank's ability to do that, the less it will need to look over its shoulder at its regulators and the more degrees of freedom it will have to innovate and create banking products 'for the people.'"
Hsu argued that banks can learn from the rise of overdraft fees — and regulators' subsequent efforts to crack down on them after the industry reported over $11 billion in overdraft income at the practices' heyday. Overdraft fee income has
Banks began offering overdraft fees in the 1990s intending them to be rarely used, but they proved popular and profitable. Their steady growth over the next few decades, Hsu argued, allowed fair lending compliance risks to build up undetected until banks publicly reported their
"What may seem manageable at first from a compliance risk perspective, can become much more complicated over time," Hsu noted. "As with overdrafts, having a clear sense of where this
fairness line is prior to the development and launch of such a product can help a bank avoid
compliance risk issues down the road, when the product has grown and consumer harms are more apparent."
Hsu also touched on what he sees as two distinct components of fairness: treatment and impact. Hsu defined disparate treatment as when a firm discriminates against or borrowers differently because of some protected characteristic, including sex, race or sexual orientation. Disparate impact, he said, is less obvious and results in disparities that by proxy disadvantage certain groups in relation to others. Hsu argued banks must mitigate both.
The regulator cited discrimination in AI and machine learning as an example. He noted banks must be hyper-vigilant to the potential that AI may replicate biases inherent to the data it is fed — something regulators
"For the banks that have been engaging in machine learning, you know — knock on wood — to date, it has been done fairly responsibly… in a way that's consistent with the model risk management guidance," he said. "The one thing I do worry about a little bit is there's intense pressure to pilot a lot of use cases."
A number of big banks have been dipping their toes into the generative AI waters. Citi, for example,
"Successfully balancing innovation and growth with safety, soundness, and fairness is getting harder [as] consumer banking and compliance are more inextricably linked now than ever," said Hsu. "This presents a challenge—and an opportunity—for banks and how they approach compliance risk management."
