CFPB Could Play AI Regulation Role, Bank Policy Institute Says
The bureau would maintain focus on consumer protection while coordinating with other agencies, according to a discussion paper on advanced technology in credit underwriting
Friday, September 13, 2019
By Ted Knutson
The Bank Policy Institute (BPI), an association representing the biggest U.S. banks, is suggesting that the Consumer Financial Protection Bureau “lead an effort” to modernize and coordinate regulation to accommodate artificial intelligence in credit underwriting.
Taking off from the assumption that AI technology can help make more credit available to more people at lower cost, and that regulatory clarity is currently lacking, BPI says that the CFPB is well positioned because of its consumer-protection mandate and because it has authority over both banks and nonbanks.
The draft signals receptivity to a central, clarifying role for the CFPB among a big-bank constituency that resisted the bureau's creation under the Dodd-Frank Act and criticized its enforcement posture during the Obama administration.
“CFPB is the federal agency with sole rule-writing and interpretive authority over a wide range of federal consumer financial protection laws, giving it the ability to assess the current regulatory approaches to implementing those laws that raise uncertainty and friction with respect to the use of AI in credit underwriting,” the discussion draft asserts in laying out principles for “responsible” regulatory modernization.
The BPI-Covington & Burling draft makes clear that the CFPB would be working “in consultation with the relevant federal agencies.” It concludes: “The advent of AI presents the CFPB and other regulators with a unique opportunity to craft a regulatory framework that enhances credit underwriting, improves credit access, and levels the playing field for all lenders and all borrowers, while simultaneously preserving and enhancing the effectiveness of the current consumer protection regulatory framework.”
The draft lists “potential principles for regulatory modernization,” including development of standards to prevent Equal Credit Opportunity Act violations; clarifying that an AI credit underwriting system “can qualify as an empirically derived, demonstrably and statistically sound, credit scoring system”; and specifying steps that lenders should take to ensure that AI credit underwriting systems comply with federal consumer financial protection laws.
The paper cites a recent CFPB statement that the combination of alternative data, increased computing power and machine learning “can potentially identify relationships not otherwise discoverable through methods that have been traditionally used in credit scoring. As a result of these innovations, some consumers who now cannot obtain favorably priced credit may see increased credit access or lower borrowing costs.”
In a section of the report delineating AI-based systems' advantages over conventional credit scoring, BPI and Covington say that the former “use more diverse data sets and credit standards compared to conventional credit scores and so allow multiple approaches to assessing a consumer's creditworthiness. Such diversification in credit underwriting should not only give underserved consumers additional opportunities to qualify for credit, but also reduce systemic risk by enabling banks to adopt different approaches to credit decisions.”
Accuracy and Explainability
Real-time monitoring and processing of large, dynamic data sets is key to capitalizing on the promise of AI in supporting credit decisions. Bank Policy Institute president and CEO Greg Baer said smart decisions based on more data inputs will tend to reduce risks.
“AI has the potential to provide access to credit to millions with no or low FICO scores,” Baer said at the conference, adding, “Banks can make more accurate credit decisions.”
With testing taking place continuously in real time, human teams must work together with the AI to guard against unfair or deceptive acts and practices, said Brad Blower, vice president of consumer practices compliance, American Express Co.
Kareem Saleh, executive vice president of machine learning risk-model developer ZestFinance, said care must be taken that AI models don't learn and propagate biases and inaccuracies. He said that regulatory agencies look for assurances that models don't pose a threat to consumers or the safety and soundness of the financial system, and that they address explainability concerns - that they can be transparently documented.
The company's website says that Zest Automated Machine Learning customers “see a 15% average increase in approval rates with no added risk.” A subprime auto lender decreased its charge-off rate 33%, a case study says.
Congressional Task Force
Small and young businesses stand to be winners with AI credit underwriting, since there are substantial regulatory constraints on banks' lending to them without collateral, said Representative French Hill, Republican of Arkansas and ranking member of the House Financial Services Committee's AI Task Force.
With the creation this year of the bipartisan task force, which is chaired by Bill Foster, Democrat of Illinois, “Congress has recognized that AI may be the next step in the evolution of credit underwriting, and that the law and regulators need to adapt to both facilitate and regulate this development,” says the BPI-Covington document. A stated goal is to “educate Congress on the opportunities and challenges posed by these technologies and what we can do to produce the best outcomes for consumers.”
The discussion paper calls attention to “asymmetry” in oversight and enforcement for banks versus nonbanks. It says the advent of AI necessitates “an updated and consistent regulatory approach for applying core regulatory principles clearly and uniformly to bank and nonbank lenders in examination and enforcement.”
The paper notes that the CFPB “has the authority to bring enforcement actions against banks and nonbanks alike, and examination authority over both large banks and certain types of nonbank lenders (known as “larger participants”). However, in practice, most nonbank lenders tend to face limited fair lending examination and enforcement from the CFPB,” and scrutiny varies at the state level.
“Similarly, oversight is strikingly different with respect to model risk management,” the BPI-Covington draft continues. “Banks and other depository institutions are subject to the Model Risk Management Guidance. Nonbank lenders do not face any comparable limitations on model development and use. While the guidance purports to be risk-based, noting that 'details may vary from bank to bank,' some banks have reported that the guidance has been applied as if it were a mandatory rule.”