Quantitative Supervision Research Paper Highlights

Banking is evolving as financial institutions reshape their branch networks. New research shows that while banks are closing branches in areas with more financially sophisticated customers who are comfortable with digital banking, they're also strategically opening new branches in these same areas to attract deposits away from competitors. This pattern highlights the changing role of physical locations in modern banking.

The rapid advancement and widespread adoption of artificial intelligence (AI) has fundamentally transformed how firms operate across industries. In the U.S. banking sector, AI applications range from customer service and fraud detection to trading and risk management. Prior research has shown that AI can boost sales, innovation, and product quality. However, there is surprisingly little evidence that it improves operational efficiency. This raises a key question: what are the hidden costs?  

Our research finds that these benefits are accompanied by significant operational risks that demand careful management. Banks with higher AI intensity incur greater operational losses than their less AI-intensive counterparts. This relationship is primarily driven by losses due to external fraud, problems with clients/customers, and system failures. AI’s negative effect is especially pronounced for banks that lack strong risk management, suggesting that robust internal control is an important prerequisite for successful AI implementation. In this Paper Highlight, we present the findings of our related research paper and implications for bank management and banking supervision.

How do economic conditions affect the ability of banks to recover losses due to operational risk? Operational losses arise from unforeseen circumstances, human or system error, inadequate or failed internal processes, as well as legal proceedings. The risk of such losses in the financial sector has emerged as a key focus for financial regulators, and research has shown that operational losses increase during economic downturns. Once an operational loss has occurred, banks may pursue various internal and external channels to recover, at least partially, some of the funds that have been lost. Frame et al. provide new evidence that the ability of bank holding companies to recover operational losses declines in downturns as well.

Bank holding companies recover about 7.6 percent of operational losses on average, and recovery rates are positively correlated with the strength of the economy. One plausible explanation for this relationship is that the risk management functions of bank holding companies face resource constraints during downturns. Bank holding companies with better risk governance have higher operational loss recovery rates during economic downturns, suggesting that they are less vulnerable to these resource constraints. These findings provide new evidence on how economic shocks transmit to banking industry losses, with implications for risk management and supervision. Losses during a downturn may ultimately depend not only on a bank holding company’s risk exposures before the downturn, but also on its ability to recoup losses after the fact. As a result, recovery plans and frameworks may need to be taken into consideration by banking supervisors when estimating losses in stress testing exercises.

Given the role that the housing market played in the financial crisis of 2007-2009, policymakers and regulators subsequently subjected mortgage lenders to greater scrutiny. For example, in 2009, the Federal Reserve issued new rules extending consumer compliance supervision to nonbank mortgage-originating subsidiaries of bank holding companies. However, these new rules did not apply to independent nonbank mortgage originators. Did this policy change affect the relative riskiness of mortgages issued by the two groups of nonbanks?

Balla et al. study this question using a large sample of mortgages originated by independent nonbanks and nonbank subsidiaries of depository institutions and bank holding companies from 2000 to 2015. Before the Fed’s policy change, mortgages by subsidiary nonbanks had a higher probability of default than mortgages issued by independent nonbanks; after the policy change, the opposite was true. Additionally, there was a small but statistically significant decrease in loan interest rates and loan-to-value ratios for subsidiary nonbanks relative to independent nonbanks after the policy change. These results hold for prime mortgages. For subprime mortgages, subsidiary nonbank originators had higher interest rates and lower loan-to-value ratios than independent nonbank mortgages after the policy change. While the authors do not identify a causal link between the changes in mortgages originated and the Fed’s policy change, their findings are consistent with bank holding companies reducing risk shifting in mortgage lending across subsidiaries after being subjected to greater regulatory scrutiny.