A decade after regulators began requiring banks to validate the effectiveness of their financial models, a talent shortfall is hampering the industry’s efforts to comply.
More than seven in ten respondents to a recent survey said a talent shortage is a challenge to carrying out the requirements, more than any other answer.
Some banks use as many as 800 different models to guide loan decisions and other determinations across their businesses, according to the poll of 62 banks, which was conducted by the Risk Management Association.
"When you look at the sheer number of banks and the number of models across the industry, you can see why there’s a real talent gap needed to validate them,” said Ed DeMarco, general counsel at the Risk Management Association.
Banks’ financial models require testing as part of a painstaking process that the Federal Reserve, the Office of the Comptroller of the Currency and later the Federal Deposit Insurance Corp. began examining more closely in the wake of the 2008 financial crisis.
Regulators found that the models much of Wall Street relied upon broke down at the height of crisis, producing loss estimates that were “ludicriously below” what was realized on subprime mortgage securities, researchers wrote in a Risk Management Association report.
Six in 10 banks said they now validate their riskiest and most complex models every two years, according to the survey.
But a boom at digital finance companies and the lure of tech giants have siphoned qualified candidates away from traditional banks, which are more often having to outsource the work, the survey found.
Specific skills are needed to conduct the validations, said Chris Nichols, director of capital markets at the $46.2 billion-asset SouthState Corp. in Winter Haven, Florida. It’s crucial to rely on tech workers who demonstrate enough creativity to solve problems that arise from frequent gaps in data, he said.
Recently, as more qualified talent has gone to major tech companies, models have also become more complex, often relying on machine learning abilities, according to Nichols.
“Model validation can be daunting,” he said.
The Independent Community Bankers of America has been in talks with federal banking agencies about allowing small banks to collaborate, share resources and personnel, and recognize common standards that could make the process more efficient, according to Michael Emancipator, the group’s vice president and regulatory counsel.
About eight in 10 community banks report paying outside companies to help handle digital products, according to the ICBA. But current guidance requires the banks to have in-house employees who know how to manage the relationships.
“This can create a certain choke point when you consider the limited number of experts in these fields, coupled with the Great Resignation now unfolding,” Emancipator said, referring to the pandemic-era wave of Americans who have quit their jobs.
In the survey, about 72% of respondents listed “talent” as a challenge to model validation, followed by 63% who cited “costs.” Some 37% of those surveyed pointed to “bank resources,” and 9% listed “technology.”
Even banks with the resources to address the model validation process head-on must pay close attention when buying or building a new model or additional data sets, Nichols said.
Bank management can face a “tough decision when they are trying to figure out the tolerance for the margin of error in their validation practice,” he added.
“Larger banks need to be more rigorous. So while the talent is more likely available, the exercise becomes more costly and relatively more time-consuming,” Nichols said.
