Regulators Scrutinize CECL Processes at Community Banks

When the new loan loss model was adopted by community banks in 2023, regulators looked for a “good faith effort.” Two years later, expectations are changing.

The accounting rule known as the current expected credit loss model, or CECL, went into effect for most community banks in 2023. It was a major shift in how institutions calculate credit losses and was often a big undertaking for smaller institutions with limited resources. In response, examiners at the Federal Reserve looked for a “good faith effort” from executives when they set their initial allowance estimates in 2023, according to an accounting roundtable hosted by the Federal Reserve Bank of St. Louis in November 2024. But as community banks gear up for 2025 exams, executives are expected to further refine and continuously improve their CECL processes and data, the Fed reported.

The phrase “good faith effort” is familiar to Mandi Simpson, a partner and the leader of the accounting advisory team at Crowe. Simpson remembers federal banking agencies, particularly the Fed, promoted the phrase when she was a professional accounting fellow in the Office of the Chief Accountant at the Office of the Comptroller of the Currency. But she recalls them saying another tagline alongside it: “Adopt and adapt.” That phrase, she says, is where bankers are today.

CECL requires banks to forecast the lifetime losses for their loans at origination, using models that include historical loss data, future economic forecasts and other variables executives deem relevant to loan performance and loss severity. The allowance for credit losses is the “biggest, most judgmental estimate” on a bank’s balance sheet, says Ashley Ensley, a partner in the financial services practice at accounting and audit firm Forvis Mazars. She says that many of her bank clients are entering a phase of “model maintenance or revisions.” These banks may not have had a final model in place until the end of 2022 and now have the time and information they need to refine that model. This can look like adjusting or incorporating economic factors or further segmenting loans to refine loss estimates. These adjustments have mostly resulted in “immaterial” changes to the estimated loss. She adds her clients who use the same vendor have also taken the intervening years to share their experiences and adjust their model or input based on feedback.

But the Fed’s accounting roundtable flagged that some banks’ allowance estimates weren’t catching the institution’s full credit risk, even as credit quality across the industry remains strong. That warning comes as regulators continue monitoring pockets of credit risk in areas like commercial real estate and consumer lending in their periodic risk publications. A survey of 2024 enforcement actions from the OCC and Federal Deposit Insurance Corp. supports the Fed’s assertion for both this and other issues; orders that mention credit run the gamut from underwriting and risk rating to loan review and the allowance calculation itself.

Ensley suspects that some institutions may need to drill down into their loans to segment them better. One area where this has happened is in commercial real estate portfolios, where executives have segmented loans into multistory office, owner-occupied, retail and warehouse categories to demonstrate appropriate risk ratings and allowance.

Executives need to properly risk-rate their institution’s loans and include how changes in underwriting could contribute to credit risk down the line. And executives should document these risks, as well as their understanding of how the risks could impact the allowance estimate, when discussing the allowance with examiners and the board.

One major area of focus for the Fed during the roundtable was the subject of qualitative factors. Qualitative factors, often called Q factors, are variables that influence losses that aren’t captured or incorporated into a loss model.

Under the prior incurred loss approach, some community banks’ allowance or loss estimates heavily relied on Q factors. Ensley says Q factors made up 80% to 90% of some of her clients’ allowance estimates. Today, her clients are “using far fewer qualitative factors” to calculate their allowances than before, and those Q factors represent “a smaller portion of the reserve itself.” She says one reason is that community banks that used simpler estimation methods in the past opted to use more complex models now from vendors to calculate CECL. Q factors remain an important way for executives to apply their judgment to a loss estimate and ensure their allowance captures all the risks they see. But in the accounting roundtable, the Fed flagged they’ve seen inadequate support for, or identification of, qualitative factors during examinations. Ensley says auditors have made similar observations.

Before adoption, Simpson says some predicted that “CECL would be the death of Q factors,” but experience has borne out that that’s not true. Going forward, Simpson and Ensley advise executives to document which of the potential Q factors listed in the standard are already captured in their base quantitative calculation, which Q factors they need to adjust outside of their model as well as why they’ve elected to exclude certain factors from the calculation.

Of course, a loss model is only a tool that can’t make up for poor credit risk management or underwriting. CECL was intended to bring credit and accounting functions together, but Simpson points out that the credit department should inform the accounting estimate. Credit teams should provide information and documentation to the accounting team about a loan’s underwriting, appraisals and risk ratings.

“Particularly for community banks, … if it’s not documented, it’s not done,” Simpson says.