By Josh Stein and Mike Gullette
ABA Viewpoint
Josh Stein is VP for accounting policy, and Mike Gullette is SVP for accounting policy, at ABA.
Community banks have been clear: the cost of CECL isn’t just the initial implementation. The recurring burden shows up every period in the same places—qualitative (“Q”) factor analysis, model validation expectations, and audit documentation requirements.
What’s striking is that these costs arise less from estimating credit losses and more from defending the estimation process itself. That distinction matters at a moment when Financial Accounting Standards Board evaluates whether CECL is delivering decision‑useful information at a reasonable cost. After all, the ongoing costs of CECL were assumed by FASB to be similar to those existing at the time.
The hidden cost driver: Q factors become the main event
In theory, CECL models produce a disciplined estimate anchored in historical loss experience plus reasonable and supportable forecasts. In practice, for most community banks, the model output is only a starting point.
As ABA has heard repeatedly, qualitative overlays and Q‑factor adjustments often represent at least as much as the base model output — and in many cases, more. When that occurs, it is not evidence of poor modeling. It is evidence that CECL, as applied, has shifted judgment from credit analysis to documentation.
CECL explicitly permits qualitative adjustments, but it does not explain — particularly for non-complex institutions — how those adjustments should be quantified or documented in a way that satisfies audit expectations. That gap has become a major practical problem.
Validation: critical at adoption, costly in steady state
Most stakeholders agree that validation at the time of adoption is important. Community banks, however, draw a clear distinction between validation at adoption and expectations once models, portfolios and outcomes are in a steady state.
Many report that even after processes stabilize and allowance levels change little, auditors continue to expect levels of validation effort that are appropriate at adoption but not on an ongoing basis. This persists even as banking examiners often apply a more proportional, risk‑based approach in their reviews.
Audit expectations shape CECL more than the standard itself
A recurring theme from ABA roundtables is that CECL’s real operating rules are often defined by audit practice, not by the accounting standard itself. Examples shared by community banks include a 14,000‑word, 28‑page document to update a Q‑factor framework and a 125‑question CECL‑specific audit checklist.
These are not examples of weak controls. They reflect an implementation environment where CECL functions as a documentation and validation exercise as much as a credit risk estimate. They reflect a costly and recurring process that adds no value to credit loss accounting.
A practical path forward requires recalibrating expectations so CECL work focuses on what matters most: changes in credit risk, not exhaustive efforts to compensate for model limitations.
In the third and final essay in this series, we’ll show how publicly available data and peer ranges can anchor CECL estimates — shifting the focus away from defending model mechanics and back toward evaluating credit risk.
