Checklist: Questions To Strengthen Your ACL Process

Audit committees must exercise oversight of the significant judgments and estimates that bank management uses in estimating allowance for credit losses (ACL). During our October 2025 “Financial Services Audit Committee Overview” webinar, Mandi Simpson, Crowe partner and accounting advisory leader, offered a checklist to equip audit committees and internal audit leaders with focused questions to apply a credible challenge to ACL estimates.

✓ Segmentation: Have we evaluated whether any pockets of risk in our portfolio warrant separate segmentation for ACL estimation purposes?

Loan segmentation under the current expected credit loss (CECL) model is not a “set it and forget it” assumption. Institutions must reassess whether loans grouped together continue to share risk characteristics, whether new risks have developed that would warrant additional segmentation, or whether old risks have dissipated such that less segmentation is appropriate.

Questions to ask:

• Have we revisited our segmentation logic since CECL adoption?
• Should any new, emerging, or changing risks, such as industry-specific pressures, prompt us to regroup or subsegment certain loans?

Insight: At the start of the pandemic, many institutions segmented hotel loans due to elevated risk associated with the lack of travel. When travel conditions improved, those loans often could be reintegrated into their original segments. For sectors currently facing evolving risks, such as agriculture or commercial real estate, now might be the time to consider if different segmentation is needed.

✓ Forecasting: Are we capturing the full scope of potential future impacts?

CECL models require institutions to consider “reasonable and supportable forecasts about the future” in their ACL calculations. However, the forecasts for some smaller, less complex institutions might incorporate only broad macroeconomic variables applicable to large portions of the portfolio without considering specific variables that are more narrowly focused but that could have a significant impact on expected credit losses.

Questions to ask:

• Which macroeconomic variables (for example, GDP or unemployment) are currently embedded in our model?
• Do those variables fully reflect any unique risks of our portfolio?
• Should we supplement the model with geographic or sector-specific forecasts?

Insight: If borrowers are being squeezed between rising production costs and falling commodity prices, as is common in agricultural lending, macro indicators alone might not capture the risk. Additional forecast elements or qualitative overlays might be needed.

✓ Qualitative factors: Do we have risks for which segmentation or specific forecasts are not possible?

Quantitative models can go only so far, so it’s important to ask what’s missing or different from the data that an institution has used in the base quantitative estimate of credit losses and to determine what adjustments need to be made on top of the base calculation. Qualitative factors can help fill those gaps by addressing differences between current loan conditions and historical data or model assumptions.

Questions to ask:

• Do we have differences in underwriting standards, portfolio mix, or terms within our current portfolio that do not reflect our historical loss information?
• Do we have any expectation that current conditions and reasonable and supportable forecasts will differ from the conditions that existed for the period over which historical information was evaluated?
• Have we identified any unique risks where additional segmentation or forecast variables are not feasible that we need to capture through a qualitative adjustment?
• Are we documenting and justifying qualitative factors with enough rigor?

Insight: Qualitative factors can be especially important when the portfolio includes loans that are too few in number to form their own segment but still carry distinct risks that need to be captured.

✓ Individually evaluated loans: What is our process for identifying loans that should be individually evaluated?

The CECL model requires individual evaluations for loans that either are collateral-dependent or do not share risk characteristics with others.

Questions to ask:

• What is our process for identifying loans that should be evaluated individually?
• Are any niche or one-off loans, such as those in unique industries or with unique risks, being missed?
• For loans that are not collateral dependent, are we considering more than just the fair value of collateral in estimating the credit loss?

Insight: Collateral-dependent loans must be individually evaluated, but not all individually evaluated loans are collateral dependent. Those in an oversight role should ensure the criteria used to identify and methods used to assess such loans are appropriate and in line with the accounting guidance.

While staying compliant with CECL is the minimum standard for any institution, the true goal is to provide robust, risk-aware oversight. These questions can guide audit committees, identify gaps, apply governance pressure, and promote sound allowance assumptions in an unpredictable credit environment.