ASC 326 Current Expected Credit Loss (“CECL”) brought many changes to the allowance process but one item remained the same: the need for qualitative factors. While many may have hoped that reliance on qualitative factors would be largely eliminated, extremely low historical loss experience and model limitations have resulted in lower-than-expected quantitative losses and supported the continued need for qualitative adjustments. CECL guidance states that the allowance calculation is a combination of historical losses, as well as adjustments for current conditions and a reasonable and supportable forecast of future losses. To the extent the bank believes the historical experience in their CECL model does not represent a bank’s expectation of future losses, banks should consider the use of qualitative factors.
2020 Interagency Policy
The 2020 Interagency Policy Statement noted that historical credit losses alone were generally not sufficient to determine the appropriate levels for the Allowance for Credit Losses. As such, management should consider qualitative adjustments relative to the institution. Additionally, reviewers like to see qualitative factors assigned to either internal or external data series where possible. Below are the nine factors included in the policy statement, potential application in CECL, and possible relevant series data:
While these factors have been suggested by regulators, banks are not required to reserve for each of them. Additionally, banks are not limited to these factors alone and may chose to apply an adjustment that is specific to their institution, region, or current conditions. Each bank should apply the qualitative factors that best reflect the current and future conditions anticipated to impact the expected losses for their portfolio. It is also important to note that some quantitative methodologies may be picking up concepts in the factors above in the model and an institution must be cautious of double-counting.
While the CECL standard is fairly open-ended, one requirement that is clear is the need to include a reasonable and supportable forecast. For many banks, a mechanical forecast may not be an option given their lack of relevant data or the resources required to complete that level of analysis. However, regulator guidance has also stated that they do not expect banks to develop a model that is beyond the size and complexity of the Bank. For many community banks, qualitative factors will likely be the mechanism used for the reasonable and supportable forecast. For those banks we suggest following the steps:
- Decide on a Forecast Source – First contemplate any forecast you may be using in another functional group to be consistent across the organization such as the ALM process. It is common to contemplate downside scenarios as economic forecasts are typically inaccurate and can change very quickly. This has been especially true in times of recession or when the forecasts are especially optimistic, yet economists are predicting a chance of recession in the next 12 months.
- Decide on the Forecast Length –Determine and document what period of time you are trying to capture in your forecast. Generally, a forecast covers 1-3 years and most commonly banks have chosen 12-24 month periods. If your forecast period is less than the forecast source, then do not contemplate the remaining forecast in your reserve. Note that the concept of reversion is not as practical with Q factor forecasting due to the fact that you are starting with the long-term average expected loss results and working backward, as compared to a mechanical forecast.
- Assign the Forecast to a Qualitative Factor – Typically, forecasts are contemplated in the qualitative factor: “Actual and expected changes in international, national, regional, and local economic business conditions and developments in which the institution operates that affect the collectability of the financial assets.” Generally, models forecasting through Q factors will be applied to expected losses based on a through- the-cycle rate result. Therefore, the forecast adjustment would be to adjust that through-the-cycle lifetime expected loss rate to account for what you would expect the variance to be during your forecast period. This adjustment could be positive or negative.
Qualitative Factor Development
Many institutions have a desire to remove some of the subjectivity that has traditionally existed with qualitative factors. It is expected that factors that are more quantitative in nature will be more defensible in an audit and are also more likely to predict future losses. That is not to say that factors cannot be subjective, as long as they are thoroughly supported and documented.
The goal is to minimize the subjectivity and QUANTIFY THE QUALITATIVE.
Guidance for the development of qualitative factors, other than use of relevant data and the documentation of assumptions, was not included in ASC 326 or regulator interagency statements. That is by design as banks are to develop a model that best fits their portfolio. Unfortunately, this leaves management questioning how to begin developing their qualitative framework under CECL.
Many banks may choose to start with their existing qualitative factors under ILM and simply modify them for CECL. This may be achieved by listing your current qualitative factors and documenting how they may apply within a CECL model as well as contemplating any new factor as shown in the example below. Banks do need to take this opportunity to consider novel approaches and updates to processes that may have been developed quite some time ago, as well as shoring up documentation and supporting their qualitative factors.
As banks begin to work through their framework, they are considering ways to anchor their qualitative factors. One highly effective approach is to develop a scale to apply to the various relevant qualitative factors. This approach would require the bank to gather historical loss rates for each segment or the overall loss rate. These historical loss rates could then be compared to each pool or cohort’s current calculated loss. An example below demonstrates how the scale development process could work:
The bank’s current calculated loss rate is .50% and will be used as a baseline. The graph above shows varying historical loss rates experienced 2007-2021, capturing a full economic cycle.
- Select a maximum expected loss rate, in this case 1.50%, the 4-year recession loss rates
- Select a moderate risk loss rate, such as the lifetime quarterly loss rate average of .80% or even consider half of the maximum expected loss rate
Next, if we assume 5 severity levels & only one Qualitative factor per cohort:
- No change would be 0%
- Significant Risk would be 100 bp (1.50% – .50%)
- Some Risk would be 50 bp (100 bp/2)
- Improvement levels should be the absolute value |x| of Risk levels
Ranges can also be developed by utilizing call report or peer data, stress test results, or CECL scenario results. If more than one qualitative factor is used, the results will need to be scaled down so the combination of results will not exceed the maximum loss rate used. Also, considering the potential volatility of a CECL model, consider reviewing or updating the scale annually.
Based on the data in the chart above and assuming 12 cohorts and 4 equally weighted qualitative factors, a $1 billion portfolio with a maximum expected loss of .50% would have risk factors of .25% and .13%.
Qualitative factors are here to stay, and many banks are finding themselves through the modeling part of the CECL process, only to now contemplate their qualitative adjustments. Rest assured CECL results are going to be under a great deal of audit and regulator scrutiny and with that comes increased emphasis on qualitative factors. Spending time now on the development and documentation of your qualitative factors will prove valuable.
About the Presenters
Shannon Morrison, Managing Director of Consulting
Shannon has almost 20 years of experience in the banking industry with a specialty in accounting and financial reporting. Shannon has managed the ALLL process and led the implementation of CECL at a variety of financial institutions on the ValuCast platform. Shannon has accounting and management experience at banks up to $21 billion in total assets. At Valuant, Shannon specializes in CECL readiness and implementation and advisory services. In addition, she is responsible for the internal training and development programs for the Client Service Team. Shannon obtained her MBA from the University of South Carolina. She attended the Graduate School of Financial Management and Bank Investments at the University of South Carolina, the SC Bankers School in Columbia, SC, and the AIB Principals of Banking in Hartsville, SC. Shannon obtained her BBA from Furman University. In addition, Shannon is a Certified Government Finance Officer.
Natalie Smith, Financial Analyst
Natalie specializes in CECL Readiness, post implementation support, and special projects related to Income Recognition, Exit Pricing, and Stress Testing for Valuant’s clients. Her experience prior to Valuant includes time as a Financial Representative at a banking institution.
Natalie attended the University of South Carolina where she earned her Bachelor of Science in Statistics with a focus in Actuarial Science and Minored in Risk Management and Economics.