Written By: Neekis Hammond

In a July 2016 webinar with more than 220 bankers attending, executives at banks and credit unions spoke up with some of their most pressing questions related to the implementation of the current expected credit loss (CECL) model.
Neekis Hammond CPA and senior risk management consultant at Abrigo addressed some of the questions during the webinar and archived the responses below for the ALLL.com audience.

For the database of loan history that’s required under CECL, should we include loans that have already been paid off, matured, prepaid, etc.? Or just loans with outstanding balances?

Neekis: At this point, it’s best to not limit the data you’ll have available, so this means not filtering your loan tape. For more information about the types of data you’re likely to need under CECL, download our CECL Data Prep Guide, which explains data points that could be important and makes some recommendations on accessibility and quality of information.

Can you use different methodologies for different segments? And do you use what you feel is most appropriate based on what you have for data, what you think is most accurate, etc.?

Neekis: Yes. The FASB has been non-prescriptive by design about which methodologies to use; instead they recommend choosing the methodology that is most reflective of your individual portfolio. Rather than use a specific methodology, calculate losses using each method, and use the method that results in the most reasonable reserve calculation for your institution. In fact, Lawrence Smith (FASB Board Member) presented (see the video at 1:33:50) changes in reserve levels due to variability in methodologies. Representatives from the AICPA, SEC and various regulatory bodies were present and made no objections.

We are evaluating a change from 36-month average losses to vintage analysis (though the question applies to any change i.e., migration, PD/LGD), what are the accounting implications of the “change in methodology”?

Neekis: Examiners have said specifically, they are looking for something that is “consistently applied”. You can change methodologies post-implementation, but you will need a supportable cause in order to execute. A few cases that offer some support include:

Moving to a more sophisticated model that should be more representative of loss. A de novo bank, for instance, would not have the data to perform a historical loss rate calculation even under GAAP today, so they could move toward a true historical loss rate after their de novo status ends.

Changing methodologies because the current methodology is not reflective. For example, if migration analysis for your bank is resulting in a reserve level that’s artificially high or low, it may require a change. To substantiate that claim, it would be important to backtest the model and see how actual compared to modeled losses.

However, because each methodology will require different data points, institutions would be well served to understand how each loan type/portfolio reacts given various methodologies. Therefore, prior to electing a specific method, comparing the results under different scenarios will prove to be a valuable exercise that will continue to pay dividends in the long-run.

What happens to FAS 114 impaired loans with individual specific impairment reserves under CECL?

Neekis: Individual impairment analysis as dictated in FAS 114/ASC 310-10 does not go away under the new CECL standard. Rather, your 114 process will still remain intact and very similar to the calculation you are using today.

ASC 326-20-35-2 and 326-20-35-4 explain: assets that no longer exhibit similar risk characteristics for a variety of reasons should be individually evaluated. Also, regardless of the initial method, when an entity determines that foreclosure is probable, the entity will need to re-measure the asset at the fair value of the collateral; similar to today’s process.