The effective date for the current expected credit loss (CECL) standard is fast approaching for all financial institutions that have not yet implemented it.
While most of us have been holding out hope for yet another extension, in November, the Financial Accounting Standards Board (FASB) all but guaranteed that CECL will move forward as planned with an effective date of Jan. 1, 2023, for non-public companies. This realization came after FASB denied a request for another two-year extension, indicating they no longer see the COVID-19 pandemic as a barrier to implementation.
As we approach the January date, banks should be past the planning stage and well on the way to implementation. The
following are a few practical considerations as we enter the home stretch:
The Process Can Be Outsourced, Not the Responsibility
Outsourced solutions have become increasingly popular, especially for smaller banks. A few years ago, discussions
centered on whether to develop a CECL model internally or purchase third-party software to handle the leg work.
As more software models hit the market, it became clear that the affordability and ease of use made this option preferable to the time and effort it would take to develop an in-house model.
While software can automate much of the process, management still bears the ultimate responsibility for the estimate. Key duties of management include evaluating and setting risk characteristics of the bank’s loan pools, developing qualitative factors and forecasts, understanding the methodology used, and identifying how changes to inputs impact the estimate. Good software can make the process easier, but it can’t replace your knowledge about your bank, your customers, and your risks.
Another important consideration when outsourcing is the control environment of the vendor providing the software.
A service organization control (SOC) audit report can be obtained from vendors providing CECL software and reviewed for areas of concern. Items to look for include issues with vendor controls that may materially impact the CECL calculation and complementary user controls identified in the report. Complimentary user controls are those the vendor
recommends the bank has in place to properly use their software. If the vendor’s SOC report identifies significant
control deficiencies, it may be necessary to reconsider the reliability of the vendor’s model.
Conduct Trial Runs
If your bank is not yet conducting trial runs of your CECL model, now is the time to start. It is best practice to run the
ALLL and CECL models concurrently for at least two quarters to assess the impact CECL will have on the allowance
leading up to implementation. This will help identify weaknesses and refine sensitivities in the calculation. Many banks may need to adjust their approach based on the initial results of the calculation.
Be Ready on Day One
The implementation date for CECL is January 1, and the impact of implementation should be reported on the first-quarter call report. Banks will need to run the CECL calculation Jan. 1, 2023. The result of the calculation is compared to the bank’s ALLL calculation Dec. 31, 2022, with the difference recorded directly to retained earnings as a cumulative-effect adjustment for the adoption of CECL. After this initial adjustment to retained earnings, future changes in the CECL estimate will be recorded through the income statement in the same manner as the ALLL.
It is also important for banks to work with their auditor on the front end to identify information required for financial
statement disclosure under the new CECL requirements. New disclosure information can be accumulated throughout
the year to ensure adequate data is available when the time comes to prepare financial statements.
Document All Components
The CECL calculation is comprised of three components: historical loss rate, qualitative factors, and reasonable and
supportable forecasts. While the historical loss rate is the base for the calculation, qualitative factors and forecasts are equally as important. FASB has identified qualitative adjustments and forecasts as significant judgments in the calculation that require proper support and documentation. They have also made it clear that there is no cookie-cutter
approach to CECL, affording banks a broad latitude in how they develop and document these two components. While this flexibility allows banks to tailor the calculation to their own risks, the lack of a standard approach can make it
difficult to evaluate these factors.
Auditors and regulators will hone in on these specific areas. Expect questions as they work through the calculation and gain an understanding of your methodology. The key is robust documentation and the ability to back up each element of the calculation.
CECL Isn’t Only About Loans
For banks, the primary focus has been the impact on the loan allowance calculation but CECL also applies to a number
of other financial instruments carried at amortized cost, including:
- Held-to-maturity debt securities
- Trade receivables
- Receivables related to repurchase agreements
- Finance leases
- Off-balance sheet credit exposures such as loan
commitments, standby letters of credit, and financial
If your bank holds these assets don’t be caught unprepared. Unlike the incurred loss model, the CECL model does not
specify a minimum threshold for recognition of an allowance. Therefore, banks will need to evaluate expected credit losses on these assets even if there is a low risk of loss. In some cases, the resulting loss may be zero.
While the CECL model does not apply to available-for-sale debt securities, the standard made targeted changes to
existing accounting for available-for-sale debt securities that are impaired. Most notably, removing the “other-than-temporary” impairment concept and requiring the use of an allowance when security is impaired as opposed to permanently writing down the cost basis. Expect expanded financial statement disclosures in this area but no significant
change in accounting for available-for-sale debt securities unless specific securities are identified as impaired.