CECL requires more than a tweak

Questions about how to implement CECL, the new accounting standard for “current expected credit losses,” are confounding many community banks.

CECL was adopted by the Financial Accounting Standards Board (FASB) in June 2016 to address concerns raised by a wide range of stakeholders following the 2008 financial crisis. The effective date for CECL implementation is 2020 for SEC-registered banks and 2021 for others.

Is 2017 year CECL “success” hinges on?

Under CECL, financial institutions will be required to include reasonable and supportable forecasts in a forward-looking credit loss estimate, rather than relying on past events and current conditions. Prior to CECL, banks could only report losses when they occurred. CECL will entail cross-functional changes to the end-to-end reserving process for financial assets measured at amortized cost.

According to the American Bankers Association (ABA), 2017 is a pivotal year when an effective implementation can be won or lost. To help community banks gain a better understanding of the processes surrounding CECL implementation, ABA invited a panel of banking professionals to discuss several issues that community banks find perplexing.

During the Feb. 1, 2017, discussion, the panel addressed:

• Regulatory expectations

• Forecasting

• Model implementation and validation

• Data challenges in the short and long run

Moderated by Mike Gullette, ABA vice-president of accounting and financial management, the panel included Bob Storch, chief accountant, FDIC; Todd Sprang, partner, CliftonLarsonAllen; and Jim Brannen, executive vice-president, chief financial officer and senior loan officer, Federal Savings Bank, Dover, N.H.

The CECL model

Laying the groundwork for the discussion, Gullette explained that the CECL model applies to expected credit losses over life of the loan or portfolio as well as to loan commitments and “held to maturity” securities.

Gullette said the expected life of the loan should be recorded at the time of origination, and it should consider prepayments and troubled debt restructurings (TDRs)—but not renewals.

As a starting part for the loss forecast, banks can take some average of historical “life of loan” loss experience and then use qualitative factors to adjust to an actual loss expectation over the remaining life of the portfolio.

Gullette noted that many institutions will find vintage analyses useful in this regard.

Regulatory expectations for CECL compliance

FDIC’s Storch stressed the importance of preparing for the new standard and making sure that the board and bank management understand what is required.

“Figure out what effective date applies to your institution,” said Storch. “Start identifying who within your institution should be involved on the team applying input on the standard. It should not be left to the accounting staff. Participation is needed across the organization.”

Storch also advised banks to leverage existing lending and credit risk management staff as the bank develops an implementation timeline, noting that the timeline should be completed sometime this year.

“Banks must determine what steps need to be taken and get started sooner rather than later,” he cautioned.

The standard provides considerable flexibility in how it can be implemented as long as the methods the bank is using to estimate lifetime expected credit losses achieve the objective or standard for its held-for-investment loans or its held-for-investment securities, not just its origination loans, he said.

Storch also noted that the financial assets to which the standard applies should not require costly modeling techniques. Compliance can be accomplished by adjusting ALLL methodologies across the organization.

“From an overall perspective, the agencies recognize that the initial effective data isn‘t going to be perfect,” said Storch. “But we do expect good faith efforts to implement the standard. Applying the standard and having a well-supported methodology is an evolutionary process so we expect it will take a few years. We also expect to see continued improvement and effort over time.”

Storch cautioned banks to continue applying the longstanding incurred loss method until 2020 or 2021.

CECL’s phased implementation

Sprang said his firm advises banks to break the implementation into phases. There should be frequent status updates with senior management, the board, and the auditors, he said.

“It’s important to get feedback along the way, and it’s also important to have frequent open communication with a lot of the parties involved so expectations are understood,” he added.

From talking to his colleagues at other community banks, Brannen has learned there are misconceptions regarding the magnitude of the project at hand.

Because CECL is scalable does not mean that it does not require a lot of work to implement.

Senior management and boards need to understand the implementation process cannot just be tweaked.

“You have to work back from your final implementation date. You have to understand the impact on capital at implementation,” Brannen said.

Brannen said community banks should begin to inventory their data now. To understand the various data points to include in your models, he recommends that banks view the demos offered by third-party providers—even if the bank expects to build its own expected loss models.

Historical risk rating information is not currently available from most systems and this is a problem, noted Brannen. He recommends that banks start to capture or preserve data sets that they can query going forward.

“Try to capture changes in risk ratings, types of loans, exceptions, credit scores, and any other data you can identify,” he advised.

Brannen also advised banks to bear in mind that the standard of reasonable and supportable forecasts can change because of market, economic, and other conditions.

Segmentation of loan portfolio

Under CECL, different loan terms and underwriting standards will result in different loss expectations, depending on the bank’s forecast of future economic conditions. This may require more segmentation of loan portfolios.

All of the panelists cautioned, however, that when a loan portfolio becomes too segmented, it becomes less useful and less cost effective.

“You need enough data to come up with a reasonable analysis, but granularity doesn’t improve the model; it can make it more difficult and not yield better results,” said Sprang.

Gullette added that banks are going to have to do the analysis either by grouping loans with specific terms in a separate portfolio or keep it in their existing call report and then do a separate qualitative analysis on the call report.

Prepayment assumptions

Analyses supporting life-of-loan loss estimates need to include prepayment assumptions, relieving bankers from needing to forecast for 30 years.

Storch further said that the reasonable supportable forecast period for longer-term loans is not the life of the loan, even after taking prepayments into consideration. The standard indicates that the bank revert to its long-term loss experience for the periods of the remaining life of the loan beyond its supportable forecast.

There is, thus, a blending of the forecasts—the bank’s adjusted loss experience within the forecast period, plus the unadjusted historical loss rates for the period covering the remaining term.

Commercial loans

While discussing a new requirement for public banks to disclose the amortized cost of loans by vintage, Gullette and Sprang agreed that the treatment for certain renewable commercial loans is one in which the industry has yet to coalesce. They also agreed that the intent of the standard is to identify when underwriting decisions were made.

“As a practical expedient, is it a new loan for accounting purposes or did we just extend it without any underwriting?” asked Sprang.

Build capital but not loan loss allowance

Storch said the regulators expect institutions to stay in accordance with GAAP so the same set of standards are used across all institutions.

If banks believe their CECL allowances will be larger, they should be holding more capital, rather than building their ALLL.

Effect on capital standards

Some community banks have asked if regulators will allow a higher percentage of the allowance to be added back to capital.

Storch explained that the impact on capital standards affects bank globally. The international regulatory body, the Basel Committee on Banking Supervision, may consider changes to bank capital based on the Expected Loss (EL) model. Global banks also are moving to an EL model and are likely to implement it sooner than U.S. banks.

In the U.S., the industry and regulators will be following Basel developments regarding capital requirements.

[A wild card in that regard is the Trump administration, which has made general negative remarks about international financial regulation, and which recently launched a review of financial regulation via executive order.]

OREO sales  

Some bankers have questioned how sales of “other real estate owned” (OREO) could be applied to determine the amount of expected chargeoffs. Gullette said loan recoveries captured in the OREO system will need to link to the bank’s loan system.

Lifetime line of credit

Some bankers are questioning how to treat demand loans in their expected loss forecasts. Storch said that, as with commercial loan renewals, he hasn’t seen any discussion papers addressing that issue and noted, “We (the industry) will need an agreed-upon approach for this.”

Forecasting

If the bank does not have local economic data, it can use state economic forecasts as a starting point in developing its own forecasts.

“The goal is to find a forecast that is germane to you and your market,” said Brannen.

Storch advises using a state-level economic forecast that reflects the level of the bank’s activity, including credit drivers and/or local university forecasts that reflect the consensus view of a number of people.

He said that these local forecasts would be more supportable than a national forecast.

Each bank needs to find a forecasting source that works with its portfolio and then look at how those forecasts are trending with other similar sources.

Model validation

All banks should already have a governance process that includes model validation. Storch said the agencies have a policy statement for ALLL about testing under the incurred loss model, and they will continue to apply it under the expected loss model.

Gullette added that the Public Company Accounting Oversight Board (PCAOB) has stated that back testing will be a big issue.

Deadline approaching

In closing, Storch offered a final word of advice: “Don’t wait till the last minute. FASB purposefully gave banks an extended period of time to do this. Use the time available and spread it across the time with staff and resources.”

You can hear the ABA event for yourself by registering for the on-demand presentation here. It is a for-pay program.