The CECL implementation approach should be broken into four phases.

NAFCU FORT LAUDERDALE, Fla

The year 2022 may sound like it’s far off, but it’s anything but when it comes to preparing to comply with the new CECL accounting standard, according to one credit union CFO.

That date is two years after banks need to be compliant (Jan. 1, 2020), and Wright noted the bank examiners are  struggling with small institutions, where data validity and model validity are issues, as is balancing reasonable expectations and resource constraints. NCUA grappling with the same issues, said Wright.While the Financial Accounting Standards Board (FASB) continues to refine some of the issues around the new Current Expected Credit Losses (CECL) standard, Doug Wright, CFO with San Diego-based Mission FCU, noted credit unions need to have a fully CECL-compliant model in place by Jan. 1, 2022.

“One recommendation I would make is to not just look at what NCUA is doing with education, but also to look to the FDIC, the OCC and the Federal Reserve to see what they are doing,” Wright told the NAFCU CFO Conference.

Four Phases

The CECL implementation approach should be broken into four phases, according to Wright.

• Phase One: Road Map. This is the assembly of the implementation team and getting people somewhat up to speed on what CECL Is, as well as beginning to gather data. “I’d say most credit unions are in phase 1 or entering Phase II.”
• Phase 2: Modeling and scenarios. “This is buy vs. build.”
• Phase 3: Final model and validation.
• Phase 4: Post implementation.

At this point, the CU should be to the point of methodology/model evaluation, according to Wright, who added that by 2021, credit unions should be at the point of a final refining of their respective models.

Data Collection

Noting there has been some “alarmist” messages to date on the amount and history of data a credit union should have collected in order to be CECL compliant, Wright said most of those messages aren’t true. “The data that is required is dependent on the methodologies you need to do,” he said, pointing to roll rate, vintage, PD/LGD, and discounted cash flow.

Many of those segments, Wright acknowledged, may lack statistically valid sample sizes. To address that, he said a CU should reduce the sample size by more statistically precise techniques, such as multi-variate regression. In addition, a sample size/lack of data can also be tackled with industry data, he said.

Meanwhile, Wright said the economic cycle “expectation: component of CECL presents some interesting questions, such as having valid data all the way back to pre-2007. “It’s likely that additional refinements will be required regarding data for the first two to three years after implementation,” said Wright.

Data Collection Recommendations

Wright offered these recommendations when it comes to collecting data:

• Identify the methodologies you are most likely to use
• Identify critical inputs required (loan specific info,  environmental and economic info)
• Do your best with historical data, but develop capabilities to retain and evaluate required data on a go-forward basis
• Collect as much “useable” info as possible on a loan-specific level
• ID how you are going to overcome small sample sizes

According to Wright, FASB is committed to allowing flexibility, but it has been largely “uninvolved” post-implementation. “This is my speculation, but I believe the industry over a period of time will start to converge down to a fewer number of acceptable methodologies and models, rather than a larger number.”

Wright further forecast:

• Regulators profess flexibility, but may gravitate to more common approaches (field examiner knowledge, institution comparability, vendor concentrations)
• Different methodologies may yield “better” results from different segments

Buying Vs. Building

If a credit union is considering either buying or building a platform, the model platform should include areas to input and describe, Wright said, including:

• Historical basis
• Current adjustments (qualitative and environmental)
• Reasonable economic forecasts
• Reversion to historical
• Other key assumptions (prepayments, contractual lives, extensions)

The model platform should be built with the future in mind, said Wright, and a credit union must ask itself how the program is going to be maintained and updated. In addition, the credit union must also consider discounted cash-flow specifics.

Modeling & Methodology Recommendations

When it comes to modeling and methodology, Wright offered these recommendations:

• Explore industry white papers/webinars on specific methodologies
• Engage with several vendors to view approaches
• If building internally, consider platform requirements, ongoing maintenance from quarter to quarter, validity/backtesting/auditability, and regulatory “defense”
• Applicability for other uses

Wright urged CUs to beware of “CECL compliant” claims and to look closely at validation/explanation capabilities. He further urged CUs to consider cost vs. “performance” trade-offs” and to not rush into a decision.

Wright said credit unions that have subprime concentrations, longer-lived assets or are operating relatively close to regulatory net worth/asset minimum (9% or below) should move sooner rather than later to begin complying with CECL.

See you in Clearwater Beach, FL 10/1-4/2019