The CECL Impact in 2023

By: Jeff Parker

The implementation of the Current Expected Credit Loss (CECL) standard in 2023 had a significant impact in reported allowance for credit loss and net worth balances. The implementation contributed to an overall increase in reported loss estimates, which increased by $5.8 billion in the first quarter of 2023 as approximately 54% of credit unions indicated adoption in Q1 2023.  The quarterly trend in credit union net worth increases from earnings totaled $4-$5 billion per quarter in 2022, and again in Q1 2023.  If it were not for the NCUA’s three-year CECL transition adjustment period, Q1 2023 net worth balances would have decreased approximately $1.1 billion as a result of the CECL adjustment to net worth.  The NCUA’s CECL adjustment provision saved the industry approximately one full quarter of earnings in 2023.  Unfortunately for the 8.7% of early adopters, this phase in approach was unavailable to them.

Regulatory Expectations for CECL

The National Credit Union Administration (NCUA) and other regulatory bodies have not prescribed specific methodologies for CECL models.  Policy statements emphasize the design, documentation, internal controls and validation of expected credit loss estimation processes.  Additionally, regulatory agencies have encouraged periodic independent validation by a party with appropriate knowledge, and technical expertise.   Suggested parties include internal audit staff, a risk management unit or a contracted third-party.

Common Problems with CECL Models

The transition to CECL has revealed several common issues that institutions may encounter around model accuracy and data management. Key challenges include:

  • Collateral Valuations: Difficulties in accurately assessing the value of collateral, often leading to discrepancies in loss forecasting.
  • Unrealistic Projection Periods/Prepayment Speeds: Projections that do not realistically reflect the length of loan lives can skew expected loss estimates.
  • Flawed Portfolio Segmentation: Segmenting loans by a loan type that does not correctly capture the underlying risk characteristics.
  • Poor Historical Data Quality: Reliance on inaccurate or incomplete historical data can lead to inaccurate model outputs.
  • Infrequent Updates of Model Inputs: When model inputs are not regularly updated, this can result in large fluctuations in loss projections, reducing the model’s reliability and increasing the volatility of the ACL.
  • Lack of Senior Lien Balance Data: Lack of accurate data on senior lien balances for loans secured by a junior lien can significantly alter the expected credit loss in the event of default.
  • Credit Tier Segmentation Redundancy: Problems arise when loans are additionally segmented by credit tier even though the model already accounts for credit tier movement in its loss estimation calculation.

What to Look for in a CECL Model

When testing a CECL model, scrutiny may be focused on the following aspects:

  • Data Inputs: Ensure the data is comprehensive and reflective of the credit portfolio. Specifically, this should include balances, charge off and recovery amounts (net charge offs), and other inputs significantly affecting the loss estimate output.
  • Projection Assumptions: Most applicable to the Weighted Average Remaining Maturity (WARM) method. It is important to understand how balances are projected to future periods and what assumptions are governing those projections.
  • Loan Segmentation: CECL requires segmentation of the loan portfolio or pooling of loans with similar risk characteristics.  An assessment of segmentation through the lens of typical borrower behaviors surrounding a given loan type is a prudent approach.  Consideration of such attributes as collateral type or occupancy status in segmentation can better segment portfolios based on similar risk characteristics.
  • Lookback Period: Most applicable to WARM and Vintage methods, assessing the lookback period for omission of periods with poor credit performance that reduces the loss estimate.
  • Collateral in Process of Liquidation: Loans where the collateral has been repossessed and held at net realizable value on the balance sheet should be excluded from the calculation data.

After much planning, anticipation, and some delays, we have now entered the CECL era!