As 2023 filers look towards CECL adoption, two questions that continue to come up are
- “What type of qualitative adjustments will be required?”
- “What type of support will be needed to justify these adjustments?”
Generally, these questions arise from filers’ previous experience with ASC 450 where many banks and other financial institutions relied heavily on qualitative adjustments as a key component of their reserve numbers. Filers must also consider the added complexity of CECL and the greater audit and regulatory scrutiny that qualitative adjustments have historically received. In this blog, we will explore these two questions.
While the regulators have tried to address the types of adjustments that financial institutions should consider in their CECL reserves in an interagency statement, the usefulness of the guidance is limited in practice because of the lack of consistency in methodology across institutions. Because CECL is a fundamentally different reserving approach than ASC450, the nature of the qualitative adjustments will also change; however, qualitative adjustments will continue to represent adjustments to the allowance calculation not considered by the quantitative calculations.
Since CECL is a new standard that is conceptual rather than a prescriptive standard, and the methodologies to address the CECL requirements are diverse, qualitative factors will be varied as well. Knowing the right qualitative adjustments is a direct result of knowing your models and model limitations. At its core, CECL requires a life of loan expected loss estimate that includes the impact of macroeconomic conditions. As a result, the process of determining the appropriate adjustments will be a result of understanding where your methodology deviates from that end and how the losses forecasted by the model deviate from loss expectations of your specific portfolio. That makes the selection of a “right-sized” model and management’s understanding of the model of paramount importance so that the management team can ensure relevance and adjust as necessary.
Once management understands the models, qualitative adjustments can be used to address model limitations and naturally flow from these limitations. For example, if a bank picks a model that does not incorporate the impact of the macroeconomic forecast, this adjustment will need to be done through a qualitative adjustment. In recent quarters, this adjustment has been material and is expected to continue to be. Another example may be for models that don’t allow for the use of prepayment speeds and rely instead on contractual maturity, an adjustment may be required to account for prepayments where it is material to the portfolio. Because of the materiality of both prepayment speeds and the macroeconomic forecast, we have incorporated these adjustments directly within the quantitative allowance for our models; however, for those models that do not incorporate these impacts, management will need to make an on-top adjustment for these items. A different type of qualitative adjustment may be needed for banks that use a peer loss rate as a basis for their methodology to account for the variances in the underlying portfolio.
When it comes to quantifying and supporting the adjustments, this is another area of management judgment and will be largely dependent on the nature of the adjustment itself. However, as best practice, management will want to ensure that any application selected to help with the quantification of the CECL reserve:
- Allows for multiple scenarios to be run each period to help truly understand the specific drivers of the reserve and potentially quantify adjustments.
- Allows for supporting documentation to be stored so that there are controls and traceability over the process and allow for a robust approval process given the risks that arise from qualitative adjustments.
For more information about the types of qualitative factors to consider and how to adjust to the new CECL changes, contact us.
Written by Theresa Meawad
Senior Director & Head of Solutions Consulting, SS&C EVOLV