Risk Management & Analytics: CECL modeling and qualitative considerations

Jeff Prelle, managing director, head of risk modeling for MountainView Financial Solutions, a Situs company, recently spoke at the RMA Rocky Mountain Risk Management Conference about CECL modeling and qualitative considerations.

Who was in the audience, and were you trying to address a specific pain point facing the audience relating to CECL modeling?

Jeff: The audience was regional banks in the Mountain and West Coast regions. I spoke about modeling considerations of CECL and the potential for additional capital hold for assets with longer duration. I also spoke on prepayment considerations and the effect on duration of assets.

What were the strategic challenges that CECL modelers need to keep in mind?

Jeff: I spoke about the economic considerations when creating these models and the potential benefit of:

1. Modeling any economic scenario models in conjunction with the loss models at larger institutions that will employ multiple economic simulations. This creates the benefit of determining the reasonable and supportable periods for CECL models and the best avenue for reversion to the mean.
2. I also spoke about the importance of testing models under multiple economic conditions, which will be relevant for all institutions regardless of size. When creating the models, the institution should be aware how the model performs under benign, adverse and positive economic conditions. Some models may function only under one or two of these conditions. When this is the case, qualitative factors will be necessary to adjust the results.

What are some of the more tactical implementation challenges that CECL modelers need to keep in mind?

Jeff: When using qualitative factors, the institution will be required to document thoroughly its rationale and support for these factor adjustments. Support and justification for these factors will require evidence that the company is not managing earnings through the Allowance for Loan and Lease Losses (ALLL), and thus the rationale is supportable.

Institutions that went through the rigors of Dodd Frank Annual Stress Testing (DFAST) documentation requirements at the peak of the regulation cycle are familiar with documentation scrutiny requirements. Remembering, “even though discussions were had but were never documented, will mean the discussions never occurred,” is a good litmus test for the documentation requirements for the auditors who will look at CECL models.

You spent a lot of time focusing on data management, including data capture and data approaches. What considerations do CECL modelers need to keep in mind related to data?

Jeff: Models are only as good as the data fed to them. Many institutions have realized once they have audited data that there are significant gaps that dictate what types of modeling approaches will be applied when creating CECL models. The quality of data can vary greatly by institution, by product type, and even sub-categories of products. This means different modeling approaches may be required for different portfolios, which can limit comparability of the losses for different products and thus the capital hold.

Determining data-capture plans going forward will be critical for institutions that really want to leverage the data for more things than just CECL compliance. Data will be able to be used to determine more information about an institution’s client base and be leveraged for more than just loss projections if data governance standards are done correctly.

At a minimum if the data does not exist, institutions should look at the effectiveness of the data-capture and retention policies and how they can get more institution-relevant data to create more accurate and relevant CECL models.

Now that the industry is further along with CECL, is there more consensus around the types of modeling used with CECL? What are some of the modeling approaches being used by institutions and why?

Jeff: Institutions are implementing a wide range of approaches, including discounted cash flow (DCF), competing risk and Age Period Cohort (APC), and many are even leveraging existing stress-testing frameworks to assist in developing the CECL frameworks. Some institutions are trying to get some benefit out of having to establish DFAST frameworks and leverage existing documentation that makes the DFAST option appealing. This harkens back to the point about documentation as well.

Likewise, institutions are taking various approaches from a compliance-only standpoint versus leveraging the CECL framework as an opportunity to improve internal processes, like modeling, model risk, and data governance and capture.

Thank you for choosing the Situs Newswatch. If you want to see your company here or have an idea for coverage, please respond to this email or email inquiries@situs.com for more information.