But Are You Really Ready for CECL?

blocks with person target light bulb gear graph
By Michael Guglielmo

5 minutes

Vital items to be aware of between now and year-end

Sponsored by Darling Consulting Group

The Financial Accounting Standards Board’s new current expected credit loss standard, or CECL, represents a profound change in how our industry accounts for future potential loan losses. While the benefits of FASB’s more forward-looking methodology are evident, efforts required to develop and implement a successful CECL modeling and governance framework have been challenging for most. Moreover, the strategic implications and potential unintended consequences are just starting to emerge.

Adoption day (currently Jan. 1, 2023) is nearing for most credit unions, and they may benefit from insights and lessons learned from those that have already been down this path. Below are some vital items to be aware of between now and year-end from an experienced CECL and credit-risk validator’s viewpoint.

Data Matters

Quality internal and external data will have an outsized impact on the accuracy and ongoing reliability of your CECL model. Many organizations have struggled to obtain accurate and reliable internal data, and for good reasons, many are migrating toward the inclusion and use of peer data. In any case, it is essential to know your “ingredients,” how you confirmed they are the best ingredients available, and how you are preparing them for consistent, ongoing use. Data deficiencies have commonly led institutions to initially choose an inferior CECL model methodology and subsequently expand their data collection and management processes for eventual model improvement.

Assumptions Matter

Like other critical models, your CECL model rely heavily on a few essential assumptions and decisions to produce its quantitative result. Take the time now to get them right and plan to routinely sensitivity test them going forward. One of the most impactful assumptions is your prepayment estimates. We have seen quantitative CECL results vary by 20% to 30% based upon this one assumption parameter. As you develop your CECL assumptions, be mindful that you may also be using similar assumption parameters for other models within your institution. For example, models relying on prepayment estimates include interest rate risk, liquidity, profitability, valuation, servicing, capital and strategic planning, and stress testing. As part of the effective challenge process (a framework for providing critical analysis that’s often a critical component of risk management), a good validation partner will investigate prepayment estimates across models and highlight the differences between your assumption approach for CECL and your other models.

Model Governance Matters

Your CECL model is only as good as the governance surrounding it. Unlike your current incurred loss modeling process, CECL requires more formalized assumption review and approval processes. Moreover, periodic sensitivity testing of your key assumptions and performance monitoring (backtesting your critical assumptions and your overall combined quantitative and qualitative results) against actual experience will also become part of your ongoing routine to ensure the model continues to work as intended. Importantly, make sure you allocate sufficient time before year-end to determine how you plan to address these vital model governance elements.

Model Documentation Matters

One of the least understood aspects of effective model risk management is the notion of “model documentation.” This concept is particularly challenging for institutions that have little to no formal model risk management program in place. Whether you are using an internally developed model, an outsourced service or a vendor software solution, every aspect of your model should be fully documented and described. Unlike the “technical” documentation you may receive from a provider, this documentation differs in that it should represent you and your use of the model. Documentary aspects should minimally include the model’s underlying theory and design, data development, transformation and testing, assumption development and support, model output, ongoing performance monitoring, governance, and procedures. Organizations commonly leave this important aspect of model management until the last minute—typically from a lack of awareness, insufficient allocation of time or resource constraints. So it can quickly evolve into a last-minute scramble, a material validation and audit finding, or regulatory criticism.

Validating Your Model in Advance of Using It Matters

While institutions with a formal model risk management program recognize the importance and necessity of validating their models (particularly critical “high-risk” models like their CECL model) in advance of their use, this crucial step is all too often left until the last minute or even after the model is already in use. It is essential to realize that these models, including the best software-based solutions, will have deficiencies or weaknesses right out of the gate, and you do not want to find out about these issues after you have already begun to use your model for your financial disclosures. Moreover, you do not want a situation in which your external auditor refuses to sign off on your financials until the model is validated. It is important to recognize that CECL models are not simply accounting models to be evaluated by a certified public accountant in relation to FASB’s principles-based standards. In fact, your CECL model is a credit loss forecasting model, and its quantitative and qualitative output is used for accounting disclosures. Accordingly, select an experienced validator who has the technical expertise, a model risk management background, and the related business experience required to test and effectively challenge all aspects of the model and its governance rigorously.

Final Thoughts

Those who implement their CECL model to seek strategic value have consistently benefited from this mindset. In addition to meeting the minimal expectations of their accounting, audit and regulatory stakeholders, several strategic benefits have emerged for many, including:

  • Improved credit-risk awareness
  • Broader improvements in risk modeling accuracy due to upgraded organizational data management practices
  • The establishment of early warning indicators that have driven preemptive strategy discussion and execution
  • More substantive credit-adjusted loan pricing
  • Informed capital and strategic planning
  • More effective portfolio and merger and acquistion valuation and decisions

Lastly, it is important to realize that the current version of your CECL model is not your last. Once the entire industry has adopted CECL, a significant amount of horizontal research and analysis will be performed by the accounting, model risk, audit and regulatory communities. As a result, leading practices and standards will emerge, the industry’s CECL modeling approaches will converge, and we will collectively be better for it.

Michael Guglielmo is managing director of Darling Consulting Group, a CUES Supplier member based in Newburyport, Massachusetts.

CUES Learning Portal