10 CECL Lessons Learned from SEC Registrants
Learn about the benefits institutions are experiencing
Abrigo’s Advisory team has worked with hundreds of financial institutions to successfully implement CECL. This paper describes some of the Advisory team’s more important findings from these first implementors for the benefit of institutions beginning or progressing in their implementation.
A few of the topics covered include:
- Model Risk Management
- Qualitative Adjustments
When the Financial Accounting Standards Board (FASB) modified the timelines for Smaller Reporting Companies and non-public entities to implement the current expected credit loss (CECL) accounting standard, it granted an additional year in most cases. This relief was not granted in light of the difficulty of implementation. Rather, it was granted to better fulfill the purpose of the original staggered timelines. Larger, public entities bore the brunt of implementation complexity and uncertainty, with the intention that CECL best practices would be established and “trickle down” to smaller entities.
This paper describes some of Abrigo Advisory Services’ more important findings from these first implementors for the benefit of institutions beginning or progressing in their implementation. The Abrigo Advisory team has already worked with hundreds of financial institutions to successfully implement CECL. The first consulting clients came on board in 2016, within a month of FASB’s finalization of the standard. The majority began their implementation project in late 2017 or 2018 as they expected then to adopt at the end of 2019. Due to the prescribed timelines for implementation, most of these clients were banking institutions; however, findings from their implementations should be applicable to all manner of preparers. It is important to keep in mind that the lens for the following “lessons learned” is primarily focused on the experience of Abrigo’s consulting clients. While we worked with a preponderance of SEC registrants, there are many we did not work with, and thus cannot speak for their experiences.
We take some pride in the degree to which our implementation methodology has delivered an efficient, successful experience for clients. In the main, a well-reasoned, practical approach has been satisfactory to external entities and sound, rational arguments have won the day. In many cases, in fact, a rational and systemic approach to calculation was more straightforward for our clients than their existing incurred loss (legacy accounting) methods for preparing the allowance. We trust the following information based on these client experiences will further ease the CECL transition for additional institutions.
CECL didn’t, in fact, require management to predict the future
A prevailing criticism of CECL is a perceived requirement to be able to predict the future. Mercifully, that requirement is not in the standard as written, and it has not surfaced in the standard as practiced.
The entirety of auditor and validator concern that surfaced during the implementation has been focused on the application of economic forecasts, and not on the forecasts themselves. Some of our clients partnered with a commercial forecast provider for this purpose. Others sourced forecasts internally, and others used public forecast inputs from, for example, the Federal Reserve. Some used a consensus or blended forecast, while others applied forecast scenarios and a weighting regime. In no case was a client criticized or challenged on their forecast inputs or choice of forecast source, even in the highly dynamic periods of early 2020.
This is well, and as it should be. The economic outlook itself is a management input and it would be inappropriate for an auditor or a prudential regulator to prescribe that outlook. In addition, fundamentally, the “correctness” of an economic outlook cannot be ascertained nor proven as of a statement date.
CECL resulted in Stepped-up expectations for Model Risk Management
As a software and services company that has been providing, among many other solutions, tools to calculate the allowance for loan and lease losses (ALLL) well before CECL was a twinkle in FASB’s eye, we are intimately familiar with the prevailing practices for creating the ALLL. The prevailing practice for commercial banking institutions, even public entities, under a certain size had been highly subjective and qualitative; auditors and regulators had been signing off on these practices (largely) without incident for many years.
While our work with clients on the ACL transition was formal and robust, we were surprised by the escalation in model development maturity expected by external entities from a mere accounting change. In a general sense, the practices developed by our consulting clients were more thoughtful and rigorous than what was in place before, but the rigor applied by validators and auditors also escalated. It remains to be seen whether this expectation penetrates to private entities in the later adoption cycle, but practitioners should be prepared to answer questions about statistical tests and other aspects of model risk management in their measurement methodology.
Later adopters should expect similar expectations for their practices; while in a general sense a given entity’s elections for inputs, parameters, and adjustments were granted some deference, the expectation throughout our client base regardless of size, auditor, regulator, or validator was that those decisions were comprehensively documented. This is a surmountable hurdle, but still requires work, no matter whether an off-the-shelf tool or home-grown solution is being used.
CECL typically brings major adjustments to qualitative adjustments
The CECL framework requires management to assess qualitative adjustments to its quantitative model, just like the prevailing practices under the legacy accounting standard. The primary divergence in practices for our clients has been the degree to which these adjustments constitute a primary portion of their allowance, especially in the commercial bank space. The new framework’s prescription to use “internal information, external information, or a combination of both” gives preparers meaningful, quantitative tools to establish a credible ACL even in the absence of observed loss in their own portfolio.
Consequently, qualitative frameworks are typically “page one rewrites,” and should take into account factors that are not accommodated in the quantitative model. Generally speaking, only a few factors can be quantitatively tied to historical credit performance, and these factors tend to be broad. However, that we cannot quantify their relationship does not mean that we reject them as important to our estimate.
In contrast to prevailing ALLL practices, in which all factors tend to be used, and all used unidirectionally (typically increasing indicated ALLL), those who have adopted CECL have tended to evaluate all factors but only make a few, comparatively minor, adjustments at a given time, resulting in a more coherent approach that is less challenging to auditors or validators. The time to generate and defend these adjustments can be generally reduced, and the qualitative framework can and should “talk to” the limitations of the quantitative framework.
Using external information may be helpful in establishing meaningful loss estimates
The standard allows for the use of external information in black-letter language. In contrast to prevailing ALLL practices, the use of this information is a first-class tool to establish a meaningful loss estimate even when no historical losses exist for a given entity. The general decision tree in Abrigo’s implementation methodology is to demonstrate the need for external information by evaluating the results of internal calculations. For higher-volume, higher-risk lenders like credit unions, internal information may often be sufficient, whereas for commercial lenders of the same size there may be no stable loss history to work with. Having established the need for external information, we delineate parameters for relevancy (yield, size, geography, etc.) and build a pool of loss experience based on those peers. The use of external information has been uncontroversial for our clients who require it in order to establish a meaningful loss estimate. The core argument being made by these clients is that “just because we have not experienced meaningful loss in this pool does not mean that we do not expect to,” which is in keeping with the guidance’s requirement to present an estimate of credit risk “even when that risk is remote.”
Timing isn’t everything when it comes to Modeling Economic Changes and Credit Losses
One of the more contentious discussions – prior to 2020 at least – was the nature of the timing relationship between economic changes and credit loss; the “lag or lead” question. For a given analysis for a given client, the timing relationship could be demonstrated based on historical experience, but we were reluctant to put too much epistemic weight on that observation. Economic change is a “natural” function beyond a given financial institution’s direct control, whereas loss events are usually a function of the entity’s policies and procedures for working with problem credits and recognizing loss. For consumer portfolios and “automatic” policies (charge-off on 180-day delinquency, for example) the timing relationship can be ascertained readily, but for commercial portfolios the emergence period for loss can vary significantly on a credit-by-credit basis. The events of 2020 have demonstrated the weakness of the timing relationship between economic events and loss recognition; as a result, many of our public clients have elected to constrain that relationship so that future losses are modeled proximately with future deterioration/improvement in economic conditions.
The fruits of CECL-Related Labor can feed other important projections for the institution
While CECL does require effort to implement, the key elements needed to compute the ACL are not standalone capabilities. Any financial institution should have an expectation for future prepayments, loss possibility under future scenarios, etc. What is novel is the application of these expectations to an activity within financial statement scope and the attendant rigors of audit standards. For our clients for whom the requirements of CECL represented a “step up” in internal models, the effort associated with their development has often been recouped by deploying these same models under different sets of assumptions to produce other, non-accounting projections such as economic stress tests, capital and budget planning exercises, and interest rate stress tests under an asset/liability management (ALM) capability.
The array of CECL Methodology options dosent have to result in paralysis
The standard allows for a broad range of approaches in calculating the ACL and mentions several specific methodologies. This flexibility was requested by preparers, but when entities are beginning their implementation, it can present a dizzying set of apparently consequential choices, especially if the preparer is intent on building their own models. The choice can be paralyzing, even for entities like our clients that have access to several different prebuilt calculation methods. By looking carefully at the need for external information and the amount of data available to an institution as “productive constraints” in methodology selection, our delivery methodology removes this roadblock and typically narrows the choice for most institutions. In our implementation experience with public entities, we have received no pushback on methodology selection, nor was a client criticized for not evaluating a different methodology than the one ultimately elected. The purpose of allowing methodological flexibility was to create an easier transition path, not a harder one, for preparers.
Our general implementation approach, which is specific to users of our tools who have ready access to multiple approaches, is to calculate a good faith, meaningful estimate using an instrument-level method with specific levers for credit and timing expectations, and then test simpler approaches to determine whether a substantially similar answer is produced for a given set of economic expectations. If the simpler approach does not yield a useful result, or is inoperable due to data constraints, it can be discarded.
Fairly broad segmentation was generally useful in many CECL calculations
Early in our implementation methodology, we performed rigorous studies of credit history to substantiate segmentation elections. Invariably, these studies never surfaced information that was new to a client, nor that couldn’t have been ascertained by uncontroversial business intuition. Consequently, we ceased performing those studies. 6 7 8 6 abrigo.com None of our public clients received pushback on segmentation elections that were supportable by pure reason. Further, segmentation ability was often constrained by availability of credit information for comparable institutions. For example, from a credit theory standpoint, there is probably a difference between a residential loan made to a borrower in one county versus another county, but an institution that has not taken losses on its residential lending will not benefit from that granularity.
Indeed, where that entity might have had one problem (establishing an allowance in a pool with no losses, at the residential-loan level) bifurcating the pool would just present two problems. As a consequence, most clients established fairly broad segmentation in order to incorporate external data in their loss estimates. In the course of the pandemic, some customers segmented out key industry codes (e.g., hotels) to provide an additional qualitative overlay, even if the quantitative credit modeling was performed at the higher level of segmentation. Segmentation is important, certainly, but it’s a lever for tuning and refinement once meaningful results have been achieved at a broader level.
Regulator deference to accounting and audit practitioners (At least in this round) was a welcome phenomenon
Throughout the implementation period, regulators had signaled a deference to the accounting and audit practitioners in how the ACL would be established for an entity, and we were pleased to see that signaled deference borne out in fact. The conversion from ALLL to ACL is, after all, an accounting standards change meant to provide information to financial statement users (including regulators) and not a prudential measure. While our clients required a degree of support for their first audits of their ACL, we have received no meaningful pushback or criticism from our clients’ first exam cycles. This was not an expected phenomenon, but it is certainly a welcome one. Our clients did receive regulatory interest in their implementation progress, but seemingly as no more than a passing “project management” concern, and definitionally the preparers that we worked with were making progress toward implementation.
Additionally, 2020 examination cycles likely had other prudential credit concerns that rightfully absorbed more focus than the accounting presentation of the credit reality. Nonetheless, we found that a meaningful and thoughtful calculation under the letter and spirit of the new guidance was sufficiently robust, even in the dynamic periods of 2020. For later adopters, who may be more likely to have a less thorough audit process than a public entity, that deference may not hold; adopting an appropriately scaled approach mirroring that of the larger entities should provide the comfort of familiarity to external reviewers.
CECL can be straightforward and provide additional useful information for management
Implementing the standard requires a certain amount of work, but it does not have to be overly burdensome if the implementation is guided by reason and focused using an entity’s constraints around loss experience or data. 9 10 7 abrigo.com
Borrowing lessons from the somewhat foggier experience of the earlier, public reporters will help avoid many of the unproductive time traps and rework issues that would otherwise loom as large risks to the project. It is our experience that once a reasonable practice for calculation is established, the financial impacts of the standard are minimal, as many community financial institutions were already presenting an ALLL that was comprehensive of lifetime loss expectation. Further, a measurement practice guided by a rational, systemic approach can significantly streamline the allowance preparation, while providing more useful outputs for the entity’s management.
One final point to keep in mind when considering the experiences of earlier CECL filers. Repeatedly, Abrigo has heard financial institutions say they wish they had started the process of CECL implementation earlier.
Community financial institutions that have yet to adopt CECL have the opportunity to adapt and scale the best practices of SEC filers to suit their own institutions. Lessons learned from these earliest adopters can help those who must begin reporting under the new standard in 2023 avoid project “black holes” that can be a drain on resources and staff. These lessons can also ensure the work pays dividends throughout the institution for years to come.
ABOUT THE AUTHOR
Managing Director, Advisory Services
Garver Moore is Managing Director of Advisory Services at Abrigo, leading a team engaged with hundreds of community financial institutions to provide valuation, credit loss modeling (stress testing & CECL), and strategic services. Prior to joining Abrigo, Garver worked with C-suite executives on technology strategy and delivery as a Managing Partner of the Orange Advisory Group and was a Technical Consultant with Accenture. He earned his bachelor’s degree in electrical and computer engineering from Duke University
Ease Your Transition to CECL
Abrigo is here to help. Our endorsed CECL solution is easy-to-use and recognized by auditors and examiners. Abrigo’s Consultants are also available to provide support for CECL transition plans, data audits, or outsourced calculations.
CECL SOFTWARE | CONSULTING SERVICES
Abrigo enables U.S. financial institutions to support their communities through technology that fights financial crime, grows loans and deposits, and optimizes risk. Abrigo’s platform centralizes the institution’s data, creates a digital user experience, ensures compliance, and delivers efficiency for scale and profitable growth. Make Big Things Happen. Get started at abrigo.com.