How to Implement CECL
This is the second in a three-part series “Getting to Know CECL” from guest blogger, Cris deRitis, PhD, Senior Director at Moody’s Analytics. In this article, we’ll explore several approaches to complying with the guideline.
Requiring lending institutions to estimate and reserve against potential loan losses earlier in the lifecycle of their loans makes intuitive sense. Shareholders and regulators want to know what the expected losses on lenders’ portfolios look like and have confidence that the allowance for loan and lease losses is adequate to cover institutions that experience a rough patch in the economy.
But it’s one thing to acknowledge the theory and quite another to put it into practice. The CECL guideline gives institutions tremendous latitude in determining how best to estimate losses. The only real requirements are specified in this excerpt:
“The measurement of expected credit losses is based on relevant information about past events, including historical experience, current conditions, and reasonable and supportable forecasts that affect the collectability of the reported amount. An entity must use judgment in determining the relevant information and estimation methods that are appropriate in its circumstances.”
In this second of our three-part series on “Getting to Know CECL”, we’ll explore several approaches to complying with the guideline.
Approaches to CECL
The CECL guidance does not prescribe any specific manner or method an institution should use to combine historical and current information along with economic forecasts to generate a loss estimate. There is no specific requirement for accuracy and the guidelines acknowledge that the forecasts – along with the loss estimates – are subject to revision as new information becomes available. The estimation of lifetime losses is not intended to be a “set it and forget it” process. Rather, the intent is to require institutions to constantly evaluate their estimates and make in-course adjustments as needed. Provided institutions make “reasonable and supportable” forecasts at the time of filing, their estimates would be in compliance with generally accepted accounting principles (GAAP) – even if they are revised significantly at a later date.
Based on our experience working with institutions that implemented stress testing and financial reporting standard IFRS9 – CECL’s international counterpart – we identified four key decision areas that institutions need to address to develop a successful CECL implementation:
- Economic scenarios
- Data availability/benchmarking
- Model methodologies
We discuss each of these in the sections that follow.
One of the key differentiators between CECL and the current incurred loss accounting process is the formal incorporation of forward-looking forecast information. Given that we are trying to predict the future payment behavior of loans and lines of credit, this forecasted information will necessarily be economic in nature given that changes in unemployment, house prices, interest rates, etc. drive borrowers’ ability to make their monthly bill payments. But what makes an economic forecast “reasonable and supportable”?
One criteria may be the source of the forecast. A forecast driven by a mathematical model based on the historical relationships and linkages between variables may be easier to explain and justify than a purely subjective judgment – even if that judgment is from a reputable source. While the CECL guidance makes no specific recommendations, institutions may prefer to use forecasts that are generated using structural macroeconomic models similar to those used by central banks like the Federal Reserve or by private forecasters such as Moody’s Analytics. These models have the benefit of being able to incorporate hundreds of economic indicators into a consistent framework. A shock to oil prices, for example, is propagated to all sectors to the economy so users can understand how the shock could impact employment, wages or house prices.
Source: Moody’s Analytics
Institutions will also consider the geographic detail of a forecast to insure consistency with their own geographic footprint. A lender concentrated in Texas, for example, may find that a national level forecast does not adequately capture the unique aspects of the Texas economy.
Source: Moody’s Analytics
Finally, institutions will want to give some consideration to the number of economic scenarios they use to derive their loss estimates. While the CECL guidelines don’t require any specific number, institutions may find that running multiple scenarios may be more informative and provide a more stable estimate of future losses. This is particularly true for consumer credit lending products whose relationship with the economy is highly nonlinear. For example, if house prices are rising at 3% per year, the impact on lowering default rates would be relatively minor if they were to accelerate to 8% per year. However, if prices were to fall by -5% per year, default rates would rise significantly as borrower equity falls.
Source: Moody’s Analytics
The estimation of future credit losses necessitates the use of some type of model. The model might be as simple as looking at historical loss rates over time for previous years’ loan originations and applying this experience to the current portfolio. This approach may be perfectly adequate for a smaller institution with a regular customer base and a stable loss experience. However, a lender that has experienced changes in its lending population or that has changed its lending practices may find this approach does not adequately reflect its current loss expectations. Other institutions may find they have incomplete, inadequate or unreliable historical performance history upon which to base an estimate.
In these instances, we find that institutions can benefit greatly from the use of external data sets either to augment their own history or to create industry-level forecasts or benchmarks. The CreditForecast.com data service, a joint solution between Moody’s Analytics and Equifax, was created precisely to meet these needs for consumer credit products. By aggregating anonymized information from consumer credit reports, analysts can track the historical performance of different credit products controlling for geography, origination quarter, credit score and loan origination term. Lenders can compare these industry trends with their own performance data to gauge their own relative performance.
Source: CreditForecast.com, Equifax, Moody’s Analytics
Users can go one step further and leverage the forecasts embedded in the service to derive loss estimates for their own portfolios based on credit loss models built on this robust set of industry performance data. With forecasts provided under variety alternative economic scenarios, users can easily compute lifetime loss estimates that can feed directly into their CECL processes or that be used to benchmark their own internal calculations.
Source: Moody’s Analytics
A Method to the Madness
The non-prescriptive nature of CECL means that institutions have a host of options when it comes to the models they use to estimate losses. The Financial Accounting Standards Board (FASB) has left it up to each institution to determine the most appropriate model to use for their specific situation. Typical methods institutions consider are the following:
- Discounted cash flow approaches
- Vintage loss rate approaches
- Expected credit loss component models (otherwise known as PD/LGD models)
- Transition matrix approaches
- Survival or discrete-time hazard models
Data availability will often limit the set of available options for an institution as will model complexity. Smaller institutions may have fewer capabilities to maintain a complex, large-scale modeling process. Notably, the homogenous nature of their lending portfolios may limit the need for a detailed model. Institutions also need to give careful consideration to their production processes and insure that they have adequate resources to run the selected models on a quarterly basis.
Regardless of the selected methodology, it is important that senior management have a sound understanding of the models being employed to perform their CECL loss estimates. Models will require a sufficient degree of transparency so that users can understand how and why estimates change given different assumptions and forecast conditions.
Reporting and Disclosures
A key area of focus in setting up a CECL process – but one that is often overlooked – is CECL’s reporting and disclosure requirement. Not only does CECL introduce a new loss estimation framework, it requires institutions to substantially increase the amount of information that is disclosed on their financial statements. These disclosures include greater detail on portfolio composition and expected performance by segment such as vintage or origination year. They also include more information on the attribution of changes in the loss estimates from period to period. If an estimate increases, what were the drivers of that increase? Did the economy perform worse? Has the outlook darkened? Have borrower delinquency rates increased?
Institutions may find it useful to work backwards from the end state of CECL – i.e. the production of quarterly disclosure reports – in designing their processes and selecting models that will ultimately support this goal. In tracing back through the process, institutions can often identify the weak links in their design process whether its lack of data, inadequate models or missing drivers in their economic scenarios.
Try, Try, Try
Perhaps the best piece of advice when it comes to implementing CECL is to get started as soon as possible. This may involve piecing together a process with data and models that are readily available. Only by attempting to execute the process can institutions identify any deficiencies and prioritize them. CECL is here to stay and will be a constantly evolving processes. No institution will get it implemented perfectly the first time around. Rather, they will evolve and fine tune their processes and models over time. As a result, systems and process need to be designed with a degree of flexibility so institutions can adapt quickly to new information.
Don’t miss this series’ first blog article, “Getting to Know CECL.”
For more information on how we can help your financial institution prepare for CECL, call your Equifax account representative, or contact us today. Additional resources can also be found on Equifax and Moody’s Analytics websites.
Recommended For You
Does “Prime” Mean Perfect? Prime rib. Prime rate. Optimus Prime. Regardless of whether you’re talking about dinner, finances or robot […]
May 2020 Market Pulse Series – FAQs Equifax and our expert guest speakers answered dozens of questions from attendees following […]
This blog post will be continually updated with timely and relevant insights gleaned from the latest Market Pulse weekly webinar […]
The economic uncertainty caused by COVID-19 has impacted the automotive industry in complex ways. The entire auto ecosystem has been […]