By Michael Gullette
The Financial Accounting Standards Board’s newly finalized Current Expected Credit Loss Accounting standard, also known as CECL, represents the biggest change to bank accounting ever. If you’re a CEO and you just tell your CFO to take care of this, you will spend a lot of money on this—and fail. Your capital will swing back and forth, and each quarter’s results will be a big guessing game. This will negatively affect your ability to serve your customers and communities. Don’t just think of CECL as an accounting change—but rather a change to how all banks manage their business.
My bet is that you have concerns about CECL. We do, too. The American Bankers Association recommended to FASB an impairment accounting model that largely reflects how you evaluate loan impairment today. Instead, FASB opted to require that both current and future losses in a portfolio be recorded. This will require an overhaul of many processes throughout your company. You will incur additional costs not only in setting up a CECL process, but also in running it on an ongoing basis thereafter.
A major change
Current accounting principles have been in place for about 40 years and loan losses are handled under “incurred loss” accounting, meaning something probably happened that caused impairment to the loan. For practical purposes, that impairment is normally measured in pools of loans and is heavily based on historic annualized charge-off rates.
In contrast, CECL is an “expected loss” notion. An event does not have to have occurred, but can be expected in the future. Further, the historical data that CECL relies upon are not annual loss rates, but life-of-loan or life-of-portfolio loss rates. This is a big difference that can be very easily misunderstood. Conceptually, I like to think of current accounting as recording the losses in your portfolio and CECL as recording the risk in your portfolio.
CECL itself is actually relatively simple. First, a life-of-loan loss expectation is effectively recorded at origination. For practical purposes, just like we do today, this will be done for pools of loans. This requires a forecast of the future, including economic indicators such as interest rates and unemployment.
Historic averages of life-of-loan losses are very important in CECL. Like today, they are used as the starting point for estimates of expected loss. Many bankers, from large banks and community banks alike, have expressed concern about their ability to forecast into the future past a couple of years. Under CECL, you will forecast as far into the future as you can (that’s what I call the forecastable future), then use unadjusted historical averages of losses past that point.
Don’t expect a lot of detailed rules in the standard of how to do it. FASB writes high-level principles with no prescribed measurement methods. Even the implementation examples I’ve seen leave out a lot of the detail needed to satisfy the intent of CECL—and the devil is in the details.
The key to understanding CECL is understanding its life-of-loan concept. Today, we usually think of annual loss rates—but if you think you will just take an annual loss rate and multiply it by the expected life of the portfolio, you will be wrong. Losses don’t occur evenly throughout the life of a portfolio.
The graph pictured nearby shows a loss curve, which demonstrates this for a specific vintage, or group of loans originated in the same year. Charge-offs on these loans, which show a pattern similar to many residential mortgages, increase the first two years and then peak, then taper off significantly after that.
Simply speaking, the area in the orange curve represents the total amount of risk that this group presents over its life, with the charge-offs on the left being the risk that has been actualized, and the allowance on right being the risk that remains. If most loans are two years old, we expect significant additional charge-offs—those in the green area.
If our loans are three years old, however, we have much less risk in front of us. The allowance would then be a lot smaller.
That’s not a terribly complicated concept. But an average portfolio normally has many vintages outstanding, and a forecast of the future then compounds the complexity, as different loan terms and maturities will react differently to the forecasted economic assumptions. At a very minimum, a CECL estimate will require more granular information and a more detailed analysis.
Top challenges for CEOs
First, you will need to communicate with your investors and management early and often about CECL. Terms like loss rates will have new meanings. You need to make sure everyone is singing off the same song sheet. The relationship of traditional credit metrics no longer continues. Today, delinquencies or nonaccrual loans go up, and so does the provision. No longer is that the case, as those provisions are supposed to be recorded at origination. Therefore, new metrics will be needed and they will need to be communicated over and over. Our industry needs to unlearn 40 years of knowledge.
Next is the data question. The vast majority of the data needed to support a life of loan loss estimate is probably not currently collected by your bank. So there will be costs in accumulating past data and collecting and analyzing it on an ongoing basis.
Remember that CECL’s life-of-loan loss notion is a much bigger playing field than we have today. Small changes in your assumptions will make very large changes to your loss estimates. When changes to those estimates mean the difference between a dividend or not, support for how you quantified those assumptions also will be important. Much more detailed analysis is required, based on much more granular data.
Smaller banks will not be expected to have as sophisticated an analysis as larger banks. But you will need the data and you will need to address the unique issues in some way.
Finally, you have to address your regulators. They will want you to align your CECL assumptions to your budgeting and planning. The expected increases in interest rates that are driving your pricing should be integrated into your CECL estimates and those results then get fed into your capital plans. CECL is about reflecting the risk in your portfolio. This is why I believe that CECL could change how you manage your bank.
CECL is effective in 2020 for Securities and Exchange Commission registrants and 2021 for everyone else, with early adoption available. Now, that gives us a decent amount of time before we need to do anything, right? Wrong. Remember that most banks do not currently collect the data needed to perform a life of loan analysis. At a minimum, you need to start collecting data as soon as possible.
Action plan
Before you make any decisions on implementation, the most important thing to do is take a step back and get educated. Education should be a priority before any decisions are made related to implementation.
ABA offers extensive background resources and webinars, and I will co-host a special CECL implementation workshop at ABA’s CFO Exchange in September. Reviewing the regulatory guidance will help, but the agencies are not recommending any specific methods or practices at this point.
Auditors and software firms are coming out with their own presentations. While many will be helpful, keep in mind that they are presented from their own point of view. What you will need may be different.
Taking action and learning more now can help your bank position itself for the big changes CECL is going to bring to your business.
