Ahh, CECL. The abbreviation for “current expected credit loss,” per an internet site that looks kinda legit, “a new credit loss estimator which replaced ALLL [allowance for loan and lease loss] estimate for loan losses. That said, the reason for the new model is we screwed up in 2008 and our next screwup needs a different acronym.”
I may have paraphrased.
The new standard currently is required to be used at all financial institutions by 2021. The idea behind the new model is to cover a glaring issue with the ALLL method: That it’s backwards-looking and doesn’t properly estimate losses as the economy falters going forward.
Because this is such a departure from old, the Financial Accounting Standards Board (FASB) introduced an easy method for implementation to minimize impacts during implementation.
Sorry, I’m laughing so hard that Starbucks came out of my nose.
In reality, FASB detailed five (yes, five) optional methods for implementing CECL (with a bonus option of more):
1. Vintage. When I hear “vintage,” I think “aged Merlot.” But for CECL “vintage” means estimating your losses for each loan by the date originated, i.e., their “vintage.”
This would mean that loans originated in say, 2009, during the crisis may have a much different loss ratio than those originated in 2014. You then get to multiply that by quantitative aspects such as the unemployment rate, GDP growth, or the chances the Cleveland Browns win a game in 2020.
2. Loss rate. This uses overlapping static pools to determine rates. In theory, this should identify business cycles and adjust accordingly.
In theory, my children should identify that they are grown and adjust accordingly as well. I doubt either will be easily done.
3. PDxLGD. This is both the acronym for a new Marvel superhero movie fight and the name of the third CECL model. It uses two pieces to determine the allowance, the “Probability of Default” (PD)” and the “Loss Given Default” (LGD).
This is superior to many methods as it can be used with discounted cash flows and assumptions on collateral better than the previous models. Also, look for the 2021 sequel, “LGD vs BSA” coming soon.
4. Roll rate. Loans are established with a “risk level” based on delinquency, current FICO, loan-to-value ratio, and so on.
Each of these levels has an expected loss rate. By seeing how often loans roll from one level to another, you can make an estimate of future losses. This is called the “roll” and should not be confused with the Pillsbury version, which is delicious.
5. Discounted cash flow. Each loan’s payment future is estimated using a variety of environmental factors and then discounted to determine its expected cash flow.
This method is complex and is similar to what large bond traders do. That’s why having large bond traders over for dinner is an incredibly bad idea.
The short of it? These issues come to mind:
Obviously, the total future losses in the portfolio will always be larger than the ALLL model—but by how much?
Mainly it’s a function of the riskiness of the loan by itself and the loan term.
The natural ramifications of dear old CECL is to avoid subprime and long-term loans, which will require a larger initial write-down than traditional loans.
And before you think you can simply raise rates to adjust, remember that losses are recognized up-front but not income. Unfortunately, this is likely to decrease lending to some of our most needy members.
Learn more about CECL and other regulatory issues in the 2019-2020 CUNA Environmental Scan
Many of us wish CECL would go home. Perhaps the best chance of that is a bill introduced by Rep. Blaine Luetkemeyer, R-Mo., that would prohibit federal financial regulators from requiring compliance with CECL.
Unfortunately, passage is uncertain. For most of us our best bet is simple:
Get used to the smell of burning CECL.