CECL May Not Increase Loan Loss Reserves—And Other Myths for 2023 Filers

By Adam Mustafa, Invictus Group CEO


The 2023 class of CECL banks is being unnecessarily conditioned to a false reality: Their loan loss reserve will need to increase under CECL. If CECL is approached correctly, this is simply not true, unless the probability that a COVID-induced downturn translates into increased loan losses significantly increases.

In this article, we will debunk several critical myths about the impact CECL will have on the loan loss reserve. It is critical that banks take this seriously; every penny of capital that is unnecessarily trapped in the loan loss reserve is destroying shareholder value and will contribute to a bank’s difficulty justifying its independence moving forward.

Myth No. 1: A bank’s reserve under CECL will have to increase because that’s what happened to the large publicly traded banks. 

Yes, it’s true: Publicly-traded banks, mostly larger institutions, saw a cumulative increase of 34 percent in their “Day 1” CECL reserve versus their incurred loss calculation, and that was all done in a pre-COVID world. However, it’s critical to keep in mind that the large banks were always projected to be disproportionately affected by CECL. Those banks have a very different portfolio construct, one that is far more weighted toward loans with higher loss rates and/or limited collateral such as credit cards and C&I. Larger banks are also under constant pressure to maximize quarterly earnings in good times, thereby placing a disincentive to lean too conservatively into the loan loss reserve. In other words, they had far less cushion in their existing reserves under the incurred loss method to absorb a transition to CECL. Many were also acquisitive; CECL is not kind to purchase accounting.

It should also be noted that in our analysis of a select group of 128 public banks that adopted CECL, 12 or nearly 10 percent, showed a negative adjustment in their Day 1 reserve. This includes Wells Fargo, Comerica, and Investors Bank. So, there is indeed precedent already for this.

Myth No. 2: Since CECL requires that banks reserve for expected losses over a longer-time horizon than the incurred loss method, the loan loss reserve must increase.

In theory, this is true. The incumbent incurred loss method only requires banks to reserve for probable losses. Most banks generally assume that probable losses would eventually emerge within one year.[1] On paper, the incurred loss method requires banks to reserve only for potential losses looking out just one year. But CECL requires banks to reserve for potentially many years of expected losses.

In practice, here is where this assumption blows up. While the above is true, what if I told you that your bank was grossly over-reserved under the incurred loss method if you followed the true letter of the guidance? While some of you are frowning as you read this, let me be very clear: If market participants were truly adhering to the incurred loss method, then 99 percent of the banks in the industry would not have been in compliance with the guidance for each of the five years prior to the pandemic.

The evidence is vast and simple. Here is an example: Think about your loan loss reserve at the end of 2017. Whatever that balance was, it will have dwarfed your net charge-offs during the following year (2018), which would represent the loss emergence period for those probable losses you estimated as of the end of 2017. In other words, those losses you said were “probable” in fact were not probable because most of them didn’t happen. In fact, let’s be even more conservative and assume it would take eight quarters for those losses to emerge, instead of four quarters. Well, go ahead and include your net charge-offs in 2019 into the calculation along with your 2018 net charge-offs, and you would still be nowhere near your estimated reserve at the end of 2017.

The reason nobody talks about this industry-wide breach of the incurred loss standard is because nobody cares. All relevant stakeholders are comfortable with the reserve being far more than it should be under the true letter of the guidance for the incurred loss method. They may argue over how much de-facto excess in the reserve they want (regulators want more, investors perhaps less), but nobody was arguing that the loan loss reserve for the average bank should be 8 or 10 basis points as opposed to 100 or 125 basis points because that was the average net charge-off rate over the past five years.

The excess in the reserve was usually calculated through qualitative factors. Most participants in the industry don’t even really understand what “q-factors” are supposed to represent. It is not technically supposed to be a fudge factor to get you to some pre-determined bogey (which is what it became), it’s supposed to represent an estimate of probable losses in your portfolio that you just don’t know about yet. In other words, the probability of some loss exists within your performing / pass-rated credits because a handful of borrowers experienced a loss-causing event during the quarter such as loss of employment, but the bank is unaware of this event at the time of reporting. In most cases, looking back at the size of the reserves, those loss-causing events did not occur nearly as much as banks implied over the five years prior to the pandemic.

As you can see, the loan loss reserve under the incurred loss method morphed into something different than a reserve for “probable” losses that would emerge within one year. This is one of the reasons why the Financial Accounting Standards Board (FASB) wanted to get rid of it. I would argue that if you do the math, many banks are sufficiently reserved for CECL already; they just don’t know it. This is why it’s so important to have the right context in mind when you compare your CECL calculation to your incurred loss estimate. If CECL is calculated properly, it should NOT be significantly larger than your existing reserve, which is likely to be overestimated. You can get rid of this overestimation in CECL with the right tools.

Myth No. 3: Knowing what we know now, the increase to a bank’s reserve under CECL would be exacerbated by the pandemic.  

Banks generally do not have material losses in good times, only in bad times. What we don’t know is when the next downward cycle is going to come. This is why everyone in the industry was okay with breaking the technical aspects of the incurred loss method to build excess into the reserve. It was a rainy-day fund for the next recession, whenever that occurred.

When you start running your CECL calculations in parallel, you will quickly discover it’s just as difficult to quantify your reserve as it is under incurred loss, unless you incorporate stress testing. If done correctly, your CECL model is just a stress test with a probability adjustment. Frankly, the same is true about your incurred loss calculation; it’s what explains the aforementioned ‘fudge factor’ – you just don’t think about it that way!

If your stress test is properly constructed, it will account for unique aspects associated with the pandemic, such as its discriminatory impact on certain industries such as hotels and retail.

As of the date of this article, we still do not have any idea on what kind of impact the pandemic will have on loan portfolios. This is because the abundance of stimulus that has already been injected and may continue to be injected into the economy has at least delayed the emergence of significant loan losses. You can still make a case that the losses are coming and 2021 could be worse than 2008, but you can also make a case that losses will be nominal due to vaccines and the upcoming stimulus.

The point is that if one were to assume there is a 50 percent chance of a recession being averted, a 35 percent chance of a moderate recession translating into an increase in loan losses, and a 15 percent chance of a severe recession with significant loan losses, then the average 2023 bank will find that their CECL parallel calculation is still very close to its incurred loss method number.

If one believes these probabilities are too ambitious or the economic sentiment worsens in the future, then the pandemic may indeed place upward pressure on the CECL reserve from a timing standpoint. However, the same could be said about the reserve calculation under the incurred loss method over time. The only difference that COVID would have on the two methods would be one of timing.

Wrap Up

CECL has created several forgone conclusions within the industry that simply aren’t true. Those banks that are willing to embrace a deeper understanding of the standard and its spirit will naturally find the right tools that will allow them to properly calculate their reserves without being swallowed up by irrational group think. My advice: Be a leader and not a follower to prevent your bank from unnecessary over (or under) reserving. It’s hard enough to make a justifiable ROE in this environment, don’t give away the lowest hanging fruit.


[1] Some banks may use a longer or shorter period for an estimated loss emergence period, but one year is most common.