There are more ways than one to skin a cat, and the weighted-average remaining maturity (WARM) method is yet another approach one can use to skin the CECL cat. WARM is a hot topic right now, and a big reason for that is because the Financial Accounting Standards Board (FASB) recently stated that the WARM method is an acceptable approach that could be used as a CECL solution. We encourage you to read through the FASB’s excellent care on the topic in their Staff Q&A to better understand the WARM method and to help when reading through this blog.
We should first point out that, yes, VE will have the WARM method available in its software in the near future, so stay tuned!
When discussing this Staff Q&A, we at Visible Equity came out of our discussion with three key points that required further analysis, which we will address in this blog. The first point is quite involved, whereas the other two points will be quick and simple.
An issue that comes up frequently in the Staff Q&A is accounting for economic adjustments for “current conditions and reasonable and supportable forecasts.” The FASB has made it abundantly clear that adjusting for current conditions and reasonable and supportable forecasts is a necessary step when determining the allowance for credit losses. The treatment of economic adjustments can be tricky business, however.
With methods like probability of default, economic adjustments are inherently built into the models, thus, no extra care is needed. However, with loss rate methods, such as vintage and WARM, the estimates for the loss rates do not take economic conditions into account (at least not always explicitly). There needs to be an additional step after calculating the loss rate to include economic adjustments.
One way of incorporating economic adjustments is with qualitative and environmental (Q&E) factors. The tricky part about Q&E adjustments is that they are based more on judgment rather than concrete data. Regardless of how one includes economic adjustments, one must be able to defend the adjustments, and while there is definite merit with using judgment (especially when taking into account changes such as underwriting standards and portfolio mix), basing adjustments on data is an easy way to defend one’s adjustments.
Another way of accounting for economic adjustments is quantitatively. This is where VE comes into the picture. Not only will VE have the WARM method in its software, but VE will also supply quantitative economic adjustments for those who wish to use them. The key with a quantitative method is that it is based on data, and we can show the math of how we arrived at our estimates, thus making it easier to defend.
So how does VE come up with economic adjustments, and how exactly are they used with WARM?
Let’s tackle the first part—how does VE come up with economic adjustments? For a given loan type (e.g. used auto loans), we take the historical charge-offs going back as far as possible and build a model that captures the relationship between the historical charge-offs and a combination of some key economic factors, such as unemployment and House Price Index (HPI) percent change. So for given values of yearly economic forecasts, we can produce estimated charge-off rates for each corresponding year. While there’s much more to this topic, this is all the information we need going forward.
Now to answer the second part of the question—how exactly are the economic adjustments used with WARM? Let’s refer back to the FASB Staff Q&A, specifically Table 2. The Q&A table presents an example of how to estimate allowances using WARM. We will use this table to illustrate incorporating our VE economic adjustments. Below is a modified copy of the table.
Columns one through five are directly taken from the table seen in the Staff Q&A. The big takeaway from these first five columns is that we multiply each projected amortized cost (A) by the average annual charge-off rate (B, calculated earlier in the Staff Q&A). This results in yearly allowances for credit losses (the fifth column).
Column six is where VE enters. Column six consists of the VE charge-off forecasts for each year (C). These are our yearly charge-off estimates using the model discussed previously that takes economic forecasts as inputs. Let’s say we used forecasts for unemployment and HPI percent change as our inputs in the model. Using these inputs, we come up with estimates for the charge-off rates for each year through 2025. So instead of using the average annual charge-off rate (0.36%), we are using economic factors to help us determine what we think the charge-off rates should be. We then multiply each VE charge-off rate by the previous year-end balance. For example, if we look at 2021, we multiply $13,980 (the previous year-end balance...or rather, the balance going into 2021) by 0.30% (the estimated charge-off rate for 2021). We do this for each year to get the new yearly estimates of allowances for credit losses. We then sum up these yearly allowances to get the total allowance for the remaining life of the pool of loans.
The WARM loss rate using the average annual charge-off rate is:
$126/$13,980 = 0.90%
The WARM loss rate using the VE charge-off forecasts is:
$108/$13,980 = 0.77%
The VE charge-off forecasts take into account economic changes over the remaining life of the pool of loans. Thus, the economic adjustment is a decrease of 13 basis points:
0.77% - 0.90% = -0.13%
Of course we will still allow for Q&E adjustments. If management has some sort of other information not taken into account with these calculations, they can still incorporate additional adjustments.
In the Staff Q&A, question four addresses other ways to perform the WARM estimation. They walk through a detailed example using the remaining life as a key component of the calculation. Using the VE charge-off forecast model does not directly work with this version of WARM the same way the other version does. Thus, it is crucial for those who wish to use this method to understand that both ways of calculating the WARM estimate are equivalent (we have proof). That is, they will both result in the same loss rate. Therefore, we can apply the economic adjustment found for the first method to the second method.
One difference between the way VE and the FASB calculate the average annual charge-off rate is that the FASB uses the averages of subsequent year-end balances. VE, on the other hand, will strictly use yearly average balances. For example, if we have balances recorded for each quarter within a year, then we will take the average of those balances. If the data are available, then using more data will make the averages more accurate.
Another issue that came up in the Staff Q&A is how far back historically one should go when calculating an estimate for the average annual charge-off rate used for WARM. We believe it is best to go back far enough to capture an entire economic cycle. Let’s say we’re looking at a pool of used autos with a remaining life of five years. Now let’s think about what has happened in the past five years economically. Will the next five years be similar? The previous five years have generally been favorable. What if we expect the next five years to worsen? Then only going back five years doesn’t really take the tougher times into account. Perhaps we should go back further. By going back further, we can see how charge-offs behaved in both better and worse times economically. This can give us a better sense of how loans will behave going forward.
In all, WARM is a relatively simple CECL solution that we at VE agree satisfies the Accounting Standards Update. We are excited about what’s to come with WARM, and we are hard at work preparing it for our software.