Answering an Inquiry on LEP
Our recent paper on loss emergence period (LEP) generated this question from a CPA firm:
“I’m a little confused on this article. If it takes us two years to identify a loss on a particular loan, and the ‘FAS 5’ loss factor on that type of collateral suggests we’ll lose on average 5%, wouldn’t the loss on that particular loan still be 5%? Why would it double to 10% just because it took a while to season to become a problem?”
Here are excerpts from answers to the inquiry that we received from the information sources for that article, Grant Thornton’s Graham Dyer and MST’s Chris Emery:Graham Dyer writes: Here’s some perspective that might be helpful. I understand this is the common view shared in the industry. The theory of the LEP, briefly, is:
Many entities use a charge-off rate-based starting point to estimate their ALLL – expressed as . The time in this equation is typically one year.
The goal of the incurred loss model (that is, “FAS 5”) is to estimate losses inherent in the portfolio but not yet charged off . . . We don’t yet have enough information to do that with only our charge-off rate because we need to know how long it takes from the incurrence of a loss (that is, the “loss event”) to the charge-off of the loss to know how many losses are sitting in our portfolio waiting to be charged-off.
For example, let’s assume we have a loan portfolio where it takes 2 years to go from loss event to charge-off. That means that we have 2 years of losses inherent in our portfolio at any given point. If we measure our annual charge-off rate at 2.5% for a portfolio of loans (expressed as 2.5%/1) and use that to estimate the ALLL, we will only have 1 year of charge-offs using the formula (Loan Balance x 2.5%/1). However, if we expand our formula to include the LEP, then we get our 2 years of losses, and the right ALLL.
Chris Emery adds: The loss rate is representative of what losses have been experienced (confirmed) on previous pools of similar loans. The LEP is the representation of how long it takes losses already incurred (what are being reserved for) within the pool to be eventually confirmed (or charged-off) by the bank. So, if you have an annualized loss rate, you are saying there is a 4 quarter lag between the loss being incurred and the loss being confirmed and charged-off by the bank. Some loan types may experience loss events much longer before the confirmation of the loss.
An easier way to think about it might be this: If you believe the LEP to be 4 quarters in a given pool, you are essentially reserving for 4 quarters worth of incurred losses already within the pool yet to be confirmed. Assuming all loans perfectly follow the LEP you’ve determined for this pool, you will have 4 sets of loans, each representing 1 quarters worth of eventual confirmed losses: Loans that just experienced a loss event, and will be confirmed in 4 quarters; loans that experienced a loss event a quarter ago, and will be confirmed in 3 quarters; loans experiencing a loss event 2 quarters ago, and will be confirmed in 2 quarters; and loans experiencing a loss event 3 quarters ago, and will be confirmed in 1 quarter.
You have no idea which loans these are, which is why the loans are being pooled together. But you’re annualizing the loss rate (multiplying by 4 essentially) because you believe there to be 4 quarters worth of unconfirmed losses within the pool. Knowing what previous quarters of confirmed losses have been, you can estimate what those 4 quarters worth of confirmed losses, your annualized loss rate, will be.
But if this pool had an LEP of 8 quarters rather than 4, you would have 8 “sets” of future confirmed losses: Loans that just experienced a loss event, and will be confirmed in 8 quarters; loans that experienced a loss event a quarter ago, and will be confirmed in 7 quarters; and so on.
In this scenario, an annualized loss rate would be insufficient as there are more than 4 quarters worth of unconfirmed losses within the pool. There are 8 quarters of unconfirmed losses within the pool, so double the amount that would be estimated for the same pool with an LEP of 4 quarters.
Graham Dyer adds: The issue we run into is that some people see the ALLL as capturing lifetime losses. It’s not – at least it’s not supposed to.
Thanks, Graham and Chris. We hope this helps to answer the inquiry and clarify the theory and importance of LEP.