Accounting for prepayments in CECL: Part 1

New Call-to-action
     

 

As we have interacted with various audiences on various topics related to CECL, we have observed a common concern: prepayments. At several points throughout the standards update, FASB makes it clear that prepayments matter in your loss calculations. They describe CECL as “...an approach for the allowance for credit losses based on management’s expectations of credit losses over the contractual life of the financial assets (considering the effect of prepayments)...[emphasis added]”.  While I don’t think there are many CECL activists out there protesting the idea that prepayments affect lifetime loss, plenty of us have spent a minute scratching our heads, wondering exactly how they should be considered in our models.

glasses.jpg




This is the first of two blogs in which we will discuss how Visible Equity is handling prepayments. In this article, we will describe prepayments in the loss curve model, and in the other, we will consider prepayments in the probability of default model. Let’s begin the loss curve discussion by looking at a statement found in the standards update:


An entity shall consider prepayments as a separate input in the method or prepayments may be embedded in the credit loss information in accordance with paragraph 326-20-30-5.


Note that paragraph 326-20-30-5 discusses expected credit loss estimates using a method other than discounted cash flows. So this statement effectively tells us that, if our model is not a discounted cash flow model, then we can either explicitly add prepayments to the model as a separate input, or we can ensure that our method implicitly incorporates prepayments. That is, depending on our model, we might not actually have to do anything special in order to check off the “prepayments” box. The rest of the article will explain how Visible Equity’s loss curve model falls into this bucket.


First, recall that the loss curve approach is simply a large collection of ratios, each of which is meant to represent the percentage of current balance expected to not be collected over the lifetime of a loan. As long as we have accurately found a ratio that does just this, we then multiply it by the loan’s current balance, and our job is finished. We can then go home and take a hot bath, and not think about prepayments. So let’s talk about that ratio. Consider a grade B auto loan (ID 123) that originated this month (February 2017), with a current balance of $10K. Assuming this loan is not under review for individual impairment, and we expect it to behave like a typical grade B auto loan, then the ratio we apply to the loan should have first year balances in the denominator, and lifetime loss amounts ($) in the numerator. As long as the total balance in the denominator includes a representative sample of loans (i.e. the right balance of matured, prepaid, and defaulted loans), then this ratio “embeds” prepayments into the loss info.


Long story short, since loss curves are just series of ratios that already factor in prepayment history, as long as future prepayment patterns are expected to remain consistent with historical ones, nothing else needs to be done. Enjoy your bath.

Rubberduckieplayskool.jpg


Rachel Messick

Product Manager/Data Scientist at Visible Equity


COMMENTS