CECL Static Pool White Paper

Transcription

1 CECL Static Pool White Paper Introduction In this white paper, we offer an in-depth description of Visible Equity s Static Pool method for collectively reviewing loans in CECL. This method is a CECL-compliant loss rate model, in which lifetime loan losses are captured as a percent of current balance for a homogeneous pool of loans. In this paper, we will include a description of the Static Pool method, examples of its use, the required data for carrying it out, and a discussion of when it should be used. The Method Static Pool is a type of loss rate method. Meaning, it seeks a rate representing the percent of current balance expected to not be collected. This is not a foreign concept to many financial institutions, as it resembles a look-back loss rate in which a loss rate is obtained by summing up all losses incurred over a historical look-back period and dividing by the average portfolio balance over that period. In both Static Pool and look-back methods, a loss rate is multiplied by the outstanding balance to obtain an allowance. However, since a look-back loss rate doesn t capture the percent of current balance expected to not be collected over the lifetime of the portfolio as required by CECL, the steps taken to obtain the loss rate differ between the two methods. For a given class in your active portfolio, steps for calculating a Static Pool loss rate are as follows: 1. Determine the remaining contractual life, L, of the class 2. Isolate a pool of loans that were active at least L years ago 3. Sum up total losses that occurred in the pool over a period of L years 4. Divide total loss by the balance of the pool at the starting point So, the rate is obtained by isolating a historic pool of loans and tracking its loss behavior over the remainder of its life. Ideally, conditions will be similar between the historic and current pools, but where that is not the case, the institution should adjust for current conditions and reasonable and supportable forecasts. 1

2 Examples Example 1 Suppose it is June of 2017, and Institution X wishes to calculate a Static Pool loss rate for its auto portfolio, which has an outstanding balance of $400M and a remaining contractual life of 6 years. The institution determines that the pool of auto loans active in June of 2011 (not to be confused with the loans that originated at that time) has reasonably similar risk characteristics as the current pool. That group of historic loans is isolated and tracked over time. Table 1 details how the pool deteriorated over the last 6 years. For example, from the table we see that, of the $250M that was outstanding as of June 2011, $98.5M was paid off, $1.5M charged off, and the remaining $150M carried over to the next period. We observe from the table that a total of $3.62M was charged off for this group of loans. Notably, other charge-offs certainly occurred over the last 6 years, but we are only interested in those from the isolated pool of loans. Table 1 Example 1 Static Pool loss rate calculation Beginning Charge-off Period Closed Ending 06/ /2012 $250M $98.5M $1.5M $150M 06/ /2013 $150M $98.8M $1.2M $50M 06/ /2014 $50M $19.5M $500,000 $30M 06/ /2015 $30M $14.7M $300,000 $15M 06/ /2016 $15M $9.9M $100,000 $5M 06/ /2017 $5M $4.98M $20, / Total $3.62M Dividing the total charge-off balance by the balance as of 06/2011, we achieve a loss rate of $3,620,000 $250,000,000 = 1.448% 2

3 In other words, of Institution X s auto balance as of 06/2011, 1.448% was charged off over the life of the portfolio. To estimate the charge-offs over the life of the current portfolio, the rate is applied to the current balance of $400M to get $400,000, % = $5,792,000 So, Institution X s loss allowance for its auto portfolio is approximately $5.8M. In this example, it was determined by the institution that risk characteristics were similar between the current portfolio and that of 06/2011. An implication of that assumption is that grade concentrations were similar between the two groups. If this is not the case, and one group carries greater risk than the other, then credit quality indicators (CQIs) should be used in segmenting the portfolio. See Example 2 for an illustration of this situation. Example 2 Suppose it is June of 2017, and Institution X wishes to calculate a Static Pool loss rate for its auto portfolio. However, the institution recognizes that underwriting standards have changed over the last 6 years, and today s auto portfolio carries more risk than the portfolio in To adjust for this, Institution X segments its auto portfolio by current grade, and performs the Static Pool calculations separately for each segment. Let s consider B-grade auto loans, which have current credit scores between 660 and 700. The outstanding balance of B-grade auto loans is $150M, and the contractual term of the segment is 6 years. The first step in calculating the loss rate is to isolate the pool of auto loans that were B-grade 6 years ago (06/2011). Table 2 outlines the deterioration of this pool over the last 6 years, showing a total of $660,000 in charge-offs. It follows that the loss rate is estimated as $660,000 $30,000,000 = 2.2% and the expected loss from the current balance of $150M (for B-grade auto loans) is $150,000, % = $3,300,000 So, in this example, differences in credit quality between current and historical data are accounted for through CQI segmentation. 3

4 Table 2 Example 2 Static Pool loss rate Grade B ( ) Beginning Charge-off Period Closed Ending 06/ /2012 $30M $9.7M $300,000 $20M 06/ /2013 $20M $9.9M $100,000 $10M 06/ /2014 $10M $4.85M $150,000 $5M 06/ /2015 $5M $2.96M $40,000 $2M 06/ /2016 $2M $940,000 $60,000 $1M 06/ /2017 $1M $990,000 $10, / Total $660,000 Required Data History is the key to Static Pool. While it doesn t require much detail, this method does require a more extensive history than other methods. Specifically, it requires monthly, loan-level balances and charge-offs going back at least the term of your current portfolio. This may be difficult for longer-term loans especially for the first few years of CECL implementation. When to Use Each method for collectively reviewing loans comes with its own strengths. The main strength of Static Pool is in its simplicity. It involves a single ratio calculation and won t be difficult to explain to regulators and other stake holders, and it is likely appropriate for a large portion of your portfolio. There are some cases, however, wherein Visible Equity recommends that you choose an alternate method. First, Static Pool will not work well for lines of credit. FASB states in the ASU that the allowance for lines of credit (with unconditionally cancellable commitments) should only consider funded balances. Meaning, only the amount of current balance expected to not be collected should be considered. For example, suppose credit card A has a current balance of $5,000, and a fortune teller informs us that the $5,000 will be paid off next month, but that sometime next year, the borrower will make new purchases on the card, which will subsequently default. For CECL, that 4

5 future loss is irrelevant to the institution today because all $5,000 of the current balance will be collected. Using a Static Pool analysis would require us to isolate a pool of loans at a historic point in time and track how that specific balance defaulted over time, not counting charge-offs from any funds added after the starting point. Another scenario in which Static Pool will be difficult is for asset classes with long terms. To properly carry out Static Pool, you need to go back at least the remaining life of the portfolio. Most institutions won t have a sufficient history to properly analyze long-term loans such as mortgages. Finally, the third case when Static Pool might not be best is when the distribution of loan ages in your portfolio has changed significantly. For example, if your historic pool of loans was mostly new with only a few seasoned loans, and your current portfolio is mostly seasoned with only a few new loans, your loss rate might be overstated since it was estimated with a newer pool of loans. 5