Calculating your allowance for loan losses involves projecting future results using historical information as a baseline. Why does this seem counter intuitive to everything we’ve learned?
Anyone presenting historical information should tell you that historical data is not necessarily indicative of future results. Our best guess on future loss reserves involve adjusting that historical information to account for changes between then and now. The primary assumptions used to do this involve the level of disaggregation, the duration of historical charge offs to use and evaluating the effect of qualitative characteristics.
FAS 5 instructs financial institutions to separate loans into homogeneous, or similar groups. Differences in interest rate, duration, collateralization and individual borrower characteristics make calling any two loans homogeneous feel like a stretch. So the question is, how much should you actually break your portfolio down in your allowance calculation?
You could think up 100 different ways to calculate your allowance for loan losses and 95 of them would be within 20% of one another. The margin of error should be smaller for larger loan portfolios (See: sample size) but you shouldn’t expect to consistently do much better than +/- 10-20%.
That being said, I think the three main objectives of your allowance account are as follows (in order of importance):
- Record a reasonable estimate
- Satisfy examiners and auditors
- Avoid throwing your computer out the window
Many times those charged with the attestation of your financial information provide assistance or even templates to help you calculate your allowance for loan losses. That poses two major problems:
- You’re responsible for this estimate
- Your auditor should be independent. Providing major assistance in this calculation jeopardizes their ability to attest to a number. (See: AICPA Code of Professional Conduct Rule 101)
Leaving Rule 101 to your auditors, let’s talk about some considerations to help you become more comfortable with your estimate.
When thinking about portfolio disaggregation and duration of historical charge offs you should think about the size and composition of your loan portfolio and the volume of charge offs.
Below is an illustration of different levels (tiers) of disaggregation:
You can break your portfolio down into many levels. Within those levels some credit unions also decide to disaggregate further based on credit risk characteristics (e.g. FICO Tier). What we’ve seen is that further disaggregation is not always better. Without an adequate sample size, excessive disaggregation can dilute or inflate your allowance and also present red flags to your auditors or examiners.
I have always felt that living in a world of extremes is the best way to prove a point. For example:
ABC Credit Union disaggregates to Tier 3. A year ago ABC had 19 used autos and 1 new auto. ABC had no used car charge offs, but the 1 new auto charges off entirely. ABC’s auto portfolio has grown and now has 50 used autos and 50 new autos. Applying one-year historical charge off percentages, ABC reserves 0% on used autos and 100% on new autos (50% of autos in the aggregate).
The change in portfolio composition has resulted in the 5% aggregate losses to be pushed forward into a 50% aggregate reserve. This calculation fails two of our three main allowance objectives.
- Fails to create a reasonable estimate
- Fails to satisfy auditors and examiners. While they don’t mention the 100% reserve on new autos, they do ask “Why don’t you have any balance reserved for your used cars???”
- Succeeds in avoiding throwing computer out window. Throws self out window instead.
Luckily, you have multiple options in circumventing this pain and suffering, both of which involve increasing the sample size of your calculation.
The first is to combine portfolio segments together. This will give you a larger sample size of loans to look at and, presumably, a more fair pool of charge offs.
I would suggest avoiding Tier 1 – Total Portfolio Aggregation. If you’ve moved to Tier 2 and still don’t have adequate volume of charge offs, congratulations! Consider pulling historical charge offs from a 24 or 36 month period. Be sure to annualize this number. For example, if 1% of your Real Estate loans charged off over a 24-month period, your historical charge off percentage would be .5%.
The sweet spot is a happy medium; disaggregated enough to account for changes in portfolio composition but also has enough information available to serve as an appropriate starting point for projections of future results.
Once you’ve set your assumptions and quantified your allowance, you may feel the need to account for what the NCUA refers to as “Qualitative Adjustments” to your allowance. I’m of the opinion that for 95% of credit unions, like an investment professional trying to “Time the Market” to create investment gains, qualitative adjustments may have a positive outcome on how close you get to actual future charge offs, in the grand scheme of things they will result in an overall less accurate position.
Qualitative adjustments are useful to credit unions who have recently implemented a new loan type or risk tier and lack the historical information necessary to create an adequate reserve. If you recently rolled out a loan product for members with sub-600 FICO scores and the product is growing at 50% annually, it would be prudent to expect some additional losses there which could be accounted for by a “Qualitative Adjustment.”
If there haven’t been any major changes, document that you have evaluated these qualitative factors and deem them to have no impact, and move on.