written by: Daniel St.Clair
FASB issued Accounting Standards Update (ASU) No. 2016-13: Financial Instruments – Credit Losses (Topic 326): Measurement of Credit Losses on Financial Instruments this past June. The ASU creates topic 326 for credit losses and pulls the related guidance from several other codification sections (mainly 310, 320 and 325) and groups them into the new section. This standard moves from the previous incurred-loss model and now requires an expected-loss approach related to the accounting for credit losses. The incurred- loss model focused on identifying and booking an allowance when potential losses become probable. The new Current Expected Credit Loss (CECL) model requires the full amount of all estimated credit losses projected over the life of the financial asset to be booked up front when recording the original financial asset at cost, resulting in a change in the timing of when the reserves are actually established.
This standard takes effect for all non-SEC filers for fiscal years beginning after December 15, 2020, while SEC filers begin the previous year. Early adoption is allowed, but not before fiscal years ending 2018. The standard effects more than just the allowance for loan loss reserve at financial institutions; it will also apply to most debt instruments, trade receivables, lease receivables and other financial assets not measured at fair value.
While most literature notes this as a significant change for financial institutions, I believe it actually aligns the guidance closer to the current methodology used in establishing the allowance for loan losses (ALLL). In the current methodology model, it appears that any specific reserves and the general reserves on impaired assets might follow the “probable” definition. However, the general reserve or old FAS 5 portion was already based on potential losses and would be hard to classify as probable under the old definition. That being said, there was a general look at recent charge offs, typically not more than the past three years, adjusted by qualitative factors such as economic changes, markets, growth and risk appetite that was recorded as the estimated allowance. Under the new CECL regulations, the institution will need to build out that loss estimate to project the expected loss over the life of a loan based on a variety of current and updated parameters.
Pooling Assets Under CECL
When applying the CECL model, an institution will need to pool their assets by certain established characteristics of similar risks, which could include geography, term, age, industry and size. They will also need to be broken down into groups by type, similar to the current reporting requirements, as the footnotes will be organized by the current groups such as commercial real estate (CRE), residential real estate (RRE), commercial and industrial (C&I), consumer and any other categories currently being used by the institution to make their disclosures meaningful. Any assets that do not share common risk characteristics for a pool will need to be assessed on an individual basis.
The new standard does not require a specific method for establishing the estimate. However, the methodology will likely include various components including historical charge-off rates, current conditions and supportable forecasts. There have been several models recommended including:
- Discounted cash flows analysis
- Average charge-off method
- Vintage analysis
- Migration analysis
- Probability-of-default method
- Regression analysis.
It is my belief that a complicated model, such as regression analysis, is usually not required and could easily muddle down the process and make it less effective. A complicated methodology is probably only necessary when you have a complicated loan that does not fit into a pool of loans with common risk characteristics.
Since the current models used by many banks already begin with historical loss information adjusted by qualitative factors such as trends in economics, risk, management and growth, I believe that is the proper starting point. The guidance notes in Section 326-20-55-3 that “Historical loss information generally provides a basis for an entity’s assessment of expected credit losses.” The real shift is in thinking that on day one, a projection of all expected losses for this type of loan (type broken down into pools of risk characteristics) should be recorded. Therefore, an estimate of the probable losses currently contained in the portfolio as currently performed is no longer adequate.
The institution must establish an estimate that projects all potential losses within the portfolio over the entire life of the loan on day one. Unfortunately, for loans with a longer life there is both a greater the risk of loss and a greater uncertainty in establishing the estimate.
The next steps include how to organize the loan types into pools with similar characteristics and verifying that the information within the core processing system can be run in a manner that easily identifies the outstanding amounts within the individual pools. The good news is that the current base used by most community banks should provide an adequate starting point upon which to build, but one must have adequate information in order to build the next layers into the model. The other good news is that there is time to establish the process. However, in order to build the pools and related projections management should start evaluating the information available as soon as possible, as you will likely need to add additional subcategories within the system to effectively isolate the information.
A Deeper Look: Breaking down the Allowance
Let’s take a look at a couple of categories as examples. The easiest may be in residential real estate. Many current allowance methodologies look at all general RRE loans the same. Some may break down more into regions, such as cities or counties, if they have slightly different profiles. Under the new CECL guidance, it would be prudent to break down that analysis further into loan sizes based on collateral and/or loan balances. In many instances, there are several levels of risks on RRE based on loan size. The middle market loans are often the most stable. For our example, we can set a floor of around $100k to $150k and a ceiling around $250k. This size of loan is often to a middle-class borrower with stable income and the collateral home is often equally stable in value. Smaller loans may have less stable borrowers or collateral in less stable neighborhoods which could increase the risks.
Even more risky are the larger jumbo loans. We have seen many examples of doctors or lawyers that had a custom built home with values over $500k, and often over a million, that ran into some kind of personal issue such as divorce. Suddenly they are not making payments on a large loan and the collateral can be very difficult to resell at the current balance on the bank’s books. With a large enough RRE portfolio, a bank should probably look at a least three risk pools per geographic area to properly comply with CECL.
Now let’s discuss commercial real estate briefly. Most lenders would agree that owner occupied CRE is one of the lower risk CRE loans. Office buildings or strip malls for lease are generally higher risk. Multi-family complexes may be higher or lower depending on condition and location, as would hospitality loans for hotels. Current methodologies may look at CRE as a single group, or at best two or three subcategories. As just noted, office buildings, apartments and hotels could be two or three subcategories each, based on the new guidance.
Specific Allowances under CECL
Any assets that do not share common risk characteristics for a pool will need to be assessed on an individual basis. This means any high risk loans, which would include loans currently classified as impaired or substandard, and any unusual loans that cannot be pooled based on risk characteristics, will be evaluated on a loan by loan basis. Obviously any loans that cannot be pooled will generally carry a higher risk profile and have a larger allowance place on the individual asset balance.
Recommendations Moving Forward
Our best recommendations are to stay calm and try not to build a model so complicated that it cannot be reasonably maintained.
- Start planning for CECL now. Gather the right people, including financial, lending and IT, to plan your strategy for implementation. Areas to discuss include the breakdown of loans into common risk pools, which methodologies make the most sense by group, and what data is currently available as well as additional data that will be needed to properly document the information required, including risk pools, historic losses, forecasts and related allowance.
- Be realistic in the current starting point of historical losses, including any additional losses that were incurred during disposition of other real estate.
- Clearly document the thought process related to the qualitative adjustments for current conditions, which is likely the most important area of the model, and should include internal risks (lending strategies and underwriting practices) and both micro and macro external risks (industry issues, business environment, geographic and economic risks, and other market conditions). Gather and maintain your support as you go, so there is an audit trail for both the regulators and your auditors.
- Lastly, don’t get trapped by the “crystal ball” idea of having to look into the future. The guidance clearly states the use of “reasonable and supportable forecasts” so avoid trying to create something you are not meant to create. This regulation, like many before it, could present challenges for some, yet they are not insurmountable. Advance planning and a measured approach will help banks navigate this latest regulatory hurdle, now and into the future.
Opinions expressed in this article are those of the author and not necessarily guidance by Briggs & Veselka.
Daniel St.Clair is an Audit Director in Briggs & Veselka’s Financial Services group, where he consults with community banks, de novo banks, credit unions and other lenders on financial reporting matters. Contact him at firstname.lastname@example.org.