Tonight, I write from Mid-Missouri, the area of the country I grew up in. My wife and I have had the pleasure in seeing our youngest son graduate with his bachelors and then are in the Show-Me State to visit my father as he is turning 90 this month. The trip has given us lots of windshield time as we drove to take my son’s beagle to him. One topic that is making the miles pass quicker is my reading on the upcoming Current Expected Credit Loss, or CECL.
CECL is an attempt by the Financial Accounting Standards Board (FASB) to make a better estimate of loan losses over the life of a loan or a portfolio. Publicly traded banks will be moving from the traditional Allowance for Loan and Lease Loss (ALLL) calculation to CECL at the end of this year. Privately held institutions and credit unions will follow suit in the next couple of years. There is a large amount of uncertainty in how to comply with the new guidelines.
One of the triggers for this move was the crash in 2008. The problems hit rapidly. Very few institutions had adequate reserves to cover loan losses since credit was deteriorating quickly and lenders were not recognizing these changes in setting aside more money for losses. The bank I worked at during this time had all the commercial lenders reanalyze the risk on each credit over one million every quarter. We tried to adjust for loan losses but even this was futile in some cases as the problems may not have surfaced in previous historic financials that were happening with the borrower at that time.
Then, as we came out of the crisis in 2010, we found that we had too much in loan loss reserves as we were attempting to do forward looking ALLL calculations based upon history. Basically, all any of us can do is to look at the past and make some predictions for the future.
The idea behind CECL is to remove the backward-looking loan loss reserve method that is used currently and gaze into the future to see what possible losses will be in a loan or portfolio throughout the entire life. FASB’s final guidance allows institutions to use different methods for CECL estimation depending upon the size and complexity of the lending. This is where the confusion lies. All institutions need to determine what models are needed, and what factors to watch to satisfy the requirement. Further confusion will come with our examiner friends as they review models which will be in place but in the early CECL years, have not been historically tested for level of success.
Foreseeable future is a vague term, and this is a concept of CECL. History is concrete in terms of loan losses. It also may be possible to have a prediction for loan loss changes with possible changes in the economy over the next 12-24 months, which could be argued as foreseeable. Once you get beyond that time, possible changes in the economy, business, and credit strength is very vague. The challenge here is that you will have credit exposure which could continue beyond the foreseeable future. Exposure within the foreseeable future can sometimes vary well beyond original predictions as we noted in North Dakota when oil prices crashed in 2015.
Models can go awry by failing to look at pertinent factors, weighing factors incorrectly, or not estimating assumptions correctly. These can also have many different factors to look at like GDP, unemployment rate, commodity prices, inflation, interest rate projections, and so forth. An institution may also have a different model for ag loans that may have a driver of commodity prices, while a model to predict housing defaults may look at something like unemployment rate, since housing defaults are more likely to occur when someone is out of work.
On the commercial side and with other loans as well, a projection for the exposure at default (EAD) is a conceptual driver behind what is needed for CECL. EAD is a combination of the probability of default (PD) and the probability of attrition (PA). PD is a measurement we do in risk rating. Based upon several factors of our choosing (e.g. Debt Service Coverage Ratio, Current Ratio, Debt/Worth) that we believe to be good indicators of financial stress in a company, what is the probability of a default in the near future? PA attempts to answer what portion of the loan balances will still be outstanding at the time of the default. Loan balances should be lower than what they are today with principal reductions from amortizations and accelerated principal paydowns from borrowers. Loans with a substantially higher interest rate compared to the current market rate will have more of a chance of being refinanced and you losing the entire loan balance. Your institution’s exposure at the time of default is a combination of the PD x PA.
The next factor to consider is what is the loss given default (LGD). When EAD hits, what will my LGD be? This differs greatly from loan to loan. A credit leveraged at 80%, secured by accounts receivable will have more LGD than a real estate loan leveraged at 40%. In some cases, perhaps like the last one, the LGD will be 0. Low LTVs on collateral that is marketable, government guarantees, and cash collateral are some factors that will lower the LGD.
Many lenders will have one risk rating scale that combines factors considering the probability of default and the possible loss if a default occurs. Given these changes with CECL, institutions may move from a single risk rating scale to one that measures the PD and a separate one that measures LGD.
The EAD and LGD are concepts to consider when starting down the road to inspect the possible CECL impact on a loan or on a portfolio. Understanding the logic here, forms a basis that expected losses can be estimated. Then you can utilize your various data points to measure the possible changes in the environment that will impact you CECL.