Unraveling the (Seeming) Complexity of CECL

Premier Insights - Industry Updates - Unraveling the (Seeming) Complexity of CECL

The new CECL standard is presumably designed to enhance the stability of the financial sector by providing more accurate assessments of loan losses.  It also requires a change from the current estimates of loan losses that are produced by most institutions today to projections or forecasts.

Unfortunately, even as we close in on implementation, there still remains a great deal of ambiguity surrounding what this actually means.  One confusing thing for many institutions is the word “Expected” in the acronym (Current Expected Credit Losses or CECL). 

When a lender makes a loan, they have evaluated the credit risk and are not expecting a loss (otherwise they would not have made the loan!).  The word “Current” is likely equally confusing as well, as, if it is the future being considered, how can it be current?    

Then there is the laundry list of different approaches and methods and only marginal guidance, thus far, from the regulatory bodies.  Additionally, the guidance that has been provided has avoided one of the main aspects of loan loss reserves that CECL was meant to help define – Q or qualitative factors. 

We have summarized CECL and have written about other facets in past articles, so we will not repeat them again here.  Instead, we will attempt to unravel a little bit of the complexity by taking a broader view that hopefully may make the specifics more sensical and easier to grasp. 

The probability of default approach, or more specifically probability of default / loss given default (PD|LGD) is the most versatile and intuitive of the models put forth to address CECL.  We discuss below some of the basic concepts in this context in an attempt to provide some simplification. 

What is meant by “Expected”

As noted, a lender typically does not make a loan under the pretense that the loan will default.  However, risk is inherent in all lending.  Although loss is not “expected” on any particular loan, a certain proportion of loans will end up defaulting.  The recognition of this fact is one of the main aspects of CECL that distinguishes it from current methods.

The best way to understand what is meant by expected is from the statistical perspective.  An expected value is akin to an average.  Think of anything that has an average associated with it – miles per gallon for a vehicle, for example.  The average is what is expected. 

Let’s assume children in the U.S. begin walking at an average age of 11 months.  If you wanted to predict when a particular child would walk, you would pick the average.  You may or may not be right on one particular child; but if you repeated the prediction for different children an infinite number of times and always chose the average, you would be correct the maximum number of times. 

Now apply that principle to a loan portfolio.  The likelihood of default may be a rare event, perhaps < 1%.  Although any one given loan in that portfolio may perform to term, on average there will be a default rate of some amount.  Again, it is the recognition of this expectation that is one of the primary differences in CECL versus current ALLL methods. 

Where current methods may estimate losses based solely on the asset’s current performance, CECL requires the recognition and quantification of the inherent risk in the entire loan pool.

In addition, particularly with the PD|LGD model, there is not reliance on a simple average for the portfolio or even a specific product.  Instead, the precision is enhanced by accounting for other factors that impact default in the model and conditioning for them. 

Opportunities to Enhance Profitability Exist With the CECL Standard

Using such an approach provides an opportunity to develop more precise loan pricing for an institution.  Quantifying hard costs such as capital and overhead are easy to incorporate into loan pricing.  The more difficult piece is adjusting for loan losses.   Typically what is done is to adjust the overall product(s) based on historic loss rates by some measure. 

The problem, for community banks in particular,  is this is not a truly precise and accurately quantified adjustment.   For example, if the institution desires a margin of 4.00% on a product, and the overall loss rate is .75%, how do the loans need to be priced to generate this return on average?  The PD|LDG model provides a basis to provide these types of calculations.

Further, although there may be an overall average default rate of .50% for a particular pool, not every loan in the pool will have that probability of default once other factors are accounted for.    Some loans may have a higher or lower rate based on particular loan attributes.  Loans should then be priced individually based on these risks to maximize returns.

Finally, the proper CECL framework is dynamic in that precision is constantly being enhanced through data gathering and analysis.  This will produce an ever improving method that will adjust to asset quality, the credit environment, and other factors as things change. 

The question institutions should ask while making decisions concerning CECL is do they want to simply do the bare minimum to comply –  OR seize the opportunity to enhance profitability  and invest in a sustainable long term strategy?  The answer to this question should govern their decisions concerning CECL.

Leave a Reply

Your email address will not be published. Required fields are marked *