JHA_HeaderImage_Blog_v6

Three Reasons CECL Results Are Disconnected from Current Allowance for Loan and Lease Losses (ALLL)

Posted by Jerry Boebel

Wed, Apr 15, 2020 @ 10:19 AM

Blog-2020-04-15-1

Since the Financial Accounting Standard Board’s (FASB) Current Expected Credit Loss (CECL) standard was announced in 2016, ProfitStars has worked with more than 180 of our clients to implement a CECL model.

As we worked with the first of our customers to go live at the end of 2019, we compiled support resources to answer the most common questions. The most frequent support topic by far has been trying to explain the gap from loan loss reserves under Financial Accounting Statement 114 (FAS 114, the current standard) and reserves determined by CECL accounting rules.

On the surface, expecting connectivity between historic losses, FAS 114 reserves and CECL reserves would seem to be reasonable. It’s not quite that simple.

In fact, the accounting treatments are so different that one should expect significant differences in CECL results against FAS 114 results and even against historic losses themselves. There are several reasons for this.

FAS 114 Results May not Reflect Actual Historic Losses – The CECL modeling process relies heavily on actual historic losses. If the FAS 114 reserve amount does not reflect history, expect a large gap. We recently encountered this scenario when installing a new client:

  • 5-year loss history of 0.02% (not two percent, point 0 two).
  • 3 consecutive years of positive losses (recovered more than was charged off)
  • Current reserves are 2.1% of balances

In this instance, current reserves are not supportable. Yet, the bank expects CECL reserves to be similar and they are challenged to cover that gap. Auditors have recognized that some financial institutions are targeting a specific outcome and then adjusting elements of their CECL model to get to that value. They are referring to it as anchoring bias and the CECL Audit Practice Aid put forth by the AICPA calls this out as a risk to be identified during the audit. If FAS 114 reserves cannot be supported by historic losses, neither will CECL reserves.

CECL is a lifetime estimation – FAS 114 reserves are meant to cover losses actually incurred or those with reasonable certainty to occur. The observed timeframe for FAS 114 losses is less than 1 year. The CECL standard applies a loss expectation to every single loan for the remaining life of each loan. Your CECL model results for a 12-month forecast may be very close to your current ALLL reserve for the same loan. But when you apply that 12-month loss rate to a multiyear horizon, the required reserve can be much larger. The loan categories impacted by lifetime calculation are obvious:

  • Long-term loans, especially residential real estate.
  • Loans subject to prepayment. To the extent that prepayments shorten the expected life of the loan and therefore reduce the required CECL reserves, it is important the lender have experience in modeling prepayments. Prepayment expectations can reduce a 30-year loan to a 9-year loan, perhaps reducing the reserve by 25-30%.
  • Lines of credit. Modeling a borrower’s pattern of draws and paydowns may be even more complicated than modeling credit losses. HELOC’s are an area of ambiguity in the standard and we are seeing several competing interpretations.
CECL allows the lender a wide range of acceptable methodologies - If a model vendor provides results from 4 different methodologies, the reserve calculations across all these models might be dispersed by a magnitude of 2-3 times. Variances of this size invite scrutiny and even disbelief of the validity of a particular model. Not only can a lender choose from many methodologies, they can use different methodologies for different categories of loans.

So how does one select the right methodology?

Ascertaining the ‘accuracy’ of a methodology is probably the wrong approach for lenders. Instead, focus on the data a model uses and understanding how calculations are performed to determine if a methodology is ‘appropriate’ for a certain type of loans. I’ll demonstrate with an example; roll rate methodology.

Roll rate models collect delinquency information and use historic patters to forecast the probability of an individual loan going delinquent and then subsequently advancing on to charge off status. Data critical to other models like loan to value, FICO score and prepayment expectations is not gathered and used in calculations. Roll rate has proven to be very predictive over short time horizons (out to 6 months) but forecast accuracy declines the farther your forecast extends.  Because the data being used is very thin, lenders may judge this method to be unsophisticated and not robust. However, if your loan category has little data to provide to begin with and the expected lifetime of the loans is less than 12 months, roll rate may be entirely appropriate and supportable.

Volatility is understandably inherent when migrating from FAS 114 reserves to CECL reserves. Banking trade groups and even CPA’s have pointed out that this accounting rules change may have the most significant impact on earnings for financial institutions over the past 25 years. To understand the effects to the financial statements early and be as prepared as possible for CECL implementation, financial institutions should focus on these implementation management guidelines:

Accumulate as much loan history as possible.

Consider not only number of months of history, but also the breadth of data available for each loan. Do not wait until a few months before adoption date with the hope that CECL will be delayed further.

Run CECL parallel with FAS 114 for an extended period.

This allows for time to research large variances, understand the cause and then plan on how to address them. Longer parallel runs also allow time to get feedback from auditors, feedback from examiners and reconcile any differences.

Review the results from multiple methodologies.

Understand their differences and then choose the model that is most reflective of your loans and that you understand and are most comfortable explaining.

Topics: Fintech

Subscribe to Email Updates

Recent Posts