Kent Fisher

written by Kent Fisher
Vavrinek, Trine, Day & Co., LLP

Joint issuance by U.S. and International Accounting Standards Boards on a new credit impairment model would overhaul current allowance for loan and lease loss (“ALLL”) accounting.

Having finally emerged from a very busy audit season and still conditioned to working Saturdays and Sundays, I spent a recent weekend reading the comment letters (Link to Comment Letters) on the Supplementary Document (“SD”) (Link to Supplementary Document) issued jointly by the International Accounting Standards Board (“IASB”) and the Financial Accounting Standards Board (“FASB”) on January 31, 2011.  The SD outlines some key elements of a proposed new credit impairment model that will likely supplant the current FAS 5 (ASC 450) and FAS 114 (ASC 310) credit impairment model.

If you’re wondering…………no it was not a fun weekend and yes I do need to get a life.  But as much as the details of this proposal are extremely laborious and difficult to understand, it is imperative that bankers and accountants get a feel for the general direction that credit impairment is headed. After all, the ALLL is the most significant management estimate in bank financial statements and it remains ground zero in the ongoing struggle between bankers and the banking regulators.

The comment period ended April 1, 2011 and now that the comments are in we can add this proposal to the fair value accounting and lease proposals as being universally loathed by the banking industry and most who serve it.  But with FASB having already backed down recently on the fair value and lease proposals, is it possible that the two Boards would allow this proposal to also go down in flames?

We can only hope so given the tangled mess that this proposal represents.  That said, recent indications are that the Boards remain undeterred by the negative comments received and so a new credit impairment model along the lines of that proposed in the SD is likely.  So grab your favorite caffeinated drink and please read on…..

This issuance represents the forced convergence of two very divergent approaches as the two Boards seemingly rushed this issuance to meet their previously announced deadline of June 30, 2011 (now delayed until late 2011).

The Boards had previously issued, in 2010, separate proposals to address credit impairment, driven primarily by how poorly the current models performed in the recent financial crisis.  Both were intent on moving from an incurred loss model to an expected loss model and the two initial proposals were similar in that both provided for a single model based on expected losses, consideration of credit losses over the life of the asset, elimination of an impairment trigger or probability threshold and changes in credit loss estimates recognized immediately.

But there were significant differences as well.  The IASB model was based on expected cash flows calculated using an operationally difficult probability-weighted approach.  These cash flows were to be reduced by credit losses estimated on day 1 of the loan with interest income recognized at a rate that discounted the expected cash flows, net of the estimated credit losses.  IASB’s proposal also allowed for consideration of future economic and market conditions when making expected cash flow estimates.

Basically, IASB wanted to gradually build the allowance for credit losses over the loan’s life through reductions in the amount of interest income recognized.  This reflected IASB’s core view that credit loss considerations are a component in the pricing of the loan and their initial proposal was generally considered to have an earnings focus.

Conversely, FASB’s focus was on the balance sheet as they were more concerned with the ability of the allowance to absorb all expected credit losses.  Their initial model called for immediate recognition of all lifetime expected losses based on estimates of cash flows that would not be collected.  FASB’s proposal did not allow for consideration of future economic and market conditions when estimating cash flows, contrary to the IASB proposal.

So how were these two disparate proposals on credit impairment cobbled together?

Well the SD reflects compromise by the two Boards on their original proposals while attempting to preserve both their respective objectives.  Unfortunately, the compromises necessary to retain both an earnings and balance sheet focus create a confused and unnecessarily complex model that left many of the commentators to the SD unimpressed.  So much so that a group of fourteen large and influential U.S. banks, including Wells Fargo, Bank of America and JPMorgan Chase put forth in their comment letters a detailed alternative proposal for the Boards to consider in lieu of the jointly proposed model.  While the details of this model are beyond the scope of this discussion, its main selling point is that it works to improve the existing impairment framework and would not require the significant operational complexity involved in implementing the jointly proposed model.  Essentially, it takes the existing model and replaces the “probable loss” threshold with a “reasonable expectation of loss” threshold.  This is very sensical, but will most likely not get serious consideration because of the need to have a converged model that is acceptable outside of the United States.  The urge to converge trumps common sense at this point I am afraid.

So what does the new model look like?  Below is the basic framework:

o Segregation of the loan portfolio into a “good book” and a “bad book”.

  • Loans are transferred from the good book to the bad book when management focus turns from receiving regular payments to recovery of all or a portion of the loan.  No bright-line criteria for when this transfer is to occur.

o  Good book allowance equal to the greater of:

  • Time-proportional expected credit loss amount (“TPA”), or
  • Expected credit losses expected to occur in the foreseeable future (“ELFF”), defined as not less than twelve months after the reporting date.

o  Bad book allowance equal to remaining lifetime expected credit losses (“RLEL”).

o  Provisions for impairment losses recognized separate from interest income in the statement of earnings.

o  Future economic and market conditions to be considered when estimating cash flows.

While a loan is in the good book, the emphasis is on accurate revenue recognition (through the TPA) thus appeasing IASB’s core objective, while providing a “floor” (through the ELFF) so that a sufficient allowance is recognized for loans that experience losses early in their life, thus appeasing FASB’s core objective.  When a loan transfers to the bad book, emphasis shifts from revenue recognition to balance sheet carrying value by recording the bad book loan at an appropriate carrying amount reflecting the RLEL.

IASB compromised by accepting the less operationally difficult TPA as the method for estimating lifetime losses and by decoupling credit losses from the effective interest rate calculation thus resulting in the separation of credit losses and interest income in the statement of earnings.  FASB compromised by limiting the “floor” on the good book calculation to the foreseeable future as opposed to the remaining life of the loan (although U.S. federal regulatory agencies recommended in their comment letter that this provision be clarified to indicate that the foreseeable future will often extend beyond the minimum 12 month period provided for in the SD).

Compromise is always nice and sometimes necessary, but in this case it’s going to mean a lot of extra work by banks to calculate both the TPA and the ELFF for the good book portion of the allowance.

It’s still a bit early to know for sure, but the ELFF appears to be an embellished version of the current FAS 5 calculation.  So that part should be manageable for most banks.  However, the SD’s definition of foreseeable future (“the period for which specific projections of events and conditions are possible and the amount of credit losses can be reasonably determined”) is considered by many to be overly broad, vague and open to interpretation.  This lack of clarity could result in reduced comparability between entities and also opens the door for increased regulatory influence on this component of the ALLL.

The TPA portion of the good book calculation is of more concern.  Many are probably wondering at this point exactly what is meant by “time-proportional expected credit losses”.

Essentially, the TPA represents the remaining lifetime expected credit losses for each loan class multiplied times the ratio of the class’s weighted-average current age (length of time outstanding since initial recognition) to the class’s weighted-average expected life (period expected to be outstanding since initial recognition).  This approach has been labeled the straight-line method and the SD allows this be calculated on a discounted or undiscounted basis.  Alternatively, entities can convert the remaining lifetime expected credit losses into annuities on the basis of the class’s expected life.  Below is an example calculation of TPA using the undiscounted straight-line method:

Portfolio Remaining Lifetime Expected Losses (“RLEL”) Weighted-Average Age To Date Weighted-Average Expected Life Annual Amount Time-Proportional Amount (“TPA”)
A B C D=A/C E=B x D
Commercial $600,000 2 years 3 years $200,000 $400,000
Real Estate $1,500,000 3 years 5 years $300,000 $900,000

It’s a relatively straight-forward calculation once you have the key inputs, but therein lies the problem. RLEL is a concern in that the proposal does not specify how entities are to calculate lifetime expected losses and bankers have not previously had to estimate losses on this basis.  Worse is the additional complexity and subjectivity of continuously establishing the weighted-average expected life for each ‘good book’ segment of an ever-changing loan portfolio, with consideration given to defaults, extensions, prepayments, renewals and any other factors that might cause the expected life to differ from the contractual life.  This information is not currently available on most core data processing systems.

With most smaller institutions likely to struggle with the TPA component, many commentators to the SD (including the U.S. federal regulatory agencies) recommended that the Boards either completely eliminate that component or provide for a practical expedient that would require only the calculation of the ELFF floor when an entity can demonstrate that the TPA would be less than the floor.  This would likely be the case for loans with short weighted-average lives or where losses occur early in the loans’ lives or when the weighted-average life is consistent with the foreseeable future time horizon.  Elimination of the TPA component would certainly make the new model more palatable, but there is no indication that the Boards are amenable to this remedy.

Below is a summary of significant additional issues and questions raised by the SD, as identified by various comment letters:

  1. Complexity and confusion over how to transfer between good and bad book.
  2. Whether the bad book should be limited to the current FAS 114 impaired loans (bankers say yes, regulators say no). How would classified, unimpaired loans be categorized?
  3. Concern about the dependence on long-term forecasts, which community banks do not typically prepare or have ready access to.
  4. Although the ELFF will most likely always be higher than the TPA, entities will still be required to perform the TPA calculation each reporting period to appease auditors and regulators.
  5. The SD only addresses open portfolios (those with new loan activity changing the composition) and does not address closed portfolios, investments in debt securities, TDR’s or purchased loans with evidence of credit impairment (“PCI”) since origination, making it difficult to fully evaluate the proposal.

All of the above makes one yearn a bit for the pre-recession days when earnings management was the primary concern with the ALLL and the SEC was pushing for a strict hand-to-mouth, incurred loss model.  Reserving for credit losses only when they are imminent sounds refreshingly simple compared to what is now looming.  But there is no turning back at this point with the occurrence of the great recession and the speed at which otherwise well-capitalized banks succumbed to their loan losses.

Although the U.S. federal regulatory agencies did have some criticisms of the SD in their comment letter, one might presume that the proposed model fits nicely with their current view of the ALLL.  They are already wielding significant influence on how institutions determine their ALLL and that influence will only grow with the increased subjectivity that would be inherent in the proposed model.  Their support for the overall direction of the proposed model certainly increases the likelihood of passage.

In that regard, FASB issued an update on May 17, 2011 (Link to FASB Update) indicating that while feedback to the SD was “mixed”, there seems to be a majority view that it would be best if the Boards could agree upon a converged solution for impairment.  Reading between the lines of this statement, it would not be surprising to see a final standard issued eventually that will include much of what is included in the SD and discussed above.

Better start preparing as the good (book), the bad (book) and the ugly (accounting) could be coming to a bank near you soon.

Tagged with:

Leave a Reply

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

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <strike> <strong>