In response to my previous post, Edith Orenstein (author of the FEI blog and fellow Ohio State alum) asked me whether I preferred the status quo "incurred," or the proposed "expected" loan loss model (CECL).
Also in that post, I stated that I would have more to say about how the FASB defined "expected" in the CECL model. As it turns out Edith's question makes for a good lead in, so I'll take on both of these questions here.
The Expected Loss Model is a Move in the Wrong Direction
Actually, Edith, both models are terrible because they both require estimates of future cash flows over a loan term of 5, 10 or even 30 years out. It's ridiculous to presume that these numbers are much better than seat-of-the-pants guesses – even for the plainest vanilla loan contracts.
But, I'm not going to evade your question: the proposed "expected" loan loss model is worse.
CECL is worse because it violates basic economic logic and reasonable accounting conventions. How can the value of a loan at its inception be anything different than the net proceeds to the borrower? This was the logic for the incurred loss model; but under the CECL model not only can the carrying amount be different than economic reality from the get go, but it must be different – and that difference must be recorded as an expense on the income statement.
When a cash outflow does not give rise to an "asset" as defined by applicable accounting standards, it is perfectly understandable that a corresponding expense must be recognized. But, this CECL contrivance is different: the cash outflow to the borrower clearly gives rise to an "asset," and there is no precedent I can think of to justify recognizing only a part of that asset.
There is also the matter of discount rates in both models. I discussed in my previous post how the CECL approach is discounting numbers that aren't actually cash flows; at least the current model at least purports to discount cash flows, but with an arbitrary discount rate. What approach is worse? Let's just say that they're tied for last in a two horse race and we'll leave it at that.
So, on a scale of 1 to 10, I rate the CECL model a minus 3, and I give the incurred loss model a solid zero. As I am about to explain, the proposed CECL model will not provide relevant information to investors and regulators. It is merely another iteration of the "anything but market values" movement driven by bankers and auditors.
Pick a number … pretty much any number.
As Edith pointed out, I was pretty "acerbic" (her word) in my criticisms of the CECL model. But, I actually saved the most insidious aspect of the FASB's loan impairment proposal for last:
"The estimate of expected credit losses is neither a 'worst case' scenario nor a 'best case' scenario, but rather reflects management's current estimate of the contractual cash flows that the entity does not expect to collect.
That helps a lot; it's management's estimate, and just like that old accounting joke about two plus two, management can pick pretty much any number it wants.*
The FASB's boilerplate approach to defining expected losses will not fix the problem of overstated loan balances during a financial downturn. But, there is a method to the madness. To see what that could be, let's tryout a saner solution: how about requiring management to provide its best estimate of future uncollectibles – i.e., the number the bank CEO would bet her own house on instead of the shareholders' and taxpayers'?
The 'problem,' with this solution is that it would put even the best intentioned auditors on the hook to do something they are not qualified to do. If management must estimate in good faith the expected loan losses (i.e., the number that is just as likely to be too high than too low), then how is an auditor supposed to provide reasonable assurance that management actually did what was required of them? Perhaps the auditor could re-perform management's task, but given its complexity and the number of judgments required, it would be unlikely to come reasonably close to management's estimate. Then what?
Alternatively, the auditor could simply review management's estimation methods and parameters, check a few calculations, and then in their infinite wisdom pronounce to the world that management's estimate "appears reasonable." The problem with either of these audit approaches is that when then next financial crisis hits, auditors' judgments will be second-guessed – and if history is any guide, rightly so.
But, if the FASB's boilerplate is finalized, auditors' lives will be improved. For starters, they will finally be able to get those persnickety PCAOB inspectors off their backs. Management can throw out the number it wants investors to behold, and auditors can keep themselves relatively protected from litigation merely by devising some rote procedures that must inevitably find that the number management picked was neither the best nor the worst case.
My point, Edith, is that the question of an expected or incurred loss model is a distraction from the real issue. By whatever model, the numbers won't mean anything – and neither will the auditor's report.
There is a better way.
These are the reasons why I have repeatedly maintained that it is essential for the carrying amounts of loans to be estimated by independent third parties and not management – regardless of the measurement basis for those carrying amounts.
Neither auditors nor management are qualified to provide unbiased and reasonably tight estimates of loan loss allowances; but even if you want to dispute that, it is indisputable that auditors can much more reliably provide assurance as to whether a third party is independent, competent, and that it did what it was supposed to do.
Which brings me full circle back to market value. If an independent third party is going to measure the carrying amount for bank loans, why should it be estimating anything other than market value?
The real question is not incurred versus expected loss models. The real question is why the FASB, despite every common sense indicator, is sticking to its mantra of 'anything but market value.'
*As an aside, I don't know what "current" is supposed to mean – as of the balance sheet date, or date of the issuance of the financial statements? It's an important distinction.