What the FASB calls “contingent liabilities” in FAS 5, the IASB terms “provisions” in IAS 37. I prefer the IASB’s language, because of its pejorative ring to my American ears: it more clearly connotes the virtually unchecked discretion issuers have abused to obfuscate liabilities and trends in earnings.
In a previous post, I described how the IASB has been working on giving issuers even more discretion to manage their provisions.At the same time, the FASB has been fighting the good fight, but getting its butt kicked in the process. Their latest defeat was the issuance of FSP FAS 141(R)-1, to back off of their requirement to measure at fair value contingent liabilities assumed in a business combination.
Part of the FASB’s problem is that they still haven’t fully come to grips with measuring the fair value of any liability, much less a contingent liability. It seems reasonable enough to aim for consistent measurement of all assets acquired and liabilities assumed; but, how does one measure the fair value of a matter under litigation? Even if it were legal to pay a third party to assume the consequences of a matter in litigation, a ‘moral hazard’ is likely an unwanted by-product. Specifically, the transferor’s incentives to act in ways that could lessen the ultimate consequences of the obligation could diminish significantly. The bottom line: score one for the issuers who fought the offending provisions in FAS 141(R) and made the FASB retreat.
But, as the FASB has retreated in one area, it is retrenching in another with the issuance of Proposed FSP No. FAS 157-f, Measuring Liabilities under FASB Statement No. 157. The FSP, among other things, addresses fair value measurement of a liability when contractual restrictions may prevent it from being transferred. The proposed FSP provides that if a quoted price in an active market for the identical liability is available, that price must be used to measure fair value. In all other circumstances, fair value would be measured by selecting the one approach, of four allowable approaches, that maximizes the use of relevant observable inputs and minimizes the use of unobservable inputs. The four allowable approaches are as follows:
The quoted price of the identical liability when traded as an asset in an active market;
The quoted price of the identical liability, or the identical liability when traded as an asset in markets that are not active;
The quoted price for similar liabilities, or similar liabilities when traded as assets in markets that are active;
Another valuation technique that is consistent with the principles of SFAS 157.
So, as best as I can tell, the issue of measuring contingent liabilities is still not resolved. Since FAS 157 defines the fair value of a liability as the price that would be paid to transfer a liability in an orderly transaction between market participants at the measurement date (para. 5), it seems that the FSP is providing no help for contingent liabilities. The first three approaches permitted by the proposed FSP are clearly inapplicable, and there is nothing new in the fourth approach that would permit the evaluation of a contingent liability from the perspective of the asset holder.
A Solution: Assume There is No Problem
As I mentioned earlier, both the FASB and the IASB have been working on the contingent liability problem in their own separate ways. The FASB has been getting lambasted for their proposals to enhance disclosures and fair value measurement, while the IASB is going on its merry way to see how they can make a new and improved version of chicken salad for issuers. They have been mulling over an approach that would recognize all ‘present obligations’ (whatever that means), regardless of likelihood of payment, if they can be measured ‘reliably’ (whatever that means). Since likelihood of payment is being taken out of the equation, whatever ‘present obligation’ means is the crux of the matter. And, as it turns out, a hypothetical case scenario that the IASB discussed during its deliberations is a convenient vehicle for me to express how I think contingent liabilities ought to be handled.
Here’s the case:
A vendor sells hamburgers in a jurisdiction where the law stipulates that the vendor must pay compensation of C100,000 to each customer who receives a contaminated hamburger. Past experience indicates that one in every one million hamburgers is contaminated. On the last day of the reporting period, the vendor sold one hamburger.
And, here are some questions, along with my responses:
Question #1: Does the vendor have a present obligation at the balance sheet date?
My response: No.
Businesses have many responsibilities under various laws: reasonably ensuring that no physical harm will come to their customers as a result of their actions, truth in advertising, not polluting, and so on and so forth. Although investors would care to know about these exposures and their potential affect on future cash flows it is simply not practical to convey this information through recognition of contingent liabilities on the balance sheet. It has been vividly demonstrated over the past 35 years that FAS 5 has been in existence that it is a pipe dream to think that reliable estimates can be expected from issuers when doing so conflicts with the interests of executives. (Note as well, that the ability to precisely state an amount for the contingency does not affect my response to this question.)
Question #2: Does the answer change if the probability of contamination is more likely than not instead of one in a million?
My response: No.
This gets at the key issue as to whether the probability of an outcome should determine whether a contingent liability is recognized. I agree with the IASB insofar that the probability of an outcome should not affect recognition. Whatever their reasoning, however, mine is that FAS 5 and IAS 37, which depend on probability estimates, have failed. Moreover, from my point of view, which is different from theirs, it is undesirable to let preparers continue to abuse the opportunities to exercise “judgment” in FAS 5 to manage earnings. The only solution is to restrict liability recognition to legally certain obligations.
Question #3: Should the answer change if the customer who purchased the hamburger claimed it was contaminated?
My response: No, unless it is established with virtual certainty that the vendor has a legal obligation.
I admit that judgment would be required to determine whether a legal obligation exists, but it is significantly less consequential than estimating probabilities and expected outcomes of contingent liabilities.
Question #4: Reverting to the original scenario, should the answer change if the vendor warranted the safety of their hamburgers, and stood ready to pay C100,000 to anyone with a valid claim?
My response: Yes.
The vendor clearly has a present obligation to stand ready to compensate customers who have been served contaminated hamburgers. That obligation is strongly analogous to an insurer’s obligation, and should be accounted for the same way.
Question #5: Reverting to the original scenario, should the answer change if there was a “constructive obligation” to compensate customers who were served with contaminated hamburgers?
My response: No.
Basically, a constructive obligation is defined in both GAAP and IFRS as a non-legal obligation inferred from business conditions indicating that if a payment would not be made, then the impairment of future business prospects would be greater than the payment saved. While there is some justification to be made on economic grounds, the practical problems of inviting preparers to exercise “judgment” once again trump.
In summary, after years of thinking about ways to improve the accounting for contingent liabilities, I am now throwing in the towel on what I have come to see as a futile exercise. By limiting liability recognition to present contractual obligations, standard setters can kill two birds with one stone: first, financial statements will become more reliable and auditable; second, some of the difficult issues in measuring the fair value of liabilities that have not yet been satisfactorily addressed would become moot.