SFAS 157: ‘Fair Value’ is Not Fair Measurement for Illiquid Assets

I am finally ready to contribute my two cents to the debate as to whether SFAS 157 is at least partly responsible for the subprime mortgage mess. I have pointed out in other contexts (here and here) the serious shortcoming of SFAS 157 as a rules-based standard, but the widespread coverage of the subprime mortgage crisis, and the allegations that accounting has caused much of it, may be the best chance I will get in a long while to explain how much better a principles-based standard would work.

The issues are well-articulated in an article that recently appeared in the New York Times (July 1, 2008), entitled “Are Bean Counters to Blame?,” by Andrew Ross Sorkin. Sorkin constructs his story around the much-publicized critiques of Stephen H. Schwarzman, co-founder of the private equity firm Blackstone Group. Basically, Schwarzman argues that SFAS 157 exacerbated the subprime mortgage crisis; because among other things, inappropriate write-downs by public companies somehow all by themselves induced companies to raise new capital at onerous rates. The example held highest is the required accounting for those collateralized debt obligations (CDOs) that were completely written off because markets in which to sell them had dried up. As quoted by Sorkin, Schwarzman says, “the concept of fair value accounting is correct and useful, but the application during periods of crisis is problematic. It’s another one of those unintended consequences of making a rule that’s supposed to be good that turns out the other way.”

The phenomena that poor quality accounting leads to poor quality decision-making by financial statement issuers has been demonstrated far too often to be plausibly denied, and problems with SFAS 157 are also self-evident. In that sense, Schwarzman’s complaint has merit; however, if he is right, it is for the wrong reasons. So, I’m going to presumptuously do him the favor of making the right argument—and at the same time show him that he is not as right as he thinks he is.

The very beginning is the right place to start: wealth is defined by economists as the ‘command over goods and services.’ A corollary principle is that the utility of an asset is derived from the amount it would cost to replace it. (For acknowledgement of this in the accounting literature, see ARB 43, on the accounting for inventories). Hence, the fundamental flaw in SFAS 157 is that its concept of ‘fair value’ is built around the notion of an ‘exit price’, but the economic definition of wealth is built around the concept of a replacement cost, or ‘entry price’.

The distinction between the ability to demand cash and command over goods and services becomes critical when an asset has value in use (or in the case of a financial asset, by holding on to it and collecting the cash flows), but little or no current value through sale. If there exists no current market in which to sell a CDO, there is no one else from whom one can currently demand cash in exchange for the asset. Is the CDO worthless? Absolutely not.

To see why the replacement cost approach is principles-based, now ask yourself what it might cost today to replace an illiquid CDO that was acquired, say, three years ago. Since it’s a financial instrument, the question could even be more general: what would it cost to replace the contracted-for stream of future cash flows from counterparties of the same creditworthiness? Asking the right question will almost certainly give you an answer that is less than the amount paid to acquire the CDO before the subprime mess hit the fan, but it is certainly greater than zero, which is the wrong answer that SFAS 157 would give.

Although this doesn’t explicitly come out in Sorkin’s article, I infer that Schwarzman would argue that balance sheet carrying amounts for CDOs would most appropriately be based on their historic costs. While I do agree with Schwarzman that zero would be wrong, historic cost-based management would be equally wrong. It is also the reality that, like SFAS 157, replacement cost-based measurement can require significant doses of judgment, but often less so. But most critically, estimating replacement cost is the only process out of which there is some chance of coming up with a logically valid result.

So, Schwarzman is right to claim that we should not be subjecting ourselves to an accounting rule that can destroy shareholder value, especially if it can be demonstrated that application of the rule will yield a ludicrous result under conditions when it is most important that it yield something better. But, he would be wrong to suggest that measuring CDOs at anything close to their historic costs would solve any problem at all. If there is any hope of gaining agreement on this point among Schwarzmans, accounting standard setters and investors, it is only through the use of clear and convincing economic logic, and not by a futile defense of the hash of rules that is SFAS 157.