Are you in compliance?

Don't miss out! Sign up today for our weekly newsletters and stay abreast of important GRC-related information and news.

Taking Measure of a Multitude of Measures

Scott Taub | January 29, 2013

For as long as I can remember, accountants have complained about the complexity of accounting standards. Congress has complained about it. Investors have complained about it. Regulators of all types have complained about it. They have complained in the past, and no doubt they will complain again in the future.

The “cost vs. fair value” dichotomy is often cited as a source of complexity, but if the “mixed-attribute model” that we're all so familiar with were limited to just two measurement attributes, things would be far simpler than they actually are. Sometimes, it seems that we have almost as many measurement attributes as there are things to measure.

I'm not going to attempt to catalog all of the measurements used in accounting, but let's stroll through the Accounting Standards Codification and I'll highlight some of the many models.

Start with accounts receivable. While the initial measurement is at cost, on an ongoing basis, receivables are measured at cost less “probable and estimable” incurred losses.

Next up are loans. You might think they are treated the same as accounts receivable, at cost less probable and estimable losses, but that would be incomplete. You need to add accrued interest, which could be based on contractual provisions or, if the receivable was purchased with a credit impairment, based on expected collections. Unless, of course, the loan has been put on non-accrual status, in which case forget the interest measure. And if the loan has been through a troubled debt restructuring, we can reset the whole thing.

On to investments in securities. Here we have the classic cost (plus accrued interest) and fair-value measures, with all kinds of variations as to how impairments are calculated and recorded, and how changes in value are tracked.

Of course, if the investor owns enough of the investment, the equity method is required, which is, I guess, cost-based, but with a fair value-based impairment test and accrual of a portion of the investee's profits and losses.

Inventory is generally measured at cost, except when impaired. While you might think the measure of impaired inventory is fair value, it isn't. It's actually “market,” which is current replacement cost, as long as that amount is between net realizable value and net realizable value reduced by normal profit margin. Net realizable value, by the way, is yet another measure, defined as estimated selling price less costs of completion and disposal.

Need I go on? Fixed assets and intangible asset are generally recorded at cost and amortized. Amortization patterns and lives reflect usage of the asset, as opposed to diminution in value, unless the value has diminished too much, in which case an impairment test requires remeasurement at fair value, which is then deemed to be cost, and amortization starts all over. However, some intangibles aren't amortized at all. Instead, they are written down to fair value if fair value declines below cost.

Flipping to the other side of the balance sheet, many liabilities are recorded at the amount the company has agreed to pay, discounted if there is a significant time delay before payment. While that sounds simple, note that it is neither a historical cost measure nor a current-value measure, but instead, this measure is based on future value.

Deferred revenue is generally measured at the amount that the company was paid to take on the obligation, which is essentially a cost-based measure. Nice and simple. Unless it's a multiple-element arrangement, in which case we allocate the revenue not based on relative fair values, but based on relative estimated selling prices, another current-value variant.

If we're ever to make accounting more understandable and easier to apply, we have to cut down on the number of ways we measure financial statement elements.

Accruals for severance that require employees to provide service are measured based on an allocation of the expected future fair value of the liability, which seems ridiculous as I write it, but makes a lot more sense read in context.

Contingent liabilities are measured at the best estimate of the amount to be paid, unless the best estimate is a range, in which case the lowest amount in the range is the measurement.

And that's all before we get into income taxes, pensions, and leases, which have their own special measurement models.

Just laying out the number of different measurement attributes doesn't completely reflect just how complex measurement is, because sometimes the same measurement attribute is applied differently to different items. While Statement 157 gave us a consistent definition of fair value, the definition of cost changes from item to item.

Contract origination and transaction costs, for example, are capitalized in the measurement of the cost of some assets, such as loans, fixed assets, and long-term service contracts, but not others such as inventory, and the treatment of internal contract costs versus external origination costs is often inconsistent. Inventory cost can be calculated on a FIFO, LIFO, or specific identification basis. Further, inventory is fully-loaded to reflect direct and indirect internal overhead costs, while many other assets are limited to direct and incremental costs.

Although many liabilities are based on the expected payment amount, discount rates are all over the map. We use the risk-free rate, the credit-adjusted risk-free rate (which is most certainly an oxymoron), the incremental borrowing rate, the interest rate inherent in the contract, and the high-quality bond rate (I probably missed a couple as well), depending on what liability it is that we're measuring.

We also have assets and liabilities that change attributes during their life. We have inconsistent measurements even when we describe the measurement attribute the same.

Compared to all of this, measurement of derivatives is remarkably simple. They're at fair value … always.

While the complexity of a multitude of measurement models is obvious, we seem to be headed toward making it worse. Let me direct you to the Financial Accounting Standards Board's recent exposure draft on “loan losses.” FASB proposed initially to measure loans at fair value, but that was rejected. So now, we have a proposal that measures loans at something like historical cost minus expected future losses—a combination of a past amount and projected future amounts. And the banks, bank regulators, and many large companies are applauding FASB for inventing another new measurement basis.

The loan-loss project is just one example of an ongoing theme. FASB often proposes standards that don't create new measurement attributes, or even eliminate some. But by and large, we tell the board in comment letters that whatever issue they're working on is somewhat unique and calls for a different kind of measurement model. So they come up with one.

So part of the reason we have all these different measurement models is because we asked for them. Part of the solution, then, is that we have to stop asking.

The difficulty is that every measurement technique we've come up with over the years was developed because it seemed logical and relevant to the topic of the standard. Even with the long list of models I just laid out, there are very few that I would argue are just completely wrong for the asset or liability to which they are applied. Each decision is understandable within the context of the particular pronouncement. It's only when we step back and look at the forest that we wish we hadn't filled it with so many different kinds of trees.

The good news is that FASB has been working on its conceptual framework, and “measurement” is one of the chapters that it plans to address. The bad news is that nothing has happened on this aspect of the conceptual framework project since 2010. There is, however, a report on FASB's Website that lays out 18 measurement basis candidates. Presumably, the project would wind up endorsing a smaller number than that as possible measurement attributes for use in standards.

It seems like it has to be possible to pick just a few measurement techniques, define them, and stick to them. We've done it with fair value. Maybe we need similar guidance on a handful of others like “historical cost,” “net realizable value” and, for those who like international standards, “value in-use.” With a handful of well-defined attributes, we could then, hopefully, have the discipline to stick with them and get away from the dozens of different measures that we have now.

I hope we do, because if we're ever to make accounting more understandable and easier to apply, we have to cut down on the number of ways we measure financial statement elements. The only alternative would be to admit that this level of complexity is what we want, and the headache I get from thinking about it all tells me that can't be the right way to go.