Playing the Accounting Game
The Weekend Edition of the Wall Street Journal was rich with interesting stuff today, including the left side of the editorial page (the neo-Neanderthal part) bashing the Bush Administration’s incompetence, and the right side featuring a piece by Barack Obama on mortgage lending issues, and a really depressing opinion on the situation in my hometown, Detroit. And, if you are into it, Phil Simms on how to throw a football.
At the risk of going over the top with shameless self-promotion (oh, gee, why not?), I was particularly gratified by the story in Section B, entitled “This Game Theory is a Cautionary Tale,” since it confirmed one of the problems I discussed in Models and Games. Discussing financial service companies, Donn Vickrey (who used to be an accounting prof) of Gradient Analytics says: “I think for a number of years they played games.” One of the alleged gaming tactics was “gain on sale” accounting, in which loans were packaged and sold to other investors.
The article doesn’t discuss it, but the issue here is really the anomaly created by accrual versus cash accounting. When you sell and ship a widget in March, you get to record the sales price as revenue in March, even though the other accounting entry is not to cash, but to accounts receivable. That receivable is an asset, and when the cash comes in later, you reduce the receivable and increase the cash (in journal entries this is debiting and crediting, but those terms don’t mean to accountants what they might mean to you and me). The reason for doing it this way is that you want to match the costs incurred in making that sale as closely as possible with the revenue from that sale. So you will also reflect in March the cost of the widget, which you record by reducing your inventory by the cost of one widget (on the balance sheet) and reflecting the cost of one widget as a “cost of goods sold” on your income (profit and loss) statement. Suffice it to say that if you do cash accounting, and you spend all the money to make the widgets, say, in 2003, and collect all the cash in 2006, your 2006 accounts will accurately reflect cash, but they won’t really tell you how profitable the sale was, because the costs are not matched to the revenues.
More than you ever wanted to know about the subtleties of accounting below the fold.
Here’s an example of where the discretion comes in. When you sell widgets, you know historically that you won’t collect, on average, 5% of the receivables. So with each sale, you record a bad debt expense of 5%, and establish a bad debt reserve (which is an asset) as a hedge against the time when you actually get stiffed. But 5% isn’t cast in stone, and you could make a case for raising it or lowering it. Note so far that not a single real dollar has changed hands. But, generally, this is how business and accounting works, and usually it works okay. There are lots of places for discretion in GAAP, which is why the WSJ article starts with the following statement: “Much of GAAP is so subjective that you could drive side-by-side snow plows through the gray areas.” And discretion or judgment doesn’t necessarily equate to BAD, it just means that we are dealing either in risk or uncertainty, and have to make judgments about the future, because the system demands a number NOW. Indeed, for insurance companies, as an example, much of the cost and liability side of the financial statements is the estimation of loss reserves, both for reported claims, and claims that you know will come in, but haven’t been reported yet. (This is known as IBNR – “incurred but not reported”).
But now we aren’t talking about selling widgets; we are talking about selling loans or loan participations. It’s not clear how these deals were structured, but apparently there was enough gray area in GAAP to allow recognition of the revenue on the sale of the loan or the loan participation, even though the receivables would be collected in the future. And the companies DID try to make allowances for delinquencies, pre-payments, and interest rate changes. But now it’s a hell of a lot more discretionary and judgmental than just applying a 5% bad debt reserve to the sale of widgets. Moreover, because the range of discretion is broader, it also makes “making the numbers” by accounting engineering – i.e. changing the assumptions or the reserves – easier. And when things go to hell in a handbasket – like there is an unexpected wave of delinquencies – those delinquencies have to be written off in a period AFTER the sale was recorded, and the reserves taken to hedge (on the books) just aren’t enough.
Vickrey says “Much of this. . .involves meeting ‘the bare minimum letter of GAAP, but not adhering to the spirit of GAAP.” I think there’s a significant hindsight bias in that judgment, because after all my years in public companies, I still don’t know if GAAP has a spirit. (Maybe it’s an essentially contested concept.) My guess is that this all has something to do with the issue I posed in the article: playing the accounting game aggressively to win, versus viewing GAAP as a model of the business, which is difficult because you get caught in a tautology that arises out of the fact that the goal of GAAP is to state the financial statements in accordance with GAAP. And that occurs because GAAP does something other than just measure the amount of cash coming in and the amount of cash going out. Knowing how to make the judgments correctly going forward (because any judgment you make stands a chance of being wrong) is a far more nuanced skill than second-guessing them when they go wrong.
One of the things that is a “spirit” of accounting, I suppose, is the principle of conservatism, which is to err on the side of minimizing your expectations of the good, and maximizing your expectations of the bad. It’s possible to be a crooked game player, but it would also be possible to be within the rules, but simply, in hindsight, not conservative enough.