Required reading for all students of investing: Warren Buffett's annual letter to shareholders. The 2007 version is as witty and entertaining as always. And Mr. Buffett often puts in his two cents on an accounting issue, too - ensuring that investors will take note of it in the year to come. This year's accounting topic: pension assumptions, an issue we've monitored for years at The Analyst's Accounting Observer.

Buffett's complaint is that the long-term pension plan earnings rate assumption that corporations use in calculating their expected earnings on pension assets is too optimistic: according to him, in 2006 the relevant S&P 500 companies averaged 8%. Our own work (Volume 16, Nos. 9 & 10, S&P 500 Benefit Plans: Will Pension Panic Resurface?) showed the median return for those companies to be even higher at 8.25% - way down from where it was just five years before at 9.0% and still with some outlandish outliers, but down nevertheless. As Mr. Buffett points out, though, the returns on the different asset classes in the plans don't make sense: 

    "... The average holdings of bonds and cash for all pension funds is about 28%, and on these assets returns can be expected to be no more than 5%. Higher yields, of course, are obtainable but they carry with them a risk of commensurate (or greater) loss.

    This means that the remaining 72% of assets – which are mostly in equities, either held directly or through vehicles such as hedge funds or private-equity investments – must earn 9.2% in order for the fund overall to achieve the postulated 8%. And that return must be delivered after all fees, which are now far higher than they have ever been.

    How  realistic is this expectation?  Let’s revisit some data I mentioned two years ago: During the 20th Century, the Dow advanced from 66 to 11,497. This gain, though it appears huge, shrinks to 5.3% when compounded annually.  An investor who owned the Dow throughout the century would also have received generous dividends  for much of the period, but only about 2% or so in the final years. It was a  wonderful century."

As usual, the man makes sense - including his larger point about the reason for the bias in the assumptions:

"What is no puzzle, however, is why CEOs opt for a high investment assumption: It lets them report higher earnings.  And if they are wrong, as I believe they are, the chickens won’t come home to roost until long after they retire."

Does anyone know what the right rate should be? Well, no - you can argue, as Mr. Buffett does convincingly, that rates are too high. But anyone who really KNOWS the real rate to come over the long term would be, more or less, a supernatural being. I don't know any in Corporate America or Wall Street. (Though I've met a number of pretenders.)

Figuring the "right rate" is not the real problem. The real problem is that there shouldn't be any expected rate of return for managers to justify in the first place. The accounting methods in place were developed back in the mid-1980's; while an improvement over the (lack of) accounting for pensions at the time, the benefit reporting standard (SFAS 87) was built with many complicated accounting devices in place to keep pension costs smooth. (We've detailed them over the years; check your back issues.) One of those permitted devices: an implicit assumption that the long-term earnings of the benefit plan would average out to some specific rate and letting managers build those assumed earnings into their net benefit cost - a "rebate," if you will, that allowed managers to report lower pension costs. It also allowed them to manipulate that rate to come up with a net benefit cost that they liked.

That "expected earnings" never produced cash for shareholders. In fact, the real returns belong to the pensioners, not the shareholders. Pension contributions are pretty much a one-way street: the company making them can't pull them back unilaterally, at will. (There are instances when they can get back some overfunded balances, but it's not simple or easy. Or common.) If the real returns belong to the pension players, why should even the real returns be reported in the income statement as measure of corporate performance? The expected returns were built into the accounting because the real returns would be quite volatile, something that companies couldn't stomach in the '80's (and wouldn't likely want to stomach in this century either.)

The real returns really do matter to the corporate sponsor, because they'll be on the hook for contributions if the plan assets fail to fund the promises made - so they belong in the investor reporting package. They just don't belong in the income statement - where the efforts of portfolio managers get mixed in with the operating performance of the company's managers.

Instead of cluttering the income statement with the "expected returns" of asset managers for assets that the firm doesn't necessarily control, the REAL performance of the plans might be better reported in accumulated other income - or some other format than what investors receive right now. The FASB is entering Phase 2 of its benefit reporting plan revamp; with the IASB, it's also tackling the revision of the whole financial reporting package. The "expected return" component of benefit costs might be gone as a result of these projects.

Good riddance, I'd say. How about you? I'm curious to know what you think and I've arranged a quick 3-question survey at this link, courtesy of SurveyMonkey . Let me know! And if there's anything you'd care to add, drop me a line.