The only thing I can imagine the SEC is investigating is the interest rate assumptions employed by Delphi. The firm says that its long-term rate of return on plan assets was 10% in 2002 and 9% in 2003. Clearly, both amounts are aggressive and facilitate suspicions of padding the income statement. But, when all is said and done, the fault appears to rest with FASB's Statement No. 87. While an improvement over its predecessor, Statement No. 87 falls short in many ways.

On the income statement the FASB declares that the pension cost is the service cost plus interest minus an expected return on plan assets plus a number of other arbitrary amortizations of transition issues and other matters. The expected return on plan assets ought to be replaced with the actual returns. The actual returns represent economic reality, not the expected returns that are merely a function of somebody's dreams.

The arbitrary amortization rules and the use of the corridor rule with respect to pension gains and losses are nonsense. When a business enterprise enjoys gains or suffers losses, it should place the whole amount into income that very year. Delaying these gains and losses not only does not report economic reality, but also deceptively reduces the reported volatility of earnings.

The balance sheet compounds the problem by netting pension assets against pension liabilities and by ignoring the effects of actuarial gains and losses that fall within the corridor and unrecognized prior service costs. Statement No. 87 even creates an intangible pension asset that belongs in fantasyland.

The one redeeming feature of Statement No. 87 is that it requires business enterprises to provide a wealth of disclosures so that the savvy analyst can obtain a better picture after making certain adjustments.

Let's consider Delphi's annual 2003 financial report. The corporation reports net income of $342 (all numbers are millions of dollars) in 2002 and $(56) in 2003. If we replace the expected returns on the plan assets ($692 and $647 respectively) with the actual returns ($(580) and $1,176), then we discover that the adjusted incomes are $(930) and $473 in 2002 and 2003. So the analyst discovers that the company's profit figures are grossly inaccurate; moreover, they overstate the two-year earnings by $743! That's a colossal error!

On the 2003 balance sheet Delphi displays a net pension liability of $693. But, we should recognize the unrecognized prior service cost liability of $1,142. We should eliminate the pension intangible asset of $1,142. And we should unnet the amounts by showing the fair value of the plan assets $7,437 on the left-hand side of the balance sheet as well as the projected benefit obligation $11,413 on the right-hand side of the balance sheet. These adjustments increase assets by $6,295, increase liabilities by $11,812, and decrease shareholders' equity by $5,567. The latter effect is interesting because it changes the reported shareholders' equity from $1,570 to $(3,997). In other words, the firm is technically insolvent.

While the disclosures in the pension footnote prove very useful in analyzing the economics of the entity, the FASB could improve financial reporting by moving these items from the footnote to the financial statements themselves. If the SEC really wants to improve pension accounting, it should put pressure on the FASB to amend the silly rules it currently has on the books.

J. EDWARD KETZ is accounting professor at The Pennsylvania State University. Dr. Ketz's teaching and research interests focus on financial accounting, accounting information systems, and accounting ethics. He is the author of Hidden Financial Risk, which explores the causes of recent accounting scandals, and columnist of The Accounting Cycle for SmartPros.com.