Telecom giant SBC Communications Inc. (SBC) last year reported a stunning $2.2 billion plunge in operating income, and the company's explanation probably seemed arcane to many investors. SBC's number-crunchers, the company said, had lowered the expected return on retirement funds by a mere percentage point. Likewise, they raised their assumption of health-care cost inflation by a single point. And they shaved their estimate of future interest rates by just 0.75 percentage points. Add it all up and it translated into a big earnings hit: $693 million sliced off the bottom line.
At the same time, many other U.S. companies were making similar adjustments. SBC was among hundreds of companies that last year updated key assumptions in their accounting for post-retirement benefits to bring estimates more into line with real-world trends. "I'm very confident that our assumptions are appropriate," says John Stephens, SBC's controller.
Such changes, and the impact they can have on profits, have caught the eye of the Securities & Exchange Commission. BusinessWeek has learned that the agency's Enforcement Division is conducting an investigation: It has identified a few dozen companies whose assumptions seem aggressive and whose pension plans are big enough that changes in the estimates could have a significant impact on the bottom line. In early October it sent letters to half a dozen of them, requesting documents, including e-mails between those who set assumptions, explaining how they arrived at their estimates. SEC Enforcement Director Stephen M. Cutler says his department is "looking to see whether managers are adjusting their assumptions with an eye to enhancing the earnings and balance-sheet numbers investors care about. That could be fraud."
The SEC, which declined to name the companies, says it has no evidence yet of wrongdoing. Pension accounting is one of many areas where executives have to make estimates (BW -- Oct. 4). Differences in the investment mix and the age of workers and retirees could explain why some assumptions vary widely from the average.
Another SEC unit, the Corporation Finance Division, has been making general warnings for some time, telling companies not to use overly rosy pension assumptions. Such estimates matter because in pension accounting, companies report their projected earnings for the plan, not actual returns. If the assumed return on pension investments exceeds the pension cost, companies can count the surplus in their earnings even when real returns stink.
Boeing Co. (BA), for instance, lost $3.3 billion on pension investments in 2002, but reported a $404 million pension gain based on its assumed 9% rate of return. That was 82% of the aerospace company's net income for the year. (Boeing notes that when assumed returns fall below actual costs, the company must report a loss in earnings, as it will have to do for 2004.) That's not an isolated case. UBS Investment Research figures about $2 of the $55 earnings per share for companies in the S&P 500 index in 2003 came from aggressive pension return assumptions.
Small shifts in projected returns can make a big difference. A study published by the National Bureau of Economic Research in May by accounting professors Daniel B. Bergstresser and Mihir A. Desai of Harvard Business School and Joshua D. Rauh of the University of Chicago Graduate School of Business found that nearly 5% of IBM's (IBM) pretax income in 2000 and 2001 resulted from an increase in the assumed rate of return from 9.5% to 10% in 2000. IBM adjusted the rate, which is meant to be a long-term projection, four times between 1991-2002. IBM says the company's decision to raise the assumed rate of return in 2000 was consistent with actual returns ranging from 14% to 21% over five previous years.
The study also offers support for speculation that some executives may deliberately monkey with pension assumptions for corporate and even personal gain. The trio analyzed accounting for 3,247 pension plans from 1991-2002 and found that companies tended to hike pension return estimates the year before making an acquisition or before a CEO exercises stock options. "If you want an easy way to manufacture earnings, this is as good as it gets," says Desai.
In December, 2002, the SEC warned companies that it might challenge rate-of-return assumptions above 9%, based on historical returns in equity and bond markets. Average return assumptions fell to around 8.38% in 2003, from 8.82% the year before. Still, 26% of the 355 companies in the S&P 500 that offer defined benefit pension plans estimated returns of 9% or higher in 2003, according to a UBS Investment Research (UBS) study. Among those using assumptions above 9%: Darden Restaurants (DRI), owner of the Red Lobster, Olive Garden, and other eateries. William R. White, vice-president and treasurer of Orlando-based Darden, says the company's 10.4% assumed return was in line with actual returns of 10.5% over the 10 years ending May, 2004. Still, Darden has lowered the rate to 9%.
Many experts think even 9% is tough to justify. Given long-term stock and bond returns, an 8.5% rate is more reasonable for a typical portfolio with 65% equities and 35% bonds, CSFB analyst David Zion argues. Higher-than-average return assumptions don't necessarily signal accounting mischief. Plans tilting more toward equity investments tend to achieve higher returns. And multinationals with big overseas workforces may invest pension money in markets with higher returns and interest rates than in the U.S.
A BLIND EYE
Companies have less discretion over another factor, the discount rate, used to calculate future pension obligations in today's dollars. The higher the rate, the lower the pension expense. The discount rate is supposed to hew closely to high-quality corporate bond yields. The average discount rate for S&P 500 companies was 6.25% in 2003, down from 6.69% in 2002, according to UBS Research. A company with an older workforce and many retirees will tend to use a lower rate than one with younger workers and fewer retirees because it must pay benefits sooner. A few of the biggest companies, however, use rates of 7% or higher.
Accounting for retiree health-care plans is just as byzantine. Some 328 of S&P 500 companies pay health-care costs for retirees. Accounting rules require companies to estimate future health-care inflation. Lowballing the trend in health-care costs minimizes the long-term benefit liability and understates expenses.
Health-care costs have been growing at a double-digit clip, hitting 14.7% in 2003, according to benefits consultant Hewitt Associates Inc. (HEW). But some companies have stuck to single-digit assumptions. Experts say such estimates are overly optimistic and suspect that auditors too often turn a blind eye. "Companies aren't willing to change these projections substantially because they don't want to create any earnings volatility and auditors aren't challenging them," says Michael Lofing, senior research analyst at Glass, Lewis & Co., a proxy advice and forensic accounting firm based in San Francisco.
Accounting experts have been lobbying for tighter rules to limit the discretion companies have over pension assumptions. The Financial Accounting Standards Board, the Norwalk (Conn.)-based private group that the SEC relies on to set accounting rules, agrees. But it could be years before the FASB takes final action. For now, the SEC's sweep serves as a warning to executives not to take advantage of the wiggle room.
By Amy Borrus and Paula Dwyer in Washington, with Michael Arndt in Chicago and David Welch in Detroit