Dr. P.V. Viswanath
The ones that get away
Accounts are increasingly more art than science
IT IS corporate earnings season once again, and investors are poring over the numbers. This week, Sprint, an American telecoms firm, announced net profits more than twice as high as a year ago. Amazon, an internet retailer, revealed net income that had fallen by 32%, due, the firm said, to tax issues. The shares of both companies rose. Should they have done?
Profits figures are meant to shed light on how a company—and its stock price—might fare in the future. But many experts worry that increasingly they don't. The scandals at Enron and WorldCom—as well as more recent accounting snafus at General Electric and a big scam at AIG—show that accounting numbers are malleable. And they are getting squishier as the use of estimates in company accounts increases.
Whether this malleability is a problem is the subject of heated debate and carries with it important consequences. Reliable numbers mean that investors can make sound decisions. Bad ones lead not just to the inefficient allocation of capital but also to a loss of confidence in the markets and, when fraud is involved, to huge shareholder losses. A study by Glass Lewis, a research outfit, found that investors lost well over $900 billion in 30 big scams between 1997 and 2004.
Ever since accounting shifted from the simple tallying of cash in and cash out to “accrual accounting”, where profits and expenses are booked when incurred, forward-looking estimates have played a critical role in measuring company profits. This role has grown as “knowledge-based” economies have begun to replace the old widget-making versions, and businesses have become more complex.
The biggest boost to estimation, however, has come from the gradual shift to “fair-value” accounting. Before, assets and liabilities were mostly carried at their historic, original cost; “fair value” is an attempt to show their current worth. Fair-value numbers are up-to-date and arguably more relevant than their static but verifiable precursors. But they also result in more volatile profits and a heavier reliance on estimates for the many items (bank loans, buildings) that may not have a ready market.
Standard-setters on both sides of the Atlantic have been urging this shift. In June America's Securities and Exchange Commission, in a long-awaited report commissioned by Congress after the Enron scandal, also endorsed fair-value accounting, which it thinks will simplify accounts and reduce firms' interest in structuring transactions to meet accounting goals.
Others are less sanguine. Even with the best will in the world, estimates can be wildly off the mark. And they are easy to manipulate. In a recent study, Daniel Bergstresser and Mihir Desai of Harvard Business School and Joshua Rauh of the University of Chicago's Graduate School of Business found ample evidence of tinkering. At delicate moments—before acquisitions and equity offerings and exercising stock options, for example—some bosses inflated the assumed rate of return on pension-fund assets, thus flattering profits.
Baruch Lev of New York University's Stern School of Business, and Siyi Li and Theodore Sougiannis, from the University of Illinois at Urbana-Champaign, harbour a deeper worry: that estimates, which are supposed to improve the relevance of financial information by giving managers a means to impart their forward-looking views (on how many customers might return their new cars, for instance), are not very useful at all. That is, they do not really help investors to predict a company's future earnings and cashflows.
The three analysed 3,500-4,500 companies a year from 1988 to 2002. They then tried to “predict” future performance (earnings or cashflow) with five models in which historic cashflows and estimates were used to different degrees. The trio found that cashflows predicted future performance robustly, but adding estimates to them was little help in predicting future performance, or in generating returns from portfolios based on these predictions. It was a “sobering result”, they concluded.
Mr Lev blames this on the difficulty of making good estimates in a fast-changing, complex world as well as on a degree of earnings manipulation. “This is not to say we should toss the accounting system, which is overwhelmingly based on estimates, into the waste bin,” he says, “but it does point to the urgent need to enhance the reliability of accounting estimates—especially given the move to fair value.”
On this point, even proponents of fair value agree. The Financial Standards Accounting Board (FASB), America's standard-setter, is planning to release guidance on how to apply fair value later this year. It has devised a “hierarchy” of items according to how hard they are to value. At the top are items that have an observable price in a deep, liquid market (eg, listed corporate debt). In the middle are items where sophisticated valuation models are based on market inputs (eg, employee stock options). At the bottom are items where valuations are based wholly on management projections (eg, Enron's most esoteric financial instruments).
It is the lower end of the hierarchy that causes concern. “At this level, there is the risk that models can be used with prejudice,” says David Bianco of UBS, a Swiss bank. Regulators, too, are worried. Research by the Federal Reserve shows that the fair value of bank loans can swing widely depending on inputs and methodology. Market values for lower-rated corporate bonds, one possible benchmark, can vary by as much as 2-5%, giving managers leeway to fiddle with numbers.
The objective is objectivity
Investors need to scrutinise the numbers harder, too. Better disclosure would help them. Mr Lev suggests that firms should say how much of key reported figures is based on facts (eg, revenues in the bag) and how much on estimates—a proposal that companies claim would tip their hand to competitors. But both FASB and its international counterpart are drafting standards requiring increased disclosure of how fair values are derived and their impact on profits and balance sheets.
Mr Lev has a more ambitious proposal. This entails forcing management to periodically “true up”—that is, disclose how their previous estimates have panned out. He argues that this would inject discipline into the system, allowing investors to penalise companies that consistently make bad estimates. Lynn Turner, a former chief accountant at the SEC, agrees. “Markets can't discipline without transparency.” Defenders of fair value say that this is confusing and unnecessary. A change in estimated value could be due to changing market conditions, not to a bad estimate. And fair-value accounting itself, some argue, is a constant “true up”. But then, there's “true” and true.