Reform never works quite the way it was intended. The short-term result of almost any change is chaos. And in the case of efforts to reform the way companies report quarterly earnings, chaos has certainly been the short-term result.
But the long-term picture isn't much more comforting. The reform efforts have cast doubt on the historical relationships between earnings and stock prices. And that has seriously weakened earnings per share as a tool for valuing stocks.
A Sea of Earnings Confusion
This quarter, reforms intended to make companies use stricter rules in reporting earnings have, in the short run, made reported earnings even harder for investors to understand. Caught midstream while the new rules are being put in place, Wall Street analysts are producing earnings projections without a consensus about which earnings numbers they should be watching.
As a result, in this quarter investors are looking at one of the most confusing and least useful sets of quarterly earnings reports delivered since the
Securities and Exchange Commission
But the problems don't end there. Even if Wall Street and investors make a successful shift from pro forma to generally accepted accounting principle numbers, the change itself will make the historical earnings record difficult to use to value stocks. What's the point of knowing that
has traded at a price-to-earnings multiple between 40 and 14 over the last 10 years, when the way earnings per share will be calculated in the future has an undefined relationship to the way earnings were calculated in the past?
Let me start by explaining the short-term chaos and gradually work toward the long-term problems.
For the short-term end of the problem, take the case of
. The deeply troubled telecommunications-equipment maker delivered some good news when it reported earnings for the first quarter of 2003 on April 24.
Nortel said it earned a penny a share, the first time in 13 quarters Nortel had reported a profit. And these results, the company proudly said, were based on the stricter GAAP rules rather than on the loose pro forma standards that give companies more wiggle room in calculating their bottom line.
That's great news, no? Especially when the 24-analyst consensus reported by Zacks Investment Research projected that the company would lose 3 cents a share this quarter. Nortel had reached profitability a quarter ahead of its own predictions.
Nortel's report sent Wall Street analysts running to their calculators -- not to see if the stock of the now-profitable company was a buy, but to see how the GAAP standards the company had used to calculate its earnings actually compared to the pro forma standards Wall Street had used to prepare its estimates.
No Agreement on Methodology
Reuters Research, the parent of earnings and estimate compiler Multex, concluded that Nortel's real comparable EPS numbers came to a loss of a penny a share after the number-crunchers shuffled some gains and losses around. This differed from Nortel's own report and the 3-cents-a-share loss under the pro forma accounting standards that Wall Street accountants used to come up with their original projections.
Confused? Welcome to the club. It's always been tough to know if Wall Street analysts and company accountants used the same rules in calculating earnings numbers. And it has always been wise to take the numbers apart before accepting the headline judgment that
beat estimates by 2 cents a share (apples to apples, it looks more like a penny surprise) or that
beat by 3 cents (the company, counting differently than the analysts, says 2 cents).
But the confusion has hit new heights this quarter. There seems to be very little agreement between companies and analysts, or even among members of those groups, on how earnings should be calculated.
Ironically, part of the blame falls on efforts to tighten accounting standards that are now being put in place. According to a new SEC rule called Regulation G that took effect last month, companies are supposed to make sure they calculate earnings using GAAP rules and give those figures top billing in their earnings reports. So far this earnings season, about 70% of the companies reporting have used GAAP earnings, according to Thomson Financial/First Call.
The problem is that Wall Street analysts have lagged behind corporate CFOs in converting their earnings estimates to GAAP. Analyst estimates for the first quarter now being reported include figures based on GAAP earnings, operating earnings, traditional pro forma measures and who knows what else. Despite being calculated using incompatible methods, those figures are simply averaged into the consensus Wall Street expectation.
The difference is significant. If you use 2002 GAAP earnings to calculate the market's price-to-earnings ratio, the
trades at 32 times earnings. According to Thomson Financial/First Call, if you use operating earnings, a loose figure that includes many different kinds of earnings methods, the price-to-earnings ratio drops to 19.
The Qualcomm Quandary
Here's how the confusion and the price-to-earnings difference plays out at a single company:
. According to Zacks Investment Research, Qualcomm beat the Wall Street consensus estimate of 35 cents a share for the March quarter when it reported earnings of 38 cents on April 23. That's a 9% positive surprise.
Of course, that consensus was based on pro forma earnings. Qualcomm led its earnings press release with its GAAP earnings of 13 cents a share.
For fiscal 2003, which ends in September, UBS Warburg projects Qualcomm will earn $1.31 a share pro forma, or 95 cents GAAP. So right now Qualcomm trades at either 24 times projected fiscal 2003 earnings per share or 33 times projected GAAP earnings per share. That's a 38% spread between the high and low multiples.
It's tempting to say, "Go with the GAAP number and the GAAP multiple because it's more accurate," but that runs into immediate problems. Investors have no idea what multiple Qualcomm's GAAP earnings have commanded in the past because the stock traded on pro forma earnings and multiples of those numbers.
On the other hand, it seems dangerous to go with pro forma numbers. Perhaps all of Qualcomm's past price-to-earnings multiples were inflated, because the pro forma numbers made the company look more profitable than it would have been by GAAP standards. In the aggregate, say some market historians, that kind of accounting inflation played a major role in creating the stock market bubble that burst in 2000.
That leaves investors in a quandary. On one hand, these numbers continue to be the measures of short-term success or failure most commonly reported and closely followed by analysts and investors. In the short term, the earnings game continues as usual: As long as the stock beats or misses some Street consensus, it will soar or dip. As flawed as these numbers are, they continue to have the power to move stocks in the short term.
Yet, we're in a period in which disagreements about how earnings should be measured are extreme. And that disagreement does make it difficult for investors to use earnings numbers to judge the trend in a company's business. That's because they first have to tear apart the reported numbers, the analyst consensus and historical reported earnings to understand the methods used to arrive at those numbers and to recalculate the results into a seamless record that can be used to judge and value stocks. I doubt most individual investors can or will go through that exercise.
The Search for 'Core Earnings'
Some Wall Street analysts have responded to this challenge by searching for better methods for calculating earnings and earnings trends. One of the most promising of these is something called "core earnings." In applying this concept, an analyst or investor tries to separate the earnings growth that's a result of core activities of the company from one-time gains from extraordinary events or gains or losses from businesses that seem peripheral to the core.
For example, looking at student-loan provider
, Smith Barney uses a definition of earnings that seeks to exclude gains from turning loans into securities and then selling them, from any trading income and from any sales of loans, and equally to exclude any charges from acquisitions. The result, Smith Barney's analyst concludes, is a number that's the best measure of the company's "economic earnings power."
Core earnings analysis is a very useful approach, especially at a time when so many manufacturing and retailing companies are announcing earnings that are pumped up by gains from financing activities that are unrelated to the company's core business. But the method is irretrievably subjective -- one analyst's core is another's periphery -- and so labor-intensive that it's not a viable replacement as a valuation tool for most investors who have relied on EPS and historical multiples.
In my opinion, finding methods that will replace those tools for most investors will require leaving earnings numbers and price-to-earnings ratios completely behind, and looking at measures such as price-to-sales and growth in cash flow per share, where the record of historical valuation hasn't been disrupted by radical changes in how the numbers have been calculated.
In the next month I'll be looking at some work on relative price-to-sales ratios that might fit the bill for more value-oriented investors. And I'll try to take some of the mystery out of valuing a stock on cash flow for growth investors who are looking for an alternative -- or perhaps just a backup -- to traditional earnings-based valuation tools.
Jim Jubak appears most Wednesdays on CNBC's "Business Center" at approximately 5:45 p.m. ET. At the time of publication, Jim Jubak owned or controlled shares in Microsoft. He does not own short positions in any stock mentioned in this column.