THE DAY AHEAD: Why bother with long-term earnings projections?

COMMENTARY--For TranSwitch, it took only a few weeks for $13 million in revenue to evaporate. For Vitesse Semiconductor, it only took a few weeks to lose more than $30 million in revenue. Palm said its current quarter sales will be $250 million short of expectations.

It's like a twisted Wall Street magic act--the revenue just went "poof."

And TranSwitch and Vitesse, two companies that make communications chips, aren't alone. The first quarter is history and plenty of tech companies--Viant, Compaq, Palm, Nortel and a host of others--are finding themselves with ugly financials.

But that's nothing new. What's notable is how Wall Street analysts are making 2002 projections when a company can't accurately predict results two weeks from now.

Where are Wall Street's sages getting this information? They're making it up. They have to be because the companies aren't saying squat.

To wit, Vitesse last week cut its first-quarter revenue target from between $150 million and $160 million to between $120 million and $125 million. The earnings target was halved to 10 cents a share to 11 cents a share.

"Near-term visibility at a majority of our customers has worsened," said Vitesse CEO Lou Tomasetta. He isn't kidding. Vitesse cut its revenue target from $180 million to $190 million on March 5.

It's the same deal with TranSwitch--three weeks and two profit warnings. The floor is falling out from under these companies and it wouldn't be prudent to give long-term projections.

Palm and Nortel hit the market with a double whammy of profit warnings. Neither company even bothered issuing an outlook for their upcoming fiscal years.

Despite this lack of visibility, analysts are giving long-term estimates the old college try.

UBS Warburg analyst David Wong recently cut his 2001 estimates on TranSwitch, Vitesse and their peers. For Vitesse's fiscal year ending in September, Wong is now expecting Vitesse to report earnings of 54 cents a share on sales of $505 million. That's a pretty steep cut from his previous estimates.

But for fiscal 2002, Wong's changes are a little less dramatic. He now sees Vitesse making 84 cents a share in 2002 on sales of $645 million. That's down from $1.14 cents a share in earnings on $785 million. He even has a guess for 2003, when Vitesse is expected to report earnings of 30 cents a share.

It's a similar drill with TranSwitch. Wong cut 2001 estimates, but still sees the company reporting 2002 sales of $224 million, down from a previous estimate of $270 million. Can you really trust a 2002 sales projection for a company that has warned twice in one quarter?

Palm estimates were cut across that board and even though the company can't predict fiscal 2002 results, analysts sure can. Nineteen analysts are predicting a profit of 3 cents a share on revenue of $2.96 billion in fiscal 2002.

Nortel, which is becoming a serial profit warner, is supposed to post a 25-cent-a-share profit on sales of $34 billion for fiscal 2001, according to First Call. There's one hang-up--Nortel isn't talking about any results beyond the current quarter.

These analysts will clearly have to cut their estimates when companies are actually brave enough to predict future results. Why not project a quarter or two and leave it at that?

Here's why: most of these stocks are valued on results that haven't been reported yet--guesses, if you will. That's why they get whacked when the fancy financial models change.

Bottom line: Wall Street's financial guesses for upcoming fiscal years are for entertainment purposes only. Why bother?TDAIN

• Day Ahead archive
•  Get The Day Ahead>