THE DAY AHEAD: The folly of forecasting

Larry Dignan
3 min read

COMMENTARY -- In recent days, you've inevitably heard of some outlook for the tech sector in 2001. Yahoo, Intel and others project a gloomy 2001, industry wonks reverse their forecasts, and cyclical sectors such as semiconductors have gone from boom to bust to boom again.

The consensus view these days is that the tech sector will rebound in the second half of 2001. But a lot of things have to go right (interest rate cuts, stronger demand, the right psychology) before a rebound. Meanwhile, executives and industry forecasters literally can't gauge demand six months in the future. That's why those earnings conference calls are so murky -- people are guessing.

It's not hard to find examples. Last week, IBM CFO John Joyce admitted his crystal ball wasn't better than anyone else's. Nevertheless, IBM is comfortable with 2001 estimates based on what it knows today. Microsoft (Nasdaq: MSFT) sees high single-digit PC growth, but can't truly predict the second half. The key word is visibility, and few companies or research firms have it.

Simply put, most tech forecasts should be used for entertainment purposes only. This isn't good news for firms that peddle their research, but it'll keep you sane. Market forecasts? PC demand forecasts? Online advertising predictions? The outlook for semiconductor revenue? Take all of them with a grain of salt.

"The forecasting of future growth leaves much to be desired," said Carl Johnson, president of INFRASTRUCTURE, a semiconductor research firm. Johnson should know -- he's been watching chip revenue forecasts for years. He even cooked up a neat slide that summarizes things.

In Johnson's industry, the forecasts are especially unreliable. "The semiconductor industry has two speeds -- pedal to the metal and complete halt," said Johnson.

In a recent research note, J.P. Morgan chip analyst Terry Ragsdale noted the wild swings for the chip revenue forecasts for 2001. "The market research community is falling all over itself lowering semiconductor forecasts for 2001," he said.

IC Insights, arguably the most accurate of the bunch, predicts 7 percent revenue growth for semiconductors in 2001 with downside risk. Dataquest recently cut its 2001 revenue forecast from 27.5 percent in October to 5 percent to 10 percent. VLSI Research reckons that the 2000 chip sales tally was really two years in one, which means 2001 is going to be ugly.

"Since forecasting overall semiconductor industry revenue growth is so treacherous and has so many moving parts, we generally confine ourselves to throwing pot-shots at others' forecasts," said Ragsdale. That's not a bad plan.

And chip forecasts are just the most obvious targets. Networking, online advertising and B2B forecasts are likely to be just as suspect.

PC demand forecasts are also on the ropes. Last month, International Data Corp. revised its forecast to call for total PC shipment growth to slip to 16.6 percent worldwide for 2001. Earlier in the year, that prediction was roughly 20 percent. Chuck some darts to guess what the real total will be.

The problem with being a sage is forecasting consumer demand, which often drives the tech sector. Talk of a slowing economy has prompted Joe PC Buyer to do an about-face. A few months ago, consumers were buying $2,000 PCs. Now, they're talking inexpensive gadgets -- maybe.

Analysts said no one expected the economy to turn this quickly. Everyone from Cisco CEO John Chambers to Hewlett-Packard CEO Carly Fiorina has used the light switch analogy. Companies such as Motorola have been dead wrong with their projections.

Put it together and these poor forecasters never had a chance -- being a sage is a thankless job fraught with peril. "You really can't go out much more than six months," said Johnson. "It's not going to work."TDAIN

• Day Ahead archive
• Get The Day Ahead>