Business Week picks small cap Net stocks

Business Week's Timothy Mullaney picks "second tier" Internet stocks in an article titled Finding Value on the Net.  His key argument: in the wake of eBay's earnings miss and stock fall, "retail investors can hedge bets on the Big Four [eBay, Amazon, Google and Yahoo] by buying smaller Net companies whose stocks are much more reasonably priced". Here are his picks:

1. Ctrip.com (CTRP)
2. Shanda Interactive Entertainment (SNDA)
3. Sina Corp. (SINA)
4. CheckFree Corp (CKFR)
5. aQuantive (AQNT)
6. Provide Commerce (PRVD)

His reasoning:

• All these companies are cheap when measured by P/E ratio to growth (P/E/G).
• Asian internet companies have faster growth rates than their U.S. counterparts, so their stocks are cheaper for a given P/E. CTRP, SNDA and SINA, which address the China market, are therefore attractive. (Mullaney likes Ctrip best of the three, because he thinks it could be acqired by IACI).
• CheckFree, a provider of online bill payment software to banks, trades at the same P/E/G ratio as the S&P 500.
• aQuantive is "more diversified than most of its rivals" and trades at a lower P/E/G ratio than the S&P 500.
• Provide Commerce runs Proflowers.com, which Mullaney argues is gaining market share from 1-800-flowers.com. "By shipping flowers directly from growers, it cuts out local florists and distributors. It says its flowers are cheaper, even with shipping - and up to a week fresher."

Quick comment: The growth rates used for P/E/G ratios are usually the long term growth rates (often five year growth rates) estimated by sell-side analysts. But those growth rates are notoriously unreliable for two reasons. First, sell-side analysts tend to focus on short-term earnings projections rather than long term growth rates. And second (related?), how on earth do you estimate a company's five year growth rate in a sector changing as fast as the Internet?

Thought experiment: guess the estimated five year growth rates in   1999 for At Home (RIP), Webvan (RIP) and eBay.

Stock conclusion: Using P/E/G ratios as a screen is interesting and useful, but ultimately insufficient.