In my column two weeks ago, I argued that Cisco (Nasdaq: CSCO) and other richly valued large-cap technology stocks were low-probability investments, but suggested that beaten-down stocks such as Apple (Nasdaq: AAPL) might be worth a look. (Now you see why I don't try to make a living predicting short-term stock movements.) In last week's column, I began my analysis of Cisco and Apple with a quick overview of the two companies. Today, I'd like to continue with an analysis of Cisco, and cover Apple next week.

Please note that my arguments against buying Cisco apply to pretty much any stock with an extreme valuation -- let's define that as 100x trailing earnings per share (EPS) or more. This still leaves quite a large universe of stocks -- mostly technology stocks -- that I think are very unlikely to provide a satisfactory long-term return. I've chosen Cisco to make my point because it's so widely known and owned, not because I think Cisco is particularly ripe for a fall. In fact, the opposite is the case: I used to own Cisco, think highly of the company, and, were I to buy a richly valued tech stock, Cisco's would be among the first I'd snap up. Thus, when I conclude that Cisco's stock is a bad bet, you can safely assume that I feel similarly, if not more strongly, about dozens of other companies that have extraordinary valuations.

To think sensibly about the returns from Cisco's stock over, say, the next five years, it helps to develop some scenarios for what the future might look like. This will help us think about the range of possible outcomes. Then, we can assign probabilities to each scenario and calculate an expected return. Keep in mind that this exercise is more art than science -- the future is inherently unpredictable -- so I encourage you to develop your own scenarios and probabilities if you disagree with mine.

There's little doubt that Cisco is priced for perfection. To its credit, the company has consistently achieved perfection for many, many years, so let's start with this scenario: Cisco continues to grow like gangbusters and achieves the very high EPS growth that analysts are projecting over the next five years. For the next 12 months ending July 2001, Cisco is projected to earn $0.74 per share, 40% higher than the last 12 months (pro forma). The next year, analysts project $0.96 per share, a 30% increase. Let's be aggressive and assume 30% growth for the three years after that, which translates into EPS of $2.11 in five years. This means net income would be $20 billion (if we assume that shares outstanding grow at 5% annually, to 9.5 billion shares). For perspective, $20 billion is 64% more than the net income earned by the most profitable U.S. company today, General Electric (NYSE: GE), which has trailing 12-month earnings of $12.2 billion.

This supercharged EPS growth likely implies even higher revenue growth, given Cisco's declining margins and increasing share count. Over the past eight years, Cisco's gross margin has fallen every year except one, from 67.6% to 64.4%. Of greater concern, operating margin has fallen every year over the same period, from 40.6% to 26.5% (pro forma, which excludes intangibles and in-process R&D). That's a big drop.

Keep in mind that Cisco's margins have become increasingly inflated due to the enormous option grants the company makes to compensate employees, the cost of which doesn't appear on the income statement. If it were, according to estimates in Cisco's 10-K, net income last year would have been 41.9% lower than reported. Two years ago, net income would have been 26.5% lower; three years ago, 19.3% lower. The problem is rapidly getting worse. This reminds me of three questions Warren Buffett asked in his 1998 annual letter to shareholders: "If options aren't a form of compensation, what are they? If compensation isn't an expense, what is it? And, if expenses shouldn't go into the calculation of earnings, where in the world should they go?"

The impact of all these options may not be immediately obvious, but they cost shareholders dearly in the form of dilution -- and will make it even tougher for Cisco to meet analysts' EPS projections. Cisco's diluted shares outstanding have increased an average of 3.7% annually since 1991, with increasing dilution in recent years (6.1% from 1998 to 1999 and 5.3% from 1999 to 2000). (Note that the increasing share count is also due to dozens of acquisitions, a strategy that has been very successful.)

My goal in pointing out the declining margins and increasing share count is not to bash Cisco -- after all, over this period, Cisco has been one of the greatest stocks of all time. Rather, they simply must be factored in to our future scenarios by assuming that Cisco's EPS growth will be at least five percentage points lower than its revenue growth. Thus, to meet the EPS targets noted above, Cisco's sales would have to grow 45% in the next year and 35% annually for the four years after that, resulting in revenue jumping almost fivefold, from $18.9 billion to $91.0 billion. Only four U.S. companies -- Exxon Mobil (NYSE: XOM), Wal-Mart (NYSE: WMT), General Motors (NYSE: GM), and Ford (NYSE: F) -- have sales greater than this. (Cisco is currently ranked No. 84.)

Of course, Cisco's growth will eventually slow and its P/E ratio will decline to reflect this, so let's assume that Cisco's P/E ratio, if all goes well, will be 75x trailing EPS in five years. This is an aggressive estimate based on Cisco today trading at 77x next year's earnings and GE -- surely one of the best-run, most-profitable businesses in existence -- trading now at "only" 42x trailing EPS.

So, in the most optimistic scenario imaginable, Cisco stock will be at $158.25 ($2.11/share x a P/E of 75) in five years, representing 22.5% annual growth. At this price, assuming diluted shares outstanding grow at 5% annually, Cisco's market cap will be $1.5 trillion. To give you a sense of how big that number is, the entire U.S. Gross Domestic Product last year was $9.3 trillion.

As improbable as all this may sound, this is the perfection scenario, so let's go with it. Thus, the beginning of our scenario chart looks like this (CAGR = compound annual growth rate):

Scenario   5-Year CAGR 
Perfection     23%

What about other scenarios? Based on the experiences of Intel (Nasdaq: INTC) and Home Depot (NYSE: HD), among many others, we can guess that if Cisco so much as stubbed its toe, its stock would fall at least 25% instantly. And heaven forbid it should run into more serious difficulties along the lines of Microsoft (Nasdaq: MSFT) or Lucent (NYSE: LU). Without going into the details of each additional scenario, here is my optimistic assessment of Cisco's future scenarios (as I noted earlier, I encourage you to come up with your own estimates):

Scenario    CAGR   Assumptions      Price in 5 Years
Perfection   23%  $2.11 EPS; 75 P/E     $158.25
As expected   9%  $2.11 EPS; 42 P/E      $88.62
Stumbles      0%  $1.64 EPS; 35 P/E      $57.40
Big trouble -10%  $1.13 EPS; 30 P/E      $33.90

Now, let's assign probabilities to each scenario. How likely is Cisco to achieve perfection? The company has certainly earned the benefit of the doubt, but it's already quite large, technologies are moving rapidly, and there are countless established and emerging competitors. I'm willing to be wildly optimistic and give Cisco a 40% chance of achieving perfection. Here's the rest of my chart, again being very optimistic:

Scenario    CAGR  Price in 5 Years  Probability
Perfection   23%    $158.25            40%
As expected   9%     $88.62            20%
Stumbles      0%     $57.40            20%
Big trouble -10%     $33.90            20%

Multiply all this out and the result is Cisco at $99.28 in five years, which represents growth of 11.6% annually. I don't know about you, but I can think of many stocks that I believe will compound at a higher rate -- using conservative assumptions, rather than highly optimistic ones.

It's easy to identify great companies with fabulous economic characteristics, strong management teams, and bright future prospects. So many investors end their analysis there and start buying, forgetting the final question: At what price? If the price you pay fully discounts even the most optimistic scenario, then you are virtually certain to do poorly. And a long investment horizon won't help -- in fact, it will work against you. In the short term, momentum might carry even an overvalued stock still higher, so you can make money if you sell quickly, but given enough time the laws of economic gravity will always prevail.

Though the bull market of the past few years has persuaded many to the contrary, I believe that the only way to consistently make money in the stock market will be the same in the future as it's always been over long periods of time in the past: Buy stocks that are undervalued. In other words, stocks that are being misunderstood and mispriced by the market. Thus, the most important question to ask yourself when you're considering buying a stock is: "What is the market underestimating about this company that is causing its stock to be significantly undervalued?" If you can't come up with a good answer, don't buy the stock.

Getting back to Cisco, many people emailed me after my last two columns to detail the company's many strengths and future opportunities, but my response remains: "What part of your argument is not widely known, and can you explain to me how, at today's valuation, the company is underappreciated?" Too often, the answer is, "Well, the stock's gone up for many years and I've made a lot of money on it, so I'm betting that it will keep going." This is the fallacy that Warren Buffett wrote about in his brilliant article last year in Fortune:

"As is so typical, investors projected out into the future what they were seeing. That's their unshakable habit: looking into the rearview mirror instead of through the windshield."

I agree that anyone looking into the rearview mirror would want to own Cisco today, but I also argue that one would reach the opposite conclusion after a careful, unemotional look through the windshield. For Cisco to be a good investment going forward, absolutely everything must go right. It's possible, but the probabilities are unfavorable. In short, it's a bad bet.

A number of people emailed me to point out that I could have made similar arguments as recently as two years ago, and the stock has nearly quadrupled since then, but Cisco was a much better bet at that time. Its trailing sales were $8.5 billion, not $18.9 billion, making high rates of future growth more likely. It also had higher margins, fewer competitors, and, most importantly, a trailing P/E ratio of 48 versus 108 today (based on pro forma EPS).

While the most widely followed, universally loved, highly valued stocks sometimes prove to be undervalued, I believe -- and there is abundant evidence to back me up -- that such stocks in general do not represent high-probability investments.

-- Whitney Tilson

Whitney Tilson is Managing Partner of Tilson Capital Partners, LLC, a New York City-based money management firm. Mr. Tilson appreciates your feedback at To read his previous guest columns in the Boring Port and other writings, click here.