Amazon and the Ethics of Belief
"If you can look into the seeds of time and say which grain will grow and which will not, speak then to me..." - Macbeth
"It is wrong always, everywhere, and for everyone to believe anything upon insufficient evidence." - W.K. Clifford "The Ethics of Belief"
"...a rule of thinking which would absolutely prevent me from acknowledging certain kinds of truths if those... truths were really there, would be an irrational rule." - William James "The Will to Believe"
One of the most common questions we get is "how do you value Internet stocks?" This question is not usually asked of momentum investors, sector rotators, thematic investors, or even garden variety growth stock investors, all of whom employ stock selection techniques that ignore, or at least minimize, valuation. It is a legitimate question, though, for value investors whose strategy rests on the purchase of stocks whose price is below some calculation of value.
Some would claim that you can't value Internet stocks, most of which have minimal sales, no earnings, negative cash flow, and unknown business prospects. These critics say the Internet is a classic bubble, a psychological phenomenon born of irrational enthusiasm for the economic potential of online activity. Such things end badly, and those who are participating in the madness are either dupes or gamblers betting against the odds. In any case, it is no place for the value investor.
Others are not so dogmatic, but point out that the early stage nature of these businesses precludes analysis based on traditional metrics such as price to earnings, book value, or cash flow. Companies which have reached the stage where they are making money, such as Yahoo or America Online, sport market capitalizations larger than any number of great American businesses including General Motors, Caterpillar and Sears. Applying standard metrics to these "new economy" companies does not yield anything approaching reasonable valuations. On these bases they look wildly over-priced (and many good investors believe they are).
Traditional metrics, when they work at all, work best with traditional businesses. The value of these analytical tools is usually a function of the historical data supporting their effectiveness. We have a lot of data about how the market has valued long-standing businesses whose economics are well understood, for example, foods and beverages, the big drug companies, retail stores, and newspapers. We know which groups have performed well early in an economic cycle, which have been most recession resistant, what historically attractive and unattractive valuations have been, and how current profitability compares with past results. All of this is absent from most Internet-related businesses, leaving the value investor at an apparent disadvantage to others who employ less rigorous methods to select securities.
It’s important to distinguish this situation from the related issue of valuing technology companies. Many value investors have chosen to ignore technology companies or maintain minimal exposure to them, despite long data trails and compelling evidence that this sector has had the ability to create substantial, long-lasting shareholder wealth. The reasons typically given are that technology is difficult to understand, that it changes rapidly, and that the stocks are usually too expensive according to standard valuation methods. All of these reasons are weak.
Ben Graham once denied that reward and risk were correlated in the stock market. (He was wrong.) He correctly said that reward was, or should be, related to the amount of work one was willing to do. If technology is difficult, it is not incomprehensible. Investors who rule out the largest sector of the stock market, and the most important driver of economic growth and progress, because it takes work to figure it out have little to cavil about when others get the rewards.
Although technology changes reasonably rapidly, it doesn't follow that such change is random or unpredictable. In many cases, it follows well-defined paths. The economics of technology and information-based businesses have been explored by economists such as Brian Arthur and Hal Varian. Their work is accessible to anyone who will take the time to study it.
Moreover, technology companies often create change and instability in other unrelated businesses that are not themselves involved in technology. Amazon is roiling the traditional book business; online brokers have forced even Merrill Lynch to dramatically alter long-standing business practices. Ignoring technology often leads to making bad investment decisions by not understanding the risks to your non-technologically based, but vulnerable, businesses. Warren Buffett found the pricing structure and demand dynamics of the encyclopedia business totally up-ended by Microsoft’s Encarta and CD-ROMs. Nobody wants the traditional World Book anymore, a business he thought had an enduring franchise.
It is true that some of the best technology companies have rarely looked attractive on traditional valuation methods, but that speaks more to the weakness of those methods than to the fundamental risk-reward relationships of those businesses. Had we understood valuation better we would have owned Microsoft and Cisco. Microsoft has gone up about 1% per week on average since it has been public. Companies don’t outperform year in and year out for over a decade unless they were mispriced to begin with, that is, undervalued. Paul Johnson, a prominent networking analyst, wrote an open letter to Warren Buffett a few years ago showing how Cisco met the criteria Buffett has so often enumerated for his investments. There is no evidence Buffett read the letter, but we read it and didn't buy Cisco. Johnson was right and we were wrong; not because the stock went up a lot, but because it was significantly undervalued and we missed that.
Undervaluations not determined by a stock's price in relation to existing or trailing earnings, book value, or cash flow, although these metrics may be evidence of a price that is below intrinsic value. Undervaluation is determined by the relation between a stock price and the present value of the free cash the underlying business will generate over one's forecast time horizon.
The past is known, and the future is not, and the only guide to the future is the past. That is why historical price-to-value relationships play such a dominant role in most value investors’ strategies. If one invested backward, instead of forward, or if the world never changed, those past relationships would be dispositive. One would not have to think about, analyze, or assess the probabilities of change in order to generate satisfactory investment returns.
The stock market is a discounting mechanism. Its prices reflect the expected value of the future, a future which at best will resemble, but not replicate, the past. Potentially disruptive technologies such as the Internet may change long-established economic relationships, providing an opportunity for those who correctly discern the changing economic patterns before they are reflected in market prices.
This, of course, is fiendishly difficult. Each individual possesses only the tiniest fraction of the information available to the market. The regulators try to assure that no one has material information before anyone else. Large changes result in large opportunities for profit, attract large interest, and often generate enthusiasms that outrun prudence. In these cases, risk far outstrips the probability of reward. Such appears to be the case with most Internet securities today. Most, not all.
The value of the publicly traded Internet securities is just over $500 billion, about 10% more than the market value of Microsoft, and about 6% of GDP. It seems reasonable that if the Internet really does "change everything", as its enthusiasts assert, the investment opportunity is probably greater than a single digit percent of GDP.
The market believes it knows where a lot of the opportunity is: the combined market value of AOL, Yahoo, and Amazon is about 40% of the total value of all Internet stocks. Two questions come to mind: first, do these values overstate or understate the intrinsic value of the businesses; second, how would you know?
The second question is logically prior to the first. Answering it is easy in theory. If the present value of, e.g., Amazon's future free cash flow is greater than $30 billion, Amazon is undervalued; if not, it is overvalued. How should one go about figuring this out?
W.K. Clifford, and many others, would say you shouldn't. Clifford, a brilliant mathematician who died at 34 in 1879, argued that making decisions without adequate justification was not just ill-advised, but wrong. If Clifford were a portfolio manager, he would argue that there is insufficient evidence on which to make a judgment about Amazon. Its business model is unproven (some would say unknown); it does nothing but report losses. The book business is mature, slow-growing, and fiercely competitive. Amazon’s new initiatives, toys, electronics, and online auctions, are certain to add to the losses through start-up risk. The new zShops effort, while promising, further fragments management’s attention. Finally, the value of Amazon not only exceeds that of Borders by 30 times, and Barnes and Noble by almost 20 times, it is greater than that of Sears, Federated, Saks, and Kmart combined! Such evidence as there is would indicate one's efforts to find value are better employed elsewhere.
Another 19th-century thinker, William James, wrote an essay to counter Clifford called "The Will to Believe." James argued in this seminal work that in many cases one was justified in believing something well in advance of what others may consider sufficient evidence. His argument carried the day philosophically, which is why no one reads Clifford anymore, while every competently trained student of the theory of knowledge has read and assimilated James.
James's argument was rich and detailed, but one of his points apposite to Amazon was that the level of evidence one needs to believe something is a function of how important it is not to be wrong. The evidence bar will be set higher the more important it is not to make a mistake. If being right has high value, and being wrong has low value, then the evidence needed for belief can be a lot lower since being wrong is not very costly, and being right has a high payoff.
This is why people play the lottery, why they buy insurance, and why both are reasonable. Some insurance events are so unlikely paying anything is usually not advised, for example, flight insurance. The lottery always has more players the higher the payoff since different people have different thresholds for betting on unlikely events.
With Amazon, we believe the payoff for being right is high. We arrive at this view by developing a model of each of Amazon's businesses, adjusting for capital employed and time to maturity. We project free cash flows using a variety of scenarios and try to assess the relative probabilities of each. This gives a range of possible values for the business. We then try to discern which scenario most closely fits the current evidence. That gives the most probable current intrinsic value.
The most you can lose is your purchase price. What is the threshold of evidence necessary to make such a bet? How will you know if you're likely to win or lose as Amazon's business develops? These are all epistemological questions. James notes that for one type of thinker, the crux of such decisions is in their principles or origin, while for another the importance is in the outcome.
One type of value investor requires the authority of the past in order to make a bet on the future. Like Clifford, they have a high threshold of evidence and wish to avoid the inevitable errors that attend trying to assess the expected value of the future. The theory of value would indicate that all of an asset's value derives from the future. Avoiding the analysis of possible futures, their probabilities and expected payoffs, may constitute the greatest error of all.
Bill Miller, CFA
Legg Mason Fund Adviser, Inc.
November 2, 1999
Share