No investing advice seems more sound than that you should buy stocks when they are “cheap” and sell when they are “expensive.” Wall Street, the financial press, and millions of investors devote countless hours and dollars to unearthing “undervalued” opportunities and panning “overvalued” ones. Business school professors forever develop and teach ever-more-refined valuation techniques. Fanatically precise analysts compute projected earnings to the penny and “intrinsic value” to the dollar.
The aggregate result of all this effort, unfortunately, is the usual one: mediocrity. No matter how committed they are to buying cheap stocks and selling expensive ones, most investors still underperform market indices. There are three reasons for this. First, as we’ve already discussed, competition is so intense that most obvious “mispricings” are quickly spotted and bought or sold away. Second, determining whether a stock is actually cheap or expensive is a lot more difficult than it seems. Third, this determination is less relevant to intermediate-term stock performance than most people think.
A share of stock is, in theory, worth the “present value of future cash flows” attributable to the share. (In practice, a share is worth what someone will pay for it, but leave that aside for a moment.) Given the confidence with which some commentators cite the theory, a casual observer might assume that the “present value of future cash flows” is an indisputable number, akin to a price tag on a can of soup. In reality, however, it is not a number but an argument, and, in most cases, it is a surprisingly imprecise argument, with a wide range of reasonable conclusions.
An analysis of the “present value of future cash flows” requires, at minimum, two components: 1) an estimate of the future cash flows; and 2) an appropriate discount rate with which to determine their present value.
Most cash-flow projections don’t extend beyond a 5- or 10-year horizon, so the analysis also usually requires a “terminal multiple” to value the company’s operations beyond the forecast period. Once one has these components, estimating present value is a matter of math. However:
1. No one really knows what the future cash flows will be; and 2. Discount rates and terminal multiples are subjective assessments based on the trading prices and outlook for other securities (and, consequently, change as the prices of the securities change).
As a result, most valuation conclusions are extraordinarily subjective, and small tweaks in assumptions can yield big changes in estimated value.
Let’s assume, for example, that we know that a company will earn $1 per share per year forever (an otherworldly assumption, but go with it). In this case, all we need to determine the “present value of future cash flows” is a discount rate. Because, in our example, the cash flows are known and guaranteed, we can use the so-called “risk-free” rate, the prevailing rate of interest that an investor can earn without risking a loss of capital. One proxy for the risk-free rate is the yield on 10-year Treasury bonds, which, as of this writing, is about 4.2 percent. Discount 150 years of earnings of $1 a year (the financial equivalent of “forever”) at this rate and—voila!—the value of our hypothetical stock is about $24. If the stock is trading at $20, we have apparently found ourselves a “bargain.”
But what if we assume that the “risk-free” rate changes, as it always does? What if, for example, we assume that the yield on 10-year Treasury bonds will regress to its long-term mean of about 7 percent, a scenario that, given enough time, is probable? Well, then our $24 stock will only be worth $14. Or what if the T-bond yield jumps to 10 percent or more, as in the inflation crisis of the early 1980s? Then the stock will be worth less than $10. In other words, even if we know for a fact that a company will earn $1 per share per year forever—something that, in practice, we will never come close to knowing—we might conclude that the stock’s “fair value” is anywhere from less than $10 to more than $30 (the T-bond yield could always drop, too), with a central value around $14 (the value generated using the bond’s long-term mean). Twenty dollars might not be such a “bargain,” after all.
Choosing discount rates, moreover, is the most precise and least subjective element of the valuation analysis. The most subjective element is predicting future cash flows. As a result, it doesn’t take much imagination to conclude that Google could be “worth” anywhere from, say, less than $20 to more than $200. If one assumes that Microsoft and Yahoo! will develop superior search services, that Google will collapse under its own arrogance, and that interest rates will rise (all reasonable possibilities), one might conclude that Google’s “fair value” is in the teens. If, on the other hand, one assumes that Google will fend off competition, develop a full-fledged portal, and sustain its breathtaking margins and growth rates (also possible), one could conclude that the IPO price is dirt cheap. Valuing a high-risk, high-growth stock like Google is far less precise than valuing, say, a mature utility, but even in the latter case, those who think they can pinpoint stock values are hallucinating.
Which brings us to the second problem with valuation: its lack of relevance as an intermediate-term price-prediction tool. Even if we could establish for certain what a stock was worth, this would be no guarantee—or even indication—that the stock would trade there in any reasonable timeframe (or ever). Over the long haul, thankfully, valuation does matter: The market is not random, stock prices do tend to regress to long-term means, and long-term investors are better off buying when stocks are cheap. As discussed in a previous piece, however, the “long term” is long (decades, not years), and valuation is not a particularly helpful prediction tool over timeframes of three months to a couple of years (not worthless—just not particularly helpful).
Take the S&P 500, for example, an index that can be viewed as a stock tied to the biggest companies in the U.S. economy. A glance at history reveals that the index’s price often takes eons to regress to its long-term mean. According to data compiled by Yale professor Robert Shiller, from 1881 to 2000 the S&P 500’s average price-earnings ratio was about 16 times earnings, with a peak of 44 times (1999) and a trough of 5 times (1920). From this data, one might conclude that the average value of the S&P 500 is 16 times earnings, and, therefore, in years when prices were significantly below or above this level, stocks were “cheap” or “expensive.” Judging from the eventual performance of the index, these conclusions would have been right. They wouldn’t, however, have allowed one to predict with confidence what the S&P 500 was going to do the following year.
In 1970, for example, the S&P 500 dropped below 16 times earnings for the first time since 1958 (“Buying opportunity!”). This apparent undervaluation did not, however, signal that prices were about to go up. In fact, 12 years later, in 1982, the index was trading at about seven times earnings (a real buying opportunity), and it did not return to “fair value” until 1986, 16 years later. Similarly, in 1986, if one had decided to sell the S&P 500 on the theory that above 16 times earnings was “overvalued,” one would have been even more frustrated. Except after the 1987 crash, the S&P 500 has remained above 16 times earnings ever since 1986. The suckers who refused to pay more than fair value, in other words, missed the last 14 years of the bull market.
Bottom line, valuation is far more difficult to determine and far less helpful in predicting intermediate-term price performance than most commentators think. Next time someone urges you to buy a stock “because it is cheap,” therefore, remember that the wise response is probably, “So?”
Next week: If valuation isn’t a good near-term price prediction tool, what is?