The Market: Too High? Too Low?
Why so many valuation models are so wrong
Wonder if the stock market is out of whack? Click onto Dismal Scientist at www.dismal.com and slide your mouse over to the market-valuation calculator. Plug in a couple of commonly used figures--a 5% growth rate for earnings, say, and a yield of 5% on the 10-year Treasury bond--and the calculator offers up some distressing news: The Standard & Poor's 500-stock index, it says, is 11.2% overvalued, even after the Mar. 23 plunge.
But if you'd like to make the worry go away, assume a higher earnings-growth rate, 7.5%, and a lower interest rate, 4.5%, and voila, the market is 2.1% overvalued, close enough to say it's fairly valued. "You have to examine the inputs" to figure out whether a model makes sense, says Mark Zandi, chief economist at Regional Financial Associates in West Chester, Pa., which runs the Dismal site on the Web.
As the stock market has made a run on Dow 10,000, it has also made fools of the folks who use statistical models to divine where the market is headed. Most of the models have been warning for months, and some for years, that the market is unsustainably high and heading for a fall. "This is a speculative bubble," cautions Ronald J. Talley, director of financial forecasting for WEFA Inc., an economic consulting firm in Eddystone, Pa., arguing that the market is still twice its fair value.DANGEROUS PLACE. But if the market continues to thumb its nose at the models' forecasts, then perhaps it's the models that are out of whack. "The traditional model is flawed if you think that people's attitudes toward risk are changing over time," says Kevin A. Hassett, a resident scholar at American Enterprise Institute. He and AEI fellow James K. Glassman have developed a model that finds the market to be sharply undervalued. Their argument: Investors don't see the stock market as the dangerous place they once did and thus will bid up stocks to far higher prices.
If the new modelers are right, the change overturns statistical relationships that have held true for decades. It would throw doubt especially on the long-held view that stocks are riskier than bonds.
Consider the so-called Fed model, which Deutsche Bank economist Edward Yardeni teased out of a report to Congress by the Federal Reserve in 1997, seven months after Chairman Alan Greenspan warned of the market's "irrational exuberance." This model compares the earnings yield on stocks to the current yield on bonds, the idea being these are competitive investments. The earnings yield is the forecast operating earnings for the stocks in the S&P 500 over the next 12 months divided by the price of the index. On Mar. 23, that was $52.59 divided by 1262, or 4.17%. Compare that with the interest rate on a 10-year U.S. Treasury bond, which is 5.16%. Under this model, Treasuries are a better buy--they yield more. Only when these two numbers are equal are stocks fairly valued. Right now, this model says stocks are 24% overvalued.
The model's beauty is its simplicity--and the fact that the variables, market level, interest rate, and consensus earnings forecast are not subject to the whim of the modeler. There's an underlying assumption in the model that the ideal earnings yield for stocks is the 10-year bond yield. But Greg A. Smith, investment strategist for Prudential Securities Inc., argues that the tight relationship between rates and earnings yield holds when the bond rate is above 6%, but is less clear at lower levels. It could be, says Smith, that in the world of low inflation and low interest rates, this model doesn't work.
That's also the rap on the Campbell-Shiller model, a valuation method advanced by John Y. Campbell of Harvard University and Robert J. Shiller, a professor at Yale University's School of Management. They use market history to come up with an average p-e ratio for stocks and argues that when p-e's are historically high--as they are now, at 33 times last year's earnings--stock prices will fall toward the long-term average of 15. The problem with this approach is that it overlooks changes in the U.S. economy. Tech stocks, for example, which have high growth rates and high p-e's, make up nearly 20% of the S&P 500. A decade ago, the S&P was dominated by autos, oil, and other cyclical stocks, while tech accounted for just 8.4% of that index.
Other models that grapple with market valuation focus on the "risk premium." The risk premium is the difference between the risk-free interest rate, usually the return on U.S. Treasury bills, and the return on a diversified stock portfolio. Over more than 70 years, the return to stocks averaged 11.2%, and T-bills, just 3.8%. The difference between the two returns, 7.4%, is the risk premium. Economists explain this extra return as investors' reward for taking on the greater risk of owning stocks. Most market watchers believe that in recent years, the premium has fallen to somewhere between 3% and 4% because of lower inflation and a long business upswing that makes corporate earnings less variable.
Looking ahead, the risk premium is critical in valuing equities. Charles M. Lee of Cornell University's Johnson Graduate School of Management closely examines the 30 stocks in the Dow Jones industrial average. He estimates a risk premium for each stock by measuring the price volatility of that stock's industry group. For instance, Lee says the risk premium for General Motors Corp. is 9.48%, while for General Electric Co. it's just 4.48%. (Auto stocks have a higher risk premium because their earnings and stock prices swing more dramatically than a diversified company such as GE.) Lee then estimates future cash flows for each Dow stock and uses a combination of the risk premium and risk-free interest rate to discount that cash to a present-day "intrinsic value."
When the Dow oversteps this value, the market is too rich. And even with the Dow dropping back from the 10,000 climes, the price-to-value ratio for Dow 9700 is 1.61, or 61% above intrinsic value. Over the past 20 years, that ratio averaged 1.08. Although this ratio peaked last April at 1.74, it's still well above the 1.41 registered before the 1987 collapse.DIRT-CHEAP DOW. But like most models, the Cornell approach looks back to forecast the future. Hassett and Glassman say that's crazy, because the risk premium is shrinking. They argue that stocks have become a lot less risky than bonds, and in fact they posit that the risk premium is heading toward zero. In any model that uses a risk premium to calculate the proper discount factor, lowering that premium from 3% to zero is the same as slashing interest rates by three full percentage points. "In time, stocks and bonds will converge," predicts Hassett. "The opportunity is being in the stock market as the market revalues stocks." That's why Hassett and Glassman argue that the Dow is dirt cheap; their forthcoming book is called Dow 36,000.
Most analysts scoff at the notion that stocks are no riskier than bonds. "There's still a lot of uncertainty in today's world," says Leah Modigliani, a Morgan Stanley Dean Witter equity strategist. "The risk premium has moved down, but it's not zero."
Forecasters everywhere concede that old models are suspect. Merrill Lynch & Co.'s quantitative analysts, for instance, look at five valuation models; only one of them suggests that the market is anything but wildly overvalued. The optimistic model is based on long-term earnings estimates by the firm's analysts, and since analysts tend to be optimists, Merrill economists take a dim view of its output. "We definitely don't think it's the best measure of valuation," says Kari E. Bayer, a quantitative strategist at Merrill. And Prudential Securities' Smith has told clients that as long as rates remain steady and earnings accelerate, they can "forget the models."
That may be sound advice. There are no market watchers and investors more humble than those who heeded the models and yanked their money out only to see the bull stampede ahead.By Joseph Weber in Toronto, with Jeffrey M. Laderman in New YorkReturn to top