A Market Barometer, Challenged on Two Fronts
By MARK HULBERT
The New York Times
December 15, 2002
TWO recent studies have found serious flaws in the Fed Model, one of the more widely used methods of valuing the stock market.
The Fed Model compares the yield of the 10-year Treasury note with the stock market's "earnings yield," defined as the inverse of the price-to-earnings ratio. When the earnings yield is lower than the Treasury yield, the model considers stocks overvalued. But when the earnings yield is greater, as it is today, the model regards stocks as undervalued.
Despite its name, the model is not endorsed by the Federal Reserve. The name was given by Ed Yardeni, now the chief investment strategist at Prudential Securities, who constructed it from comments in a paragraph of a Fed report to Congress on July 22, 1997. The paragraph noted that the yield on the 10-year Treasury note at that time far exceeded stocks' earnings yield. But the Fed did not say in the report that stocks were overvalued.
At its core, the model forecasts that P/E ratios will be higher when interest rates are lower, implying a higher price level for the market. That makes intuitive sense, because stocks and bonds are competing assets.
But the model has had a poor record over the last 75 years in forecasting long-term returns, according to Clifford S. Asness, managing principal of AQR Capital Management, a quantitative research firm. On average, Mr. Asness found, 10-year periods that began with high P/E ratios were followed by low returns over the decade. That was true whether interest rates were high or low at the beginning of those periods.
Consider stocks' valuation at the end of 1964. The earnings yield of the Standard & Poor's 500 was 5.4 percent, and the 10-year Treasury note yielded 4.2 percent. The Fed Model would interpret stocks' advantage of 1.2 percentage points as strong evidence of undervaluation. Yet over the next 10 years, the S.& P. 500's annualized return, after inflation, was minus 3.9 percent.
The Fed Model would interpret the current stock market as being about as undervalued as it would have in 1964. But Mr. Asness's research gives us no reason to believe the market will perform any better over the next 10 years than it did after 1964.
Why does the Fed Model fail? One shortcoming is that it does not take risk into account. Stocks must provide a higher long-term return than Treasury bonds in order to entice investors into incurring equities' greater risk. A higher earnings yield may not be evidence of undervaluation but instead may reflect nothing more than stocks' risk premium over bonds.
A more subtle weakness of the Fed Model, according to another recent study, is its failure to account properly for inflation — an error that economists refer to as the "money illusion." The study, by two finance professors, Jay R. Ritter of the University of Florida and Richard S. Warr of North Carolina State, has found that investors systematically price stocks too cheaply when inflation is high and place too rich a valuation on them when inflation is low. (The study appeared in the March 2002 issue of the Journal of Financial and Quantitative Analysis.)
The professors said investors generally fail to account for inflation's effect on earnings. For example, they erroneously assume that nominal earnings will not grow faster when inflation rises, which means that they think real earnings — earnings after inflation — will grow at a slower rate. If that assumption were accurate, stock prices would come down as inflation worsened.
But the professors say nominal earnings historically have grown faster when inflation is rising, keeping the growth rate of real earnings more or less constant.
Mr. Asness reminds us that stocks are a good long-term inflation hedge. The Fed Model implicitly assumes that this is not the case. As he puts it, "A pundit who believes in the Fed Model, but also believes equities are a good long-term inflation hedge, is not being consistent."
nytimes.com |
