The Fed Model Impels, but Does Not Compel

Indicating stocks were undervalued in the mid-1990s, at the cusp of the biggest rally in history;

Hitting overvaluation in the summers of 1997 and 1998, prior to sharp drops in share prices, and a steep undervaluation in the fall of 1998, at the cusp of the blowoff phase of the biggest rally in history. In January 2000, the Fed model said stocks were overvalued by a record, and extraordinary, 70%.

More recently, and famously thanks to a Barron's cover story, the Fed model showed shares were significantly undervalued in late September 2001, then swung to overvalued early this year. In late July 2002, the Fed model said stocks were undervalued again, right around the July 22 bottom. However, they stayed undervalued prior to the September 2002 swoon, reaching more extreme levels of undervaluation in early October.

"The Fed model is not a perfect market-timing tool," as an over- or undervalued market can become even more so, wrote Yardeni, who is widely credited (including by Yardeni himself) with coining the term "Fed model" to describe something the central bank has never formally endorsed. "However, the Fed model does have a good track record of showing whether stocks are cheap or expensive."

Critics, naturally, contend otherwise.

Unsafe at Any Level

"I hate the Fed model for a couple of reasons," said Thomas McManus, equity portfolio strategist at Banc of America Securities.

First and foremost, earnings yield of the S&P 500 is "not representative of the market," McManus said, recalling that so-called average stocks were dramatically undervalued in March 2000, even as major averages such as the S&P 500 were vastly overvalued.

Late 1999 and early 2000 was a point of "crisis valuation" for stocks, he said, noting the Chinese characters for crisis is a combination of danger and opportunity. "The Fed model failed to tell you there was danger in Cisco (CSCO) at $82 and opportunity in Conagra (CAG) at $15."

Second, "the math used in the Fed model is w-r-o-n-g," McManus argued. Rather than as a ratio, stocks' earnings yield should be compared with Treasuries on a spread basis, he said, similar to how corporate bond yields are discussed vs. Treasuries.

Third, and finally, McManus prefers a model which values nominal yields -- inflation expectations plus real interest rates -- to determine the expected return on common stocks. (Even Yardeni concedes the Fed model is a "very simple stock valuation model" and "should be only one of several inputs investors use.")

Having said all that, McManus believes the average stock remains attractive at current levels based on his proprietary valuation model, which he declined to specify further.

Bill Gross, managing director at Pimco, which has over $300 billion under management, takes an even harsher view of the Fed model.

"Should the Street really be using 10-year Treasuries as the convenient bogey to beat?," Gross mused in a report on Pimco's Web site earlier this month. "After all, Treasuries are risk-free and stocks anything but." (The idea that the S&P's earnings yield should be equal to the 10-year's is controversial, although the average spread between the two is only 25 basis points since 1979, according to Yardeni.)

Instead of comparing shares to the 10-year's yield, Gross suggested using Baa-rated corporate bond yields, "a pretty fair measure of the composite debt of S&P 500 companies."

Using this "apples-to-apples" metric, stocks are about 30% overvalued compared with the near-7% yields on Baa corporates, he wrote. "In the end analysis, it is fair to ask who is the true fool -- the eternal 'glass more than half full' stock optimists or the bond guy with a supposed axe to grind."

As always, investors are encouraged to draw their own conclusions about that statement, as well as the value of the Fed model. My conclusion is the Fed model is a decent guide for the relative attractiveness of the S&P 500, but is far from an infallible indicator.

>To order reprints of this article, click here: Reprints