SEVERAL editors of investment newsletters have contended that stocks are undervalued, based on the market's current dividend yield. The argument, however, does not stand up to historical scrutiny.

Dividend yield is one of the oldest market-timing tools, having first gained wide acceptance nearly 100 years ago in the United States. For most of the last century, market timers generally considered the stock market to be near a top whenever its yield dropped to around 3 percent, and near a bottom when its yield rose to about 6 percent.

The model fell out of favor during the 1990's, however, owing to its persistent bearishness. The yield of the Standard & Poor's 500-stock index dropped to less than 3 percent in late 1991, for example, and continued to fall, reaching 1.1 percent in 2000. Yet the stock market in the 1990's turned in one of its strongest performances ever.

The newsletter editors who are now turning to the dividend yield model are arguing in relative, not absolute, terms. The current yield of the S.& P. 500, at 1.8 percent, is still well below levels that indicated a market top before the 1990's. But the yield is now higher than short-term interest rates for the first time in more than 40 years.

Louis Navellier, the editor of the MPT Review and Blue Chip Growth Letter advisory services, says that this development means that stocks are undervalued. Mr. Navellier, whose company also manages some no-load mutual funds, said recently that he had "no doubt" that we're "at the beginning of a great bull market."

But should a comparison of the dividend yield and short-term interest rates offer a good reason for bullishness? I don't think so.

As Jeremy J. Siegel, a finance professor at the Wharton School of the University of Pennsylvania, argues, the stock market is an instrument for long-term investment. As a result, it should not be viewed as a substitute for short-term bonds. A more appropriate comparison would be to the rates paid by the 10-year Treasury note or 30-year bond; the yield of the S.& P. 500 is still lower than both. (Don't confuse the argument being made by Mr. Navellier and other newsletter editors with the so-called Fed Model, a popular but flawed market-timing model that compares the 10-year note rate with the stock market's earnings yield — the inverse of the price-to-earnings ratio.)

Professor Siegel allowed that his theoretical argument should be tested, however, because it was conceivable that some short-term traders might consider dividend-paying stocks a substitute for short-term bonds.

Following his suggestion, I analyzed the stock market from 1871 to the end of 2002, seeking correlations between the market's relative yield and its subsequent performance. I focused on returns over periods as short as three months and as long as five years. I based my analysis on historical dividend yield data from Professor Siegel and on short-term interest rate data from Robert J. Shiller, a Yale economics professor.

None of my tests found any meaningful support for the notion that the stock market offers above-average gains when its yield is above short-term rates.

But the tests did find that smaller dividend yields are generally followed by lower long-term returns in the stock market, confirming what many researchers have found. The dividend-yield model has been only moderately successful in predicting the stock market's short-term direction. But even considering its failure for several years in the 1990's, its record at predicting five-year market moves has generally been impressive. Before the 1990's, for example, the market's record low yield was reached just before the 1987 crash. Its previous low was at the start of the 1973-74 bear market.

Because dividend yields are so low, the model is bearish for now. My tests of that model predict that over the next five years, the stock market is unlikely to produce annualized returns as high as the market's long-term average of about 10 percent. 

Questions:

1. "For most of the last century, market timers generally considered the stock market to be near a top whenever its yield dropped to around 3 percent, and near a bottom when its yield rose to about 6 percent." In using this dividend yield model as a market-timing tool, what must one assume?
2. Did this version of the "dividend yield" model (which one may call absolute) work until the early 1990s?  If yes, why do you think so?  If not, on what basis do you say so?
3. On what basis does Navellier, according to you, put forward the new version of the dividend yield model (which one may term "relative" -- to interest rates)?
4. The "Efficient Markets Hypothesis" says that security prices (at least in liquid markets) incorporate all relevant information.  Are the absolute or the relative versions of the Dividend Yield model consistent with the Efficient Markets Hypothesis?
5. Why does Professor Siegel say that the stock market is an instrument for long-term investment?  Is this a normative statement (i.e. this is what it should be) or a positive statement (i.e. this is what it happens to be)?  (Is the answer to this obvious from the article?)
6. The author says that "smaller dividend yields are generally followed by lower long-term returns in the stock market."  Can you explain why this should be so?  Is it an instance of mispricing by the market?