It’s been known for some time–decades, really–that relatively high dividend yields tend to precede relatively high returns in subsequent years. Graham and Dodd’s Security Analysis suggests as much about the relationship between valuation and return. More rigorous studies of the valuation phenomenon (of which dividends are only one measure) arrived in the 1980s, when a number of new research efforts found a strong relationship between relatively undervalued equities and higher prospective return.
One review of the possibilities came in a 1984 Journal of Portfolio Management article: “Dividend yields are equity risk premiums,” by Michael Rozeff. He explains that “the evidence dictates” that dividend yields can be used to time purchases. He warns against reading too much from specific dividend levels for establishing absolute buy and sell signals. He also counsels readers away from trying to profit from dividend-yield signals in the short term. Nonetheless, the basic premise, if not exactly original, reflects economic common sense, Rozeff argues:
My evidence indicates that returns increase continuously and monotonically as dividend yield in the prior year increases. My theory that the dividend yield is a measure of the ex ante risk premium explains why this is so. High returns tend to occur when the environment is perceived to be so risky that investors demand a high premium for holding stocks. Low returns tend to occur when the environment is perceived to hold such little risk that investors demand a low risk premium for holding stocks.
Subsequent studies lend support to the idea that valuation overall matters. For example, Robert Shiller, in Irrational Exuberance, argues in favor of return predictability based on valuation parameters. One example comes by way of a diagram in the book that plots price-earnings ratios against subsequent 10-year returns based on buying the S&P Composite Index (a proxy for U.S. stocks) at a given p/e ratio. The relationship, which draws on more than 100 years of market history through 1989, shows a “moderately strong” link between low p/e ratios and relatively high returns, and vice versa, according to Shiller.