No, dividend yield is not a reliable predictor of future returns

Many investors still work under the assumption that dividend yield is a good measure of value and that stocks with higher dividend yields should outperform stocks with lower yields in the long run. The years of outperformance in low-dividend growth stocks between 2010 and 2021 may just have been an aberration, but if you want to assess the value of stocks, dividend yield is not a reliable measure, as Yang Bei from the University of Missouri has demonstrated.

I have written about Yang Bei’s work before. Back then, I wrote about a study of his, where he created the longest possible time series of US and UK market returns and tested the predictive powers of an entire battery of valuation measures. On a one-year horizon, the vast majority of these valuation measures failed miserably, though I remain of the opinion that for longer time horizons of 5 to 10 years, valuation measures still play an important role in assessing risks and opportunities in financial markets.

To double down on the inability of dividend yields to forecast one-year returns, though, Yang Bei used the data of the Macrohistory Lab that provides equity prices, dividend yields and metrics about bond and housing markets for 17 countries since 1870.

Using that data, he could test the predictive power of dividend yields in these 17 countries over 150 years. And lo and behold, in sample, five out of these 17 countries showed a significant relationship between dividend yields and equity market returns. But that is using regression analysis for past data. When he tried to forecast market returns out of sample for future, unknown data, only one out of 17 countries (the UK) showed significant forecasting power. But note that if you run the same statistical test 20 times, on average you would expect one out of these 20 times to show a significant result. So, one country out of 17 having a significant forecasting result is the same as saying that the forecasts of dividend yield are no better than chance.

These results are not earth-shattering for many investors, but it once again shows that valuation measures are a poor predictor of one-year ahead performance, no matter where in the world you invest. If you are a value investor, you need to have a multi-year holding period because only over those longer time frames does value investing create outperformance.