Predicting the Equity Premium With Dividend Ratios

Our paper reexamines the forecasting regressions which predict annual aggregate stock market returns net of the risk-free rate with lagged aggregate dividend-yield ratios and dividend-price ratios. Prior to 1990, the conditional dividend yield could reliably outperform the historical equity premium mean in predicting future equity premia *in-sample*. But our paper shows that the dividend ratios could not outperform the prevailing unconditional mean *out-of-sample*, plus any residual power was directly related to only two years, 1974 and 1975. As of 2000, even this in-sample predictive ability has disappeared. Our paper also documents changes in the time-series processes of the dividends themselves and shows that an increasing persistence of dividend-price ratio is largely responsible for weak stock return predictability.