Dividend Yields for Forecasting Stock Market Returns

This paper empirically assesses the ability of dividend yields to predict future stock returns in Germany assuming efficient markets and rational expectations. Since the order of integration of regressors are not exactly known, a bounds procedure, namely an autoregressive distributed lag (ARDL) model, is applied to test for cointegrating re-lationships among future stock returns and today’s dividend yield. It is also capable of dealing with the controversial issue of exogeneity of the dividend yield. ARDL and error-correction models are estimated for (future) stock returns and the dividend yield based on consistent estimates and standard normal asymptotic theory.