Generalized financial ratios to predict the equity premium

•Current values of market prices and dividends can predict the equity risk premium.•The predictive relationship evolves due to changes in macro-economic conditions.•This causes a disconnection between the P/D ratio and the equity premium.•The generalized P/D ratio learns from the log-log regression of prices on dividends.•Generalizing the P/D, P/E and P/B leads to more accurate equity premium forecasts.

Empirical evidence for the price-dividend ratio to be a predictor of the equity premium is weak. We argue that changes in the economic conditions and market composition lead to a time-varying relationship between prices, dividends and the equity premium. Exploiting the information in the rolling window log-log regression of stock prices on dividends, we obtain the Generalized Price-Dividend Ratio (GPDR), that compares the price per share with a time-varying transformation of the dividend per share. The GPDR leads to economic and statistical gains when forecasting the equity premium of the S&P 500 at the 1, 3, 6 and 12 month horizon, as compared to using the classical price-dividend ratio or the prevailing historical average excess market return. Similar improvements are obtained for Generalized Financial Ratios based on the corporate earnings and book value.