Stock Return Predictability : Using the Cyclical Component of the Price Ratio

This paper considers whether the cyclical component of the log dividend-price and price-earnings ratios contain forecast power for stock returns. While the levels of these series contain slow moving information for predicting long horizon returns, for short-horizon returns it is the relative movement between prices and fundamental that matters for investors, and whether prices are accelerating away or converging with fundamentals. We use three approaches to extract the cyclical component of these ratios and conduct a range of in-sample and out-of-sample tests. In addition to the cyclical components, we include further predictive variables that account for economic growth and the relation between stocks and bonds. In-sample estimation using the ratio levels reveals results that do not accord with economic intuition. In contrast, using the cyclical component leads to economically sensible values, as well as improved in-sample fit. Out of-sample forecasting reveals that in comparison to a historical mean model, the performance of our predictive models is generally better, although that depends on metrics used to evaluate the forecasts. Moreover, the cyclical component models outperform the levels based models. Notably, the historical mean model is preferred using standard mean absolute and squared errors measures but the predictive models perform better using Mincer-Zarnowitz and related encompassing regressions. Notably, when using economic based forecast evaluation, the predictive models are clearly preferred, with a stronger ability to predict the future direction of return movements and in obtaining higher trading returns. A further examination of the results reveals that this greater performance largely arises from a superior ability to predict future negative returns