Predicting Stock Returns in the Presence of Uncertain Structural Changes and Sample Noise

Abstract

The predictive power of the dividend-price ratio has been the subject of intense scrutiny. Most studies on return predictability assume that predictor variables follow stationary processes with constant long-run means. Following recent evidence of the role of structural breaks in the dividend-price ratio mean, we propose an estimation method that explicitly incorporates the uncertainty about the location and magnitude of structural breaks in the predictor that extracts the regime mean component of the dividend-price ratio. Adjusting for structural changes in the ratio's mean and estimation error significantly improves the predictive power of the dividend-price ratio as well as other standard predictors in-sample and out-of-sample.

Keywords: Return Predictability; Structural Breaks; Dividend-Price Ratio; Estimation Error

JEL Classification: G170