Stock Return Predictability: Evaluation Based on Prediction Intervals

Abstract

This paper evaluates the predictability of monthly stock return using out-of-sample (multi-step ahead and dynamic) prediction intervals. Past studies have exclusively used point forecasts, which are of limited value since they carry no information about the intrinsic predictive uncertainty associated. We compare empirical performances of alternative prediction intervals for stock return generated from a naive model, univariate autoregressive model, and multivariate model (predictive regression and VAR), using the U.S. data from 1926. For evaluation free from data snooping bias, we adopt moving sub-sample windows of different lengths. It is found that the naive model often provides the most informative prediction intervals, outperforming those generated from the univariate model and multivariate models incorporating a range of economic and financial predictors. This strongly suggests that the U.S. stock market has been informationally efficient in the weak-form as well as in the semi-strong form, subject to the information set considered in this study