On the Economic Value of Return Predictability

Recent studies provide strong statistical evidence challenging the existence of out-of-sample return predictability. The economic significance of return predictability is also controversial. In this paper, we find significant economic gains for dynamic trading strategies based on return predictability when appropriate portfolio constraints are imposed. We find that imposing appropriate portfolio constraints is critical for obtaining economic profits, which seems to explain the contradictory findings about economic significance in the literature. We also compare the performance of several predictive models including the VAR, the VAR-GARCH, and the (semi)nonparametric models and find that the simple VAR model performs similarly to other more complex models.