Predicting returns and dividend growth — The role of non-Gaussian innovations

https://doi.org/10.1016/j.frl.2021.102315Get rights and content
Under a Creative Commons license
open access

Highlights

  • Departures from Gaussianity can impact return and dividend-growth predictions.

  • A Bayesian VAR framework is used allowing for flexible modelling of the innovations.

  • Stochastic volatility has a non-negligible impact on the predictability evidence.

  • Skewness and fat tails are secondary, once controlling for stochastic volatility.

  • Out-of-sample improvements occur mainly for density forecasts.

Abstract

In this paper we assess whether flexible modelling of innovations impact the predictive performance of the dividend price ratio for returns and dividend growth. Using Bayesian vector autoregressions we allow for stochastic volatility, heavy tails and skewness in the innovations. Our results suggest that point forecasts are barely affected by these features, suggesting that workhorse models on predictability are sufficient. For density forecasts, however, we find that stochastic volatility substantially improves the forecasting performance.

JEL classification

C11
C58
G12

Keywords

Bayesian VAR
Dividend growth predictability
Predictive regression
Return predictability

Cited by (0)