| Abstract : | The issue of predicting equity returns is one of the most widely discussed topics infinancial economics. Yet no consensus exists on the fundamental questions: whetherpredictability exists and which variables show best predictive performance. Workingin a time series framework we employ the data used by Goyal and Welch (2008). Wefirst purpose to identify the variables which show significant predictive ability relyingon a variety of suitable econometric models. Secondly, based on the best in sampleperformers, we propose a range of stationary models aiming to forecast equitypremium returns and we attempt to evaluate their out-of-sample performance byutilizing time series cross-validation with a rolling origin.
|
|---|
Copyright © 2013 Library AUEB. All rights reserved.