Stock return predictability and variance risk premia

: a frequency domain analysis

Abstract

The main objective of this thesis is to analyze the out-of-sample equity return forecasting power of the variance risk premia and its frequency components. The variance risk premia (VRP) is represented by the difference between the risk neutral (implied variance) and physical (realized variance) expectations of the variance. In the literature, a considerable number of variables present strong in- and out of- sample performances, being one of them the variance risk premia. Likewise, by decomposing some variables into their frequencies, their out-of-sample performances increases. Therefore, in order to study the behavior of this variable, we decompose the time series of the variance risk premia into frequencies. The main result of this thesis is that the original time series and its medium frequency component demonstrate a remarkable out-of-sample performance when predicting the equity excess of return. We also show that, although the time series presents a better statistical performance (i.e., a higher out-of-sample R2), in economic terms its medium frequency component delivers higher gains.

Date of Award	15 Jul 2021
Original language	English
Awarding Institution	Universidade Católica Portuguesa
Supervisor	Gonçalo Faria (Supervisor) & Fabio Verona (Co-Supervisor)