The role of variance risk premium in predicting excess stock market return: out-of-sample evidences

Abstract

This paper examines the out-of-sample performance of variance risk premium in predicting excess stock market returns across nine international markets. We assess the out-of-sample predictability through statistical and economic significance tests and find that the variance risk premium has strong forecasting power at the 4-month horizon for most of the international markets considered in this study. In addition, we find the predictability is even stronger during the recent financial crisis period.

Keywords:

• variance risk premium
• out-of-sample prediction
• stock return predictability

JEL Classifications:

• G11
• G12
• G13

Acknowledgements

We are grateful to Jerry Coakley, Hao Zhou, Hui Guo, Hai Lin, Xiaoquan Liu and seminar participants at Xiamen University and the University of Essex for many helpful comments.

Notes

¹ These international markets are selected due to the data availability.

² Following Campbell and Thompson (2008) and Rapach et al. (2010), we choose the g value of 3. The empirical results are qualitatively similar for other reasonable γ values.

³ In view of the empirical findings of Welch and Goyal (2008) that out-of-sample forecasting performance may depend on the choice of sample periods, we divide our sample into two sub-samples.

⁴ In unreported tables, we find that the VRP has stronger out-of-sample predictability than that of economic variables documented in Welch and Goyal (2008). These results are available upon request.

⁵ We also check the out-of-sample predictive performance at other monthly horizons for both VRP measures and confirm that the predictability is strongest at the four-month horizon. We thank the referee for this suggestion.

Additional information

Funding

Jian Chen thanks the financial support from National Natural Science Foundation of China [71201136].