The forecasting power of EPU for crude oil return volatility

Rufei Ma,Changfeng Zhou,Huan Cai,Chengtao Deng

doi:10.1016/j.egyr.2019.07.002

Rufei Ma, Changfeng Zhou + Show 2 more

Open Access

https://doi.org/10.1016/j.egyr.2019.07.002

Copy DOI

Abstract

Economic policy uncertainty (EPU) has important implications for crude oil market. To explore the implications, this paper investigates the impact of EPU on the crude oil return volatility and which EPU index has the most forecasting power in crude oil market. To this end, we employ the GARCH-MIDAS model which can incorporate lower frequency EPU index variable with higher frequency crude oil return variable effectively. We find that EPU has a positive and significant impact on the crude oil return volatility, but the effect is short-lived and the decay period is about one year. Particularly, our results show that the US EPU index has the best forecasting power for crude oil return volatility over the long-term, whereas China EPU index has the best forecasting performance in the past one year. Our findings have important implications on risk management for investors in crude oil market.

Highlights

As one of the most important commodities in the world, crude oil is an energy product, and a financial asset, which plays an important role in world economy
We find that most models combine with Economic policy uncertainty (EPU) and Oil Volatility Index (OVX) have the better performance than the models only with OVX, confirming that the EPU index has a good forecasting power in crude oil return volatility
We examine the impacts of economic policy uncertainty in different countries on the crude oil return volatility, and investigates which EPU index has the most forecasting power in crude oil market

Summary

Introduction

As one of the most important commodities in the world, crude oil is an energy product, and a financial asset, which plays an important role in world economy. We use the EPU index proposed by Baker et al (2013) to examine the impact of economic policy uncertainty on crude oil return volatility and investigate the predictive performance of different EPU indices. The second strand of literature uses realized volatility models (Tian and Hamori, 2015) Most of these models have difficulties in dealing with the different frequencies between crude oil return volatility and the macroeconomic covariates. The frequency of the crude oil return data is on a daily basis, while the macroeconomic covariates are obtained monthly or on a even lower frequency For solving this problem, Engle et al (2013) develop the GARCH-MIDAS model. We employ GARCH-MIDAS model to examine the impact of the monthly EPU indices on the daily crude oil return volatility.

Model specification

Performance test

GARCH-MIDAS estimation

Model prediction and evaluation

Conclusion

Declaration of competing interest