Leverage effect in energy futures revisited

Abstract

The objective of this paper is to replicate the results in Kristoufek (2014) on the leverage effect in energy futures and to analyze its robustness to both the methodology and the type of returns used. We first apply correlation-based tools for detecting both conditional heteroscedasticity and leverage effect. Then, we estimate asymmetric and long memory GARCH-type models using the data provided by Kristoufek (2014) by considering different software and the possibility that innovations follow a non-Gaussian distribution. Our findings confirm most of the results in the replicated paper. In particular, we can strongly confirm there is a significant leverage effect in the return series of WTI (West Texas Intermediate) and Brent crude oils. For the heating oil and the natural gas series, the statistical significance of the leverage effect depends on both the methodology and the type of returns used.