19 - A Copula Approach to Dependence Structure in Petroleum Markets

Abstract

A copula-generalized autoregressive conditional heteroskedasticity (GARCH) model is used to investigate the conditional dependence structure in the petroleum markets over the recent period 2005–2011. Our results show evidence of a significant and symmetric dependence between the prices of crude oil, gasoline and heating oil. They also indicate that Student- t copulas provide the best fit to the data and that the use of a copula-GARCH model leads to an improvement of the out-of-sample value at risk forecasting accuracy