All countries consume crude oil or oil products. Both producers and consumers are highly concerned about crude oil prices. The crude oil prices are being directly affecting by several factors such as economic, political, geopolitical, technological, oil reserves, available stocks and weather conditions, among others. On other hand the crude oil prices fluctuations influence directly the world economy. Compared to the financial assets the crude oil prices have had an elevated volatility in recent years. Therefore, studies of crude oil prices movements and co-movements are highly complex. So the academics and practitioners are developing many studies about themes related with crude oil prices. The economic agents indirectly involved with crude oil negotiations, such as the planners of firms or governments, are looking for related petroleum prices forecasting models construction studies, while the agents directly involved are looking for the hedge strategies studies as well. The hedge strategies allows negotiators that have short and long positions in the market protection against prices fluctuations. This paper examines the performance of two bivariate volatility models for the crude oil spot and futures returns of the Western Texas Intermediate – WTI type barrels prices. Besides the volatility of spot and future crude oil barrel returns time series, the hedge ratio strategy is examined though the hedge effectiveness. The hedge strategy, or the hedge ratio, is implemented using two methodology approaches presented and compared in this study. Thus this study compares hedge strategies built using two methodologies applied in the variance modeling of returns of crude oil prices in the spot, and future markets, and covariance between these two markets returns, which correspond to the inputs of the hedge strategy shown in this work. These two bivariate models are a classical model and a Bayesian model. The classical model used is a bivariate Generalized Autoregressive Heteroskedasticity (GARCH) in a Diagonal VECH representation, or specification, while the Bayesian model used here is a hierarchical model with a random effect. Those models are appropriate to estimate volatility, so they are suitable for refining hedge strategies and to reduce potential market risk. The methodologies used here take into consideration assumptions for the returns distribution more realistic than other methods that have been used in the financial literature: the heteroskedasticity and the non normality. The data used is logarithm returns of daily prices quoted in dollars per barrel from November 2008 to May 2010 for spot and future contracts, in particular the June contract.
Keywords: Crude Oil Prices; Future Market; Volatility Models; Hedging Effectiveness
Biography: DSc in Production Engineering (Operational Research and Statistics) from Federal University of Rio de Janeiro. His research areas are applied econometric methods and operational research in finance. He is currently associate professor in Federal University of Rio de Janeiro at the Industrial Engineering Department of the Polytechnic School. He is involved of the following scientific societies in his country: Brazilian Statistics Association (ABE); Brazilian Society of Operational Research (SOBRAPO);and Brazilian Production Engineering Association (ABEPRO).