Authors: Julien Chevallier
Addresses: IPAG Business School, IPAG Lab. 184, Boulevard Saint-Germain, 75006 Paris, France; Université Paris 8 (LED), 2 rue de la Liberté, 93526 Saint-Denis, France
Abstract: Using tick-by-tick data for the WTI crude oil (2001-2010) market, this paper relies on the recent bivariate model by Maheu and McCurdy and compares the forecast accuracy of the density of returns with the HAR-RV model at different horizons upto 60 days. Our results provide evidence of the incremental information for density forecasting embedded in intraday data when the model is compared with the univariate EGARCH model. Turning to the comparison between the forecasting power of the realised volatility, which includes the total quadratic variation, vs. Bipower Variation (BPV) and median realised volatility, which only estimates the diffusive component, it is shown that the additional information contained in the jump component is significant on average. The findings for WTI crude oil futures confirm the importance of considering the continuous/jump decomposition for density forecasting.
Keywords: WTI crude oil market; density forecasting; bivariate modelling; market returns; crude oil futures; futures markets; crude oil markets; volatility.
International Journal of Global Energy Issues, 2015 Vol.38 No.4/5/6, pp.201 - 231
Available online: 20 Jun 2015 *Full-text access for editors Access for subscribers Purchase this article Comment on this article