Title: Evaluation of VaR models' forecasting performance: the case of oil markets

Authors: Med Imen Gallali; Raggad Zahraa

Addresses: Business School of Tunis, ESCT, La Manouba, Tunisia. ' Management School of Tunis, ISG, Tunis, Tunisia

Abstract: This paper highlights the importance of Value-at-Risk (VaR) methodology in managing oil market risks of three international crude oil rates (Brent, OPEP and WTI). Comparing between the conventional VaR models proposed by the literature (non-parametric models, hybrid models and conditional and unconditional parametric models), we point to the supremacy of conditional GARCH-type models (GARCH-T) or hybrid models (Filtered Historical Simulation). In contrast, the unconditional models or those based on the normality hypothesis are the least performing. In general, there is a tendency to prefer the conditional models as they allow integrating the dynamic nature of volatility and distributions flat tails.

Keywords: risk management; oil markets; VaR; value-at-risk; variance-covariance method; historical simulation; conditional models; backtesting; forecasting performance; crude oil rates; hybrid models; volatility; flat tails.

DOI: 10.1504/IJFSM.2012.046935

International Journal of Financial Services Management, 2012 Vol.5 No.3, pp.197 - 215

Received: 04 Apr 2010
Accepted: 05 Feb 2011
Published online: 29 Nov 2014 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article