This investigation estimates the Value-at-Risk (VaR) for the return on crude oil for WTI (West Texas Intermediate) via the RiskMetrics and AR-GARCH models using the rolling bootstrapping method. The empirical results demonstrate that the bootstrapping method is superior to no-bootstrapping method. The average quadratic loss function in bootstrapping method is significantly lower than in no-bootstrapping. It is evidence that the no-bootstrapping forecast is too confident and the bootstrapping forecast is too cautious for 1% VaR estimates. Moreover, the AR-GARCH model performs better than the RiskMetric model in no-bootstrapping, but not in bootstrapping. Finally, the longer the observation periods, the forecasting performance is improving.
International Research Journal of Finance and Economics 40, pp.7-19