(Hot New) Sexy Japan

VaR is central to financial risk management. In this study we use three methods to calculate VaR, namely the industry standard RiskMetrics, Monte Carlo simulation and Filtered Historical simulation. To model the return we use GARCH and NGARCH processes augmented with normally distributed jumps. We examine the implication for In-Sample testing in the presence of jumps and find that the inclusion of jumps generally leads to fewer VaR violations. This finding is consistent with the findings of other researchers. However, at higher confidence intervals and long horizon dates all methods and models fail In-Sample testing.