Realized volatility models and alternative Value-at-Risk prediction strategies》.pdf

Economic Modelling 40 (2014) 101–116 Contents lists available at ScienceDirect Economic Modelling j ournal homepage : www.el /l oc ate/ ecmod Realized volatility models and alternative Value-at-Risk prediction strategies Dimitrios P. Louzis a,b,⁎, Spyros Xanthopoulos-Sisinis a, Apostolos P. Refenes a a Financial Engineering Research Unit, Department of Management Science and Technology, Athens University of Economics and Business, 47A Evelpidon Str., 11362 Athens, Greece b Bank of Greece, Financial Stability Department, 3 Amerikis Str., 105 64 Athens, Greece a r t i c l e i n f o a b s t r a c t Article history: We assess the Value-at-Risk (VaR) forecasting performance of recently proposed realized volatility (RV) models Accepted 7 March 2014 combined with alternative parametric and semi-parametric quantile estimation methods. A benchmark inter- Available online 21 April 2014 daily GJR-GARCH model is also employed. Based on four asset classes, i.e. equity, FOREX, fixed income and com- modity, and a turbulent six year out-of-sample period (2007–2013), we find that statistical accuracy and regula- Keywords: tory compliance is essentially improved when we use quantile methods which account for the fat tails and the High frequency intra-day data Filtered historical simulation asymmetry of the innovations distribution. In particular, empirical analysis gives evidence in favor of the skewed Extreme value theory