Estimating and Testing Risk Approaches: A Technical Analysis using Affine Term Structure Models, Monte Carlo Simulation and GARCH Method

Sammanfattning: This paper investigates if the Log-Normal Mean-Reverting Ornstein-Uhlenbeck spot price (LNMROU) and the Vasicek (1977) process can forecast Value-at-Risk (VaR) using the Monte Carlo method. The results from LNMROU are validated against Delta-Normal-GARCH (DNG) and Historical Simulation (HS) which are well known approaches for VaR estimations. The backtested results indicated that HS and DNG are good measures for VaR-estimation and that LNMROU failed in capturing price changes in the stock index market. The Vasicek (1977) proved to be a good model for forecasting VaR.