Considering the recent financial crises, risk management becomes a more important issue. Value-at-risk (VaR) has been a popular tool to measure financial risk. Disputes about the appropriateness of how to calculate the VaR have arisen more seriously since the market for subprime mortgage securities collapsed. These problems include the presence of fat-tailedness and its impact on the VaR. This paper develops two new methods based on a hidden Markov model (HMM) in which the distribution function is not deterministic but rather stochastic. We evaluate and compare the performance of several di erent VaR methodologies for two testing sample periods one of which covers particularly the Global Financial Crisis. The results show that the market’s extreme movements can be evaluated more successfully with the HMM based methods.
Keywords: Value-at-risk, Hidden Markov model, Monte-Carlo, Risk management
