VaR And CVaR Estimation Method As Well As Risk Management

As financial market develops, risk grows much more complicated. Risk is future's uncertain loss, that is, volatility of the intending income. And risk can be caused by the volatility of both the unpredictable assets and the debt. How to measure the risk accurately is worthing of further research. Financial market risk is the most important one among all the risks.The emergence of Value at Risk (VaR) enables us to quantify the maximum loss of financial portfolios in certain period. And it has become the fundation of financial risk measurement system up to now. Firstly, the estimating precision of VaR has to be improved. Especially when the financial market grows more complex, an accurate VaR value could help company better evade the risk and distribute the capital better. Secondly, current studies on VaR focuse on estimating short term VaR, for example, one day. However, usually there are many long term portfolios in insurance and investment field. Because of the absence of low frequency data, methods in literatures can not be used to estimate the long term portfolios directly. Therefore, the study of methods of estimating the long term VaR is very important. Finally, the asset or portfolio is considered separately in traditional methods. In fact, because of the complexity of the financial market, the risk of the financial asset is usually influenced by other factors. And those factors that influence the assets should be considered synchronously when VaR is estimated. The so called Conditional Value at Risk was introduced to solve such problem.Several main methods were presented to deal with the problems listed above in this paper. The structure and the innovation of this dissertation are described as the follow:In the first part, the estimating methods of VaR were discussed. In chapter 1, the definition and traditional estimating methods were summarized. In chapter 2, Bootstrap and random weighting methods were applied to estimate VaR, which helped to get more accurate estimator of VaR. Since the distributions of financial data are heavy-tail generally, the distorted function was introduced to modify the normal...