| With the globalization of economy, integration of finance, intensification of competition, relaxation of restriction, and innovation of technology, basic and structural changes have taken place in the world financial market, with an enlarging scale and increasing efficiency. Faced with stronger fluctuation and more system risks, management and measurement of the financial risks have become key abilities for financial institutions and industrial and commercial enterprises in competition and also make the major content in finance engineering and modern finance theories. Financial risk management is a complex process, consisting of risk identification, risk measurement, decision making and implementation of risk management as well as risk control, among which risk measurement plays the most important role. As traditional risk measurement methods, the applications of variance and coefficient methods fail to reflect the direction of the return rate deviated from the expectation and cannot provide an accurate measurement of the lost. However, VaR models have been proved to be an effective method people are looking for as a way to measure the downside risk, focusing on the left side of the distribution of probability of the return rate. A formal definition of VaR is the predicted maximum lost of one portfolio within a given period of time, with normal market conditions and confidence level. In another word, the probability of the lost which exceeds VaR equals the given probability with a given time span and market condition. Based on a precise statistics theory, VaR models are able to quantify the general market risk of various financial tools and portfolio into just a number to briefly and clearly indicate the degree of the market risk. Due to these advantages, VaR models have been widely recognized and supported in the field of finance globally. There are two methods in the VaR measurement. One is to analyze on the basis of partial estimation which is represented by variance-covariance method, the other is on the basis of holistic estimation characterized by historic-simulation method, Monte Carlo simulation and stress testing method. This article will take the method of analysis as focus of discussion. According to the difference between the approximate relationships of portfolio and market factors, the analysis method can be divided into two types of Delta and Gamma. However when the assets managers want to measure the market risks they are facing in making decision in financial practice, it is not enough to know only about the general risk condition of the assets portfolio, each component in the portfolio and what the adjustments and changes of the components will do to the portfolio will have to be taken into consideration. So in addition to VaR, marginal VaR, component VaR and incremental VaR are also introduced in this article. These are some key concepts of risk information and calculating methods which will be helpful in identifying the source of the risks in whole risks exposure, improving the integral risk condition, assessing investment opportunities, analyzing the influence of asset adjustment on the combination risk and setting the position quota. Considering the heteroskedasticity of financial time series, we also introduce the GARCH(1,1) model into the VaR models to reflect the dynamic influence on the risk measurement caused by new information and new impacts. The variance-covariance method always supposes that the return rate abides by normal distribution. Actually, according to numerous study, the distribution of the data with high frequency features pyramid with characteristic of volatility clustering, high peak and fat tail. Six different methods have been adopted to deal with the questions mentioned. They are: 1.using EWMA and ARCH model to deal with questions concerning volatility clustering and to estimate the volatility effectively; 2. with the help of different distribution of pyramid characteristics, such as t-distribution, mixed formal distribution and generaliz...
|
PDF Full Text Request