The VaR Study Based On Non-symmetric GARCH Family Model And FHS Technology

Under the global background of financial crisis, the market crisis attached great importance to by both financial investment institutions and individual investors gains increasing attention from the population. Compared with developed countries in Europe and the United States in terms of the financial sector, China is in low leveled financial openness, but the present financial crisis still affects China's financial industry a lot. In this circumstance, it is quite necessary to study into the risk management in the capital market in our country. VaR, one of the instruments measuring market risk, if made use properly, may well expose the degree of market risk and help the financial institutions and individual investors manage the position risk of the capital. This thesis, based on the research into the VaR of SSE composite index and HuShen-300 index, proposes some suggestions on how to manage the market risk when the fund companies manage the index fund or the investors invest the stock index futures market.The thesis is divided into two parts. The first part studies the characteristics of the rate of return and the rate of volatility in our country's stock market. The second part is to study the present index in terms of the risk measurements of VaR by using the results in the first part and also FHS technology. Later in this part, the author makes a comparative study between this kind of methodology and the traditional VaR method. The study in the first part found that there is a clear non-normal characteristics about china's stock marker index ruturns.Compared with foreign markets, there is also obviously "Peak-Thick tail" characteristics in China markets.The t distribution and GED distribution is better to explain this characteristics. And, there exists an obvious volatility clustering and non-symmetrical effect in,while HuShen-300 index exits only volatility clustering effect. Based on the normal distribution, t distribution and GED distribution, these two features about voltility of the SSE composite index can be well explained by t EGARCH and TARCH models, and the GED-EGARCH model fitting results are the best.as to HuShen-300 index, the t-GARCH model have a good work to explain the features of its volatility.The second part of this thesis first introduces the calculation methods of VaR, mainly focusing on Historical Simulation Method (HS), Monte Carlo Simulation Method and FHS technology. Differing from the traditional calculation method for VaR, FHS technology will be used with GARCH models together. GARCH model s reflect the features of existing fluctuant rate, and forecast the future rate of return and fluctuant rate through FHS technology. The result from this is more realistic a nd further meets the requirement of risk management. The testing results of Kupie c failure reveal that all of these three methods can reflect the market risk respective ly, but the results from Historical Simulation Method and Monte Carlo Simulation Method are too rough, with the failure days too close to the upper limit. However t he result from FHS technology based on GARCH models are better, with less failu re days. When the market fluctuates too strongly, these three methods cannot reflec t the market risk completely. The Var result will underestimate the risk. A severe an d abnormal fluctuation...