The Shanghai A-share Index Var Model Comparison And Empirical Research

With the rapid development of globalization of economy and finance, the financial stability has descended, and the financial market has become much fluctuant than ever before. Under this condition, the management of financial risks becomes more and more important, and it becomes the key issue of financial administrations and financial institutions.These days, Chinese financial institutions use VaR to manage the market risks. There are many models to calculate the VaR, but there isn't the answer that which model is suit for Chinese rules of market run.To use VaR approaches to estimate market risk of a portfolio, the key issue is how to depict the distributing of financial and capital yield. The paper makes some empirical analysis of the statistical properties of A-share of Shanghai Stock Exchange. Then based on an analysis of existed VaR models, the paper proposes a new VaR model which based on kernel estimation and local linear estimation. The main results of this paper can be summarized as follows:1. The paper begins with the research of the properties of A-share of Shanghai Stock Exchange, and then analyzes the elementary statistical properties of the fluctuation of the stock yield with some descriptive statistical variables, such as mean, standard deviations, skewness, and kurtosis, heavy tail and normality of the series of yield. The paper concludes that the non-normality of the series of yield of A-share of Shanghai Stock Exchange has a proper of peaked kurtosis and heavy tail, the series of yield has a very weak relation, and the mean of the series of yield has2. Having compared the exited VaR models, the paper brings forward to estimate VaR by using the model of kernel estimation and local linear estimation.3. The paper makes some empirical analysis of the series of yield of A-share of Shanghai Stock Exchange, and calculates the value of risk of the series of yield with the exited VaR models, kernel estimation and local linear estimation. The paper verifies all kinds of VaR models with the back-test which stipulated by Basel Committee, appraises the advantages and disadvantages of all kinds of VaR models. The conclusion of the paper is that the VaR models of kernel estimationand local linear estimation can depict the risk curve of the series of yield of A-share of Shanghai Stock Exchange very well, but the model of kernel estimation maybe underestimate the market risk, and that the model of local linear estimation can overrate the market risk in some sort.At the same time, the real cost which has exceeded the value of VaR in the verified samples is accorded with the corresponding left probability.