Risk Measure Analysis Of Financial Market Based On VaR Model

With the information and technology developing continuously, the pace of the global economy and financial integration are accelerating, and financial market volatility and risk is also increasing, which bring exhaustive effect on countries and companies, so the financial risk management has been paid more attention by financial institutions and investors. The techniques of measuring and analyzing risk are developing very rapidly. VaR (Value at Risk) originated in the1980s, and it is a statistical approach to measure financial market risk, which is widely used in banks, securities companies, commodities and other trade organizations. The main feature of this research is to measure and analyze VaR of the market risk in the financial markets, which has an important position in modem financial risk management. The application of VaR is gradually developing and consummating in China, and there are many studies on this method. Its main contents are predicting VaR of financial losses that may arise in the future, and identifying how the distribution of risk factors affecting the distribution of portfolios, and providing reliable information for the portfolio optimization and risk management.The traditional VaR has some disadvantages, such as the assumption of normal distribution, dissatisfying consistency axiom, the lack of considerations of extreme events and tail dependence between financial assets and so on. These shortcomings will affect the risk and effects of portfolio investment, and affect the accuracy of VaR. The Copula theory has theoretical advantages compared to traditional methods, and Copula models can organically integrate the research about correlation and related patterns. We can capture the nonlinear, asymmetric and tail dependence of variables by Copula function. Copula theory provides a new way of thinking to analyze financial time series of multivariate variables.The main idea of this research is establishing GARCH model to measure financial market risks through considering the volatility clustering of the financial markets, introducing Copula function aiming on the shortcomings of VaR, and assuming that variables are obeying normally distributed, t distribution and GED distribution because of leptokurtosis of financial data in the marginal distribution. We can determine the optimal marginal distribution model and the optimal Copula function through a series of tests. Combining optimal marginal distribution GARCH model with optimal Copula function, then we can measure VaR of CSI index portfolio by the Monte Carlo method.This paper is divided into six parts, and the main contents are as follows.The first chapter is an introduction, which make a brief overview of the background and significance of the research, review research status about VaR and measuring VaR by Copula approach of local and abroad, and summarize the contents, methods, innovation and disadvantages of this study.The second chapter is the theory about financial risk. It briefly generalizes the definition of financial risk and traditional methods of financial risk metrics.The third chapter is the theoretical basis of VaR. Firstly, the paper totally discourses the realistic background, concepts about the VaR theory, and introduce holding period and choices of confidence level about VaR. Secondly, it discusses three traditional methods to measure VaR, including Delta-normal approach, historical simulation approach, Monte Carlo simulation approach. Then, it introduces the model for calculating the VaR and method validation. Finally, we make a simple introduction to the advantages and disadvantages of VaR.The fourth chapter is the related knowledge about Copula function, which explains the definition, related properties, common functions and related theorems of Copula function in detail. Then it makes a brief description of the correlation measure of the Copula, namely Kendall’s tau and Spearman’s rho. Meanwhile, the tail dependence is briefly described. Finally this paper elaborates the measure of VaR based on Copula-GARCH model and Monte Carlo simulation approach.The fifth Chapter is empirical analysis. This paper constructs an equally weighted portfolio of CSI Index from January1,2008to April3,2013with1278sets of data.Taking the logarithm yields to make empirical analysis, first we estimate the marginal distribution model by analyzing data characteristics, volatility, stability, autocorrelation-partial autocorrelation and ARCH effects, then conduct regulation respectively in the normal distribution, t distribution, GED distribution, and determine the optimal model named the GARCH (1,1)-t model through further analysis.We can determine that the t-Copula function is the most appropriate Copula function to describe the correlation between variables through the K-S test, and use rank correlation coefficient for verification. This article will measure the risk of the securities market including CSI Index by combining t-Copula Functions with GARCH-t model based on Monte Carlo simulations approach. From the empirical results we can see that measuring VaR with the t-Copula-GARCH-t model is effective.The sixth chapter is conclusion. First it summarizes this paper, including the principles, methods and models, and then presents several policy recommendations for future research.The innovation of this paper is that we establish Copula-GARCH model to calculate VaR, and apply the model to the analyze the risk measure of the CSI index portfolio, which greatly improve the measurement accuracy of VaR. Monte Carlo simulation approach is based on historical data, and it uses a series of testing and statistical methods to find the best distribution function which can better portray the characteristics of the data, so that this approach simulates the better effect.The shortage of this paper is that the investment ratio of the CSI index portfolio is fixed without dynamic investment optimization. Copula estimates difficultly, so binary Copula model uses quite often, but binary Copula obviously can’t be better used for analysis of multiple assets.