Securities Market Risk Measurement Based On VaR And CVaR Models And Empirical Research

The stock market risk is the main risk which the negotiable securities investors face in the investment process, is one of focal point questions which the multitudinous investors and the theory researchers pay attention, also is the securities regulatory authorities in all countries to management focus. Stock market risk measurement as the core of risk management and prevention, it has directly decided the risk management and the guard is effective. Therefore, the risk management of the securities market research is of great theoretical and practical value.Based on the various stock market risk measurement methods for system analysis and research, and ultimately selected VaR (Value at Risk) and CVaR method as the research methods. The article highlighted the VaR model and the basic principles of CVaR and VaR, as well as the methods of calculation, and take from 2002 to 2007 the Shanghai Composite Index and Shenzhen component index as the research object, has carried on the empirical analysis using VaR and the CVaR methods in the securities markets of our country, and the sequence proceeds stability and relevance. Empirical Test achieving better results.This article fully in the stock market than on the basis of the data carries on processing through Eviews statistical analysis software and the econometrics methods, Empirical Analysis of the results obtained: The financial assets yield sequence distribution of the characteristics of peak thick tail, and has obvious GARCH effects and the release lever effect, its volatility has the gathered and when degeneration (conditions heteroscedasticity). Based on the normal distribution of CVaR estimates may underestimate the risks, based on the t distribution of CVaR estimates may overestimate risk, based on the generalized error distribution (Generalized Error Distribution, or GED) of CVaR the results more accurate estimates. Empirical research shows that the GARCH race model can yield a good characterization of the residual sequence heteroscedasticity.