Apply The VaR To Measure The Risk

In today’s market economy, competition in the financial industry is very fierce and has the greatest risk. Financial activities associated with financial risks all the time.In today’s market economy with economic globalization and financial liberalization, how to measure and control the financial risks, protect the financial security, has become the primary issue of financial risk management.This paper aims to build models to measure and compare the risk of Shanghai Composite Index and Shenzhen Component Index, then determine the model whether is valid or not by the test of effectiveness of model.If the model is valid, then compare and analyze the model that passed the test to identify the most appropriate one which can best measure the risk of Shanghai Composite Index and Shenzhen Component Index.This paper calculates VaR (Value at Risk) through the variance-covariance analysis and quantile regression, and compares the merits and demerits of the model by Kupiec test. Choose Shanghai Composite index and Shenzhen Component Index as the research object and divide the sample data into the sub-samples of estimation sample and evaluation sample.This paper chooses Shanghai Composite index from Demcember 19th,1990 to March 24th,2010 totally 4722 data as the estimation sample and sets aside from March 25th,2010 to May 13th,2014 totally 1000 data as th evaluation sample. Through normality test, stability test, ARCH effect test, heteroscedasticity test and correlation test for the evaluation sample logarithmic return rate srquence finds that it does not obey the normal distribution, but has the characteristic of spikes tail, volatility clustering and the non-symmetry. In the variance-covariance analysis method,using the general GARCH family models and GARCH-M family models at different confidence levels (95% and 99%) and under the conditions of normal distribution, t distribution and GED distribution calculates the mean VaR, standard deviation of VaR, failure frequence and failure rates of VaR.In the quantile regression method, build the five-order lag autoregression model to predict the expected return rate and use the average rate of return on behalf of the minimum return rate,then according to the definition of VaR, that is VaR= E(R)-R’to calculate the VaR. Finally, uses failure frequency test method that put forward by Kupeic to test the efectiveness of the model.That is to say,compare the number of failures with the table of accepted domains given by Kupiec,if the number of failures in the accepted domains,indicating the model is good, vice versa, the model is not.More precisely,if the number of failures in the left side of the accepted domains,indicating that the model overestimates the probability of occurrence of loss, it is conservative.If the number of failures in the right side of the accepted domain,it shows that the model underestimates the probability of occurrence of loss.By the results of variance-covariance analysis and quantile regression finds that the risk of Shanghai stock market is slightly higher than the risk of the Shenzhen stock market.For the Shanghai Composite Index, under the 95% confidence level, EGARCH-M with t distribution model is better, under the 99% confidence level, GARCH-M with GED distribution model is better. For the shenzhen Component Index, under the 95% confidence level, PARCH-M with t distribution is better, under the confidence level of 99%, GARCH model with GED distribution is best. Quantile regression passes the test for both the Shanghai Composite Index and Shenzhen Component Index under the confidence level of 95% and 99%.