Study On The Measurement Of Interest Rate Risk Of Commercial Banks Based On GARCH-VaR And GARCH-CVaR Model

Along with the advancement of globalization,management pattern of commercial Banks' in our country have realizeed the international standards and commercial Banks in our country were to fully open,global financial integration has accelerated the degree and process of China's liberalization of interest rates,Changes in interest rates will be more depends on the market rule of factors.In 1996,the central bank in our country has opened the interbank lending rates,which formally opened the prologue of the reform of interest rate liberalization in our country,after 20 years of the reform of interest rate liberalization,interest rates are expected to have an increased susceptibility to financial environment,commercial banks in our country have basically realized interest the reform of rate liberalization.With the full realization of Chinese market-oriented interest rates,interest rate clearly increased violation frequency and range.,most of the assets of commercial banks were financial assets,whose value will change while the interest rate violate.,the interest rate risk is having a greater influence oncommercial bank,more and more financial assets will be in the interest rate risk,lead commercial banks facing more and more interest rate risk,then the rate risk management is becoming more and more important for commercial banks in our country.The core of the interest rate risk management is to evade,eliminate or transfer the risk of earnings or the value,which caused by interest rate changes,in order to effectively supervise interest rate risk of commercial banks that control and prevent interest rate risk,first and foremost task is to have a good method to measure interest rate risk.Then how to apply scientific methods to measure interest rate risk has become the core issue of Chinese commercial banks.Therefore researching the operating mechanism,the characteristics and the interest rate risk management of SHIBOR has important theoretical significance and practical significance.We Selected the O/N and Week lending rates data of SHIBOR,ranged from 2011.1.4 to 2016.12.31,as the object for our study,On the basis of existing methods of VaR model,measure the risk of CVaR model is introduced into the commercial bank interest rate risk measurement,we established a number of VaR and CVaR methods based on GARCH models to measure the risk of SHIBOR at different confidence levels.Through the research we can draw some conclusions as follows.First,the SHIBOR market presence of volatility clustering and rush fat-tailed features,the O/N and Week lending rates data of China's SHIBOR logarithm yield sequence sequence has the relevance and conditional heteroscedasticity effect,which means the SHIBOR has not not implemented completely the liberalization of interest rates.Second,GARCH models can effectively characterize the earnings volatility of the O/N and Week lending rates data of China's SHIBOR sequence,the General error distribution is more suitable for characterizing the distribution of China's SHIBOR sequence than the distribution of Normal and T,while the T-distribution is not proper in characterizing the distribution of China's SHIBOR sequence.Furthermore,model estimation result shows O/N and Week lending rates data of China's SHIBOR sequence has reverse leverage effect and long memory effect,the good news has greater impact on the interest rate volatility than bad news has.Third,CVaR can measure a wider coverage of the left tail risk than VaR,while the VaR model fails to measure effectively the risk of SHIBOR,the CVaR model overcome the defect of the VaR model and can effectively measure the real risk of loss,so we can choose to use CVaR model instead of the VaR model for risk measurement.Fourthly,from the study of O/N lending rates data of China's SHIBOR sequence we can find,at the 95-percent confidence level,all models passed the Kupiec test,but among them,ARMA(2,1)-EGARCH(2,1)-N and ARMA(2,1)-EGARCH(2,1)-G can be better to simulate the fluctuations in earnings of the of O/N lending rates data of China's SHIBOR sequence,while the model prediction effect of ARMA(2,1)-GARCH(2,1)-N and ARMA(1,1)-GARCH(2,1)-M-G are the worst of all.At the 99-percent confidence level,ARMA(1,1)-EGARCH(2,1)-G and ARMA(1,1)-TARCH(1,2)-G are better better than any others at measuring the interest rate risk of O/N lending rates data of China's SHIBOR sequence,while the other models failed the Kupiec test.,can not accurately measure the interest rate risk of the sequence,the CVaR model can overcome the defect of the VaR model and can effectively measure the real risk of loss,so we can choose to use CVaR model instead of the VaR model for risk measurement.Fifthly,from the study of Week lending rates data of China's SHIBOR sequence we can find,at the 95-percent confidence level,except of AR(1)-PARCH(2,2)-G model,VaR under other models all pass the Kupiec test,which proves good risk-management-effect,ARMA(2,1)-EGARCH(1,2)-N and ARMA(2,1)-TARCH(1,1)-G are better better than any others at measuring the interest rate risk of Week lending rates data of China's SHIBOR sequence,while the ARMA(2,1)-TARCH(2,1)-N model is the worst of them.At the 99-percent confidence level,only VaR calculated under ARMA(2,1)-EGARCH(1,2)-N,ARMA(2,1)-EGARCH(1,2)-G,ARMA(2,1)-TARCH(1,1)-G and ARMA(2,1)-GARCH(1,1)-G passes the Kupiec test,among them,ARMA(2,1)-EGARCH(1,2)-G and ARMA(2,1)-TARCH(1,1)-G are better better than any others at measuring the interest rate risk of O/N lending rates data of China's SHIBOR sequence,while the AR(1)-PARCH(2,2)is not suitable to measure the interest rate risk of the sequence.while the other models failed the Kupiec test,can not accurately measure the interest rate risk of the sequence,the CVaR model can overcome the defect of the VaR model and can effectively measure the real risk of loss,so we can choose to use CVaR model instead of the VaR model for risk measurement.