Research On Inter-bank Interest Rate Risk Measurement Of Commercial Bank Based On Var Model

Positions in risk assets expose commercial banks in China to interest rate risk,which refers to the possibility of loss in the market value of the positions dued to adverse changes in interest rate. Along with the liberalization reformof interest rate in our country, especially after the explosion of financial crisis in 2008,the fluctuation of interest rate is becoming more and more frequently and sensitively. Interest rate of inter-bank, which has become the most marketizational interest in China, is more sensitive increasingly. The inter-bank interest rate risk which our commercial banks faces is stronger gradually. How to improve the level of interest rate risk management attracts more and more attentionof commercial bank in China. The key to measure of interest rate risk accurately is the which measurement to use. Currently, the most widely used international risk measurement of interest rate is VaR model. Whether VaR model can predict the interest rate risk accurately is very important tocommercial banks . Based on the research of foreign and domestic researchers, we introduced the ARIMA-GARCH models into the research on measurement of interest rate risk of inter-bank in the paper.Firstly ,we analysed the definition and form of interest rate risk, and characteristics of inter-bank interest rate risk. We take a comparative analysis on traditional static measurement of interest rate risk and VaR model. Connected with the characteristics of inter-bank interest rate risk nowadays, we think that VaR model is fit for measurement of inter-bank interest rate risk. Secondly, this paper will introduce the class of ARMA-GARCH model, and propose an idea on evaluating the validity of the VaR model basedg on domestic and foreign research. Thirdly, this paper will chose Chinese interbank market data as the research object and introduce the ARMA-GARCH models and the methods recommended by the Basel II into measurement of the interest rate risk. CombineD with different residual distribution hypothesis, we set up 12 specific methods of interest rate models to estimate the risk. Unconditional coverage and conditional coverage back testing are used to analysized the accuracy of VaR models estimation and independence of exceptional value in the paper. We also made a comparative analysis on the specific measurements of VaR models by Back-testing. The results showed that under 95% confidence level, under GED distribution the GARCH models are more effective than the normal distribution assumption of the GARCH models, and asymmetric GARCH models estimation are more effective than the symmetric GARCH models. Through GED distribution tail parameter and asymmetric GARCH model leverage analysis, we find that income of inter-bank lending distribution series appear "fat tail", And fluctuation of the income has a "negative leverage effect" feature. Under 90% confidence level, all measurements of VaR models overestimate the risk of interest rate, while measurements under the GED distribution of asymmetric GARCH models and historical simulation method are more accurater relatively. Under 99% confidence level, all models have passed the mixed Kupiec test, but through the view of exception, the GARCH models under normal distribution was more validity than the GARCH models under GED Distribution.