Research On The Interest Rate Risk Of Commercial Bank Basing On VaR Model

The management of interest rate risk in commercial banks has become a challenging topic in financial field. Because the development of the interest rate marketization in our country is relatively late, and commercial banks are not independent in related policy as well. Most of the commercial banks don’t have complete interest rate management department which is full of rich experience to forecast, manage and evade interest rate risk. As the requirement of Basel agreement improves continuously, and our government promotes market-oriented interest rate risk reform, it is important and practical to improve the interest rate risk management and measurement technology in the process of commercial bank operation. Especially on July20th this year, People’s bank of China decided to open up the lending rate control of financial institutions, which brings great challenges to interest rate risk management of our commercial bank. The VaR model as the mainstream risk measurement tool in western country is widely used in securities and banking. This model has become an effective tool in risk prediction and prevention for financial institutions and regulation department. In this paper, I make a measurement and prediction in interest rate risk using VaR model, in order to establish a model which describes the characteristics of financial data better in accordance with Chinese national condition. In my opinion, this paper has vital theoretical and practical significance in protecting and controlling interest rate risk of commercial bank.Basing on the SHIBOR overnight data from December31,2009to December31,2012, this paper estimates and predicts interest rate risk by establishing GARCH-SKST model, E-GARCH-GED model, TARCH-GED model and PARHC-GED model, and combining different level of significance. The first part is introduction, which contains the background of this paper and research status in risk measurement and VaR method of Chinese and foreign scholars, as well as the innovation and shortcoming. In the second part, it shows the basic information of the interest rate and problems which commercial banks face. Meanwhile, it illustrates the different measurement methods as well as the advantages and disadvantages. By comparing to the advantages and disadvantages of various measurement tools, it is concluded that the VaR method is the more intuitive risk indicator, which is also easy to master and operate by regulators. The third part is the empirical part. By contrasting, it is concluded that GARCH-SKST model can make a better combination in skewness and kurtosis of financial data. Because of the higher accuracy in risk measurement, it has more advantage for commercial bank in risk controlling and prediction. The fourth part is the conclusion which puts forward new proposal to perfect our commercial bank interest rate risk system.In the empirical part of this article, because influence of interest rate risk which is effected by the long and short of SHIBOR is opposite, it divided the data into two parts. The first750data are used to compute short and long VaR value, and the left200data are used to predict and test the validity of the model by Kupiec inspection and relative error test. By data fitting, GARCH-SKST model can forecast effectively under three kinds of significant level. Under the confidence level of99%, PARCH-GED model is failure in short, TARCH-GED model underestimate the interest rate risk in long, because it predicts there is no failure date. EGARCH-GED model is totally ineffective in long and short under95%confidence level. Meanwhile, it can’t be effective under90%confidence level in prediction for long.There are two main conclusions in this paper. On one hand, Peak, thick tail and asymmetric is characteristics of financial data which don’t meet the normal distribution. The key point for commercial bank to predict and control interest rate risk effectively is to find the suitable distribution of risk factors and valid model which is fitted financial data better. Although some scholars hold the point that GED distribution can describe the thick tail, GED itself is symmetrical distribution which can’t combine the asymmetry of financial data. While SKST distribution can solve the data peak, thick tail and asymmetric problem. At the same time, choosing the appropriate confidence level is vital for risk forecast and management.On the other, by establishing the asymmetric model and data fitting, it shows SHIBOR data has a "lever" which indicates the volatility of interest rate is impacted stronger by good news than bad news. Therefore, in the future risk prediction and controlling process, I suggest that, according to the actual situation, our commercial bank may use VaR model and SKST distribution to describe risk factors, so that it is better to improve risk measurement accuracy.