The Shibor’s Market Risk Study Based On The Nonlinear Time Series Analysis And Exterme Value Theory

Interest rate marketization is becoming more and more popular, so we have tofocus on the market risk of interest. Choosing the appropriate model to measure theShibor’s risk has important significance because Shibor is an important Chibor inChina.Focusing on Shibor, one of the financial time series, we not only consider the peakfat-tail and volatility clustering characteristic，but also note that it has asymmetry,cyclicity and jumping characteristic. So we use extreme value theory and GARCHmodel to describe the peak fat-tail and volatility clustering characteristic and use TARmodel, one of the non-linear time series model to describe the other characteristic.Combining the three model, we get TAR-GARCH-POT model to simulate the Shibor.Comparing with the AR-GARCH-POT model and declustering in extreme valuetheory and computing the VaR and CVaR, by the back test technique, we can get theconclusion that TAR-GARCH-POT model is better to describe the Shibor’s risk thanother method.