Empirical Analysis Of Chinese Bond Market Based On Garch And SV Models

The volatility is existing in all kinds of financial time series. In fact, the volatility is a core question in financial research field. At present, there are two kinds of models for the study of the fluctuation in financial market:one of them is auto-regressive conditional heteroscedastic (ARCH) model, and the other is stochastic volatility(SV) model. These two kinds of models have been further development in recent years of empirical research.For example,GARCH models,is one of the expanding models of the ARCH mode and SV-N model、SV-M model、SV-MN model、 S V-MT model and SV-Leverage model are the expanding models of the SV model. There has been some research in China which simulated the stock market and options market with these two types of mode. But, for the bond market which is also of significance to our economy, there are really very few research. Therefore, we simulated the financial times series of bond market with GARCH model and SV model. In the end, we evaluated the accuracy of the forecasting ability of these two types of models by introducing a series of evaluation indices.First, we analyzed the TGARCH model and EGARCH model based on the data of returns ration series of Shanghai’s index of treasury bonds、corporate debt index and financial bond index. Then, SV-N model、SV-M model. SV-MN model、SV-MT model and SV-Leverage model were analyzed according to Markov chain Monte Calo algorithm procedure. After these empirical research, we found that financial time series in bond market displayed strong cluster effect and leptokurtic effect. But the leverage effect was not significant as in stock market.The evaluation indices which we used to evaluate the forecasting ability are RMSE, MAE, LL indices. The comparison result is SV models are superior than GARCH model in most time. Especially, SV-MT model performed best in forcasting the volatility of return ration of nation debt.