Research Of Realized GARCH-SGED Model And Its Application In Risk Measurement

For the universal existence of financial asset returns' sharp peak and fat tail,this paper constructed a new Realized GARCH-SGED model with the residual distribution of Realized GARCH model has been extended to the thick tail characteristics of the form of the skewed generalized error distribution and the power of leverage function is relaxed to estimated parameter.The parameters of Realized GARCH-SGED model have been estimated by MLE method,and it is necessary to relax the power of the lever function.Then,Monte Carlo method was used to simulate Realized GARCH-SGED model,and the results showed that the mean square error of all parameter estimation is smaller,and the estimation result is ideal.Secondly,the high frequency data of SSE 50 index for 5 minutes as the research object,the error distribution of normal distribution,generalized error distribution and skewed generalized error distribution of Realized GARCH model predict VaR estimation,using Kupic test compute and compare the accuracy of the prediction.The results show that compared with Realized GARCH model,Realized GARCH-GED model,Realized GARCH-SGED model can describe volatility leverage more accurate,to a certain extent promoted the tail risk measurement accuracy.Realized GARCH-SGED model to measure the effect is better than that of Realized GARCH-GED model,that has been achieved although Realized GARCH model can produce certain skewness,but the skewness is not satisfied the requirements of data,require additional distribution to generate skewness,so should consider errors for fat tailed distribution deviation.Finally,the different realized measures are combined with each model.The parameters estimation and risk measurement analysis of the model under different realized measures have been obtained by using the data of the SZSE 100 index 5 minutes frequency data.In addition to the use of back testing methods of risk measurement and analysis of model,also introduces the loss function and the SPA test method based on risk management,compares the empirical effect of each model in risk measurement,obtained in the introduction of different measures have achieved prediction accuracy of risk measurement are improved,different choice of measure has been implemented have a great impact on the effect of risk measurement.We provided a new risk prediction model of financial risk measurement of Realized GARCH-SGED model.In a certain extent has enriched the theory and methodology of risk management in financial markets.