Realized GARCH-Copula Model And Correlation Measure Based On High-frequency Data

In recent years,with the continuous progress of economic globalization,the world financial situation has undergone tremendous changes.In the meanwhile the potential risks of finance are gradually released.In China,the obvious performance of that is the violent turbulence of A-share marke in 2015,and as a milestone in the history of Chinese capital market opening up,the operation of Shanghai-Hong Kong Stock Connect program and the recently starting up of Shenzhen-Hong Kong Stock Connect program also led to a series of risks,such as the risks of market volatility and the risks of investment.This once again reminds the majority of investors and regulators that the importance of risk management and inter-market correlation measure in the financial market research can not be ignored.The research on market volatility has been an important place in the field of risk management.With the increasing availability of financial high-frequency data,the realized measure based on high frequency data has become a hot issue in volatility research.Based on this situation,this paper promotes the Realized GARCH model,extends it to the heavy tail distribution,and compares it with the traditional GARCH model.At the same time,taking into account the impact of different realized measures on the effect of the model uses four hierarchical high-frequency data realized measures to form comparative models of tail risk measurement.In the comparison of the effect of tail risk measurement,the effect of VaR and the two loss functions which under the angle of the financial market risk management that further improve the robustness of the model comparison results.For the measure of correlation between the financial markets,this paper establishes Realized GARCH Copula modle which uses the extended Realized GARCH model as the marginal distribution function to integrate the more volatility information in the high-frequency data,while it uses Copula function to fit the correlation between the financial markets and describe its related structure.The empirical results based on the high-risk data of the Shenzhen Stock Index and the Heng Sheng Index show that the risk measurement' S effect which of the Realized GARCH model based on the student distribution is better than that of the traditional GARCH model.The selection of different realized measures has a significant effect on the model's risk measuremeant effect,and at the different risks level,the performance of each measure is different.The Realized GARCH Copula model constructed in this paper is robust to the correlation measure.Overall,the Shenzhen stock market is not closely related to the Hong Kong stock market before Shenzhen-Hong Kong Stock Connect program.