EEMD Based Research On Macro-economic Factors And Var Of Stock Market

China's stock market has made a great achievement Since it was established in 1990s.It has played a very important role in China's economy and become the basis of prosperity and stability of China's economy.However,due to the characteristics of high return with high risk,and the immature mechanism as well,China's stock market often has abnormal fluctuations.Therefore,it is very important to study the causes of stock market volatility and stock market risk measurement.Some scholars believe that there is a close relationship between stock price volatility and macro-economic factors.With the gradual opening of China's capital market,the financial market,especially the stock market is facing more and more risk which is very complex.Therefore,it is more and more important to improve risk prevention,control and supervision.Deep research on the stock market risk is very needed to prevent the stock market risk and keep China's stock market stable and safe.Norden E Huang,who is a academician of NAE of America,created a new time-series decomposition method called Empirical Mode Decomposition which does not need to preset the basis function,but just to decompose signals based on itself characteristic of time scale.Compared with the Fourier decomposition and Wavelet decomposition method,EMD has a stronger partial performance ability,and can describe the physical characteristics of the original signal,which made EMD very suitable to decompose nonlinear and non-stationary time series.Ensemble empirical mode decomposition(EEMD)has solve the problem of mode mix,with which can accurately decompose time series.In this paper,we propose a method to learn the impact of some macro-economic factors on stock market volatility based on EEMD.First,we use EEMD to decompose China's stock market index time series,in order to get the intrinsic mode function(IMF)of several period.Then we use the Exhaustive method to choose several applicable IMFs,which will be reconstructed into a new time series.Then the new time series,with five macro-economic such as CPI,IAV,M1,IR and ER,are to be modeled by a Vector Auto-Regression model(VAR)to learn the relation between them and stock market,with the application of Co-integration Analysis and Variance decomposition.With the continuous development of the stock market and the increasing complexity of the stock market risk,the demand for the measurement of financial risk in the stock market has become more and more needed,Value at risk(VaR)is one of the most popular model in the field of international risk management model.It has a practical significance to innovate VaR method.In this paper,based on EEMD,we propose a VaR metric method ensembled with Neural Network and Quantile Regression(QRNN).QRNN model not only does not need to take into account the distribution of financial series data,but also does not rely on the measuring of volatility to calculate VaR,which is very suitable to adapt to the characteristics of Chinese stock market,whose financial data has a leptokurtosis feature.The empirical study in this paper shows that China's macro-economic factors has corresponding impact on the stock market volatility,which is consistent with the past theory.And the results show that EEMD-QRNN method works well when it is used to measure the stock market VaR.