Research On The Tail Correlation Of Industry Segments In Hu-shen Stock Return

With the accelerated pace of economic globalization,the role of correlation analysis in the financial field is more and more important.In particular,the correlation analysis of the stock market is becoming a hot topic in the financial field,because it has important reference value for decision making such as investment strategy,market supervision process and macro financial policy and so on.Having carefully reviewed the literature on the correlation analysis of financial assets at home and abroad,we find 3 points to be further studied: firstly,in the research content,the tail dependence of financial variables describes the probability that the other financial variables will fluctuate sharply when a financial variable fluctuates violently.It has great significance in analysis of the liquidity risk which the financial institutions is facing.However,the existing literatures which study the tail dependence are very few.Secondly,in the research method,Pearson correlation coefficient and Granger causality test are the most commonly used methods,but these methods have its own limitations such as correlation coefficient method is only suitable for the description of the linear relationship between variables.So it is necessary to find a more flexible and comprehensive measurement method.Finally,in the object of study,most of the literature focuses on the relationship between overall market or the "plate effect",but the research of correlation between plates of stock market which affecting the investment portfolio construction is less.In view of the above 3 points,this paper constructs the M'-Copula-t-GARCH model based on the M-Copula-t-GARCH model.This model shows good characteristics in measuring the tail correlation of financial variables.In this paper,the model is applied to the tail correlation study of the 16 plates in Shanghai and Shenzhen stock markets: using t-GARCH model to fit the return series of 16 plates,estimating the parameters of M'-Copula model,and then getting the tail correlation coefficient between plate yield series,based on the above,we draw a cluster spectrum to make a qualitative analysis of the tail correlation between plates,and finally use the portfolio theory to give the optimal portfolio of the 16 plates.Results show that:(1)The clustering results of upper and lower tail correlation of 16 segments are of some difference.The linkage is stronger maong segments when stock shoots up and the market is more sensitive to good news.(2)There is a big difference in the optimalportfolio when markets boomed or plunged,but in either case,the Shenzhen I that is the segment of information transmission,software and information technology services should hold a large proportion of the,at least about 20%.