Based On VAR And Nonlinear Error Correction Model For Shanghai Shenzhen And Hong Kong Stock Index Linkage Studies

Now the world financial markets exist between interdependence and mutual dependence, which is already well known, but how the dependence of specific, we unable to agree on which is right, public opinions are divergent. Understanding of the linkage between Chinese capital market plays a very important role in the development of Chinese economic decision-making, but the long-term equilibrium relationship between variables that cointegration analysis method can well reflect this linkage relationship, if you can correct understanding of our country stock market price long-term relationship, this can be a very good further promote the healthy development of Chinese stock market itself and the more perfect, most studies using traditional linear cointegration theory it is not comprehensive and lack of thoroughness. This paper first consideration of traditional cointegration relationship, to further explore the logistic smooth transition autoregressive (LSTAR) existence of nonlinear cointegration model, using R programming language and EVivews software to estimate the parameter, according to the references to construct the practical test program and test statistics suitable, and linear cointegration Shanghai,Shenzhen and Hong Kong stocks, which as well as the nonlinear cointegration analysis through the actual data of Chinese stock market.The main content of this paper is divided into the following four chapters, the first chapter is the introduction, mainly introduces the motivation, background and practical value, research status at home and abroad were reviewed and VAR model analysis method and the nonlinear cointegration relationship, more comprehensive summary of the previous research results of researchers, necessity then this paperThe second chapter is the theory and method are introduced, which mainly discusses the recognition and binding form of VAR model, VAR model, determine the number of lags, Granger causality test and impulse response function for routine analysis of nonlinear cointegration theory, the existence of nonlinear cointegration test method, construction and estimation method.The third chapter is empirical research through the second chapter of the theory and method of introduction, this will be the first on the Shanghai and Shenzhen stock indexes set up VAR model, to test its stability, and then determine the lag order, Granger causality test, impulse response function, cointegration test on the VAR model to determine the good, then the Shanghai, Shenzhen and Hong Kong stock indexes refers to a more comprehensive analysis of the linkage, through the empirical study of nonlinear each index, and then use logic smooth transition autoregressive (LSTAR) method for constructing nonlinear cointegration model, further for a more comprehensive analysis of cointegration between the stock indexes.The fourth chapter is the summary of this thesis, and to summary the third chapter. It gives some of the main conclusions and policy recommendations based on the empirical results, the results show that there are exists a linear cointegration between Shanghai and Shenzhen stock index, but also the existence of logic in the smooth transition autoregressive (LSTAR) nonlinear relationship, for Shanghai, Shenzhen and Hong Kong three stock index, not only they do not exist linear cointegration between logistic smooth transition autoregressive but also does not exist (LSTAR) nonlinear relationship, finally puts forward the problems reflected in the empirical process, and the future direction of further research to progress.Finally, the innovation of this paper is mainly found there are exists a linear cointegration between Shanghai and Shenzhen, we based on the theoretical basis, according to the tested structure, nonlinear LSTAR model type, the results show that Shanghai, Shenzhen composite index exists a nonlinear LSTAR model structure, and then the stock index prediction for them using the nonlinear error correction model and linear error correction model, the prediction of nonlinear error correction model better than linear error correction model results.The main shortcomings of this paper are the following two points, the first is this paper only discusses the smooth transition autoregressive (STAR) which is a nonlinear structural model, Markov regime cointegration model, structural break cointegration model, the conversion of threshold cointegration model and smooth conversion cointegration model, the four modeles is the mature theory for some nonlinear structural model research results in modern, for so many research results, this article is only take into account the smooth transition autoregressive model, this paper is a drop in the sea, only can peep a little cannot know everything. It has not very convincing, so I decide to research it more comprehensive and in-depth,and want to show them more comprehensive, more systematic from internal relations.The second although the Shanghai, Shenzhen and Hong Kong stocks that were studied, but this is very superficial, through the stock index research to understand the inland and Hongkong area of economic and trade contacts closely, it could not perform deeply and comprehensively. If we want to compare the thorough understanding of cross-strait economic relations, we should study deeply on the cross straits economic system, economic indicators and other aspects of a more comprehensive.