| The purpose of this dissertation is to clarify perplexities in the application of Granger-Causality and develop Granger-Causality Test in the sense of multivariate.Originalities of this thesis lie in the following five aspects.1.As the mathematical form of the cause was not defined in Granger's definition of causality, the cause can be an individual variable or a set of variables. It is one of the reasons why applications of Granger-Causality restrained to two-variable system. It is also the theoretical basis to develop Granger-Causality Test in the sense of multivariate.2.Being a widely used Econometrical instrument, there are still many critiques of Granger-causality's utilization. By analyzing the systemic feature of causality, we draw the conclusion that Granger-causality's defect lies in curt simplification of cause.3.On the basis of the idea of two-variable Granger-causality Test, we expand Granger-causality model in the form of VAR and make use of likelihood-ratio test to identify the causality among many variables. This development makes Granger-causality Test a more perfect econometrical instrument.4.In the research of stock market, Granger-causality is widely used to study the relationship between volatilities of two markets. While complicated correlations lies among many markets not just between two markets, it is more reasonable to study them as a co-movement system. By studying the Granger-causality between information and stock-price, the viewpoint is put forward that the result of Granger-causality Test can demonstrate the structure and efficiency of stock market. 5.The Chinese stock market consists of A-bond, B-bond and H-bond.While these three markets are separated to each other, there are correlations among their volatilities. We make use of multivariate Granger-causality test to analyze the Causality among their volatility. On the basis of empirical results, the market structure and the process of information-transfer among the three markets are analyzed and some conclusions are reached which is helpful to policy-making and policy evaluation.The dissertation is divided into four parts: preface, text, epilogue and accessories. The plan of the dissertation is as follows. The preface consists of three sections. In Section 1,the background of the studying will be introduced. Section 2 is the review of many related literature. Section 3 briefly describes the method and frame of the studying.Chapter 1:Granger Causality—concept and question. The concept of causality, Granger-Causality and Granger-Causality Test is firstly introduced. Then a particular study is considered which illustrate some perplexities in the application of Granger-Causality.Chapter 2: Multivariate Granger-causality. This chapter is divided into four sections. In the first section, the theoretical basis on which to develop Granger-Causality is discussed. In the second section, a specific study of the form of Granger-causality's expression is made, which demonstrate that VAR is a better model to express Granger-causality. In the third section, by analyzing the method of Granger-causality test, the method of multivariate Granger-causality test is deduced make using of Likelihood-ration test. In the last section, the problem that we should pay attention to in multivariate Granger-causality is discussed. Chapter 3: Research on the application of Granger-causality. In this chapter, four perplexities in the application of Granger-causality are clarified. Firstly, the difference between causality and Granger-causality is discussed. Secondly, a particular study is considered which demonstratethat Non-Granger-causality is not equal to exogeneity, but the result of Granger-causality test is helpful to distinguish whether a variable is exogenous or not. Thirdly, the relationship of co-integration and Granger-causality is discussed. At last, by analyzing the process of information-transfer and the causality between stock-price and information, the viewpoint is put forward that Granger-causality... |
PDF Full Text Request