The Study Of Factors Influencing Capital Structure Of Listed Enterprises In China——Based On The Dynamic Bayesian Model In Complicated System

Most of the traditional econometric models are founded on the basis of linear and Gaussian distribution. However, in practice, financial data is always not only non-linear and non-Gaussian, but connected with latent variable and variable coefficient, which results in more restrictions in the application of traditional tools and new development of constructing econometric models. In the analysis of time series, academics introduced time-varying parameter model, also called State Space Model, to deal with conditions where parameters are varying as time changes. At the same time, Structural Equation Model has been rapidly popularized and developed in recent years because of the advantages in analyzing latent variable. In this paper, the writer tried to combine the good points of SSM and SEM, to build a dynamic econometric model which can include the time-varying effect as well as background effect and handle all the problem mentioned above, and to analyzing the conclusion using Bayesian method. On this foundation, we can check how to apply the model in practice to example analysis at last.This paper is made up of seven sections. In first section, the background, research target and innovation will be introduced. Then, in second section, the writer will review the literatures related to SSM and SEM in recent years. On the basis of previous two parts, the third section will analyze in detail how to build SSM and estimate state variables and unknown parameters. Afterwards, this paper will take SEM as a study approach into analysis from the fourth section, where we can find the introduction of construction process, strengths, categories and estimation methods, including Maximum Likelihood Estimation and Bayesian Method.In fifth section, the writer select to build a dynamic Bayesian econometric model according to the conclusion of fourth section. Besides, there also includes the introduction of generalization procedure, features and scope of application, solution and estimation, as well as usage method of the Bayesian model in this part. And the sixth section will analyze what may influent the capital structure of listed company in China making use of our model and test to prove the availability of this brand new econometric modeling. The last section concludes the achievements of this paper and puts forward some possible research areas for the sake of academics with interest.On the foundation of empirical analysis, we can reach the following conclusions: Accordingly, the scale of enterprise, growth ability and the value of guarantee will positively affect the capital structure, while profitability, debt paying ability and non-debt tax shield will affect the capital structure in negative way. In addition, this paper has been proven flexible and summative enough to be widely applied and extended.