The Growth Analysis Of Company Based On Nonlinear Structural Equation Model

Structural equation model is widely used in the disciplines of psychology, medicine, sociology and economics, and it is an important tool to study the relationship between the Manifest variable and latent variable. But so far, most research of structural equation model is about latent variable linear structural equation. This year, in some complex situation, many researchers recognize that the non-linear relationship between the latent variable is very important to establish a more meaningful and more accurate model. For example:Jonsson (1998), Ping (1996), Kenny and Judd (1984) all mentioned the importance of latent variable secondary and the interaction term effects in their actual study. With the development of computer technology, now more statistical software can be used to analyze the structural equation model, such as:AMOS, LISREL, EQS6, MPLUS and WinBUGS etc. As a result, in this dissertation, we will systematically discuss the nonlinear Structural equation model and a Bayesian method to analyze nonlinear Structural equation model. The main content of this paper can be summarized as follows:1. A Bayesian method is developed to analyze nonlinear structural equation models. After setting prior distribution of the parameters, we calculate the posterior distribution of the unknown parameters in nonlinear structural equation model. A hybrid algorithm combining the Gibbs sampler and the MH algorithm is used to obtain the joint Bayesian estimates of structural parameters, latent variables and their standard deviation estimates. A procedure calculating the Bayesian factor for model comparison is given via path sampling. A goodness-of-fit statistic is proposed to assess the plausibility of the posited model.2. A nonlinear structural equation model with squared terms and interaction terms is given in the study of the company’s growth. A Bayesian method is used to estimate parameters of the nonlinear structural equation model. Financial statements of200companies, which are listed in the Shenzhen Stock Exchange, are selected to estimate the Model parameter. Winbugs is used for data analysis. Winbugs is often used in the Bayesian econometrics. The Bayesian method considers the prior information of samples. It is very meaningful and useful in economic research. In summary, this dissertation has systematically investigated the nonlinear structural equation model. By employing a hybrid algorithm combining Gibbs sampling and the MH algorithm, Bayesian analysis for nonlinear structural equation model is conducted. It is proved that the company’s profitability and the company’s capital is positive to the company growth.