In this paper,we first outline the development process in the credibility theory and the status quo, then introduced the credibility theory of Bühlmann-Straub's model and the regression model, as well as estimated structural parameters in the model. Considering the model estimated parameters's posteriori risk in the mean-square loss model. The credibility regression models are extensions to the usual credibility models, which in statistics fall into the framework of linear estimators for linear regressions with random parameters. In the context of empirical Bayes theory, we investigate the approximate optimality of the estimators of the parameters in credibility regression models. The sufficient conditions for the approximate optimality are suggested and proved, and verify the estimated parameters in used with this optimality. |