The Research On Estimation Of Ridge Parameter Of Ridge Regression In Econometric Model

In multiple regression it is shown that variables of the model have a high probability of multicollinearity,which may lead to incorrect parameter estimates,if prediction vectors of the multiple regression are not orthogonal.For many methods to deal with the multicollinearity,their common weakness is not to make full use of the information of the variables.Introduced is the ridge regression by Arther Hoerl,a method can retain most of useful information of the variables with solving multicollinearity.Thus the ridge regression is an important solution to multicollinearity,and making some research on ridge regression is meaningful.In the procedure of using ridge regression,calculating ridge parameter is an important issue.Three jobs had been done in the paper.Firstly,the paper had computed ridge parameter used by genetic algorithm in terms of decreasing variance inflation factor(VIF)of explanatory variables according to theories related to ridge regression.Nextly,the paper had validated good performance of the method through Monte Carlo simulation.Finally,the paper had illustrated the strengths of the method by empirical application regarding Chinese total health expenditure.The ridge regression based on the new method to calculate the ridge parameter can have stable coefficients and control the variance inflation factor in a proper interval to solve the problem of multicollinearity effectively.In addition the method can decrease the mean square error of coefficients significantly and improve accuracy of parameter estimates notably.And in the empirical application,the paper had made research upon the factors that affect the total health expenditure through the new method to compute ridge parameter to build ridge regression model of Chinese total health expenditure.