Research On Auto Insurance Pricing And Customer Classification Management Based On Quantile Regression

With the intensification of competition and the reduction of profit margins in the insurance industry,the demand for customer refinement management is growing.In order to achieve customer's refined classification management,companies need more precise pricing models as support.Inaccurate pricing that lacks risk discriminatory ability in the traditional models commonly used in insurance industry will result in premiums paid by customers that are inconsistent with their actual risk levels,making low-risk users or insurance companies bear the additional costs of high-risk users,and result in the loss of low-risk customers and increase in high-risk customers.This paper introduces quantile regression model.Firstly,we compare the auto insurance pricing results of the two models through simulation experiments.It is proved that the accuracy of the quantifier regression for pure risk premium estimation is higher than that of the generalized linear model.Then through the quantile regression model,an analysis of policy data in three years for china continent insurance company was conducted to explore the impact of each risk variable under different variate levels of the dependent variable.It is found that in the auto insurance pricing process,the influence of all variables can be divided into three categories which include the variables with positive influence such as purchase price of a new car and the duration of insurance renewal,the variables with negative influence such as the owner's gender,no indemnity preferential treatment coefficient,the age of the owner,seat number,and such variables that influence the action fluctuates greatly and the trend is uncertain,such as the year of accident,region,car series and age.Therefore,we have a clear and intuitive understanding of the impact mechanism of risk factor.According to the analysis results,it is suggested to set a differentiated pricing plan for customers with different risks.And in customer classification management,we must pay attention to customer screening and establish a quality customer circulation system to maintain strong competitiveness.The research results have important practical significance for the Chinese insurance companies to develop more differentiated customer classification management strategies and differentiated and precise pricing.