A Study Of Car Insurance Rate Making Based On Copula

Since the restoration of motor vehicle insurance business in China in 1980,with the economic growth and the continuous increase of national income,the auto insurance business has continued to grow and develop.After 1988,the premium income of auto insurance business became the first major insurance of property insurance till now.However,due to the backwardness of domestic auto insurance research and the unreasonable pricing strategy,the auto insurance premium income is increasing,while the auto insurance business yield rate continues to be sluggish.Therefore,how to properly price auto insurance products has become a key concern in the actuarial industry.The application of Generalized Linear Models(GLM)in auto insurance pricing has a long history.When using generalized linear models for rate making and auto insurance claim data analysis,it is often carried out under the assumption that the claim severity and the number of claims are independent.However,in the real world,there is a strong interdependence structure between them.So,independent hypothesis will result in an underestimation of the amount of losses.In order to better reflect the interdependence between the two,this paper introduces a hybrid Copula to describe its dependence.Compared with the model using the assumption of independent hypothesis and single dependence,the model used in this paper has better ability to interpret actual data.Empirical analysis shows that using the model proposed in this paper can obtain more accurate claims prediction.In addition,in the field of auto insurance rate making,it has been a long time to adjust the motor vehicle premiums using the Bonus-Malus system(BMS).The purpose of the system is to evaluate each individual and every policy more accurately.The BMS system can help determine the premiums that are more in line with each individual's risk level by embodying unquantifiable levels of risk in the GLM.However,at present,the BMS only uses the prior probability to adjust the reward and punishment grading,which obviously cannot fully consider the complex interdependence in the auto insurance loss.Therefore,the vine Copula model is introduced in this paper.When using the auto insurance vertical claim data for empirical analysis,the model used in this paper is better predictive than the model using a single Copula.In summary,there are obvious deficiencies in the auto insurance pricing model used in current insurance practice.The improved model proposed in this paper can describe a variety of complex interdependence relationships,is more suitable for the actual situation,and has broad research prospects.