Resrach On Auto Insurance Premium Pricing Based On Dependency Model

At the present stage,the Chinese automobile industry is developing at a high speed,contributing a strong driving force for the development and expansion of the Chinese auto insurance market,and constantly promoting the development and progress of the Chinese auto insurance industry;at the same time,auto insurance has become the leading business in property insurance and occupies nearly 70% market share in the property insurance market.Therefore,it has become the competitive advantage and value pursuit of insurance companies to develop actuarial methods for auto insurance rate determination and premium pricing optimization,so as to make auto insurance premium pricing more scientific and reliable.Based on the existing literature research,this thesis studies the pricing methods of auto insurance premiums from two aspects: model application and research perspective.Firstly,the characteristics of the influencing factors of the observed data are analyzed.It is found that the individual continuous variables have nonlinear effects on the response variables and the classification level of the classification variables is more.Secondly,based on the characteristic analysis of the influencing factors,the claim frequency and claim intensity models are established respectively.Finally,under the condition that claim frequency and claim intensity are independent and interdependent,the insurance premium of auto insurance data is estimated,and the corresponding methods and suggestions are put forward according to the analysis results.The innovation of this thesis is mainly reflected in:(1)The generalized additive model and generalized additive hybrid model about location,scale and shape parameters are used to model and analyze the claim frequency and claim intensity.Not only the linear and nonlinear effects of the rate factors are considered,but also the multi-level factors are set as random effects to further optimize the fitting effect of the model.(2)Different from the common assumption that claim frequency and claim intensity are independent of each other,this thesis takes the nonlinear dependence relationship of claim frequency and claim intensity into consideration of premium pricing,builds a model combined with the two-stage method,and then analyzes and discusses the estimation of pure premium under the independent and dependent conditions.Based on the empirical analysis of a group of actual auto insurance loss data,the research conclusions are as follows:(1)Claim occurrence probability,claim frequency and claim intensity were respectively used as response variables for modeling analysis.It was found that the impact of risk variables such as vehicle brand,driver age,gender,residence area,vehicle value and vehicle service life on different response variables was different to some extent.(2)The generalized additive-mixed model is applied to find that,first,the generalized additive-mixed model established after treating the multi-level factors as random effects and considering the nonlinear effects of continuous variables has a good fitting effect.Second,the impact of continuous variable vehicle value on the intensity of claim shows a nonlinear trend of decreasing first and then increasing.For the claim frequency,the vehicle value almost presents a linear effect,that is,the higher the vehicle value,the lower the claim frequency,which may be related to the more expensive the vehicle owners pay more attention to safe driving and other factors.(3)Through the modeling analysis under the independent and dependent conditions,it is not only concluded that the claim frequency and claim intensity are correlated,but also found that the pure premium prediction results considering the dependence of the two are more ideal than the prediction results under the independent assumption.Therefore,in the field of auto insurance premium pricing,the correlation between claim frequency and claim intensity plays a pivotal role in premium pricing.