Research On Automobile Insurance Fraud And Its Recognition In China

As the highest proportion of premium income in property insurance in China,automobile insurance is directly related to the stability of property insurance companies.However,in China and even in the international insurance market,there are serious fraudulent claims in automobile insurance.Automobile insurance fraud will not only affect the normal operation of the insurance market,endanger the interests of honest policyholders,but also intentionally create traffic accidents,damage other people's property and other acts seriously disrupt social order.In view of this phenomenon,this paper makes a theoretical analysis of automobile insurance fraud and effectively recognizes fraud cases,which is practically and theoretically meaningful.Firstly,this paper defines the concept of automobile insurance fraud,then introduces the current international popular recognition methods of automobile insurance fraud,as well as the information asymmetry,incomplete contract,the principle of utmost good faith and cost-benefit theory used in the analysis of automobile insurance fraud.Next,aiming at the current situation of fraud and identification of automobile insurance in China,this paper summarizes five kinds of common fraudulent means of automobile insurance in China,and analyses the harm of automobile insurance fraud to insurance companies,honest policyholders and national and social subjects,and puts forward the causations for the emergence and prevalence of automobile insurance fraud in China,including insurer's limited awareness and lacking of integrity,the identification methods and technologies of insurance institutions are backward,the relevant laws and regulations are not perfect,and lacking of fraud recognition information sharing platform.Finally,in view of the current situation of fraud recognition in China's automobile insurance and various fraud recognition methods' advantages,disadvantages and applicable conditions,the combination of Logistic regression and BP neural network model is selected as the empirical analysis method in this paper.On this basis,this paper makes an empirical analysis of 302 automobile insurance claims in P Branch of a large property insurance company in China(including 200 cases of honesty and 102 cases of fraud).In the selection of variable indicators,this paper collects 17 indicators of accident information,party information and automobile information as independent variables,and takes the type of case(honest or fraudulent)as the dependent variable.The first step is to use SPSS 22.0 software to test the case correlation of these 17 indicators,then logistic regression analysis is made on 12 significant indicators,and finally 7 indicators are determined for automobile fraud cases,including "reporting time","type of traffic accident","whether to report the case on the same day","whether to report the case on the third party","location of the accident","amount of claim" and " automobile use".In the second step,according to the experience and the characteristics of the input and output data of the neural network,this paper constructs a three-layer BP neural network model,chooses the appropriate transfer function and training function,trains the neural network with 262 samples,and uses 40 samples as test samples to analyze the effectiveness of the BP neural network model in recognizing automobile insurance fraud claims.Finally,the average correct rate of the BP neural network model for recognizing the sample of automobile insurance claims with refined indicators is 91.25%.Among them,the correct rate for recognizing honest cases is 93.75%,and for fraud cases is 88.75%.Finally,this paper puts forward 7 preventive measures for China's automobile insurance fraud,including paying attention to the audit of insurance entrance,enriching the content of investigation information,refining the classification of indicators,improving the professional skills of claims settling personnel,strengthening internal cooperation of the company,making extensive use of external law enforcement agencies,establishing fraud information sharing platform,establishing joint anti-fraud organizations,improving relevant laws and regulations,and intensifying fraud punishment and relevant education.In the conclusion part,this paper summarizes the shortcomings in the process of samples' statistics and empirical analysis,and prospects the future work of automobile insurance fraud recognition and anti-automobile insurance fraud in China.