An Application Study On The Neural Network Model Used In Early Warning Model Of Solvency Of Life Insurance Company

The core of this paper is to build an early warning model of life insurance company solvency."China Risk Oriented Solvency System"(C-ROSS)has strengthened the management of life insurance companies' solvency.However,in recent years some life insurance companies still experience a sudden drop in the solvency adequacy ratio in the quarterly report.This decline in solvency is a risk on itself.For the company's reputation,the insured's capital security and the entire insurance industry have had an adverse impact.Whether it is to point the way for the improvement of the solvency level of life insurance companies or lay a solid foundation for the supervision of the solvency of life insurance companies,the establishment of an early warning mechanism for solvency of life insurance companies can play an important role.Firstly,this paper analyzes and collates domestic and foreign literatures on the early warning model and solvency research.After defining the research direction,this paper studies the solvency status of China's life insurance companies and their related early warning theories,and proposes the characteristics of artificial neural networks.Then I combine the advantages of artificial neural networks with the solvency warning index system to recalculate the level of solvency before 2016.It's necessary to determine specific indicators and early warning mechanisms,and use the role of neural networks in model building.Finally,By adopting the business data of 13 life insurance companies from 2011 to 2016,using neruosolutions neural network tools,according to new solvency adequacy ratio calculated in the third chapter,the solvency early-warning model of life insurance companies in future under "The Construction Plan of China's Second Generation Solvency Supervision System" will be trained in the fourth chapter by selecting four aspects and nine major early warning monitoring indicators.This model experienced 3859 times of update and replacement,and the optimal probabilistic neural network was obtained.The solvency early warning model was tested and tested,and the early warning model test results were obtained.The research results show that the solvency adequacy ratio under the second generation of compensation and the solvency adequacy ratio under the original supervisory system have changed significantly,and the solvency adequacy ratio of life insurance companies as a whole has declined slightly;the neural network model can improve the construction of life insurance.The deficiency of the company's solvency early-warning model is due to the high degree of fitness of the neural network,the speed of learning,and the high fault tolerance rate;the test sample test of the final probabilistic neural network solvency early-warning model output shows that the police index is in the negative solvency and medium-low The solvency data interval is more accurate,and the artificial neural network model passed the test of the case.This shows that the prediction accuracy of this model is high and can be used to early warning the solvency of life insurance companies in China.