Font Size: a A A

Pricing Of Joint Long-term Care Insurance Based On Neural Network Model

Posted on:2023-03-21Degree:MasterType:Thesis
Country:ChinaCandidate:G J HouFull Text:PDF
GTID:2569306806992969Subject:Insurance
Abstract/Summary:PDF Full Text Request
In recent years,China’s social aging process has accelerated.By the end of 2021,China’s population aged 65 and over reached 201 million,accounting for 14.2% of the total population,and has entered the stage of "aging society".The acceleration of the process of population aging not only shows the reduction of China’s labor force,but also means the aggravation of the pension burden of the labor force.The pension problem is closely related to everyone,which has aroused the general concern of the society.With the continuous increase of the proportion of the elderly population in the total population,the rise of this group will lead to the increase of social demand for care.According to the 2018-2019 China long term care research report,there are nearly 40 million disabled people in the elderly group,of which the demand for care accounts for about 11.8%,and they need the care of others.While the nursing power available in reality is limited,so meeting the nursing needs of the elderly is an urgent problem to be solved.As the "sixth insurance" in social security insurance,long-term care insurance is an important way to solve this problem.It ensures that the subsidy and related professional nursing services of the disabled insured in long-term disability care.However,due to the slow start of China’s long-term care insurance related system,which is currently in the pilot stage,the popularization of relevant knowledge is not sufficient.At the same time,its long-term characteristics lead to its high insurance risk,low underwriting willingness of commercial insurance and serious product homogenization.In the current context of health big data,it brings development prospects to the long-term care insurance market.Therefore,based on the family micro data of CHARLS,this paper studies the combined long-term care insurance,so as to provide a theoretical basis and ideas for the longterm care insurance to adapt to the pricing under the background of health big data,and to provide suggestions and innovations for accelerating the construction of long-term care insurance related systems,and enrich the types of long-term care insurance.After summarizing the joint long-term care insurance,in order to better meet all nursing needs of the current society,this paper divides the long-term care status with reference to many aspects and dimensions such as ADLs,including cognitive ability,mental state and syndrome,which is different from the traditional definition of ADLs;Using XGboost model to map the influence characteristics of long-term care status to the question data in the questionnaire;Then the appropriate characteristic number is obtained according to the above to carry out BP_Adaboost neural network model is trained to calculate the individual transition probability matrix;Finally,the joint transition probability matrix is obtained by copula function,and the single premium rate of combined long-term care insurance is calculated according to the single premium calculation formula.In the process of long-term health care,it can be found that the rate of single life insurance is far greater than that of single life insurance for men and women.Therefore,in practice,this paper regularly considers the disability,cognitive ability,mental state and chronic disease to determine the long-term care state,which is more in line with the specific interests of the insured and improves the depth of insurance;Based on this,eight important indicators affecting the status of long-term care are mapped,which provides ideas for commercial insurance to innovate personalized insurance products,long-term care risk identification and evaluation,and realize actuarial balance;In terms of statistical analysis,the joint longterm care insurance rate obtained by connecting the marginal probability through copula function is significantly less than the sum of male rate and female rate in different initial states,and the correlation between the consortia is fully considered: on the one hand,this is because the consortia is more likely to have healthy living conditions and emotional sustenance in the state of cohabitation in daily life.On the other hand,when the long-term care state occurs,the consortium’s mutual care and spiritual and emotional support can help the disabled party recover from the disability state as soon as possible to a certain extent.Therefore,the joint long-term care insurance can greatly reduce the total premium of double life long-term care insurance and reduce the insurance threshold.
Keywords/Search Tags:long-term care insurance, Xgboost, Transition probability matrix, copula, Neural network model
PDF Full Text Request
Related items