基于长记忆特性的死亡率模型研究  

Research on mortality model based on long memory characteristic

作　　者：王莹莹 季彦颋 闵蓥宵 房启全 WANG Yingying;JI Yanting;MIN Yingxiao;FANG Qiquan(School of Science,Zhejiang University of Science and Technology,Hangzhou 310023,Zhejiang,China)

机构地区：[1]浙江科技学院理学院,杭州310023

出　　处：《浙江科技学院学报》2023年第1期81-88,共8页Journal of Zhejiang University of Science and Technology

基　　金：国家自然科学基金项目(11701513)。

摘　　要：【目的】为了在实际应用中准确估计死亡率,提出基于长记忆特性的死亡率模型。【方法】选取个体死亡率数据,构建长记忆性死亡率模型进行研究。首先根据R/S分析(rescaled range analysis,重标极差分析)法估计死亡率队列的Hurst指数;然后利用长记忆性Milevsky-Promislow死亡率模型和Milevsky-Promislow死亡率模型对个体死亡率数据进行拟合对比;最后采用长记忆性死亡率模型预测个体死亡率,并将其应用到中国寿险业经验生命表中。【结果】能够捕捉长记忆性的死亡率模型对个体死亡率的拟合效果更好,队列的初始年龄、性别因素对拟合效果有一定的影响,且该模型对死亡率的预测较为准确。【结论】本研究通过构建长记忆性死亡率模型,为提高死亡率拟合预测效果提供了理论方法。[Objective] In order to accurately estimate mortality in practical applications, mortality model based on long memory characteristic was proposed. [Method] The individual mortality data were selected to construct a long-memory mortality model to carry out the research. First, the Hurst index of mortality cohort was estimated by means of R/S analysis(rescaled range analysis);then, the long-memory Milevsky-Promislow mortality model and the Milevsky-Promislow mortality model were employed to fit and compare the individual mortality data;finally, the long-memory mortality model was used to predict the individual mortality and was applied to the experience life table of China life insurance industry. [Result] The mortality model capable of capturing the long memory has a better fitting effect on the individual mortality, and the initial age and gender of the cohort have a certain influence on the fitting effect, and the model has a relatively accurate prediction of mortality. [Conclusion] This study can provide relevant theoretical methods for improving mortality fitting prediction by constructing the long-memory mortality model.

关 键 词：长记忆性 死亡率模型 HURST指数 个体死亡率 拟合 预测 

分 类 号：F840.323[经济管理—保险]