模型平均下基于稀疏导向学习的寿险公司偿付能力影响因素分析  

Assessing the Influential Factors for Solvency of Life Insurance Companies--A Sparsity Oriented Importance Learning Approach Based on Model Averaging

作　　者：杨娟[1] 钱振伟[2] YANG Juan;QIAN Zhen-wei(School of Statistics and Mathematics,Yunnan University of Finance and Economics,Kunming 650221,China;School of Finance,Yunnan University of Finance and Economics,Kunming 650221,China)

机构地区：[1]云南财经大学统计与数学学院,云南昆明650221 [2]云南财经大学金融学院,云南昆明650221

出　　处：《数理统计与管理》2024年第5期800-810,共11页Journal of Applied Statistics and Management

基　　金：国家重点研发计划项目(2018YFC1508904)。

摘　　要：自第二代偿付能力监管制度体系正式实施以来,中国银保监会对保险公司的偿付能力有了新的规定,在此情况下,研究偿付能力影响因素具有重要的现实意义。现有研究多集中于面板数据模型等传统的假设检验方法,没有考虑多模型推断。作为模型选择的重要推广,模型平均将不同备选模型的估计进行加权平均,可以减少遗失有用信息,且在模型平均框架下亦可研究变量的重要性。本文基于我国2016-2020年67家寿险公司的面板数据,在模型平均和稀疏导向学习(Sparsity Oriented Importance Learning,SOIL)的框架下,分析偿付能力重要影响因素及其影响程度。研究结果发现,在控制GDP增速、利率水平等10个变量影响的基础上,债券投资占比、资产负债率等变量对偿付能力充足率有重要影响,并基于此对改进偿付能力监管提出相应的政策建议。As the China Risk Oriented Solvency System was formally introduced,China Banking and Insurance Regulatory Commission(CBIRC)has a new regulation on the solvency of insurance companies,in this case,it is very important to study the infuential factors.To the best of our knowledge,most of the existing studies are based on traditional hypothesis test and the multi-model inference is rarely considered.As an important extension of model selection,model averaging can reduce the loss of useful information by combining estimators from different candidate models and one can also study the importance of variables.Within the framework of model averaging and sparsity oriented importance learning(SOIL),we consider the panel data collected from 67 life insurance companies in China from 2016 to 2020.Our study contribute as follows,bonds investment ratio,the asset-liability ratio and other variables are identified as important variables.Based on this,this paper puts forward suggestions for improving solvency supervision.

关 键 词：偿付能力 模型平均 稀疏性导向学习法 

分 类 号：F840[经济管理—保险] O212[理学—概率论与数理统计]