非寿险准备金风险度量模型与方法研究

发布时间：2018-03-15 09:11

本文选题：非寿险准备金风险　切入点：一年期观点　出处：《天津财经大学》2016年博士论文　论文类型：学位论文

【摘要】：2016年1月1日,历经5年之久,对整个欧洲保险业具有约束力的SolvencyⅡ(欧盟偿付能力Ⅱ)正式生效,与此同时中国保险业进入"偿二代"实施过渡期也满一年。以风险为导向的偿付能力监管体系要求对保险公司面临的风险进行全面、精细的度量,准备金风险作为保险风险的重要组成部分,其准确计量对增强保险公司偿付能力,提高保险公司风险管理水平具有重要意义。国内外保险监管机构和保险公司越来越重视准备金风险的度量工作。然而,现有研究较多关注准备金估计模型与方法,较少涉及准备金风险度量问题。本文从影响准备金风险度量的因素出发,结合已有的准备金评估模型与方法,紧紧围绕非寿险准备金风险这一主题开展研究。具体而言,本文重点解决四个问题:(1)非寿险一年期准备金风险度量;(2)残差相关条件下的非寿险准备金风险度量;(3)赔款数据相关条件下非寿险准备金风险度量;(4)考虑模型不确定性对准备金风险的影响。首先,本文通过一个直观实例阐述非寿险一年期准备金风险的含义,引出一年期准备金风险的刻画工具:赔付进展结果,并探讨在贝叶斯对数正态模型下获得赔付进展结果预测分布的随机模拟方法。其次,为避免由残差相关性引发的准备金风险度量偏误问题,本文基于多重假设检验理论与错误发现率控制过程,给出检验和解决残差相关性问题的方法,对残差相关性引起的准备金风险波动进行分析。残差相关的本质是赔款数据相关,本文进一步在独立泊松模型基础上,添加相关随机效应,构建了描述赔款数据相关性的条件自回归泊松模型,并采用贝叶斯方法对模型进行估计。最后,本文研究了模型不确定性对非寿险准备金风险的影响,以Loglogistic增长曲线模型和Weibull增长曲线模型为例,创新性地提出利用贝叶斯模型平均对两个模型的结果进行加权平均,不仅得到综合两个模型结果的准备金估计值,而且还能得出模型不确定性视角下的准备金风险度量值。本文开展的一系列工作,有望丰富非寿险准备金风险度量的研究成果,进一步拓展准备金评估的研究领域,并为监管机构和保险公司提供理论支持和实践参考。
[Abstract]:In January 1st 2016, after five years, Solvency II (European Union solvency II), which is binding on the European insurance industry, came into effect. At the same time, the Chinese insurance industry has entered a transitional period of "second generation compensation". The risk-oriented solvency regulatory system requires a comprehensive and precise measurement of the risks faced by insurance companies. As an important part of insurance risk, the accurate measurement of reserve risk can enhance the solvency of insurance company. It is of great significance to improve the risk management level of insurance companies. Domestic and foreign insurance regulators and insurance companies pay more and more attention to the measurement of reserve risk. This paper, based on the factors that affect the risk measurement of reserves, and combining the existing models and methods of reserve assessment, focuses on the topic of non-life insurance reserve risk. This paper focuses on solving four problems: 1) Non-life insurance reserve risk measurement for one year (2) Non-life insurance reserve risk measurement under the condition of residual correlation. (3) Non-life insurance reserve risk measurement under the condition of compensation data correlation. Qualitative impact on reserve risk. First, This paper expounds the meaning of the non-life insurance one-year reserve risk through an intuitionistic example, and leads to the description of one-year reserve risk: the result of compensation progress. The stochastic simulation method to obtain the prediction distribution of compensation progress under Bayesian logarithmic normal model is discussed. Secondly, in order to avoid the bias of reserve risk measurement caused by residual correlation, Based on the theory of multiple hypothesis testing and the control process of error detection rate, this paper gives a method to test and solve the problem of residual correlation, and analyzes the fluctuation of reserve risk caused by residual correlation. The essence of residual correlation is compensation data correlation. On the basis of independent Poisson model, the conditional autoregressive Poisson model is constructed to describe the correlation of compensation data, and Bayesian method is used to estimate the model. In this paper, the influence of model uncertainty on the risk of non-life insurance reserve is studied. Taking Loglogistic growth curve model and Weibull growth curve model as examples, the paper innovatively proposes to use Bayesian model to average the results of the two models. Not only can the reserve estimate of the results of the two models be obtained, but also the reserve risk measures under the uncertainty of the model can be obtained. A series of works carried out in this paper are expected to enrich the research results on the risk measurement of non-life insurance reserves. Further expand the research field of reserve assessment, and provide theoretical support and practical reference for regulators and insurance companies.
【学位授予单位】：天津财经大学
【学位级别】：博士
【学位授予年份】：2016
【分类号】：F224;F841.3

【相似文献】