经济危机时退保风险测度与管理研究

Distribution of surrender rates during an economic crisis tends to be very different from that in normal situation. To manage great surrender risk, it is a must for life insurance companies to consider this difference while making assumptions on surrender. Therefore there is a need to do researches on both measuring and managing surrender risk. In our research, we have accomplished the following. Firstly, we applied complex data analysis method to investigate the impact factors of surrender, proposing an statistical metric model describing the behavior of surrenders instead of the classical S-shaped deterministic curve (in function of the spread). We assume that surrenders are conditionally independent with repect to the impact factors of surrender.We suggest and develop correlation function to construct conditional probability model, which depicts the distribution of surrender rates when economic crisis occurs. Secondly, we develop the non-Gaussian approach to surrender risk management. And based on this method, we construct a model for measuring surrender risk at economic crisis using relative surrender rate and tail risk measurement based on the distribution of surrender in economic crisis. Thirdly, in our research on managing surrender risk we focused on two issues, pricing surrender option and calculating its solvency capital. Modeling asset price with exponential Levy process, we construct a pricing model for surrender option when abnormal behavior of surrender took place. We also constructs a model to calculate reasonable solvency capital in case of great surrender risk. In a word, our research on surrender risk is of great significance not only in theory but also in practice, which may offer meaningful guidance on both risk management of life insurance company and solvency supervision of regulators.

经济危机下寿险合同退保率较高且公司间差别大，这显著不同于正常时的退保率分布，改变了寿险公司经营决策的退保率假设，使得退保风险突出，亟需研究危机下退保率分布、退保风险测度及相关管理问题。 本研究以复杂数据分析理论与方法研究经济危机时保单持有人退保的驱动因素，构建退保行为计量模型，以此为条件，基于退保人退保行为的关联函数建立了条件概率模型，以刻画经济危机下的退保行为，构建了危机时的退保风险分布；研究了非高斯假设下的风险度量理论，从相对退保率和尾部风险度量两方面建立危机时退保风险测度模型；从退保权定价和偿付能力资本管理两方面研究退保风险管理，一是引入危机下的退保行为刻画模型，以指数形式Levy过程刻画标的资产的价格变化，建立基于危机情况下异常退保行为的退保权定价模型，二是建立针对退保风险的偿付能力资本要求模型。 这对退保风险研究具有重要的理论意义，对退保风险管理和偿付能力监管都有很强的实际意义。

经济危机情况下退保率分布、退保风险测度及其相关管理问题是当前亟需解决的问题。本研究首先利用2001-2013年中国寿险公司全行业数据，分别建立基于公司层面的经济危机和正常年份的退保行为影响因素计量模型，对比退保影响因素的差异，同时检验退保影响因素假说的有效性。其次，考虑危机时退保率分布的非高斯性，研究非高斯分布下的风险度量工具，并根据中国寿险公司退保率实际数据，拟合危机时退保风险分布，基于退保率尾部风险度量测度经济危机时的退保风险。然后在经济危机可能改变资产收益率分布的情形下，利用广义极值分布刻画资产收益率分布，研究累积分红保险内嵌退保权的定价模型，将结果与收益率服从正态分布的结果进行了对比。最后，考虑保险公司投资多种资产以及不同资产间的关联性的情况，基于MEGB2 Copula函数，设计了MEGB2分布下的累积分红保险内嵌退保权的定价理论模型。.结果表明，“支付贬值”假说和“利率替代”假说能够解释中国寿险市场退保率的变化，经济危机时保险公司的经营状况是保单持有人更为关心的影响退保率的因素，在经济正常时，保单带来的收益情况是影响保单持有人退保决策的主要因素，因此对保险公司而言，既要设法保持其经营的稳健性，增加消费者对公司的信心，又要在风险可控的情况下提高投资的收益率，以此增加产品的吸引力；Gamma分布用于拟合正常年份的退保率分布较为合适，而广义极值分布较为适合拟合危机年份的退保率分布；在收益率服从广义极值分布假设下，退保权的价值对最低保证利率、年度分红比例、无风险利率这几个定价参数的变化较为敏感。保险公司在设计保险产品时，需要合理确定这些定价参数的值，防止出现退保权价值过高，进而引起退保率的升高。大多数情况下广义极值分布下退保权的价值高于正态分布假设下。这说明了选择合适的分布对资产收益率进行刻画对于精确定价的重要性。以上结论对退保风险研究具有重要的理论意义，对寿险公司的退保风险管理和监管机构的偿付能力监管都有很强的应用价值。