动态死亡率建模与年金产品长寿风险的度量——基于有限数据条件下的贝叶斯方法

摘要

本文使用贝叶斯方法通过MCMC抽样对Currie模型的参数进行估计，在此基础上，运用该模型对我国人口未来死亡率进行预测，最后对年金产品的长寿风险进行度量。研究表明，贝叶斯方法能够更好地拟合我国人口死亡统计数据；如果不考虑人口死亡率的变化，而只使用现有的生命表为年金产品定价，保险公司将会面临较大的承保风险；由于死亡率变化的不确定性，保险公司为年金持有的长寿风险偿付能力资本要求为其年金均值的2．3％。%This paper estimates the parameters of Currie mortality model based on Bayes MCMC method, uses the model to predict the future mortality, and measures longevity risk of annuity. The paper concludes the following points, un- der the conditions of relatively lack of demographic data, the Bayesian approach fits better death statistics of Chinese population; if not taking into account changes in the mortality trends and only using existing life table for annuity product pricing, insurance companies will face great underwriting risk; due to the uncertainty of mortality, solvency capital requirements of longevity risk for the annuity equals to its mean of 2.3 %.