中文摘要：

本文采用1996-2010年国家统计局公布的死亡率数据,以70岁男性人口作为高龄人口的代表,基于中国人口死亡率数据较少的特点,突破了传统Lee-Carter模型的框架,直接从死亡率改善产生的原因入手,采取Monte Carlo方法建立中国高龄人口死亡率随机波动趋势模型.通过对不同死亡率改善原因进行组合,从中选取最优模型来探究死亡率的随机趋势性与波动性的关系,更好地克服了死亡率普遍被低估的事实,使得对未来死亡率的预测更加准确、可信.

英文摘要：

This paper takes the males aged 70 as a representative of the elderly and presents an empirical analysis of stochastic volatility and trend of mortality based on the data of the year 1996 -2010 from the National Bureau of Statistics. Considering the limited and incomplete mortality date in China, we give up the traditional Lee-Carter model and use Monte Carlo method to build stochastic mortality models by analyzing the reasons of mortality improvement. By comparing combination models with different mortality improvement reasons, we study on the relationship between stochastic mortality volatility and trend, and then select the best model as mortality projection model. It can overcome the mortality underestimate and makes the future mortality prediction more accurate and reliable.