中文摘要：

Lee-Carter模型是当今世界上最流行的死亡率预测模型,传统的Lee-Carter模型在样本量很大时才能得到较好的效果,而中国的死亡率数据量较少,且部分年限的数据缺失,难以达到较好的预测效果。本文基于Li等（2004）提出的有限数据死亡率建模方法,同时考虑样本量不足的影响,采用韩猛等（2010）提出的＂双随机过程＂建模,构建了有限数据下中国人口死亡率的预测模型,并用于对未来死亡率变动趋势和人口寿命的预测,最后将预测结果与保险公司采用的死亡率改善因子以及社会养老保险个人账户中采用的计发月数进行对比分析,给出了相关结论和有关死亡率风险管理的建议。

英文摘要：

Lee-Carter model is the most popular method to model and forecast mortality in the world,and classical Lee-Carter model needs large amount of sample data to get better results.But there are limited and incomplete mortality sample data in China,it is difficult to apply the model directly to forecast the mortality.In this paper,based on the method which was proposed by Li et al（2004）,considering the limited data and applying ＂two-stochastic process＂ which was proposed by Han Meng et al（2010）,we build a Chinese mortality model with limited data and forecast the mortality trend and life expectancy in the future,then by comparing the predicted results with the annual rates of mortality reduction in the insurance companies and annuity factor for individual accounts in social pension,we give some conclusions and suggestions about the mortality risk management.