高年龄段死亡率模型研究

任何国家在编制生命表时都面临着高年龄段可以观察到的数据量较小的局限。本文首先对我国第五次人口普查数据及日本死亡率数据进行了初步分析,同时对死亡率的基本特点进行了比较,在此基础上,选择80～94岁的死亡率数据作为建模依据;并对不同参数模型进行拟合,选择了拟合度较好的 Logistic2、Logistic3、Kannisto 和 HP模型作为高年龄段死亡率趋势外推模型;最后利用前面讨论的模型对我国寿险业1990～1993和2000～2003生命表进行高年龄段的趋势预测,并比较了不同模型下预期寿命的差异。

This paper made an initial comparison of the results of China＇s 5th census with the mortality data of Japan, including a comparison of their respective mortality characteristics. Then it used the mortality data for the 80-94 age group as its modeling basis. It also uses different parameters to model this trend, and come out with the Logistic2, Logistic3, Kannisto and HP models as the tendency-inference mortality model for the high age group. Finally, the paper employed the aforesaid models to make a high-age trend projection for the 1990-1993 and 2000-2003 life tables of China, and compared the differences of life expectancy under different models.