两人群引力模型框架下的中国死亡率预测①--结合美国相应数据

摘要

Lee-Carter model and its subsequent extensions are the dynamic models in the modeling and prediction of the mortality of a population. Because of the poor quality of sampling data for the mortality in China, The accuracy of most models in population prediction is lower than that of foreign literatures. In this paper, the model of mortality data in China and the United States is modeled in the framework of the gravity model of two people and the results are compared with the single group model. The gravity model combined with APC model can improve the accuracy of prediction greatly and it shows that China's old dependency ratio is going to rise as high as 23.32%, male mating population is going to be 20.79 million more than the female, infant and juvenile cumulative death number is to be about 2.93 million.%Lee－Carter系列模型是对一个人群的死亡率动态建模和预测的模型。由于中国死亡率抽样数据的质量问题导致模型预测的效果不如国外文献所反映的那么精确。本文在两人群引力模型框架下结合中国和美国同期死亡率数据建模，并将结果与相应的单人群模型比较。研究表明，引力模型与APC模型相结合取得了最好的效果，在此基础上本文预测2025年老年抚养比会急剧上升到23．32％，2030年的婚配男性人口超出女性约2079万，婴幼青少年20年间累计死亡人数约293万。