人口预测的随机方法：基于Leslie矩阵和ARMA模型  

Stochastic Model for Population Forecast:Based on Leslie Matrix and ARMA Model

作　　者：任强[1] 侯大道[2,3] 

机构地区：[1]北京大学人口研究所 [2]北京大学数学科学学院 [3]香港中文大学社会学系

出　　处：《人口研究》2011年第2期28-42,共15页Population Research

摘　　要：本文探讨了人口预测的一种随机方法。文章回顾了经典的Leslie矩阵并结合中国的人口统计数据,用时间序列的ARMA模型对未来的生育率、死亡率进行估计,并由此构造Les-lie矩阵,经时间序列的数据中心化,根据自相关函数、偏自相关函数的截尾性或拖尾性,以及贝叶斯信息准则函数方法对模型定阶,实现对ARMA模型的识别。在中国人口预测方面的应用证明,基于Leslie矩阵和ARMA模型的人口随机预测方法是稳健的,具有很强的适用性。由于统计数据可获得性的局限,对模型做了不少假设和近似。随着人口数据的积累,未来将会在此方面有所改进。This paper develops a stochastic model for population forecast with applications to China. Forecasting fertility and mortality with ARMA model creates a Leslie matrix. The classical Leslie matrix is adjusted with Chinese data; and the specific ARMA model is identified through centralization of time - series data, determining the order of model by auto - correlation function, the properties of cut - off or tail - off of partial auto - correlation function, and Bayesian information crilerion fundion. Applications to China＇ s population forecast indicate that this stochastic model based on Leslie matrix and ARMA model is robust and adaptable. We have made some assumptions and approximation clue to limitation of data availability, and this could be improved when appropriate data are available.

关 键 词：人口预测 Leslie矩阵 ARMA模型 随机模拟 

分 类 号：C921[社会学—人口学]