摘要
对于人口未来变动趋势的准确判断一直是人口学领域的研究重点。概率人口预测方法近年来得到逐步发展,其中,联合国人口的贝叶斯分层模型最为典型。该方法基于各参数的先验分布确定其未来的变化区间。与传统的队列要素法等人口预测方法相比,概率人口预测能够获得更加客观的人口预测结果。本文对联合国人口司《世界人口展望》2019年的修订版所使用的贝叶斯分层模型概率人口预测方法进行介绍。内容包括总和生育率和出生预期寿命的贝叶斯分层概率预测模型构建的理论基础和方法,年龄别生育率和死亡率、总迁移率、年龄别迁移率的预测方法,总人口的概率预测方法以及确定性人口预测方案等。以中国为例对贝叶斯分层模型概率人口预测方法的应用进行研究。联合国概率人口预测方法充分考虑不确定性对未来人口发展产生的影响,丰富了现有的人口预测方法。但是,联合国人口司现有的贝叶斯分层概率人口预测方法仍有待进一步完善,比如,在人口迁移预测和对艾滋病高发国家预测的应用方面,如何建构较稳健和可信的预测模型仍然存在挑战;在现有的总和生育率和出生预期寿命的预测模型中,如何考虑各种危机、冲突、灾害、疾病大流行等不可预测事件对未来人口发展的影响,科学地解决不确定性带来的偏差和局限性。
How to project a more accurate future trend of population has always been a hot theme in demo⁃graphic research.Recently probabilistic population projection has gained an impetus.One of popular meth⁃ods is the Bayesian hierarchical modeling used by the United Nations Population Division,which predicts the future possible range of each demographic parameter in projection models based on its prior distribu⁃tion.Overall,the probabilistic population projection produces more objective outcomes than the convention⁃al cohort component method.This paper briefs the method of the Bayesian hierarchical modeling used in the 2019 Revision of the World Population Prospects by the United Nations Population Division.The meth⁃od includes probabilistic projections for total fertility rate and life expectancy at birth.The paper also intro⁃duces the method for projecting age-specific fertility and mortality.China is used as an illustration.The pa⁃per concludes that although the probabilistic population projection method used by the United Nations has advanced the population projection in general,it still needs some improvements for its existing projection methods for total fertility rate and life expectancy at birth to incorporate sub-regional variations and shocks by crises,disasters,and pandemics.Development of robust projection models for international migration and for countries with high HIV prevalence is also warranted.
作者
盛亦男
顾大男
SHENG Yinan;GU Danan(Population and Economic Research Institute,Capital University of Economic and Business,Beijing,100070,China;Independent Researcher,New York,10017,USA)
出处
《人口学刊》
CSSCI
北大核心
2020年第5期31-46,共16页
Population Journal
基金
国家社会科学基金项目:京津冀城市群流动人口集聚机制与态势研究(19CRK020)。
关键词
人口预测
概率人口预测
贝叶斯分层模型
确定性预测
情景预测
Population Projection
Probabilistic Population Projection
Bayesian Hierarchical Model
De⁃terministic Projection
Projection Scenarios