小区域贫困指标估计方法的比较与模拟  

Comparison and Simulation of Poverty Index Estimation Methods in Small Regions

作　　者：于力超 Yu Lichao(College of Science,Minzu University of China,Beijing 100081,China)

机构地区：[1]中央民族大学理学院,北京100081

出　　处：《统计与决策》2020年第11期32-36,共5页Statistics & Decision

基　　金：国家社会科学基金青年项目(18CTJ011);全国统计科学研究重点项目(2017LZ01)。

摘　　要：精准扶贫战略思想已是我国的一项基本国策,其中扶贫对象精准是基础,首要工作是测度区域贫困指标,了解贫困人群的分布状况,随着区域划分精度的提高,一些区域调查样本量不足,需要借助小域估计方法提高指标估计精度。文章总结了四种适用于区域贫困指标估计的小域估计方法(即直接估计法、世界银行法、经验贝叶斯法和Fay-Herriot模型法)及其适用情形,并通过数值模拟,从无偏性和有效性两方面研究了四种方法估计区域贫困率指标的效果。The strategic thought of targeted poverty alleviation has become one of China’s basic state policies, in which the accuracy of poverty alleviation target is the foundation, and the first task is to measure the regional poverty index and understand the distribution of poor people. With the improvement of the precision of regional division, the sample size of some regions is insufficient, thus small region estimation method is needed to improve the precision of index estimation. This paper summarizes four kinds of small area estimation methods(direct estimation method, world banking method, empirical Bayesian method and Fay-Herriot model method) applicable to regional poverty index estimation, and their application situations as well. And through numerical simulation, the paper also makes a study on the effects of four methods to estimate regional poverty rate from two aspects of unbiasedness and validity.

关 键 词：区域贫困指标 小区域估计 经验贝叶斯法 Fay-Herriot模型 

分 类 号：C811[社会学—统计学]