SAE - A Stata Package for Unit Level Small Area Estimation

Nguyen, Minh Cong ; Corral, Paul ; Azevedo, Joao Pedro ; Zhao, Qinghua

World Bank, Washington, DC

关键词: SMALL AREA ESTIMATION;    POVERTY MAPPING;    BIG DATA;    GEOSPATIAL ECONOMICS;    STATA;   

DOI  :  10.1596/1813-9450-8630
RP-ID  :  WPS8630

学科分类：社会科学、人文和艺术（综合）

来源: World Bank Open Knowledge Repository

【 摘 要 】

This paper presents a new family ofStata functions devoted to small area estimation. Small areamethods attempt to solve low representativeness of surveyswithin areas, or the lack of data for specificareas/sub-populations. This is accomplished by incorporatinginformation from outside sources. Such target data sets arebecoming increasingly available and can take the form of atraditional population census, but also large scaleadministrative records from tax administrations, orgeospatial information produced using remote sensing. Thestrength of these target data sets is their granularity onthe subpopulations of interest, however, in many cases theylack the ability to collect analytically relevant variablessuch as welfare or caloric intake. The family of functionsintroduced follow a modular design to have the flexibilitywith which these can be expanded in the future. This can beaccomplished by the authors and/or other collaborators fromthe Stata community. Thus far, a major limitation of suchanalysis in Stata has been the large size of target datasets. The package introduces new mata functions and a pluginused to circumvent memory limitations that inevitably arisewhen working with big data. From an estimation perspective,the paper starts by implementing a methodology that has beenwidely used for the production of several poverty maps.