Poverty mapping is one of most common applications in small area estimation. Many examples are available for the proportion of population living below the international or national poverty line (indicators 1.1.1 and 1.2.1).
World Bank applications
The World Bank proposed a poverty mapping process that was conducted in several countries. Based on surveys and additional data sources, various poverty and inequality estimates such as the Foster-Greer-Thorbecke poverty estimates and the Gini coefficient were derived.
The report More than a pretty picture - Using poverty maps to design better policies and interventions published in 2007 shows case studies for the countries Albania, Bolivia, Bulgaria, Cambodia, Yunnan Province (China), Ecuador, Indonesia, Mexico, Morocco, Sri Lanka, Thailand and Vietnam that describe all poverty mapping steps and also lessons learned. Hence this can be a good starting point for a new poverty mapping study.
In 2005, the World Bank provided technical assistance to the Philippine national statistical system to leverage on small area estimation techniques to produce municipality- and city-level poverty statistics. The Philippine Statistics Authority conducts the Family Income and Expenditure Survey (FIES), which is the main source of official poverty statistics in the country, every three years. The small area estimation technique used in the Philippines is based on the ELL method. It entails regressing (log) per capita household income from the FIES with auxiliary information from the FIES, the Labor Force Survey, and the Census of Population and Housing. The model regressors include survey-obtainable variables such as educational attainment of the household head and other household characteristics, and census-derivable information like average family size in a village, and other village-level information. Since small area poverty statistics became available in 2005, numerous government agencies have used these data as inputs for formulating and implementing poverty reduction programs. For example, the Philippine Department of Social Welfare and Development (DSWD) used the estimates to identify poor municipalities for its National Household Targeting System for Poverty Reduction (NHTS-PR) data collection.
Indicators | Disaggregation dimension | Data availability | Estimation approach | Model |
---|
1.1.1/1.2.1 | Spatial | Unit-level survey and auxiliary data | Model-based estimation | ELL |
Poverty estimation in Chilean comunas
To improve fund allocations among comunas, the Chilean Ministerio de Desarrollo Social (in the following the ministry) is required to provide poverty estimates for all 345 comunas in Chile which is the smallest territorial entity. After the evaluation of various options, the ministry decided to combine the National Socioeconomic Characterization Survey (CASEN), which is Chile's official data source for poverty statistics, with relevant administrative records. Since 2011, model-based poverty statistics are obtained for Chilean comunas.
Indicators | Disaggregation dimension | Data availability | Estimation approach | Model |
---|
1.2.1 | Spatial | Unit-level survey and area-level auxiliary data | Model-based estimation | Arcsin-transformed area-level model |
Small Area Income and Poverty Estimates (SAIPE) program by the U.S. Census Bureau
The SAIPE program produces small area estimates of income and poverty statistics for all school districts, counties, and states. The estimates are based on several data sources such as the American Community Survey and Federal Income Tax Returns. The produced indicators do not exactly follow the definition of the SDGs but the example is added since the SAIPE program is continuously improving their approach and the disaggregation is not only spatial but also by age groups.
Indicators | Disaggregation dimension | Data availability | Estimation approach | Model |
---|
| Spatial and age | Unit-level survey and area-level auxiliary data | Model-based estimation | Log-transformed area-level model |
Mean income in Middle-layer Super Output Areas in England and Wales
Super Output Areas (SOAs) are a geographic hierarchy introduced for the reporting of small area estimates. The mean population of Middle-layer SOAs (MSOAs) ranges from a minimum of 5,000 to 7,200. To obtain different income estimates (e.g., equivalised and unequivalised) for the MSOAs, data from the Family Resources Data is combined with additional data including Census information, energy consumption and house price statistics using a linear mixed model.
The case study is explained on the homepage with more methodological details in the technical report.
Indicators | Disaggregation dimension | Data availability | Estimation approach | Model |
---|
| Spatial | Unit-level survey and area-level auxiliary data | Model-based estimation | Log-transformed unit-level model |