Southern African Regional Poverty Network (SARPN) SARPN thematic photo
Country analysis > South Africa Last update: 2020-11-27  

 Related documents

Experiences with the development and use of poverty maps:
Case study note for South Africa

Contributors: Miriam Babita (Statistics South Africa), Berk Цzler (World Bank)
Contact: ,

Edited by Mathilde Snel, Miriam Babita, Norbert Henninger

A summary of all case studies can be found at: or
[Download complete version - 135Kb ~ 1 min (6 pages)]     [ Share with a friend  ]

  1. Background information on the poverty mapping initiative

    The development of poverty maps in South Africa was originally initiated due to an interest in exploring their possible use in facilitating the allocation of municipal grants. Since 1998, the Division of Revenue Act (#28) has required the equitable distribution of nationally raised revenue to municipalities based on poverty levels. Under this act, unrestricted municipal grants are allotted based on the number of poor households and costs associated with providing household-level basic services. Originally, census data on income were used to determine the number of poor households in each municipality. However, the use of these data raised concerns, as the South African census, unlike most income surveys, did not use a detailed module, but rather relied on one brief question on income (Alderman et al. 2001).

    In 1999, Harold Alderman, resident World Bank staff in South Africa, approached the then Deputy Director General of Statistics South Africa (Statistics SA) about using a new methodology combining census and survey data to estimate the number of poor households and generate highly disaggregated poverty maps (Hentschel et al. 2000). The process would involve two stages: 1) determining whether poverty estimates based on census data were comparable to those based on the Income and Expenditure Survey (IES) data (the best available data on income in South Africa at the time, but limited to provincial aggregates only), and 2) producing a highly disaggregated poverty map based on the better measure of poverty. If the outcome of the first stage revealed that census income data were a poor measure of household welfare compared to IES income data, then a highly disaggregated poverty map would be created by imputing expenditure from the census data using IES data. (In general, consumption expenditure produces more reliable household welfare descriptions than household income.) Initially, there was some reluctance to conduct this assessment, because of data compatibility concerns (e.g., between the 1991 and 1996 censuses) and data access issues (e.g., releasing census data to an external organization) (see Section 2). However, senior-level support was eventually obtained.

Octoplus Information Solutions Top of page | Home | Contact SARPN | Disclaimer