|
In recent years a technique has been developed to obtain generate poverty estimates for small geographical areas where the available data is not representative. This estimation method, in popular terms called poverty mapping, combines data sources of the population in total, with data from surveys of household income and consumption. A population wide census covers the entire population but only provides a limited amount of information, while data from household surveys provide detailed information, but only for of a limited sample.
It is possible to obtain poverty estimates based on a household survey, however the level of disaggregation is limited by the survey’s sample size. This means that poverty estimates based on the 1991 Household Budget Survey (HBS) for Tanzania are limited to Dar es Salaam, larger towns, small cities and rural areas. For the data collected during 2000/01, poverty estimates were generated for urban and rural areas for each of the regions. Using the small area estimation method (poverty mapping), the 2005 Poverty and Human Development Report (PHDR)1 published the first attempt to obtain poverty estimates below the regional level for Tanzania, being at the district level2. To generate these estimates the 2000/01 Household Budget Survey was used in conjunction with the 2002 Population Census.
While the poverty mapping technique was originally developed to obtain poverty estimates at lower levels of spatial aggregation, it is can also being used to obtain estimates over time. The latter entails combining a survey with containing consumption information with other surveys at different points in time. The high costs involved in Household Budget Surveys limits the frequency of collecting information on household expenditure, making it appealing to use survey-to-survey poverty mapping to obtain poverty trends. The survey-to-survey small area estimation method was applied by the International Food and Policy Research Institute (IFPRI)3 (2006) for Tanzania. This study combined the 1991/92 Household Budget Survey with four different Demographic and Health Surveys (DHS)4.
The main assumption of the IFPRI approach is that ‘the model for predicting income based on household characteristics is valid over a range of years covered by the DHS surveys’ (IFPRI 2006). This discussion paper assesses the validity of this assumption.
Footnotes:
-
The Research and Analysis Working Group, (United Republic of Tanzania), Poverty and Human Development Report 2005 (PHDR), Mkuki na Nyota, Dar es Salaam
-
The 2002 Population and Housing Census had basic demographic information for the full 100% of the population, but more detailed population and household information was collected for a 20% sample. This led to just having poverty estimates at district level, and not as preferred at division or ward level.
-
International Food and Policy Research Institute (2006), Poverty and malnutrition in Tanzania: New approaches for examining trends and spatial patterns, Total Design, Dar es Salaam
-
In the IFPRI study the 1991/92 and the 1996 Demographic and Health Survey, the 1999 Reproductive and Child Health Survey and the 2003 Tanzania HIV/AIDS Indicator Survey were used. The last survey was not an actual demographic and health survey, but methodology, sampling, as well as basic questions were similar to the other surveys used.
|
|
Related documents
- 316Kb ~ 2 min (8 pages)
