District poverty in Mozambique
Small area estimates for all 157 districts
In progress
In plain language
National surveys provide a good picture of poverty in Mozambique as a whole and in each province. However, decisions about schools, clinics, roads, and cash transfers are often made at district level, and the surveys do not interview enough households in each district to produce reliable estimates on their own.
This project aims to fill that gap. It combines the latest Demographic and Health Survey with open geospatial data, such as satellite measures of night-time lights, land cover, and access to services, to estimate poverty in each of Mozambique’s 157 districts. Every estimate is reported with a measure of its uncertainty, which makes it clear where the figures are robust and where more data would be needed.
Approach
- Data: the 2022–23 Mozambique Demographic and Health Survey, linked to open geospatial covariates at district level.
- Model: spatial Fay-Herriot small area models fitted in R-INLA, which borrow strength from neighbouring districts and from the covariates.
- Uncertainty: calibrated intervals for every district, not just a point estimate.
- Testing: a simulation study checks that the method recovers known values before it is applied to real data.
- Data protection: only aggregated results and code are published, and no survey microdata or confidential locations are released.
Interactive map
Open the map full screen, or see the code and full method on GitHub.