Poverty mapping in Uganda: Extrapolating household expenditure data using environmental data and regression technique

Abstract

This ‘Working Paper’ series pulls together different strands of work on the wide range of topics covered by the Animal Production and Health Division with the aim of providing ‘fresh’ information on developments in various regions of the globe, some of which is hoped may contribute to foster sustainable and equitable livestock sector development.The work described in this paper follows directly on from earlier attempts to develop a novel approach to mapping poverty using environmental data. The aim was to get closer to understanding some of the underlying causes of poverty – something that is unlikely to be feasible using approaches based only on socio-economic data such as the traditional small area estimate (SAE) techniques. The environmental poverty mapping technique involved modelling geo-registered household expenditure estimates in Uganda, available from household surveys, using discriminant analysis of a range of environmental data – mostly derived from satellite remote sensing. This analysis was successful, resulting in a series of poverty maps and lists of environmental variables that were strongly correlated with poverty at different spatial resolutions.