Environmental Approaches to Poverty Mapping: an example from Uganda

Abstract

To be of real value to governments and development agencies, poverty maps should go beyond describing the distribution of poverty, to help explain and thence predict its spatial distribution. Poverty maps are traditionally produced by exploiting links between extensive census data and intensive socio-economic household survey data; relationships found within the survey data are extended to census data, through variables common to both data sets. Many of the dimensions of poverty are environmentally related; people are poor because they are unhealthy, or underfed, or without access to fuel and water etc. We suggest that a more useful approach to poverty mapping might be first to identify its (environmental) causes. In this analysis, we explore a novel approach that combines household survey data from Uganda with a suite of environmental variables that are either direct measures of key climatic variables (such as temperature), descriptor variables of key ingredients of poverty-generating processes (such as agricultural production systems) or proxies for constraints on the health and well-being of the human populations (such as disease-causing pathogens). This potentially allows us to move beyond description, to explain and then predict the distribution of poverty at the spatial resolution of the predictor variables. Whilst correlation obviously does not automatically imply causation, we suggest an environmental approach to poverty mapping is more likely to reveal causes than the traditional, small area approaches. This paper takes the first steps towards establishing the predictive accuracy of an environmental approach to poverty mapping.