The application of a spatial regression model to the analysis and mapping poverty
By Alessandra Petrucci, Nicola Salvati, Chiara Seghieri
University of Florence
Poverty mapping in developing countries has become an important tool in identifying ways to improve living standards. The methods used to generate poverty maps have come under closer scrutiny as their policy implications become more apparent. Those most commonly used until now have used econometric models to generate local indicators of poverty.
Most of these econometric models do not take into account the spatial dependence that may exist in human societies with regard to income distribution. For example, poor households are more likely to be close to other poor households than they are to be close to higher income households.
In this report, the authors use spatial regression to model more accurately the distribution of poverty across regions in Ecuador. The difference between results that are adjusted for spatial patterns and the unadjusted results is statistically significant. Although the absolute differences are not dramatic, they do provide policy planners with greater confidence that the results reflect the real situation in that country.
Although the geographic focus of this paper is on Ecuador, its main contribution is methodological, mainly the comparison of results from models that apply spatial regression techniques with those that do not.
FAO is grateful to the Government of Norway for its support to this work.