SEARCH

SEARCH BY CITATION

Keywords:

  • Africa;
  • malnutrition;
  • hunger;
  • spatial data analysis;
  • spatial autocorrelation

Abstract

Hunger is a major problem in sub-Saharan Africa, and unlike other world regions the problem seems to be worsening. This paper seeks to determine if, when controlling for income and the health conditions, biophysical and geographical variables help to explain variation in the rates of child malnutrition. The literature on this subject is inconclusive precisely because the survey data, upon which most studies of malnutrition rely, rarely include variables that measure these factors. Examples of biophysical and geographic factors often cited in the literature include the balance of rainfall to evapotranspiration, the productivity of agricultural lands, distance to urban areas, topography, the frequency of drought, malaria endemicity, and market access through road networks. This paper introduces these variables into an analysis of 367 sub-national units in Africa based on household demographic survey strata. Running a spatial error model, and controlling for income, only three variables were found to be significantly correlated with child malnutrition: drought prevalence, the percentage of households with piped water, and diarrheal disease prevalence. This contrasts with the findings of earlier studies that sought to introduce similar biophysical/geographical variables into analyses of child malnutrition and mortality, as well as with ordinary least squares regression results presented in this paper, which found much larger sets of biophysical and geographical variables to be significantly correlated with malnutrition, thus highlighting the importance of addressing spatial autocorrelation in studies of this kind. The paper concludes with policy recommendations and a brief assessment of the strengths and shortcomings of continental scale analyses on the drivers of hunger and poverty. Copyright © 2009 John Wiley & Sons, Ltd.