GDEST 2008 Conference Sessions
Theme 1 - Observing Africa
Geographically-Informed Policy Response and Intervention: Modeling Spatial Non-Stationarity in the Analysis of Poverty Determinants
Paul Belanger, UNECA, Ethiopia
Both in models of economic growth and poverty alleviation, conventional modeling techniques, esp. regression, using spatial units as observations too often accept that the functional relationships between the process being model and its predictors are spatially invariant. This leads uncritically to inferences about policy and various operational and strategic interventions that may be more or less relevant in particular locales. Contrasting a global model with a spatially-aware model cast in a geographically-weighted regression framework, this paper seeks to demonstrate how the latter can empirically recover spatially varying relationships and, thus, suggest geographically tailored policy responses and interventions that are sensitive to the significant intra-unit variance present in most African member States.