This paper uses recently released município-level data on rural poverty in Brazil to identify and analyze spatial patterns of rural poverty in the São Francisco River Basin (SFRB). Moran's I statistics are generated and used to test for spatial autocorrelation, and to prepare cluster maps that locate rural poverty 'hot spots' and 'cold spots'. Our results indicate that poverty reduction policies in the SFRB should take into account the spatial distribution of poverty. Not only is poverty in the SFRB clustered spatially, but the bulk of the basin's poor resides in municípios that comprise the poverty 'hot spots' we identified. These clusters did not correspond to state-level boundaries (the political delineations often used to measure poverty and to manage poverty reduction programs), so scope may exist for geographically refocusing poverty reduction efforts to make them more efficient. Maybe more importantly the results set the stage for the use of spatial econometrics for a future multivariate analysis of rural poverty in the basin.
rural poverty; cluster analysis; spatial autocorrelation; São Francisco River Basin
