This study combines household survey and census data to generate estimates poverty and inequality at the provincial, district, and commune levels in Vietnam. It then explores the ability of agro-climatic variables and market access to "explain" the spatial variation in poverty. Finally, it carries out a spatially-weighted local regression analysis that reveals how the geographic determinants of poverty vary across the country.

The results indicate that poverty is greatest in the north and in rural upland areas and least in urban areas, with the delta areas having intermediate poverty rates. The results generally match conventional wisdom, though they differ markedly from district poverty estimates currently used by the government to target anti-poverty programs. Inequality is greater in urban areas than in the rural delta regions, but it is surprisingly high in the rural upland areas. Rural poverty is significantly related to the type of land cover, the average slope, soil type, and distance to small urban centers. Distance to large urban centers, however, was not statistically significant. The local regression analysis reveals that the effect of agro-climatic variables and market access on poverty varies significantly from one region to another. The implications of these findings for policy are discussed.