Mapping Poverty Hotspots and Infrastructure Deficits in Ghana: A Spatial Analysis of Household Welfare Using GLSS 7

Abstract This study examines the spatial distribution of poverty in Ghana and the geographical concentrations of deprivation. Drawing on the seventh round of the Ghana Living Standards Survey (GLSS 7), complemented by OpenStreetMap spatial data and district-level boundary shapefiles, the analysis employs a multi-stage empirical strategy covering GIS-based poverty mapping, global and local spatial autocorrelation tests, Getis-Ord Gi* hotspot identification, and spatial regression modelling. Global Moran's I statistics confirm strong positive spatial autocorrelation in poverty rates (I = 0.47, p < 0.001), rejecting the null hypothesis of random distribution. Local Moran's I clustering reveals statistically significant High-High poverty clusters concentrated in the three Northern regions, Upper East, and Upper West. Getis-Ord Gi* hotspot analysis corroborates these findings, identifying a contiguous northern poverty belt covering about 42 per cent of all districts. Spatial regression results (from both spatial lag and spatial error models) show that road network density, proximity to health facilities, market accessibility, and school density together explain 68.3 per cent of cross-district variation in poverty incidence (spatial error model), with road density and health facility access emerging as the strongest predictors. Robustness checks using alternative poverty measures and spatial weights matrices confirm the stability of these estimates. The findings suggest that spatially targeted infrastructure investment, particularly in transport connectivity and primary healthcare, is necessary for dismantling Ghana's persistent north-south welfare divide. investigates whether infrastructure deficits are associated with