Spatial education inequality for attainment indicators in sub-saharan Africa and spillovers effects

Space plays a prominent role on educational inequalities. Spatially proximate communities are likely to behave and perform similarly than spatially distant communities because educational processes, demand and supply factors, are often location specific within a country, with educational outcomes and educational inequalities being spatially dependent. Yet, studies on monitoring education inequalities linked to SDG4 indicators have ignored the crucial role of spatial dependence and failed to look at granular educational inequality beyond standard urban/rural and country's regions classifications. In this paper, we account for social dependence among communities to assess spatial education inequalities for the sub-Saharan Africa (SSA) region by relying on the geo-localisation of 16,000 communities for 29 countries based on DHS surveys. We use an array of education indicators across the lifecourse (completion rates from primary to tertiary) and measures of attainment and for risk of dropout (primary over -age), allowing us to measure how spatial dependence of educational outcomes changes at varying levels of education. We employ mapping, spatial correlations statistics and spatial regression models to account for the spatial dependence and endogeneity among communities' educational performance shaped by their contextual factors and to derive education spillovers. Our study's findings for the SSA region can be grouped as: space matters for communities educational performance, even after accounting for various community -level observables; educational spatial dependence operates more powerfully in marginalised communities; and that space matters indirectly through contextual factors of nearby communities in the form of educational externalities. The overreaching implication of our study is that commonly used geographical categories of rural-urban, or regions within countries are not adequate to address educational challenges and studies should place more emphasis on GIS-based analysis.