This ePaper investigates the non-linear effects of geo-referenced World Bank aid projects on economic development at the sub-national level, measured as night-time luminosity. The data framework is based on a grid cell structure at a 0.5 x 0.5 decimal degree resolution, covering approximately 10,600 grid cells across 54 African countries, over the period of 1992 to 2014. This approach addresses endogeneity concerns associated with sample selection and reverse causality. Using a fixed effects quantile regression approach, I estimate the impact of foreign aid at distinct levels of development within countries. Overall, the results suggest a positive and statistically significant effect of aid on night-time luminosity, with the largest impact observed within relatively poorer grid cells. In addition, there is evidence of spill-over aid effects from neighbouring grid cells. These findings are however sensitive to different model specifications and variable transformations.