We examine the impact of geo-referenced World Bank development programs on subnational development using equally sized grid cells with a spatial resolution of 0.5 decimal degrees latitude × longitude as the unit of investigation. The proposed grid cell approach solves a number of endogeneity problems discussed in the aid effectiveness literature that make it difficult to identify the true effect of foreign aid on development outcomes due to the presence of unobserved heterogeneity, lack of key country-level controls, aggregation bias, simultaneity and/or the presence of reverse causality in the association between foreign aid and economic growth, measurement errors, and endogenous sample selection bias. The estimates reveal that World Bank foreign aid projects contribute significantly to grid cell economic activity measured by night-time lights growth. This finding is robust to the presence of unobserved country-year and grid-cell-specific unobserved heterogeneity, and to the inclusion of a full set of grid-cell-specific socioeconomic, demographic, conflict-related, biogeographic, and climatic controls. Additional sensitivity tests confirm the robustness of the main findings to various econometric estimators, alternative model specifications, and different spatial aggregation levels.