Industry Agglomeration in a Developing Economy: Evidence from India

Transport corridor investments absorb enormous capital in settings where fiscal space is scarce, yet rigorous evidence on their economic returns remains thin -- particularly in Sub-Saharan Africa. We estimate the local economic effects of upgrading two segments of Tanzania’s Central Corridor, using a staggered difference-in-differences strategy on a panel of 10 km × 10 km grid cells covering 2003-2018. We take satellite-derived per capita consumption (Huber and Mayoral, 2024) as our primary outcome and define control units as those grid cells proximate to other major highways that were not exposed to upgrades. We find that the two upgrades have considerably different effects: relative to the control-group trend, consumption near the Singida-Shelui segment (completed 2009) increases steadily until it is 14 percent higher after eight years, while consumption near the Lusahunga-Kagera segment (completed 2013) falls around 3 percent over the five years we observe post upgrade. Our central finding is that neither effect is homogeneous within its catchment. Initially larger and richer cells capture the bulk of the gains along Singida-Shelui, while smaller and poorer cells account for all of the negative effects along the Lusahunga-Kagera improvement. Better roads may therefore not merely distribute gains unequally: where the local economy is thin, improved connectivity can leave an area absolutely worse off. Secondary findings include the fact that population falls persistently near both segments and in rich and poor areas alike, which is consistent with out-migration toward newly accessible urban centres. The results hold against an alternative control group drawn from cells near planned-but-uncompleted projects, while a synthetic difference-in-differences specification preserves the Singida-Shelui gains while rendering the Lusahunga-Kagera decline statistically insignificant.