Can migration reduce benefits of spatial programs ? A model of congestion externalities with evidence from South Africa ∗

This paper demonstrates that migration lowers welfare gains from new local public goods (like electricity, hospitals or schools) by congesting inelastically supplied local amenities. Our simple model shows that the structure of the market for local amenities matters for welfare: Program-induced migration will be higher and congestion externalities larger when markets for local amenities (like land) are missing. To illustrate, we estimate welfare bounds for a location-based program using a recent example that provides consistent estimates of income and population effects of household electrification in South Africa. We provide new evidence that this program led to significant congestion in local schools. Then, we use our model to show that migration in response to household electrification reduced local welfare gains by half. Our paper argues that because of congestion externalities, evaluations of spatial programs should account for migration. Moreover, we show how this might be done without land price data and when markets for land and other local amenities are missing. [160 words]