Group-Level Policy with Network Spillovers: Theory and an Application to Entrepreneurship

We develop a framework for group-based intervention targeting when learning spillovers are mediated by networks and apply it to place-based entrepreneurship policy using a new dataset linking U.S. entrepreneurs on Facebook to the firms they own. Motivated by evidence of disparities across aspiring entrepreneurs’ networks, the model combines network position with place-specific spillovers to predict local treatment effects. We estimate spillovers using a new quasi-experimental design that exploits the timing of friends’ migration. The estimated model implies that policies aimed at reducing geographic deficits in firm creation face an equity-efficiency trade-off due to large positive spillovers in high-entrepreneurship places.