Chasing opportunity: spillovers and drivers of US state population growth

Abstract

We examine the drivers and spatial diffusion of US state population growth using a dynamic spatial panel model over the period 1965–2017. Methodologically, the spatial network is recovered from the data rather than imposed a priori, and estimation framework permits heterogeneous slopes and interactive fixed effects. Population growth displays heterogeneous conditional convergence: around three-quarters of states converge, while a small high-growth group diverges mildly. Core drivers such as amenities, labour income and migration frictions are robust across network specifications, whereas productivity effects arise only under data-inferred networks. Spatial spillovers are economically meaningful, accounting for roughly one-third of total effects and extending beyond contiguous neighbours.