The optimal distribution of population across cities

The received economic wisdom is that cities are too big and that public policy should limit their sizes. This wisdom assumes, unrealistically, that city sites are homogeneous, migration is unfettered, land is given freely to incoming migrants, and federal taxes are neutral. Should those assumptions not hold, large cities may be inefficiently small. We prove this claim in a system of cities with heterogeneous sites and either free mobility or local governments, where agglomeration economies, congestion, federal taxation, and land ownership create wedges. A quantitative version of our model suggests that cities may well be too numerous and underpopulated for a wide range of plausible parameter values. The welfare costs of free migration equilibria appear small, whereas they seem substantialwhen local governments control city size.