Do suburbs need cities?

In this paper I examine the relationship between city and suburban growth over the last three decades for a sample of U.S. metropolitan areas. I develop a structural empirical model relating city income growth to suburban growth in income, population, and house values. The model allows for bidirectional effects of cities on suburbs and suburbs on cities, as well as for unobserved factors affecting both city and suburbs. The simultaneous, latent-variable model is identified using a combination of exclusion and covariance restrictions. Instrumental estimation results indicate that income growth in large cities enhances suburban growth; but income growth in small cities has little effect.