Agglomeration: A Dynamic Approach

This paper studies the sources of agglomeration economies in cities. We begin by introducing a simple dynamic spatial equilibrium model that incorporates spillovers within and across industries, as well as city-size effects. The model generates a dynamic panel-data estimation equation. We implement the approach using detailed new data describing the industry composition of 31 English cities from 1851-1911. We find that industries grow faster in cities where they have more local suppliers or other occupationally-similar industries. Industries do not grow more rapidly in locations in which they are already large, though there can be exceptions. Thus, dynamic agglomeration appears to be driven by cross-industry effects. Once we control for these cross-industry agglomeration effects, we find a strong negative relationship between city size and city-industry growth. This allows us to construct the first estimate of the aggregate strength of the cross-industry agglomeration forces. Our results suggest a lower bound estimate of the overall strength of agglomeration forces equivalent to a city-size divergence rate of 2.1-3.3 % per decade.