What Causes Industry Agglomeration? Evidence from Coagglomeration Patterns

TLDR

Firms often cluster geographically, yet the underlying drivers of such agglomeration remain unclear. The study tests Marshall’s theories by identifying which manufacturing industries coagglomerate. Coagglomeration indices were built from US Economic Census data, linked to shared goods, labor, or ideas, and validated using UK data and non‑collocated US regions to mitigate reverse causality. All three of Marshall’s agglomeration theories are supported, especially the input‑output linkage theory. JEL codes: L14, L60, O33, R23, R32.

Abstract

Why do firms cluster near one another? We test Marshall's theories of industrial agglomeration by examining which industries locate near one another, or coagglomerate. We construct pairwise coagglomeration indices for US manufacturing industries from the Economic Census. We then relate coagglomeration levels to the degree to which industry pairs share goods, labor, or ideas. To reduce reverse causality, where collocation drives input-output linkages or hiring patterns, we use data from UK industries and from US areas where the two industries are not collocated. All three of Marshall's theories of agglomeration are supported, with input-output linkages particularly important. (JEL L14, L60, O33, R23, R32)