Geographic concentration, observational equivalence and the source of industrial agglomeration

Ellison and Glaeser's (1997) index of geographical concentration distinguishes between natural advantages and spillovers as a source of industrial agglomeration, but the well-known 'observational equivalence' means little is known about the relative importance of these. This paper uses the difference in the temporal scope of the agglomeration source to decompose the index, and sets out a methodology for measuring each of these using the frequency estimator approach of Maurel and Sédillot (1999). When applied to a dataset on foreign investment it shows spillovers decay and on average extend over three to five years. An implication is that the geographic concentration index will mainly reflect natural advantages, revealing comparatively little about spillovers.