An empirical tool to classify industries by regional concentration and spatial polarization

Traditional measures of geographical concentration of industries based on regional data (such as Gini, Herfindhal, and Ellison-Glaeser indices) do not consider the information about the spatial positions of regions. This implies their insensitivity to regions’ spatial order and inability to account for neighboring effects. As an attempt to cope with this limitation, a recent streamof literature [7, 4, 9, among others] has focused on developing measures that quantify the degree of concentration of an industry while adjusting for spatial connections among regions. Following the idea that a single measure cannot fully describe the characteristics of an industry in terms of both concentration and spatial interactions, this paper proposes an alternative approach that measures the two dimensions jointly and allows for the classification of economic sectors into meaningful types of geographical configurations.