Do birds of a feather flock together?: Economic linkage and geographic proximity

Spatial association patterns of manufacturing activities are examined in this paper with the corresponding economic linkage patterns. Four specifications are used to measure the spatial association pattern: intraindustry/intracounty (locational Gini), intraindustry/intercounty (Moran’s I), interindustry/ intracounty (correlation coefficient) and interindustry/intercounty (“spatial” correlation coefficient). Two sets of spatial specification were used for the different locational context: unconstrained (full dataset) and constrained (dataset without zeros). The result on the 3,110 US counties for 361 manufacturing sectors revealed that, in the intraindustry context, there is little proof that stronger economic linkage results in and/or from a more concentrated pattern of the industry. However, interindustry economic linkage reflected and/or was reflected from the spatial distribution pattern in a significantly positive way. There was a pattern in the intraindustry model that the industry that showed clustering at a county scale had relatively weaker spatial concentration at a multi-county scale and vice versa. Results of the unconstrained and constrained specification of the data revealed important differences, implying that special care should be taken in the spatial specification to be used and how the results are to be interpreted. However, the relationship between economic linkage and spatial proximity did not substantially change between two cases except Moran’s I that had conflicting signs.