Local and global agglomeration patterns: Two econometrics-based indicators

We propose two model-based measures of the intensity of local and global agglomeration patterns at the industry level. Spatial plant dispersion in a given industry is modeled as the outcome of an error components process that provides a convenient framework for the simultaneous treatment of strictly localized and distance-related agglomeration forces. Having dealt with maximum likelihood inference on model parameters, tackling also the issue of data censoring, the methodology is subsequently implemented on Italian census data for both manufacturing and service industries and a comparison of findings with respect to standard agglomeration and spatial autocorrelation measures is undertaken.