Disproportionality measures of concentration, specialization, and polarization

The paper extends the methodological toolbox of measures of industrial concentration and regional specialization. First, a taxonomy is proposed which gives rise to a modular construction system for disproportionality measures based on three characteristic features: the projection function, the reference distribution, and the weighting scheme. The taxonomy helps reduce the mismatch between research purpose, data and statistical measure which has been one of the major obstacles to reliable inferences in the literature. Second, the taxonomy is extended (i) to measures of polarization which evaluate specialization and concentration simultaneously and allow for a nested analysis at different spatial and industrial scales, and (ii) to spatial concentration measures which deal with the checkerboard problem and MAUP by taking into account information from neighboring regions.