Testing for Localization with Entropy-Based Measures

This paper aims to give statistical significance to the measurement of spatial concentration in the context of entropy-based approaches. We simulate confidence intervals based on a null hypothesis able to capture systematic spatial concentration of firms from random patterns, and dissimilarities between the distributions of firms and employees. We implement this two-step methodology to the European manufacturing economy, and we find a substantive spatial clustering of establishments whereby the spatial divergence between employees and firms is significant both for small-scale industries typically considered as localized because of industry-specific Marshallian external economies and for those industries characterized by considerable internal scale economies. We suggest that a high heterogeneity in firm size may have positive implications for aggregate competitiveness at the sectoral level.