The Determinants of Localization: A Conditional Distance-Based Approach

Do pairs of plants with ‘close or similar’ input-output linkages, types of workers, and that use or exchange similar technology locate near one another in space? To answer these questions, I propose a new non-parametric approach to measuring the localization of ‘closely related’ multiple industries – i.e., a multidimensional way to assess coagglomeration – in continuous space. More precisely, I combine the measurement approach of localization in continuous space with a coagglomeration approach, and then relate them to the degree to which industries share goods, people, and ideas. My results show that plants which belong to manufacturing industries with similar input-output linkages or workforces tend to locate near one another. I find little evidence that plants that share similar technologies cluster geographically.