Assessing the tendency of Spanish manufacturing industries to cluster : co-localization and establishment size

In this paper, we evaluate the spatial location patterns of Spanish manufacturing firms and we assess the different tendencies to cluster in each industry relative to the whole of manufacturing. To do this, we use a distance-based method (Marcon and Puech, 2003; Duranton and Overman, 2005), more concretely the Ripley's K function, which measures concentration by counting the average number of neighbours of each firm within a circle of a given radius. This method allows us to treat space as continuous, analysing simultaneously multiple spatial scales and avoiding the shortcomings of the administrative scale. In addition, we employ a polygonal envelope to improve the delimitation of our area of study, substituting the rectangular shape used by other authors and thus avoiding the nuisance of empty spaces. We apply this method to Spanish manufacturing sectors at two-digit and four-digit level, isolating like this the different behaviours of spatial distribution of each subsector caused by 'spillovers' characteristic of each activity and also preventing compensation effects due to previous aggregation. Furthermore, we examine the co-localization between horizontally-linked and vertically-linked industries to assess the importance of these spillovers across industries and, finally, we try to answer what type of establishment, depending on its size, is the driver of the Spanish industrial agglomeration.