Measuring Agglomeration by Spatial Effects: a Proposal

In this paper we examine the influence of spatial dependence on the estimation of the effect of standard measures of agglomeration on local growth. The hypothesis is that the forces of agglomeration do not use up their effects inside a single local area but extend to wider geographic areas, particularly on the adjacent territories. To account for these spatial effects, we propose a modification of the traditional agglomeration index, suggested by Maurel and Sédillot (1999), including both agglomeration of economic activity within an area and neighbour effects, that is, the effect of agglomeration that comes from neighbouring areas. In the empirical exercise we compute the a-spatial and the spatial indices for the set of Sicilian local labour systems (LLSs) using 21 manufacturing and services industries from the NACE-CLIO sectorial disaggregation. The results of the exploratory spatial analysis confirm the agglomeration index varies notably when spatial effects across areas are included: both the level of the index and the degree of spatial dependence increase when it is computed accounting for the spillovers across local areas. In addition, results from the estimation of a model of the local employment dynamics reveal the most of the residual spatial dependence - observed in this type of specification - can be accounted by the spatial measure of agglomeration. This suggests the existence of spatial spillovers that are not bounded to the single production systems. Moreover, we distinguish the different contribution of the traditional and spatial measure of agglomeration on the growth process of the employment, together with other productive and structural characteristics of the selected geographical units.