KIBS and the city: GIS evidence from Milan

In this paper we assess the existence, and the magnitude, of agglomeration economies affecting the KIBS industry located in the city of Milan. By exploiting a rich firm-level dataset, we calculate distance-based measures of spatial agglomeration, in which we distinguish between own-industry, other-industry and related-industry neighbouring firms. Our OLS estimates show that TFP is positively affected by proximity to own-industry and related-industry firms, whereas it decreases with proximity to other-industry firms. We also observe that localization economies are strong at low distances, and rapidly decay over space. Differently, related variety effects materialize after one kilometre, but are more persistent over distance. Our quantile regressions show that these results depend also on the position of the firm over the TFP distribution. In particular, when looking at KIBS with the lowest performance, we find that only localization economies are at work, and on a very limited geographical scale. Differently, top-performance KIBS benefit from proximity to technologically-related firms, and across a wider spatial scale.