Economic agglomeration in Italy before and after the Great Recession

This dissertation examines agglomerations in Italy, analyzing and measuring their extent and their patterns of change during the five-year period characterized by the Great Recession. An innovative distance-based method – Marcon and Puech’s M (2010) – is used to measure agglomeration in a way that aims to overcome the multiple issues (notably among them the Modifiable Areal Unit Problem) that affect more conventional measures of agglomeration, while still avoiding the unmanageable computational intensity that would be required by a standard implementation of M. Indeed, by approximating the geographic position of each plant to the centroid of the municipality where it is actually located, we believe that not much is lost in terms of accuracy, especially when considering the relatively small size of Italian municipalities. Such an accurate and innovative measure of agglomeration for Italy would constitute a significant attainment by itself, but even more so when considering the opportunity it provides for evaluating the change in spatial economic patterns during a peculiar period such as the Great Recession. After a reconstruction of the most relevant literature concerning the spatial distribution of economic activities – ranging from the pioneering works of Von Thünen to Evolutionary Economic Geography – and a synthesis of distance-based methods, we rely on ISTAT’s ASIA dataset to correctly identify each plant’s geographic location and number of employees for both manufacturing and services, producing what we believe is a reliable measure of agglomeration for each industry in the entire country for 2007 and 2012. We then proceed to analyze the change that occurred in such a period of crisis, and to investigate correlations with a certain number of variables. Finally, we compute the same index again for each one of four different macro-regions: the rich northern part of the country, the peninsular and more heterogeneous Center-South, and the two major islands of Sicily and Sardinia. Each macro-region is treated as a separated entity, in order to evaluate whether determinants behind the agglomeration results obtained at the country-level might be hidden by opposite-sign patterns that would instead become apparent when areas are treated separately. The inevitable limits of a work based on such an immense amount of data and computations are both methodological and analytical, and we aim to confront them in future research. Indeed, although the results are surely interesting and provide a useful ranking of agglomeration for Italian industries, the next step will necessarily involve the estimation of its statistical significance. Such a task is certainly beyond the scope of this dissertation, because of the massive amount of time and processing capabilities required to perform the necessary tests. Assessing the significance of the results will certainly help when performing a deeper analysis of agglomeration determinants and when studying the change occurred during the Great Recession. Finally, we also aim to assess the reliability of our methodology by computing M without any approximation whatsoever for an area that might be large enough to be economically representative, but small enough to be computationally dealt with: we believe that the area of choice could be our land, Sardinia.