Methods for detecting spatial clustering of economic activities using micro-geographic data

This PhD thesis consists of three self-contained but related essays on the topic of empirical assessment of spatial clusters of economic activities within a micro-geographic framework. The tendency of economic activities to be concentrated in a specific territory is well recognized, starting at least from the seminal studies by Alfred Marshall (Marshall, 1920). This spatial behaviour is not fortuitous; by concentrating in some areas firms enjoy a number of advantages, which then have implications for local economic growth and regional disparities and, as a consequence, are object of study in the fields of economics, geography and policy making. It has been recognized, however, that a major obstacle to further comprehension of the agglomeration phenomena of firms is the lack of a method to properly measure their spatial concentration. The most traditional measures employed by economists, indeed, are not completely reliable. Their most relevant methodological limit lies in the use of regional aggregates, which are built by referring to arbitrary definitions of the spatial units (such as provinces, regions or municipalities) and hence introduce a statistical bias arising from the chosen notion of space. This methodological problem can be tackled by using a continuous approach to space, where data are collected at the maximum level of spatial disaggregation, i.e. each firm is identified by its geographic coordinates, say (x, y), and spatial concentration is detected by referring to the distribution of distances amongst economic activities. The main purpose of the dissertation is to contribute to the development of this kind of continuous space-based measures of spatial clustering. The scientific context and motivation are outlined in depth in the first three chapters. Then the first essay introduces the space–time K-function empirical tool, proposed in spatial statistical literature, into economic literature in order to detect the geographic concentration of industries while controlling for the temporal dynamics that characterize the localization processes of firms. The proposed methodology allows to explore the possibility that the spatial and temporal phenomena, producing the observed pattern of firms at a given moment of time, interact to provide space–time clustering. The presence of significant space–time interaction implies that an observed pattern cannot be explained only by static factors but that we should also consider the dynamic evolution of the spatial concentration phenomenon. Indeed, for example, new firm settlements may display no spatial concentration if we look separately at each moment of time and yet they may present a remarkable agglomeration if we look at the overall resulting spatial distribution after a certain time period. In general, without knowing the temporal evolution of the phenomenon under study it is not possible to identify the mechanism generating its spatial structure. As a matter of fact, different underlying space–time processes can lead to resulting spatial patterns which look the same. The methodology is illustrated with an application to the analysis of the spatial distribution of the ICT industries in Rome (Italy), in the long period 1920–2005. The problem of disentangling spatial heterogeneity and spatial dependence phenomena when detecting for spatial clusters of firms is the topic of the second essay, “Measuring industrial agglomeration with inhomogeneous K-function: the case of ICT firms in Milan (Italy)”. Spatial clusters of economic activities can be the result of two distinct broad classes of phenomena: spatial heterogeneity and spatial dependence. The former arises when exogenous factors lead firms to locate in certain specific geographical zones. For instance, firms may group together in certain areas in order to exploit favourable local conditions, such as the presence of useful infrastructures, the proximity to the communication routes or more convenient local taxation systems. The phenomenon of spatial dependence, which is often of direct scientific interest, occurs instead when the presence of an economic activity in a given area attracts other firms to locate nearby. For instance, the presence of firms with a leading role encouraging the settlement of firms producing intermediate goods in the same area or the incidence of knowledge spillovers driving industrial agglomerations. This essay suggests a parametric approach based on the inhomogeneous K-function that allows to assess the endogenous effects of interaction among economic agents, namely spatial dependence, while adjusting for the exogenous effects of the characteristics of the study area, namely spatial heterogeneity. The approach is also illustrated with a case study on the spatial distribution of the ICT manufacturing industry in Milan (Italy). The third paper is titled “Weighting Ripley’s K-function to account for the firm dimension in the analysis of spatial concentration”. In the methodological context of the continuous space-based measures of spatial clustering, firms are identified as dimensionless points distributed in a planar space. In realistic circumstances, however, firms are generally far from being dimensionless and are conversely characterized by different dimension in terms of the number of employees, the product, the capital and so on. This implies that a high level of spatial concentration can occur, for example, because many small firms cluster in space, or few large firms (in the limit just one firm) cluster in space. A proper test for the presence of spatial clusters of firms should thus consider the impact of the firm dimension on industrial agglomeration. For this respect, the third essay develops a methodology based on an extension of the K-function considering firm size as a weight attached to each of the points representing the firms’ locations.