III: Modeling the spatial locational choices of economic agents
Preliminary definitions and concepts in point pattern analysis
This chapter introduces various important spatial concepts relating to points whose spatial location is considered to be random. It introduces the general concept of a point pattern, of complete spatial randomness (to be used as a benchmark) and the ideas of cluster and inhibitory point patterns. It further presents various indices of spatial randomness and statistical testing procedures to verify the significance of the departure from the complete spatial randomness (CSR) case (quadrat counting and the Clarke-Evans test)
Spatial point patterns of economic agents
The aim of this chapter is to introduce the necessary framework, definitions and concepts to identify the spatial location patterns of economic agents. The typical dataset providing the micro-geographical distribution of economic events and activities in a given study area consists of a set of spatial coordinates indicating their locations. In the terminology of spatial statistics, if the observed locations of economic agents (such as firms’ establishments and customers’ houses) have a negligible physical size with respect to the geographical dimension of the whole study area, the set of their geographical coordinates is called a “spatial point pattern”.
Spatial point patterns, like any other types of data, can be analyzed using statistical methods designed to assess why the observed points are arranged in a given pattern. This can be of importance tor two main reasons: firstly, an analysis of the spatial distribution of points can be a valid approach to identifying the behavioral models in all those cases where we have prior knowledge of the underlying data generating mechanisms. This is the case, tor example, of the spatial distribution of firms over a given area, where economic theories may point to the existence of certain particular configurations, as we shall see a little later on. Secondly, when our knowledge of the phenomenon is rather limited, as in the case of a preliminary exploratory analysis, the evidence gathered from an analysis of the point pattern may constitute the initial phase of the description and formalization of the phenomenon itself.
Figure 6.1 shows the spatial point pattern of the 164 start-up firms in the tourism and hospitality sector in the main island of Sicily in 2010. When observing
Figure 6.1 Locations of start-up firms in the tourism and hospitality sector in Sicily, 2010.
this map, the first question that comes to mind is whether or not the economic activities are purely randomly scattered in space or whether, on the contrary, they follow a regular pattern which may be regarded as having some economic importance. For example, firms may tend to cluster, or to localize, in particular locations, such as the seaside. As it is often the case in econometric analysis, the first step towards the identification of any interesting regularities involves a process of abstracting the phenomenon we wish to study from any disturbing factors. In this context, we can proceed by considering the economic activities as if they were located in a homogeneous space. This process can only be partially legitimized in that it invariably eliminates other factors which may prove to be of importance in our analysis of localization processes (such as, for example, the presence of useful infrastructure or proximity to communication routes). However, it does enable us to make a start.