Spatial econometric modelling

Typically, social and economic data have some spatial dimension. Unemployment is recorded by local unemployment agencies and tracked by government, accumulations of hazardous waste occur in proximity to specific human populations, crimes are committed at a location, consumers purchase goods at stores located in certain places and social inequality is spatially situated. While human geography, regional science and urban planning have a long tradition of observing the spatial patterns of various phenomena and using these to develop and test explanatory models of social interaction and urbanization (Frank, 2003: 147), conventional procedures of social data analysis, particularly in economics, often do not make use of this important locational information. The urban sociologists (and later the criminologists) at the Chicago School (Park et al., 1925; Park, 1936; Hawley, 1950), from which theories of human ecology evolved, stressed that social facts are located facts, situated in time and place, and that social life cannot be fully understood: ‘without understanding the arrangement of actors at particular social times and places’ (Abbott, 1997: 1152). In sociology and economics there has been a renewed interest in: · models of social interaction and dependence among economic agents (Durlauf, 2003); · spatial spillovers (Topa, 2001); and · knowledge externalities and agglomeration economies (Banerjee, 1992). In such models, information about the location of economic agents is essential to correctly predict the nature and magnitude of outcomes generated. (For a summary of these developments see Goodchild et al., 2000: 141.