Spatial Econometric Models: Theory

Accounting for the presence of spatial linkages is extremely important in regional economics. These linkages may follow, for example, when policies implemented by a specific geographical unit spread also to its proximate locations. In this regard, spatial analysis provides different tools to detect the presence of spatial interdependencies. The failure to acknowledge for these externalities would result in a mis-specification of the empirical model, with an incorrect representation and understanding of the true causal processes at work. Readers interested in understanding the way this transmission mechanism takes place, and how different regions react to external shocks, would highly benefit from learning the basics of spatial econometrics. To this aim, the chapter starts from a brief description of the Exploratory Spatial Data Analysis (ESDA), providing global and local indexes to control for the presence of spatial patterns, verifying whether there is a clustering in the space of high or low outcomes for the regional variable under observation. The chapter then illustrates the main models of spatial interaction: depending on the type of interaction between observations of neighbouring units, different specifications may be followed (SAR, SEM, SDM, SARAR and so forth) according to the results of model selection tests (usually Likelihood-based tests).