Analyzing local incomes in Greece by fitting space with an econometrics driven method

Abstract

The article puts forward a method that improves the grouping of observations in empirical analyses by comparing the properties and results of two econometric specifications regarding the temporal evolution of personal incomes across Greece's 932 local districts during 2001–08. More specifically, it organizes the disaggregated data via categorical variables constructed after the (a) usual sub-national partition of the country, and (b) spatial patterns of the residuals; and finds that the second approach yields a better fit, ameliorates the misspecification problem, and reveals considerable asymmetries across space and within the usual partitions of the county in the form of clusters, strings of localities, and individual districts associated with substantial positive and negative coefficients (effects). The areas associated with positive effects supply potentially useful development examples, while the areas associated with negative effects may be in need of policy intervention. It turns out that if the conventional sub-national organization is not employed it does not emerge at all. The findings open up a useful research trail in economic analyses, especially to those interested in identifying through statistics retail, health, education or other policy areas.