Comparative examination of spatial econometrics software

Abstract

In this work, the latest libraries in the field of spatial econometrics are examined. Their difference to the classic econometrics approach lies in the existence of the spatial weight matrix which introduces computational problems in the estimation of the models due to its high dimension. We discuss the first approach appeared in literature, as functions in Gauss, and compare the current approaches present in MatLab, R, Python, GeoDa and Stata based on their capability to reading spatial data, to create/use spatial weight matrices, their cost, their functionality and the existence of active community supporting their use, including repositories. The comparison is limited to models developed for cross sectional data and is accompanied by a test case of a simple model in all platforms by using data freely available.