Exploring scan methods to test spatial structure with an application to housing prices in Madrid

Abstract

We evaluate the usefulness of the spatial scan test in specification testing for spatial econometric models. The null hypothesis assumes equality of the mean values of a variable in all the locations of a geo-referenced data set. The alternative hypothesis relies on the existence of one (or more) spatial cluster(s) where mean values differ from those of the rest of the sample. First, we conduct a Monte-Carlo simulation study to analyse the properties of this test when applied to regression residuals. Second, we illustrate this test with an empirical application on housing prices in Madrid.