sphet: Spatial Models with Heteroskedastic Innovations in R

This introduction to the R package sphet is a (slightly) modified version of Piras (2010), published in the Journal of Statistical Software. sphet is a package for estimating and testing spatial models with heteroskedastic innovations. We implement recent generalized moments estimators and semiparametric methods for the estimation of the coefficients variance-covariance matrix. This paper is a general description of sphet and all functionalities are illustrated by application to the popular Boston housing dataset. The package in its current version is limited to the estimators based on Arraiz, Drukker, Kelejian, and Prucha (2010); Kelejian and Prucha (2007, 2010). The estimation functions implemented in sphet are able to deal with virtually any sample size.