Separate estimation of spatial dependence parameters and variance parameters in a spatial model

Abstract

This paper suggests a two step estimation procedure for a spatial model with different kinds of spatial dependence and heteroscedastic innovations. Since maximum likelihood estimation is cumbersome due to the large number of parameters, we use a generalized method of moments approach to estimate the parameters of spatial correlation which does not need the large number of variance parameters to be known. For illustration purposes, we apply our estimation procedure to daily stock returns of the Euro Stoxx 50 members. JEL subject classifications: C13, C51, G12