BAYESIAN-ESTIMATION OF SPATIAL AUTOREGRESSIVE MODELS

Spatial econometrics has relied extensively on spatial autoregressive models. Anselin (1988) developed a taxonomy of these models using a re gression model framework and maximum likelihood estimation methods. A Bayesian approach to estimating these models based on Gibbs sampling i s introduced here. It allows for non-constant variance over space taki ng an unspecified form and outliers in the sample data. In addition, e stimates of the non-constant variance at each point in space allow inf erences regarding the spatial nature of heteroskedasticity and the pos ition of outliers.