Estimation of Spatial Panel Data Models Using a Minimum Distance Estimator: Application

This paper is concerned with modelling and estimating panel data autoregressive spatial processes in the framework of minimum distance methods. A contiguity matrix based on distance between points relates observations spatially. The model is estimated in two stages. First, the cross-section parameters are consistently estimated by maximum likelihood, and a consistent asymptotic covariance matrix is computed for the second stage. Minimum distance estimators are derived under fixed slopes and all identical parameters restrictions. We used this specification to examine empirically spatial patterns of residential water demand for the French department of "Moselle", including electricity price effects.