Some Issues on the Concept of Causality in Spatial Econometric Models

A specific feature of spatial econometric models is the simultaneity of the relations, which complicates the distinction between causes and effects. However, the notion of cause is of paramount importance in any specification. In fact, a model is a statement pointing that a variable, called endogenous, reacts to the variables that appear in the right hand side, the regressors. Our impression is that this problem has been scarcely treated in the Spatial Econometrics literature. The content of the paper focuses on questions related to the specification procedure for spatial models. We examine what may be called the 'current traditional practice' and discuss the role that the concepts of identification and causality should play. Our purpose is to claim for the development of clear econometric guidelines to help the users to improve the theoretical foundations of their specifications. An application to the relation between per capita income and weight of the agricultural sector in the Spanish provinces illustrates our discussion.