Modeling and Estimation Issues in Spatial Simultaneous Equation Models

Spatial dependence is one of the main problems in stochastic processes and can be caused by a variety of measurement problems that are associated with the arbitrary delineation of spatial units of observation (such as counties boundaries, census tracts), problems of spatial aggregation, and the presence of spatial externalities and spillover effects. The existence of spatial dependence would then mean that the observations contain less information than if there had been spatial independence. Consequently, hypothesis tests and the statistical properties for estimators in the standard econometric approach will not hold. Thus, in order to obtain approximately the same information as in the case of spatial independence, the spatial dependence needs to be explicitly quantified and modeled. Although advances in spatial econometrics provide researchers with new avenues to address regression problems that are associated with the existence of spatial dependence in regional data sets, most of the applications have, however, been in single-equation frame-works. Yet, for many economic problems there are both multiple endogenous variables and data on observations that interact across space. Therefore, researchers have been in the undesirable position of having to choose between modeling spatial interactions in a single equation frame-work, or using multiple equations but losing the advantage of a spatial econometric approach. In an attempt to address this undesirable position, this research work deals with the modeling and estimation issues in spatial simultaneous equations models. The first part discusses modeling issues in multi-equation Spatial Lag, Spatial Error, and Spatial Autoregressive Models in both cross sectional and panel data sets. Whereas, the second part deals with estimation issues in spatial simultaneous equations models in both cross sectional and panel data sets. Finally, issues related specification tests in spatial simultaneous equations models are discussed.