On the specification of spatial econometric models
In this paper we discuss the specification process of a spatial econometric model. There are a lot of results available in a time series context, which form a very well structured phase, but little has been said when the domain of the model is the Space in relation to this question. There are several points that merit some special attention. For example, it is not clear if the concepts of integration and cointegration are also relevant in a spatial context. Moreover, at this moment we do not have any specific criteria in order to solve the selection of the variables. This weakness lead us to another question, which is related to the correlation problem: Which kind of cross-sectional dynamics could be allowed in a spatial specification? At present, in case of finding spatial autocorrelation in a model we try to model it immediatly, without considerer that if it may reflect a previous specification error. Finally, there is the problem of the outliers which are of paramount importance in time series but just unkown in a spatial context. We will discuss each one of this topics making reference to a general specification and looking also at the final stage of model selection.
Please report citation or reference errors to , or , if you are the registered author of the cited work, log in to your RePEc Author Service profile, click on "citations" and make appropriate adjustments.:
- Granger, C. W. J. & Newbold, P., 1974. "Spurious regressions in econometrics," Journal of Econometrics, Elsevier, vol. 2(2), pages 111-120, July.
- Nelson, Charles R. & Plosser, Charles I., 1982. "Trends and random walks in macroeconmic time series : Some evidence and implications," Journal of Monetary Economics, Elsevier, vol. 10(2), pages 139-162.