Selección de modelos espacio-temporales con datos de panel en Matlab y R

Abstract

The unstoppable interest in spatial econometrics, new statistical techniques, increased availability of panel data and advances in software technology have led to development of new packages for modeling spatio-temporal dependence in a panel of data. The objective of this paper is to select the best spatio-temporal panel data model using two software packages: MATLAB and R for further comparison. The statistical methodology used takes into account the neighborhood relations between the spatial units over time. The models finally selected were validated by two measures of goodness of fit: the coefficient of determination and the residual variance. In addition, their coefficients were interpreted including direct and indirect effects (spatial spillover). The data panel used corresponds to the mortality of 26 European countries for the period 1990-2009.