No label defined (Q4396375)

The well-known econometric model \[ y_t=X_t\beta+\varepsilon_t,\quad t=1,2,\dots \] is investigated in the case where the error component is of the form \[ \varepsilon_t= (\rho W+\tau V)\varepsilon_{t-1} +\lambda W\varepsilon_t +\sigma v_t,\quad t=1,2,\dots, \] where \(W\) and \(V\) are known spatial matrices, \(v_t\) is a standard Gaussian white noise, \(\rho,\tau,\lambda,\sigma\) are parameters to be estimated. The spatial autoregressive term \(\lambda W\varepsilon_t\) describes the contemporaneous spatial effects. Temporal dependence is incorporated into the autoregressive term \((\rho W+\tau V)\varepsilon_{t-1}\). Weak consistency and asymptotic normality of maximum likelihood estimates of the unknown parameters are proved.