Padé approximation in economics (Q1410478)

The main purpose of this work is to show that Padé approximation (PA) and its properties can be used to forecast time series models. The applications of some well-known techniques closely related to PA (corner method, epsilon, \(rs\) and \(qd\)-algorithms), to the rational characterization for the ARMA models in the univariate case [see specially \textit{A. Berlinet} and \textit{C. Francq}, Comput. Stat . 9, No 2, 117--133 (1994; Zbl 0937.62088)], are previously reviewed. These methods are then treated (in the multivariate case) for identifying VARMA models in the context of causal and expectation TF models.