Abstract

The echelon form of a VARMA (vector autoregressive moving average) model is considered. Its advantages over other identified VARMA representations are discussed. Furthermore, a general strategy for specifying echelon-form VARMA models from data is presented. Specifically procedures for choosing the Kronecker indices that characterize an echelon form are reviewed. The feasibility of the method is demonstrated by analyzing a well-known set of flour-price time series and the term structure of German interest rates.