A Proposal for Estimation of the Parameters of Multivariate Moving-average Models

A method is proposed for the preliminary estimation of the parameters of multivariate moving-average processes. Initially, m independent white-noise series are generated. Then the parameters are estimated through an iterative procedure. The algorithm is fast and simple since it does not require specialized software for time series but a least squares routine. The estimators are consistent and a simulation experiment shows good performance in finite samples. An applicati on to real data is also illustrated