Backward and forward closed solutions of multivariate models

Economic models that incorporate expectations require non causal time series theory. We provide a general method useful to solve in closed form any forward linear rational expectations multivariate model. An anticipative VARMA model is likely to explain a behavioral relation were a tentative future guides the today action. The work develops general conditions to get the unique stationary closed solution, backward or forward, so extends over the well known accepted results on causal invertible multivariate models and shows that to incorporate non causal models one should rely on Complex Analysis.