In reduced rank regression (RRR), the dependent variables are first submitted to a PCA and the scores of the units are then used as dependent variables in a series of stafldard MLR's where the original independent variables are used as predictors (a procedure akin to an inverse principal component regression).