Who's afraid of reduced-rank parameterizations of multivariate models? Theory and example

Abstrakt

Reduced-rank restrictions can add useful parsimony to coefficient matrices of multivariate models, but their use is limited by the daunting complexity of the methods and their theory