Abstract
Janssen (1987) proposed a method for the incorporation of time delays (and causality) in MFD models for the purpose of equation error identification. This method is shown not to be completely adequate, since it retains linearity in the parameters by sacrificing optimality of the solution.