On the validity of the supermultiplet model