The author defines the concept of identification and explains what 'goes wrong' with some nonrecursive models to make them nonidentified. He provides various tests which can be used to determine whether a nonrecursive model is identified, and reviews common techniques for estimating the parameters of an identified model.
How do you feel?