SOAS Research Online

Abstract

This paper examines how “confluence” and “autonomy,” two key concepts introduced by Frisch around 1930, have disappeared in econometrics textbooks and why only some fragments of the two have survived mainstream econometrics. It relates the disappearance to the defect in the textbook position of equating a priori theoretical models as correct structural models unequivocally. It shows how the confluence-autonomy pair in unity reflects the complexity of econometricians’ goal to find and verify robust and economically meaningful models out of numerous interdependent variables observable from the real and open world. The complexity deems it essential for applied research to have a set of model design rules combining both a priori substantive reasoning and a posteriori statistical testing. Such a need also puts the task of ensuring an adequately minimum model closure to top priority for applied modellers, a task much more important than the textbook task of parameter estimation.