On measurable uncertainty and the fight for taking uncertainty seriously in economics

This paper discusses the engagement of economists with the issue of the measurability of uncertainty. Since Knight’s seminal distinction between risk, intended as measurable uncertainty, and unmeasurable uncertainty, the question has been to what extent the extension of the theory of choice from certainty to risk through von Neumann and Morgenstern’s expected utility hypothesis would allow dealing with uncertain events. The paper develops from a study of the rationale underlying the theories of those authors who objected to the mainstream view that the axiomatic approach developed in the early 1950s, mainly through Leonard Savage’s generalization of expected utility, makes it, indeed, possible to reduce uncertainty to risk. After a summary of the meaning attributed by authors such as Knight, Keynes, Shackle and Ellsberg to the contention that uncertainty is irreducible to risk and unmeasurable, the paper aims to investigate why this view did not emerge as a significant alternative to the mainstream up until recently. A main reason, at times alluded to but never openly discussed in the literature, is shown to be the close link between Savage and the group of decision theorists at the Cowles Commission for Research in Economics under the directorship of Jacob Marschak and Tjalling Koopmans. Archival evidence suggests that arguing that a theory of decision under uncertainty could be developed on the basis of “axioms that seem unobjectionable,” as Koopmans put it, was indeed an integral part of the attempt undertook at Cowles to move forward in economic theory by prioritizing scientific rigour in the form of mathematical models engaging with new mathematical tools