Appendix 1: Fundamental uncertainty and the bankruptcy of classical and neoclassical theories of rational agent choice
In building his theory of agent choice on the assumption that the future is fundamentally uncertain and therefore unknowable in the present, Keynes raised basic questions about the very meaning of "rationality" in New Classical and neoclassical theories of "rational" agent choice, and in so doing created a little-noticed transformation of the theory of agent choice.
In New Classical theory, agents are assumed to have probabilistic knowledge of future states of the economy that, by some unexplained miracle, is actually complete and correct. In the illogical language of the theory, agents have "rational expectations," though only an irrational person could hold such views. Future economic states are assumed not just to be known in the present, but also to be stationary; they will not change in response to current and future agent decisions or exogenous shocks to these probability distributions. We might think of such models as macrofounded because, since the future must be determined before agents make current decisions, the task of the agents individually and collectively is not to influence or create future economic states, but rather to adjust their own decisions so as to make them consistent with pre-given future economic states - no micro-foundations here. If we remove the absurd assumption that agents have complete and correct knowledge of the stationary future in New Classical theory, it is unclear how rational agents would select among alternative courses of action. Indeed, there would be no definition of what a rational choice is.
In the subjective probability models of agent choice used in traditional neoclassical micro theory, it is assumed that agents believe they know the true probability distributions that represent future states of the economy even though it is acknowledged by those who created the theory that it is impossible for anyone to have true knowledge of the future.
"For the subjectivist, in fact, probabilistic knowledge does not necessarily correspond to anything in objective reality" (Lawson 1988, p. 41). The problem here is that it is impossible for a rational agent to assume he or she has perfect knowledge of the future if in fact the basis for perfect knowledge is formally assumed to be unavailable to him or her. The neoclassical theory of rational agent choice thus implicitly assumes that agents are irrational because they believe they have infallible knowledge of the future when the theory explicitly assumes they cannot possess such knowledge. This problem would, of course, be solved by introducing Keynes's concept of the degree of confidence agents have in the truth content of their fallible expectations, but this would destroy the foundation of the theory of rational and optimal choice. It is remarkable that neoclassical theorists seem unaware of this striking logical contradiction at the heart of their models.In both theories, agents are assumed to have a known objective function (at least implicitly assumed not to change during the time that elapses from when a choice is made until the outcome is determined) and perfect stochastic knowledge about the relation between current choice and the future results of that choice. An agent is thus considered to be rational if he or she performs the optimization math correctly and irrational if he or she does not. If we replace the obviously unrealistic perfect-futureknowledge assumption used is these theories with Keynes's obviously correct assumption of fundamental uncertainty, mainstream theories have absolutely nothing to say about how sensible agents make choices.
Robert Lucas believes that, "in cases of uncertainty, economic reasoning will be of no value" (Lucas 1981, p. 224). But in the case of Keynes's radical uncertainty, this is simply not true. Keynes offers an alternative behavioral theory of what might be called "sensible" agent choice under radical uncertainty in chapter 12 based on his development of "conventional" expectation and confidence formation.
This micro theory, which is a central part of his analysis of financial markets and the capital investment decision, is a crucial building block of his macro-theoretical apparatus. This theory of sensible agent choice under fundamental uncertainty should have created a revolution in mainstream micro theory, but alas, it did not.One might think that modern behavioral economic theory, which has been used to resolve anomalies in the standard theory of efficient markets, has incorporated most of the core insights of Keynes's theory of agent choice under fundamental uncertainty, but one would be wrong to do so. These theories are not intended to be substitutes for the vision incorporated in neoclassical theory. Rather, they are meant to be amendments to it that do not challenge its dominant position in the profession. "The behavioral finance literature... simply adopts the neoclassical view with biases added (e.g., overshooting, undershooting, framing, etc." (Findlay and Williams 2008, p. 224). Camerer and Lowenstein insist that behavioral finance does not seek "a wholesale rejection of the neoclassical approach to economics based on utility maximization, equilibrium and efficiency" (Camerer and Lowenstein 2004, p. 1, emphasis added). Matthew Rabin, a star in the behaviorist camp, argues that their research program is "not only built on the premise that mainstream economic methods are great, but also that most mainstream economic assumptions are great" (Rabin 2002, p. 658, emphasis in original).
In my opinion, the core problem in mainstream theories of rational agent choice is not the standard assumption that all agents are rational, though we know people may act irrationally at times. Rather, it is that agents' unavoidable ignorance of future economic conditions makes assuredly optimal choice - the foundation of mainstream micro theory - impossible. That is, the damage done to neoclassical micro theory by the assumption of fundamental uncertainty is far greater than that caused by agent irrationality.9