Commentary on rational expectations since the 2011 Nobel Prize in economics was awarded to Tom Sargent and Chris Sims continues to pour in. Last week, in addition to a post of my own, John Kay posted a hostile assessment of rational expectations on his website, prompting a half-hearted defense of sorts from Matt Yglesias likening the rational expectations revolution to the Copernican Revolution, a comparison of which Noah Smith, launching a renewed attack on rational expectations and its claims to have transformed economics into a hard science, heartily disapproves. In today’s Financial Times, John Kay returns to the attack with an ironic jab at the Nobel committee for avoiding any mention of rational expectations in its citation of Sargent and Sims, and some well-aimed barbs at the pretensions of rational expectations to a sort of infallible knowledge.
I agree with these criticisms of rational expectations, though as I suggested, the rational-expectations hypothesis can be, and often is, a useful one. Rational expectations only becomes dangerous when transformed from an empirical hypothesis and a way of testing the coherence of a model into a methodological principle and an axiom purporting to embody a necessary attribute of the world, on the order of Newton’s laws of motion. Thus, through a program of methodological imperialism, rational expectations has been imposed on much of mainstream economics as the only acceptable modeling strategy.
The spirit of this methodological imperialism is captured remarkably well by the abstract to the entry on “calibration” contributed by Edward Prescott and Graham Candler to the New Palgrave Dictionary of Economics, second edition.
The methodologies used in aerospace engineering and macroeconomics to make quantitative predictions are remarkably similar now that macroeconomics has developed into a hard science. Theory provides engineers with the equations, with many constants that are not well measured. Theory provides macroeconomists with the structure of preference and technology and many parameters that are not well measured. The procedures that are used to select the parameters of the agreed upon structures are what have come to be called ‘calibration’ in macroeconomics.
I would like to comment on another aspect of the methodological imperialism of rational expectations which is its reductionism. (The reductionism of rational expectations has, to my knowledge, been noted only by Allesandro Vercelli in a paper “Keynes, Schumpeter and Beyond: A Non Reductionist Perspective” in A Second Edition of the General Theory, edited by G. C. Harcourt.) In philosophy, reductionism refers generally to the idea that all higher level (i.e., more particular theories) are (can be? will be?) ultimately derived from and translatable into (can be reduced to”) lower-level, deeper, theories. Thus, the laws of chemistry, geology, astronomy, meteorology, biology (?) can ultimately be “reduced to” the laws of physics.
One field in particular in which the idea of reductionism is very much alive and discussed is in the nexus between brain science and the philosophy of mind — the old mind-body problem. Many philosophers and brain scientists argue that all mental states are “reducible to” physical states. The problem with reductionism in brain science and the philosophy of mind is that, unlike chemistry and physics, where the reduction is to some (unknown to me) extent already worked out, there is not even a glimmer of understanding of the physics that would allow one to derive mental states from physical states. Thus, reductionism in brain science and the philosophy of mind is a purely metaphysical notion that has no basis in known science. (Admittedly, I am speaking largely out of relative ignorance of the relevant philosophy and total ignorance of the relevant science, but my impression is that there are philosophers and scientists who would agree with my assertions about the current state of brain and physical science.) Moreover, even if we had all the science worked out, it is still not clear that mental states would in fact be reducible to physical states because there is not necessarily any way of bridging the chasm between mental and physical.
Rational expectations as a methodological principle requires that agents in every model base their decisions on the expected values of the stochastic variables, ruling out divergent expectations among agents, thereby promoting the adoption of representative agent models. Macroeconomic models not derived from explicit utility or wealth-maximizing assumptions and rational expectations are inadmissible. This sharply limits the possibility for deriving anything like what Keynes referred to as involuntary unemployment.
Thus, involuntary unemployment is to rational expectations theorists what consciousness is to mental reductionists. Even though we all have direct experience of consciousness, mental reductionists deny that there is any such thing as consciousness (which obviously is a far stronger claim than that consciousness is a figment of our imagination). Similarly, rational expectationists deny that there is such a thing as involuntary unemployment even though we seem to have plausible evidence that there are people who would be willing to work at the prevailing wage but are unable to find work. One can imagine that such evidence for involuntary unemployment might be discounted, but generally to do so, one would have to be able to provide a plausible reinterpretation of the evidence combined with compelling independent evidence that supported an alternative theoretical conclusion. But where is the compelling alternative evidence of the rational expectationists? And what is the plausible reinterpretation of the plausible evidence that there is involuntary unemployment? Rather than confront these issues squarely, rational expectationists simply insist that they are following the dictates of rigorous science. But their conception of rigorous science confuses an axiomatic method designed to deduce logical inferences from a set of assumptions and definitions with an attempt to compare the implications of alternative hypotheses with the evidence. The latter it seems to me is closer to true science than the former, unless your conception of true science is very close to that of Ludwig von Mises.