Instrumental rationality, cognitive rationality, and the structure of long-period expectations

New Classical Economics and Keynes's stochastic processes

The 'rational expectations hypothesis' (REH) appeared around the beginning of the 1970s as a new approach to rationality and uncertainty. It formed part of the analysis of equilibrium business cycles by the New Classical Economists. This hypothesis was supposed to satisfy the need for a theory of rational choice under conditions of uncertainty as the concept is defined in neoclassical theory. The central idea of this approach is that if individuals are presumed to be rational, then they should also be rational in their search for information and in their formation of expectations. Efficient expectations will thus be formed on the basis of the set of available information. A second hypothesis is taken from 'Walrasian' and 'classical' theory: the assumption of 'market clearing' (Lucas and Sargent, 1981, p. 304). A third assumption characterizes the nature of the movement of economic systems as stationary, stochastic processes (see conversation with Robert Lucas, in Klamer, 1984, p. 44). This latter assumption produces the regularity of the economic process that is at the basis of a succession of cycles.

The analytical structure defined by these assumptions inevitably produces an extremely simplified representation of the economic process. It is a abstraction in which 'Nature' undertakes 'independent drawings from a fixed cumulative probability distribution function' (Lucas and Sargent, 1981, p. xii), and each agent has the possibility of maximizing income on the basis of the mathematical forecast generated by given distribution functions that correspond to the conditional probabilities associated with the stochastic process that is supposed to describe the movement of the economy. The result of this maximization procedure is that 'on average' agents make optimal forecasts given the current and future states of the world (Lucas and Sargent, 1981, p. xiv). Following the approach indicated by Muth (1961), the problem of the formation of expectations becomes the identification of subjective probabilities with the observed frequency of the forecast events.

This new approach can also be evaluated from the point of view of the reformulation of Keynes's position in terms of stochastic processes given above. For example, as Davidson has argued, the most important characteristic of the stochastic processes assumed by the New Classical Economists is not that they are stationary but that they are ergodic. Then,

for the REH to provide a theory of expectations formation which provides forecasts which are efficient, unbiased, and without persistent errors, not only must the subjective and objective functions be equal at any given point of time, but these functions must be derived from an ergodic process. In other words, the average expectation of future outcomes determined at any point in time will not be persistently different than the time average of future outcomes only if the stochastic process is ergodic.

(Davidson, 1982-83, p. 185)

However, as emphasized in the first section, the Keynesian definition of the 'real world' is of a non-ergodic process. As a consequence, the forecasts obtained from the existing distribution functions can diverge consistently from the time averages obtained as a result of measuring the frequency of the actual observations of the events which are being forecast. As a result, Post Keynesians consider the assumptions of rational expectations and stationary processes as incompatible with analysis of the 'real world'.

Despite these attempts to clarify the importance of uncertainty in the sense given the term by Keynes and Knight for empirical analysis, mainstream evaluation of the question remains that represented by Walliser (1985, p. 17), who writes that Keynes's ideas on uncertainty have 'remained at the preformal stage, they do not seem to have had any general "cultural" influence on economic thinking except as complications or useless subtleties'. Indeed, for most economists better-known Keynesian concepts such as 'animal spirits' are considered to be purely subjective and thus irrational and unscientific. As a result, in the contemporary literature Keynes's analysis is either criticized or credited as an approach to macroeconomics that is incompatible with the traditional concept of individual behaviour based on the assumption of economic rational- ity.^{5} This situation is in part linked to the identification of Keynesian uncertainty with radical subjectivism, and the associated scientific and theoretical nihilism. This is clearly the interpretation of Keynes adopted by Lucas, who writes that 'in cases of uncertainty, economic reasoning will be of no value' (Lucas, 1981, p. 224). It must be admitted that some Post Keynesian authors have given support to such an interpretation. For example, Shackle has claimed to be a 'nihilist', and applied the same term to Keynes (see Shackle, 1984, p. 391).

Further, as we have already pointed out, the definition of uncertainty associated with Keynes and Knight implies that it is impossible to provide exhaustive rendering of all possible future events ex ante. Thus the representation of uncertainty by means of probability encounters difficulty, not so much because of the estimation of the probability of any given event but in the representation of the set of possible future events. The use of the assumption that the agent can specify a 'residual even' appears as a completely arbitrary solution to the problem. As Shackle has pointed out,

if the list of hypotheses in answer to some question is acknowledged to be endless and incapable of completion ... the language of subjective probability may seem unsuitable. That language distributes a total, representing the certainty of inclusiveness, over a finite number of specified rival answers. Or if not, then instead it includes a Black Box, a residual hypothesis of unknown content.

(Shackle, 1972, p. 23)

While this criticism makes it clear that this approach is incompatible with Keynesian uncertainty and non-ergodic processes, its force has been ignored because it appears to be based on radical subjectivism and theoretical nihilism.