Expectations

John T. Harvey (Forthcoming, entry in The Encyclopedia of Political Economy, edited by Philip O'Hara)

All but the most trivial of economic activities require that individuals predict the outcomes associated with the choices available to them and choose a course of action based on those forecasts. Both processes are manifestations of reasoning on the part of the economic agent. Thus, expectations, choice, and RATIONALITY are very closely related concepts in economic theory and in reality.

That expectations play such a central role is well recognized by all schools of thought. However, the relative emphasis placed on modeling them (directly or indirectly) varies considerably with both the context of the study and the characteristics and biases of the research program in question. It is the latter that is of most interest here.

The neoclassical view of expectations (and decision making) is strongly influenced by their belief that market systems are, unless proven otherwise, optimal and efficient. They further contend that Homo sapiens are by nature rational, utility maximizers. This characterization of human behavior is rather open ended; however, their confidence in the efficacy of free markets leads them to define rational utility maximization in a unique manner consistent with the assumption that market solutions are optimal.

For example, orthodox economists contend that rational economic agents who are free to make their own decisions will, on average, make the "right" one. In other words, they will choose the option that would maximize their utility given the average outcome. This view is formalized in expected utility theory (also known as rational choice or subjective utility theory) and can be illustrated with a simple example. If faced with a choice between one alternative that promised a ten-percent chance of a gain of $500 (and a ninety-percent chance of no gain) and another that gave a twenty-percent chance of a gain of $2000 (and an eighty-percent chance of no gain), the rational actor would choose the second. This is so because, assuming a large enough number of trials, the first choice would yield a gain of $50 on average (one gain of $500 per ten trials) while the second would give $400 (one gain of $2000 per five trials).

Their view of expectations is very similar, also assuming that rational economic agents will be "right" on average. The theory (rational expectations; see Lucas and Sargent 1981 or Muth 1961) intentionally makes no attempt to explain how expectations are formed, but contends that whatever the process, rationality combined with the discipline of the market must (after a period of learning) lead to a situation in which any persistent forecasting errors are eliminated. This is so because if the expectations were biased by a repeated error, the error would represent profitable information that could, and would, be exploited. So any remaining errors must be entirely random.

Rational expectations and expected utility have become vital empirical and theoretical research tools in neoclassical economics (though there are criticisms from within neoclassicism). It is no coincidence that they are consistent with the basic themes of the paradigm within which they were developed (a characteristic of normal scientific research that is certainly not limited to neoclassicism). They justify a market system in that they show that rational, utility maximizing individuals who are given the freedom to choose will, on average, select the alternatives most consistent with their welfare.

Post Keynesian economists have been highly critical of the so-called "rational expectations revolution." Because they draw their inspiration from KEYNES, Post Keynesians have tended to focus on factors consistent with their belief that market systems are inherently unstable and prone to collapse. With respect to expectations, the most important factors in this regard are UNCERTAINTY and nonergodicity.

Paul Davidson argues that in order for rational expectations to hold true it must be possible to create probability distributions of future events based on those from the past [Davidson 1982-83]. But this can only be accomplished if we can assume that the processes that create events in the economy (if that is valid analogy in the first place) do not change over time, i.e., if the economic world is ergodic. If the world is in fact nonergodic and the processes do change over time, then it is impossible to draw accurate inferences from the past about the future, especially with sufficient precision to build probability distributions. This is exactly what Sir John Hicks and G.L.S. Shackle have argued, the latter in terms of the "crucial decision." When a person (for example, an entrepreneur) makes a choice that forever changes the circumstances in which future choices must be made, that person has made a crucial decision. "In other words, crucial choice involves, by definition, situations where the very performance of choice destroys the existing distribution functions" [Davidson 1982-83, p.192]. This makes rational expectations impossible.

In building their explanation of expectations, Post Keynesians argue that the future is uncertain. Uncertainty causes the expectations of economic agents to be tenuous and subject to sudden and violent change. In general, it is argued that fundamental uncertainty forces agents to adopt different methods of decision making than would be appropriate under circumstances typically assumed in neoclassicism (i.e., absolute or probabilistic certainty). Because they lack adequate information for decision making, market participants are forced by the necessity of action to rely on present circumstances as being a reliable guide to those in the future (even though they know they are not). Only insofar as they have specific reasons to expect otherwise do they stray from this convention. The basis for such specific reasons can never be more than vague, however (again because of uncertainty), and thus sudden and violent changes in the conventional valuation are quite possible. Keynes' view is summarized by Geoff Hodgson:

Actions flow from judgements about the future (which often lack a firm, objective empirical foundation) as well as from observation of 'the convention' that is formed by the action of others (Hodgson 1985, p.13).

The sort of calm rationality described by expected utility theory cannot exist under uncertainty. Keynes goes so far as to say that if "mathematical expectation" were the only source of direction that "enterprise will fade and die" (Keynes 1964, p.162). Only our sense of spontaneous optimism, or ANIMAL SPIRITS, prompts us to action despite the serious lack of information regarding the likely outcomes.

Institutionalist economists have no argument with the Post Keynesian stand, except that they see it as incomplete. On this matter Hodgson writes:

In Keynes' work there is also a failure to consider the processes through which expectations are formed and the social culture and structures which give them colour and substance. This omission is not untypical of the overwhelming majority of economic theorists, but it is all the more acute for Keynes who made uncertainty and expectations central to his analysis (Hodgson 1985, p.16).

In particular, institutionalists see the lack of emphasis on society and CULTURE as a weakness. "The nature and structure of economic institutions is at least as relevant as 'psychology' in the determination of expectations" (Hodgson 1985 p.17). At this stage, little has been done to address this shortcoming, but it is nonetheless interesting that the institutionalist predilection with the social organization of economic activity is apparent in their view of expectations.

Marx did not write a great deal about expectations, nor have his followers. Nevertheless, as Caludio Sardoni has argued, Marxist theory must at least implicitly assume Keynes-like expectations, especially with respect to the existence of uncertainty. The role that this plays in Marx's theory is, as in the Post Keynesian, in making the economy unstable. In particular, it creates the possibility of hoarding, thus breaking Say's Law (Sardoni 1991, pp.219-239).

For all four schools of thought, central paradigmatic themes play an important role in the modeling of expectations. Perhaps this should not be surprising given the importance of expectations to economic activity. Indeed, neoclassicism would have a difficult time explaining the ability of independent agents to achieve optimality in the presence of uncertainty, as would Post Keynesians in modeling a world of volatility within the framework of rational choice and expectations. Which is most accurate? Obviously, that is not a question that can be easily answered. It is interesting to note, however, that the extensive theoretical and empirical work done in psychology on the subject of expectations and decision making shares far more with the Post Keynesian and Institutionalist traditions than with the neoclassical [Kahneman and Tversky 1979].