This chapter is a revised version of a paper originally published as 'Alternative Analyses of Uncertainty and Rationality: Keynes and Modern Economics', in
S. Marzetti Dall'Aste Brandolini and R. Scazzieri (eds), La probabilita in Keynes: premesse e influenze, Bologna: CLUEB (1999), pp. 115-37.
- 1. While there is general agreement that both economists recognized the importance of the distinction between risk and uncertainty, there is less agreement about whether they had similar conceptions of uncertainty. See Hoogduin (1987) and Schmidt (1996).
- 2. As emphasized by Davidson (1991), the review of the literature by Machina (Machina, 1987), which is intended to give an exhaustive survey of the perspectives of choice under uncertainty, is revealing of the position of traditional economics on this question. While Machina does mention models of subjective probability, he never mentions the existence of other analyses of uncertainty that might be added to the 'probabilistic tool box' of traditional theory.
- 3. Some Post Keynesian authors such as M. Lavoie (1985, pp. 499-500) consider cruciality rather than the uniqueness of a decision—as in Knight, for example—as the fact that alters the future economic environment. Crucial decisions are, in effect, very common (for example, investment). As noted by Davidson (1982-83, p. 192), the uniqueness of an event in a finite realization could simply mean that the event has a very low probability.
- 4. See the passage from Keynes's Quarterly Journal of Economics article (1973e ) cited above.
- 5. This is emphasized by Richard Arena (1989).
- 6. See Keynes's judgement of Ramsey's works in Keynes (1972 ), pp. 338-9.
- 7. Keynes writes the probability relation thus defined as ah = a.
- 8. J. R. Hicks (1979, p. 114). In contrast to the traditional formulation of this axiom which states that given a certain set of information, of two propositions, either one is more probable than the other, or they are equally probable.
- 9. GT (p. 173). It is necessary to point out that the notion of 'weight' poses a very delicate problem of the determination of the point in time at which a decision will be taken. Keynes in fact excludes the idea of a maximum weight that would correspond to the existing set of available information (to put the point in the language of rational expectations). Rather, Keynes argues that it is not possible to equate the cost of additional information with an increase in weight in such a way that traditional theory relates an increase in information with an increase in certainty: '[t]here clearly comes a point when it is no longer worth while to spend trouble, before acting, in the acquisition of further information, and there is no evident principle by which to determine how far we ought to carry our maxim of strengthening the weight of our argument' (TP, p. 83.)
- 10. See Keynes' analysis in Chapter 12 of GT.
- 11. Since the mid-1990s Joseph Stiglitz has led a group of authors called New Keynesian Economists in developing an approach in which agents face two types of uncertainty. First is 'exogenous' uncertainty, defined as being independent of the actions of agents. This is similar to simple probabilistic risk in which it is possible to know both the set of possible states of the world and their probability distributions. This probabilistic risk is not considered to have a great impact on the behaviour of agents or markets. More important for this approach is the second type of uncertainty, which is considered to be capable of impeding the operation of supply and demand and market clearing. This uncertainty is based on asymmetries in information. The rejection of the assumption of the equal distribution of information across agents is often considered the basis of the contributions of the New Keynesian Economists and the source of their differences with the New Classical Economists (see Stiglitz, 1992, pp. 38-86). Information asymmetry creates uncertainty that can be considered as endogenous to the extent that results depend on the actions of the best-informed agents in the model. However, this uncertainty does not appear to be the equivalent of the true uncertainty defined above by Keynes and Knight. This can be seen by recognizing that, although information is asymmetrically distributed, it remains true that full information, in the sense of probability distributions over all possible future states, is available to at least some of the agents in the economy, and further it is known that the present and future behaviour of market is determined by this distribution. As a result the criticisms that have been made of the New Classical approach in conditions of uncertainty also apply to the New Keynesian Economists. This judgement is reinforced by the fact that the latter group appears to accept the rational expectations hypothesis.