I find Shackle’s work to be discursive rather than technical or logical, and I do not find it easy to follow the reasoning and do not feel confident that I have understood his work enough to apply it, but tend to appreciate the examples and broadly to agree with the conclusions.
A collection of papers, 1939-1953. Quite repetitive. The key to understanding Shackle seems to be in this observation:
‘The things among which a man is free to choose are not satisfactions themselves, but actions designed to secure for him some sort of satisfaction … .’
and this question:
‘[W]hat measure does a decision-maker use of the strength of a claim of a hypothesis to be true … ?
His approach to utility starts with:
‘What sort of conditions must be satisfied in order that [utility maximization is reasonable? They] can be epitomized by: …
- The frequency-ratios, unless derived a priori, must have been obtained from a set of performances sufficiently uniform and sufficiently numerous. …
- The experiment … must be … divisible … . [i.e.] its result will be obtained by adding together the results of a series of separate performances, and … these performances are going to be sufficiently uniform in their circumstances, and sufficiently numerous, for frequency-ratios derived from a past series of similar performances to be applied.’
Thus Shackle regards utility maximization as reasonable when one is going to toss a coin many times and gamble on the proportion of heads, but not when the coin is to be tossed only once.
Shackle suppose that probability is meaningless when one has a ‘crucial experiment’:
‘By crucial experiment I mean one where [it is possible that] the very act of performing the experiment may destroy for ever the circumstances in which it is performed.’
Choosing a partner may be crucial, but an affordable gambling habit may not be:
‘… when we feel ourselves to be living in an assured and stable environment the routine of private life is very largely amenable to probability calculations … .’
This the appropriateness of probability theory depends not just on the decision to be taken but our relationship to it. (Thus one might expect a decision-taker, exposed to the full risk, and an adviser, somewhat insulated, may have different views on the nature of uncertainty.)
Shackle notes that for most people choosing a holiday is an intermediate case, since one does not get many opportunities.
Shackle introduces the notion of (potential) surprise, which is the psychological surprise that one would feel if some hypothesis were falsified. It is a form of disbelief. This is not additive, and so fails to meet the desiderata of probabilists, but Shackle argues that it is more useful: surprise depends only on the hypothesis and evidence, so that the discovery of another credible hypothesis does not lead to any re-calculation.
Shackles conditions resemble those that one might use for any heuristic: that it is to be used in circumstances that resemble some comparator, where one has had sufficient experience with a good outcome. Thus we might regard an outcome as surprising if it had rarely or never happened in the comparator. Surprise would then be an inverse to Keynes’ likelihood, so that less likely outcomes would be more surprising. Thus Shackle’s ideas could be seen as a contribution to the problem of induction. Shackle does not seem to suggest any other interpretation.
Shackle doesn’t regard utility maximization as sensible for a single coin toss, but – to me – it seems reasonable, like Keynes, to define a probability as the thing that should inform decision making, even if it is not utility maximization. But that may be more than just a number.
With regard to crucial experiments, it is not clear to me whether Shackle supposes that no utility function exists, or whether there is no way to determine what it is.
Shackle explains his notion of ‘time’, mirroring Whitehead‘s notion of epochs, but from a psychological perspective. He also attempts to explain ‘surprise’, giving an pseudo-mathematical account. This seems to me to be better at indicating the problem than the solution.
Epistemics and economics: a critique of economic doctrines. 1972/1992
Shackle uses the term ‘unknowledge’ for knowing what you cannot know or do.
‘The history-to-come which will flow from men’s decisions is non-existent until those decisions themselves are made. What does not yet exist cannot now be known.’
‘The ultimate indispensible permissive condition of knowledge is the repetition of recognizable configurations. These … form a hierarchy in our minds.’
‘The economic theoretician ought to be concerned as to the general nature of the world he is, consciously or unconsciously, assuming … .
Choice is made amongst the invented, subjective creations of thought.
‘[T]he freedom of men to choose destroys their power to know.’
‘The paradox of probability … is that there is required to be a variation of detail within bounds which remain constant.’
‘There must not … be any tendency towards a self-reinforcing or ‘cumulative’ process. Such self-reinforcing process could conceivably take the form of a spreading ring of repercussions involving a different set of details at each successive performance. … The system … must have no inherent tendency to explode. Neither must it be inherently evolutionary. … the determination of probabilities … requires a stable, constant and given design of the system.’
‘[W]hat of a world where … the systems cannot stay the same because they involve, and partly consist of, our knowledge concerning them?’
‘Beyond the purely physical, probability seems to drown in a sea of unanswerable difficulties.’
Shackle’s notion of hierarchies of repeatable configurations resemble Whitehead’s, which may explain why his notion of uncertainty mirrors that of Whitehead and Keynes, and a concern for reflexive uncertainties.
Shackle was a student of Hayek, and strongly against any notion of ‘scientific’ prediction in economics. He thinks Keynes’ notion of uncertainty is critical, but is more radical than Keynes about its implications.
“[T]here will always be shocks and things that really upset all calculations. I can’t really quite believe in the idea of steady improvement, you know. After all, some of these men, they’re very clever entrepreneurs, are not all working together, they’re trying to undermine each other’s positions, they’re working against each other and trying to outdo each other.“
“Your list of choosable things has to be constructed or composed by yourself before you can choose.“
“I see Man as a “beginning,” a chooser which cannot be fully explained. He is an uncaused cause.”
“You might make measurements which are all right for today. But, there are countless people whose interest it is to make nonsense of those measurements tomorrow.”
He agrees with “all we ever have is the historical record and what was historically relevant in the past may not be for our period.”
“I’ve been saying for almost forty years that economics isn’t a science, and we ought not to call it a science.“
Shackle emphasizes the game-like, rather than the gambling, aspect of life, and the creative rather than the merely routine.
A memorial to Shackle provides a description of his work that may be more accessible.
“How then does one handle the unknowledgeable future? How does one make choices? For Shackle the answer was clear; one used one’s knowledge of the past plus one’s imagination pertaining to all things considered not only probable but also possible. One kept in mind any number of possible scenarios, and as news of events came in (over time) one could easily and quite likely probably shift one’s choice of the presently preferred scenario to one more suited to the recently arrived ‘news.’ An even better way to put the point is to quote the beautiful prose of the final paragraph of George’s paper, ‘The Bounds of Unknowledge,’ where he suggested, in effect that the action-chooser “can tell himself where possibility reaches its limits, he can trace upon his thought map of imagined things a boundary-fence beyond which, given his resources, given his time-horizon for reaping the fruits of his action, no action of his can reach. In a sense which has some degree of paradox, his knowledge is of the non-possible. It is beyond the fence that everything can be tagged with effective certainty as not belonging to the attainable. Unknowledge for him consists in the plurality of rival possibilities. The bounds of the possible are bounds of unknowledge.”
Shackle’s uncertainty concerns focus on the future, rather than current ambiguities. The model of decision-making that emerges from Shackle is consistent with utility maximization and probability, where they are justified, but more generally recognizes uncertainty. One is concerned with what might happen and what one might be able to achieve, considering various imagined scenarios and trying to influence things so that the worst possible outcomes are avoided and some good outcomes remain attainable, despite shocks.
Shackle describes rather than defines an approach to ‘measuring’ uncertainty: The work of Keynes and Good would provide more detail.
To me, the value of Shackle is in stimulating thought, rather than in providing an ‘actionable’ guidance.