Economic Expectations

BIBLIOGRAPHY

Economic actions are chosen with a view to imagined consequences assigned to some more or less distant future date or stretch of time. Such imagined and temporally projected consequences are what we mean by economic expectations. The decision maker, choosing among the rival schemes he has in mind for using his resources, must find some principle and basis for assigning to each such scheme a statement of its meaning to him in possible outcomes. The outcomes that he connects with any one scheme may be plural and mutually exclusive. In this case, he is uncertain what will be the outcome of the scheme. He must then adopt some procedure for valuing (in a general and not necessarily monetary sense) as a whole the array of rival conceived outcomes. This valuation has as its purpose, which governs the choice of the basis and method of valuation, a comparison of the rival schemes in order to choose one of them and act upon it.

The principal matters that have to be considered by a theory of economic expectations are the following :

(1) What kinds of evidence or data play a part in the decision maker’s task of conceiving the outcomes he connects with any specified scheme of action?

(2) Can a set of rules be found that are, or that ought to be, invariably used by any decision maker in this task?

(3) Is this task a process of pure reasoning, capable, if correctly performed, of logical justification; or do elements of judgment, of invention or inspiration, or of ex nihilo creative origination enter into it?

(4) When the outcomes imagined for any action scheme are plural and mutually exclusive, what procedure of valuation of the scheme and its array of rival conjectured outcomes as a whole is employed for the purpose of ordering the schemes or of selecting the best?

(5) Is the choice of such a procedure made in-dependently of the evidence available for forming the expectations, or does this evidence influence or govern it?

(6) The array of imagined outcomes that the decision maker connects with any contemplated scheme of action will influence his valuation of that action in two ways: by the character of each such outcome individually and by the structure composed by all these hypotheses taken together. What are the relative strengths and respective effects of these influences, and what is the nature of their interaction?

(7) If any such expectation structure is the out-growth of fragmentary, enigmatic, elusive, or self-contradictory evidence; if, as must be the case when several rival hypotheses are entertained concerning the result of any one action, the evidence is insufficient to determine a unique outcome; and if, in consequence, the evidence possessed by the decision maker at any moment is liable to be supplemented, canceled, differently illuminated, partly contradicted, or reinterpreted, what effect upon his valuations of different action schemes will this inherent and recognized fragility of his expectations have?

Logical teeth, however sharp, do nothing toward the nourishment and development of the body of practical knowledge until they have bitten upon fact. Some classification and cross classification of impressions reaching the mind from outside, some sorting of records and measurements, some emergence of repetitive associations and configurations are indispensable to the formation of tools for coping with the environment. The knowledge of “fact” that is appropriate for expectation forming is of two sorts. First, there must be knowledge of pervasive, repeated stereotypes of structure. These are answers to the following questions: How, by the ineluctable testimony of nature (including human nature), do things hang together? What things do we constantly find in association, either at one moment or in temporal sequence? How do things work? What can happen? These are statements of the principles of the world we live in, not the logical but the factual principles. Given such facts, logic, especially mathematics, may reveal a vast body of entailments that then assume the rank of fact without having been actually observed. Second, when we have some idea of where the state of affairs can go from a given situation, we need to know what situation, at the moment of seeking to form expectations, is given. This is the other sort of fact.

Historical determinism. The most fundamental question, on which the expectation-former determined to start from first principles would need some opinion, is whether the course of world history, in detail as well as in the large, is at each moment the uniquely inevitable sequel of the past. There are at least two hypotheses, either of which if true would require a negative answer to the fore-going question. One is an essential randomness in the origin of events at subatomic levels, or at the level of genetic mutation. The other is the nonexistence of any necessitating link between circumstance, past and present, and the thoughts that a man may have. If a thought can be in some sense, as to some part of its character, “uncaused,” we have plainly the possibility that history from time to time and from place to place makes a fresh start. One further feature would need to be true of the world if randomness in nature, or essential novelty of thought in man, were to defeat that statistical determinacy, which, for example, makes possible the laws of thermodynamics. This extra feature is the principle of self-reinforcing processes, by which a “small” event (such as a “favorable” genetic mutation) releases larger forces, such as the rise of a species adapted to a hitherto barren environment, and so leads to biological events on the largest terrestrial scale. Having, for the sake of completeness, tried to suggest the nature of these ultimate questions, we have to admit their virtual irrelevance for the practical business of forming expectations. This irrelevance arises from the fact that, whether or not it is in the nature of things (and, in principle, possible) to predict the future from the past, it is plainly inconceivable that the requisite information about the past could be possessed and processed by any individual, more especially since the time available for such processing, if it were to give any useful answer, would be strictly limited.

Expectations that reach beyond the immediate threshold of the future or that concern themselves with any but the most direct physical effects of the proposed action must accordingly always be uncertain. This uncertainty is responsible for the whole difficulty that expectations bring into economic analysis; and it is the source of much the greater part of the difficulty arising in economics from considerations about time, under any of its aspects. Uncertainty of expectation gives rise to agreeable or disagreeable states of mind, and can thus have an economic valuation and enter directly into the play of economic influences. Indeed, it is itself one of the most powerful of such influences and explains the possibility of large-scale unemployment in a world of omnipresent scarcity. It is the chief explanation of the existence of rates of interest. A service is rendered to society by those who, by embarking their resources in enterprises looking to a distant future, expose themselves to uncertainty to an especial degree; and if, on the whole, the rendering of this service entails discomfort, some compensating hope of profit will be needed to elicit this service. Such a hope might not sustain itself in the minds of those concerned unless there were a record of some degree of fulfillment of these hopes.

The stereotype. The venture of forming economic expectations depends upon an extension of the stereotype from the plane of the physical and technical to that of the psychical, social, and historical. By “stereotype,” we mean a fixed and stable pattern or configuration, composed of specific antecedent circumstances, or kinds of circumstance, and a specific sequel or end situation, or a specific train of consequences. By calling it stable, we mean that it occurs repetitively and can be frequently observed, or may even appear to be a universal and infrangible aspect of the natural order. It is upon such stereotypes (whether or not they have been combined into a larger, more general, and comprehensive system of knowledge) that all applied science or technology depends. When such stereotypes are used in isolation, they are rules of thumb. When combined into a great structure of interlocking ideas, they constitute a science or an art. In a bridge, the conformation, scantlings, and materials are together the source of strength; in a chemical process, the reaction is the consequence of juxtaposing certain substances under certain conditions of temperature and pressure; in biology a crossing of certain breeds may yield a predictable kind of offspring. When we look at events on a cosmical scale, the idea of the stereotype takes on a somewhat different aspect. If astronomers agree that the outer galaxies are receding from us, what is repeated is not their positions, as of planets in their orbits, but the recession at a given acceleration, that is, at a speed proportional to distance attained. Thus even an evolutionary, irreversible transformation can exhibit repetitiveness and the essence of the notion of the stereotype. It is when we come to events in the human and social sphere that the application of this tool, although still inescapable, becomes altogether more difficult and hazardous. Although Spengler (1918–1922), Toynbee (1934–1961), and Rostow (1960) have sought to demonstrate repetitiveness in history (as the only hope of making history a predictive science), still there are immense difficulties, dangers, and a widespread skepticism even (or perhaps especially) among historians.

These difficulties lead the businessman to distinguish two types of questions (Keirstead 1953): First, how will the general posture of affairs, describable by large aggregative statistical measurements such as the general level of prices, the monthly or annual expenditure of the community on all retail goods together or on goods of particular types, the percentage of the labor force unemployed, the rate of income tax or profits tax, and so on develop over the coming months or years? Second, what are the prospects for a particular enterprise—exactly specified as to the nature of its product; the type, design, and scale of its equipment; its location, marketing policy, research policy, management recruitment policy, and so on —given the particular industry’s record of fast or slow innovation, its composition of many small or a few large firms, etc.? For his answer to the first of these questions, he may rely on a particularly simple form of that permanence or stability in the world of nature and of man upon which all possibility of prognostication depends—namely, on what we can alternatively call inertia or momentum in affairs. Cataclysmic change can occur: a crucial battle can be lost; a revolution can break out. But, save for such upheavals, some of the great public aggregates cannot change very abruptly. Governments and political parties are committed to certain policies and programs and could not abandon them without electoral disaster; thus, the budget may leave little room for choice. To a lesser degree, the same may apply, although for different reasons, to the investment policies of large firms and publicly owned industries. Government expenditure and the investment expenditure of business enterprises together account for a large part of the Keynesian multiplicand, which, with a Keynesian multiplier deriving its numerical value and its stability from the habitual or conventional behavior of large masses of people in electing what proportions of their incomes to spend and what to save, largely determines national prosperity or depression. Alongside such reliance on the present as an image of the near future, businessmen pay great heed to each other’s opinions (Keynes 1937) without caring to explain to themselves how these can be better founded than their own.

Keynes (1936) has suggested the extreme mutability, fragility, and instability, with possible explosive or cumulative self-reinforcing consequences, that characterize such “conventional” opinions or assumptions about the future. Indeed, he paints a picture of the businessman’s suppressed and latent desperation, his urge to clutch at straws, and thus, finally, to swing in behind any apparent coalescence of opinion or hint of leadership and turn it into a mass movement. The bear and bull markets of the stock exchanges have their rationale but can partake also of hysteria. Since it matters nothing to the speculator whether or not a rise in share prices has any basis in the genuine situation of the firm concerned, he needs only to know that bulls of the shares are more numerous, wealthy, or convinced than the bears; then he can safely join the bulls in their self-justifying operation, which itself drives up the price. The game of trying to guess what the other man is guessing, what he is guessing that you are guessing, and so on to the nth degree, is plainly unrelated to accounting facts or the real state of the economy and leads therefore to inherently restless markets, which generate movements out of nothing or next to nothing. Moreover, no speculative profit can be made in any market unless there is a change of opinion, bringing about a change of price; and there exists at all times, therefore, a motive for some speculators to set afloat ideas that will cause such changes of opinion. Genuine investment decisions in the economist’s sense, that is, decisions by firms to order plant and equipment, are somewhat insulated from such speculative markets. But it is still true, as Keynes (1936, p. 151) indicates, that a firm that can acquire another firm as a going concern at a depressed stock exchange price may find this a more profitable use of its reserve funds than establishing a mint-new plant of its own. Such a choice tends, of course, to depress employment, unless the sellers of the shares use the proceeds themselves to order newly made equipment.

It was, at one time, common to speak of the “business cycle,” meaning by this a repetition of alternate phases of increasing and decreasing prosperity, as measured by national general output or income, employment, the general level of prices, and so on. The record of the nineteenth century and early twentieth century seemed to show a length, from crest to crest or from trough to trough, of from 7 to 11 years. The amplitude of these fluctuations seemed to increase greatly after World War I and to disappear completely after World War ii, being then replaced in the United States and Great Britain by a much shorter and milder cycle. At no time, however, could any such pattern have offered a sure guide to general business conditions for a number of years ahead. If there is a cyclical mechanism, its effect can be transformed or obliterated by the kind of extreme political and social upheavals that have occurred in our time. It has not been the experience of our generation that general history, in the large, can be foreseen.

It is in seeking an answer to the second type of question, concerning the prospects of a closely specified projected enterprise, that the businessman can more plausibly have recourse to stereotypes of business and investment situations. Such stereotypes will be quite distinct from case studies, although possibly founded on them. The stereotype will need to select and combine features from many such studies. The information a stereotype affords will be like that derived by a yachtsman about the likely performance of a sailing boat in a race, from a knowledge of the boat’s design. This knowledge may tell him how the boat will perform in this set of conditions or in that set, but can, of course, tell him nothing as to which set will in fact prevail. Such stereotypes may be derived by what we may call a clinical method, that is, by so close an examination of a few particular instances, recorded or observed, that insight is gained into the precise interplay of policy and circumstance and a grasp of the reasons for success or failure in each case obtained; or they may be derived by a statistical method, where only the probabilities of success, in this and that degree, of each given design of enterprise is studied and no attempt is made to penetrate the mechanism or rationale of this success. Naturally, these methods can be combined.

Probability. For an enterprise of given design, it will be natural to suppose that each distinct hypothetical degree of success will be associated with a different degree of probability. The pattern of association of these two variables, if statistically derived, will be expressed in a frequency table. If derived by some procedure of intuitive judgment, it will be expressed as a subjective probability distribution. We have to ask whether these two kinds of expression can, in practice, be so sharply distinguished. Many meanings have been proposed for the word “probability”; but these, for the most part, converge on the idea of recorded, or else potential, relative frequency, that is to say, on the notion of the counting of cases. To say that a given outcome Q has a probability of £ is to say that if a trial or performance, the nature and circumstances of which lie within strictly stated bounds, is many times repeated, the result will be Q in approximately J of the instances. Now when this trial or performance consists in tossing a coin or throwing dice, it may be easy to specify that the coin or dice must preserve their physical character exactly throughout the series of tosses and that each of these must be made by hand, in still air, etc. But what of the proposed setting-up of a business enterprise? The nature of the product; the type, design, and scale of the physical equipment; the location; and even the particular persons composing the initial cadre of directors and managers can no doubt be specified. The policy of the enterprise can be laid down in broad terms. But who can guarantee that the political, fiscal, social, commercial, fashionable, technological, and epistemic circumstances within which the enterprise will have to operate will remain unchanged? And if they change, what becomes of the meaning of probability? The relevance and applicability of any given statistical record of the past to the future experience of a new firm can be only a matter of judgment, that is, of conjecture, on the part of the enterpriser, since this relevance involves a double stretch of time in which the circumstances have evolved and will evolve further. Thus, in our context, there is no clear dividing line between statistical and subjective probability. But there are deeper questions.

Frequency ratios give us knowledge, namely, knowledge of the proportions in which the total list of individual results—one result for each of many repetitions of some kind of trial, a kind of trial exactly specified as to the range of permitted variation of its circumstances—have fallen under this or that head, or into this or that range of the variable whose values are the form taken by the results. This knowledge is applicable also to contemplated further repetitions of the same kind of trial, specified precisely as before. Let us call the complete list of contemplated further trials, considered as one whole, a divisible experiment. Then, a properly derived frequency table gives us knowledge of what will be the outcome of a divisible experiment, provided that experiment can in fact be performed. With a tossed coin or thrown dice, with any game sufficiently constrained by explicit rules, there is no question but that it can in fact be performed. But what of the establishment of a new business or a new plant? Here the tide of history itself is, willy nilly, changing some highly relevant circumstances all the time, rapidly or slowly. The question is whether the establishment of precisely this kind of business, in a given location and facing given conditions in the market for its product and in the market for its factors of production, can conceivably be repeated enough times before the whole scene has changed. If not, the statistical frequency table ceases to provide knowledge of what will happen and merely offers a tipster’s suggestion of which horse is the most promising, on past form.

Potential surprise. The enterpriser who tries to go beyond statistics and to gain insight into the conditions of success still needs a means of ex-pressing a degree of solidity, of seriousness, which he adjudges to this and to that imagined (hypothetical) degree of success or of misfortune of some given project. The rival (mutually exclusive) suppositions that he makes concerning the degree of success of some project, when these suppositions are treated as values of one variable, need to be associated with values of a second variable that measures the strength of the claim of each supposition to be taken seriously. Such a measure is afforded by the notion of subjective probability, which is the formal husk of statistical probability filled with quite different content; or it is afforded, alternatively, by a concept differing essentially from that of probability not only in meaning but in form, namely, a measure of possibility. The actual occurrence of something that we had supposed impossible causes us surprise. Some surprise is caused by an occurrence that we had only with difficulty been able to think of as feasible. Thus, a measure of adjudged possibility, or rather of degree of difficulty in supposing a given outcome possible, is provided by the potential surprise (Shackle 1949; 1955; 1958; 1961) to which the individual is exposed by his attitude to that outcome. Subjective probability, on one hand, and possibility or potential surprise, on the other, may each be called an uncertainty variable. They differ radically in nature. Probabilities must be assigned, in effect, to a list of contingencies looked on as exhaustive and complete; or, if probability is expressed as probability density, the function connecting these densities with values of the outcome variable must be supposed to rest on exhaustive and stable analysis of the underlying conceivable outcome situations. Since the exhaustiveness of the list implies that the factual outcome, when it shall emerge, will necessarily be found in one or another of the contingencies of the list, the probability to be assigned to the list as a unified whole is the equivalent of certainty, represented by the number 1. This probability is distributed over the listed contingencies in fractions that, therefore, sum to unity; and so the present writer calls subjective probability a distributional uncertainty variable. Possibility, or potential surprise, by contrast, can be assigned in mutually independent degrees to an indeterminate number of contingencies, the list of which, therefore, has no need to be looked on at any moment as complete. Potential surprise, measuring adjudged imperfection of possibility, is a nondistributional uncertainty variable. Methods of applying the two sorts of uncertainty variables must, like the nature of the variables themselves, differ radically.

Whereas a statistical frequency table can be applied directly, as it stands, to express the result of a divisible experiment, subjective probability, on the other hand, being concerned essentially with the outcome of a single, special, and nondivisible experiment, needs to be summarized to yield any usable message. This summary will take the form of a mathematical expectation, the result of multiplying each contingency (for example, each hypothetical amount of monetary profit) by its adjudged probability and adding together the results. Other moments of the distribution may also be used. The second moment, for example, indicates the dispersion of those “frequencies” that, although meaning little in the application of subjective probability as a guide to policy, are nonetheless the form in which the subjective probability judgments are expressed. The dispersion may be looked upon as a measure of the relevance, or lack of it, of the mathematical expectation. The doubtful relevance of any measure that seeks to average a collection of rival and mutually exclusive (that is, mutually contradictory) hypotheses is a chief reason for adopting the other type of uncertainty variable and using it as a basis of focus values. These are the two constrained maxima of a function that increases with the numerical size of supposed gain or loss and decreases with increase of the potential surprise associated with the various hypotheses. Such a function, or surface, may be visualized as follows. Imagine a seacoast where the straight shore line represents that degree of potential surprise corresponding to absolute rejection of hypotheses as impossible, while somewhat inland a line parallel to the shore represents zero potential surprise, the judgment that hypotheses are perfectly possible. A straight arm of the sea runs inland at right angles to the shore; and on either side of it near the sea, the sloping hillsides swing away to merge with those that front the sea. Sea level, along the shore edge and the central fiord, now stands for zero ascendancy, a zero degree of the “interestingness” of hypotheses, the power of any one imagined or hypothetical outcome of a contemplated action to arrest attention. Ascendancy will be nil for hypotheses adjudged impossible, and it will be nil, we are supposing, for those that mean no change in the decision maker’s situation, these being the hypotheses represented by points along the central fiord. The greater the improvement or deterioration represented by any hypothetical outcome (the greater the distance from the central fiord), the more interesting, the more ascendant, or attention commanding, the hypothesis will be. But the more nearly impossible (the nearer the sea) it is, the less will be its ascendancy over the decision maker’s mind. Now upon this landscape is superimposed a hill path, crossing the central fiord at its inland end, where potential surprise is nil, climbing on either hand across the bluffs where they swing round from the fiord to the sea, and descending again to sea level some little way along the coast. This path, in its shape in the horizontal plane, represents the adjudgment of potential surprise to diverse hypotheses of the outcome of some specified action. The highest altitudes attainable without quitting this path are the two constrained maxima of the ascendancy function, the focus points or, looked upon as purely scalar quantities of gain or loss, the focus values referred to above. A pair of focus values may be held to reconcile, as far as may be, the conflicting needs of the enterprisers’ situation: to settle upon one policy or course of action in face of an essentially plural or uncertain vision of the consequences.

Uncertainty of expectation is not a contingent, curable disability from which human beings will some day be rescued by the advance of science. To suppose it would be like supposing that the physical horizon of the ocean can be brought nearer by sailing toward it. Frank Knight has said that consciousness itself involves uncertainty. This surely must be so. Consciousness is the continual apprehension of subjectively new things, circumstances, and conjunctures that were hitherto not known to exist or to be imminent. To know the future would destroy the possibility of this stream of continually fresh perceptions. But at a somewhat less basic level, we may say that uncertainty is the price of hope, and hope is the sustaining force of human endeavor. Who would exert himself if the result were a foregone conclusion? How could the result be a foregone conclusion if no one exerted himself? Uncertainty is the price of hope, for only by exposing ourselves to possible loss can we expose ourselves to possible gain. The practical question is, How can this price be most explicitly, incisively, unmistakably expressed? How can it be made easy to judge and to grasp, so that a comparison can be made between what is paid in anxiety and what is gained in imagination of success? Can this best be done by throwing into the cauldron all the mutually contradictory hypotheses that the individual can think of, each multiplied by some subjective probability that apes the gestures but altogether lacks the meaning of statistical probability? Or can it be done in a more clear-cut fashion by setting clearly in view what the decision maker stands to lose by each given available choice of action? Both methods involve a judgment, a judgment, moreover, that by its nature quite escapes the categories, in P. W. Bridgman’s sense, of “operational” or “non-operational.” If measurement were possible, there would be no need for judgment; moreover, each method involves giving some weight to two or more possibilities, only one of which can come true. We are here concerned with a once-for-all, nondivisible, experiment.

G. L. S. Shackle

[See alsoDecision making, article oneconomic aspects; Probability, article oninterpretations.]

BIBLIOGRAPHY

Arrow, Kenneth J. 1951 Alternative Approaches to the Theory of Choice in Risk-taking Situations. Econometrica 19:404-437.

Bowman, Mary Jean 1958 Introduction. Pages 1-10 in Social Science Research Council, Committee on Business Enterprise Research, Expectations, Uncertainty and Business Behavior. Edited by Mary Jean Bowman. New York: The Council.

Carter, Charles F. 1950 Expectation in Economics. Economic Journal 60:92-105.

Carter, Charles F. (1953) 1957 A Revised Theory of Expectations. Pages 50-59 in Charles F. Carter, G. P. Meredith, and G. L. S. Shackle (editors), Uncertainty and Business Decisions. 2d ed. Liverpool Univ. Press. → First published in Volume 63 of the Economic Journal.

Carter, Charles F. 1957 The Present State of the Theory of Decisions Under Uncertainty. Pages 142-152 in Charles F. Carter, G. P. Meredith, and G. L. S. Shackle (editors), Uncertainty and Business Decisions. 2d ed. Liverpool Univ. Press.

Cohen, John 1960 Chance, Skill and Luck. Harmonds-worth (England): Penguin.

Edwards, Ward 1954 The Theory of Decision-making. Psychological Bulletin 51:380-417.

Egerton, R. A. D. 1960 Investment Decisions Under Uncertainty. Liverpool Univ. Press.

Georgescu-Roegen, Nicholas 1958 The Nature of Expectation and Uncertainty. Pages 11-29 in Social Science Research Council, Committee on Business Enterprise Research, Expectations, Uncertainty and Business Behavior. Edited by Mary Jean Bowman. New York: The Council.

Hart, Albert G. (1940) 1951 Anticipations, Uncertainty and Dynamic Planning. New York: Kelley.

Jouvenel, Bertrand de 1964 L’art de la conjecture. Monaco: Editions du Rocher.

Keirstead, Burton S. 1953 An Essay in the Theory of Profits and Income Distribution. Oxford: Blackwell.

Keynes, John Maynard 1936 The General Theory of Employment, Interest and Money. London: Macmillan. → A paperback edition was published in 1965 by Harcourt.

Keynes, John Maynard 1937 The General Theory of Employment. Quarterly Journal of Economics 51:209-223.

Lachmann, Ludvig M. 1959 Professor Shackle on the Economic Significance of Time. Metroeconomica 11: 64-73.

Marschak, Jacob 1950 Rational Behavior, Uncertain Prospects and Measurable Utility. Econometrica 18: 111-141.

Meredith, G. Patrick 1957 Methodological Considerations in the Study of Human Anticipation. Pages 37-49 in Charles F. Carter, G. P. Meredith, and G. L. S. Shackle (editors), Uncertainty and Business Decisions. 2d ed. Liverpool Univ. Press.

Niehans, JÜrg1959 Reflections on Shackle, Probability and Our Uncertainty About Uncertainty. Metroeconomica 11:74-88.

Rostow, Walt W. (1960) 1963 The Stages of Economic Growth: A Non-Communist Manifesto. Cambridge Univ. Press.

Shackle, G. L. S. (1949) 1952 Expectation in Economics. 2d ed. Cambridge Univ. Press.

Shackle, G. L. S. 1955 Uncertainty in Economics and Other Reflections. Cambridge Univ. Press.

Shackle, G. L. S. 1958 Time in Economics. Amsterdam: North-Holland Publishing.

Shackle, G. L. S. 1961 Decision, Order and Time in Human Affairs. Cambridge Univ. Press.

Spengler, Oswald (1918–1922) 1926–1928 The Decline of the West. 2 vols. Authorized translation with notes by Charles F. Atkinson. New York: Knopf. →; Volume 1: Form and Actuality. Volume 2: Perspectives of World History. First published as Der Unter-gang des Abendlandes.

Toynbee, Arnold J. 1934–1961 A Study of History. 12 vols. Oxford Univ. Press.

Watkins, J. W. N. (1955). 1957 Decisions and Uncertainty. Pages 107-121 in Charles F. Carter, G. P. Meredith, and G. L. S. Shackle (editors), Uncertainty and Business Decisions. 2d ed. Liverpool Univ. Press.

Weckstein, Richard S. 1952-1953 On the Use of the Theory of Probability in Economics. Review of Economic Studies 20:191-198.