Economics, Economists and Expectations

William Darity, Robert Leeson, Warren Young

• Publisher: Routledge
• Year: 2004
• ISBN:
  9781134886234
• Edition: 1
• Topic: Business
• Subtopic: Business General
• Index: Business

The concept of rational expectations has played a hugely important role in economics over the years. Dealing with the origins and development of modern approaches to expectations in micro and macroeconomics, this book makes use of primary sources and previously unpublished material from such figures as Hicks, Hawtrey and Hart. The accounts of the '

1 From Hayek to Vernon Smith

Prices, the cobweb, and game theory

Hayek, Hicks, Kaldor, and Morgenstern

In the decade between 1928 and 1937, Hayek developed an approach to expectations which still stands as one of the most original – albeit controversial – aspects of his overall approach to economic analysis. Much has been written about Hayek and Hayekian economics, and also about general aspects of his treatment of expectations, and it is not our intention to survey this material here. What has not been surveyed, however, are the specific characteristics of his treatment of expectations which highlight the originality of his approach, that is the distinction he made between level of expectation, the way in which he distinguished between types of foresight and the link he made between equilibrium, foresight, and expectations (Hayek 1928; 1933). For example in Monetary Theory and the Trade Cycle (1933) (originally published in German in 1928) Hayek asserted that given a large number of independent producers with individual views regarding future price, then their errors of pessimism and optimism would cancel each other out thus producing equilibrium output. While this somewhat “impossible result” was later criticized by, among others, Rosenstein-Rodan (1936) and Coase and Fowler (1937: 73), Hayek was perhaps the first to differentiate in this and subsequent work (Hayek 1935) between the individual agent (micro) and the aggregate (macro) aspects of expectations. For example, in an overlooked paper published in 1935 in Nationalokonomisk Tidskrift, Hayek distinguished between equilibrium and foresight regarding “complete” economic systems as against “certain” prices, such as interest.

In his now classic 1937 paper “Economics and knowledge”, Hayek went even further by establishing a connection between what he called “correct foresight” and equilibrium, treating the former as “the defining characteristic of a state of equilibrium” and distinguishing it from the case of “perfect foresight”. Moreover, Hayek provided a framework for what could be considered an “expectational equilibrium” by asserting that equilibrium existed only during the period when expectations were correct (1937: 36, 41–2). Furthermore, Hayek delineated an additional category, which he termed “relevant foresight”, asserting that for a state of equilibrium to be maintained, expectations needed “to be correct only on those points which are relevant for the decisions of the individuals” (1937: 2). Finally, Hayek focused on the role of expectations as primus inter pares with regard to equilibrium analysis and the problem of “constancy of data”. He said that in order to “include changes which occur periodically or perhaps even changes which proceed at a constant rate” the only way of defining “constancy” was with reference to expectations” (47–8). But, as in the other focal points of his treatment of economics, Hayek’s approach to expectations was overshadowed by that of Keynes, and thus, even though it was taken over and synthesized by Hicks (1933, 1939a) into a Walrasian general equilibrium framework, as will be seen below, it remains until today a somewhat overlooked aspect of his contribution to economics.

Hicks, Kaldor, and Morgenstern

As in Hayek’s case, the general treatment of expectations put forward by Hicks in Value and Capital (1939a) [below V&C] is well known and will not be repeated here. What has not received attention, however, is the specific approach to expectations he proposed in his 1933 paper “Gleichgewicht und Konjunktur” (“Equilibrium and the trade cycle” (Hicks 1980)), its relationship to Hayek (1928) and Tinbergen (1932), and Morgenstern’s (1935) overlooked critique of it on the one hand, and his unpublished debate in correspondence with Hawtrey over his 1939 V&C treatment of expectations, on the other hand.

In his 1933 paper among other things Hicks focused on the link between equilibrium and expectations. Indeed, to get around the “ ‘famous fiction’ of the Stationary State” as characterized in the general equilibrium system, Hicks pointed to the work of Knight (1921), Hayek (1928) and Tinbergen (1932) in which account was taken, in the production processes they described, “of the influence of future (expected) as well as current prices” on behavior. Moreover, according to Hicks, by “confining attention to stationary equilibrium, we can set future prices and present prices equal to one another, and so make the equilibrium determinate”. He went on to say, “however, the economic data vary, there will always be a set of prices which, if it is foreseen [our emphasis], can be carried through without supplies and demands ever becoming unequal to one another and so without expectations ever being mistaken. The condition for equilibrium, in this widest sense, is Perfect Foresight. Disequilibrium is the Disappointment of Expectations” (1933 [1980], 525–6).

In his own seminal RES paper entitled “A classificatory note on the determinateness of equilibrium”, Kaldor (1933–34) also dealt with Hicks’ 1933 treatment of expectations and especially the linkage between foresight, anticipations, equilibrium, stability, and “static” as against “dynamic” analysis. Kaldor identified what he took to be the two basic and “implicit” assumptions of “static analysis” as “all independent variables remain constant through time” and “all individuals expect the prices actually ruling to remain in force permanently: no price changes are anticipated” (1933–34: 123). To this he added in a footnote, referring to Hicks’ 1933 paper

Just because the dependence of equilibrium on anticipations is not always clearly realised, this assumption is hardly ever expressly stated although it is inherent in any type of static analysis which aims at demonstrating the tendency towards equilibrium independently of the degree of foresight. The only alternative assumption consistent with the degree of abstractness necessary for the generalisations of pure theory would be the assumption of complete foresight [italics in original]: that everybody foresees correctly the future course of prices. In this latter case, however, there is no need to assume constancy of the independent variables in order to show the determinateness of equilibrium: and consequently this latter assumption can be more conveniently adopted as the basis of a “dynamic” as distinct from a “static” type of analysis.

Kaldor concluded that “instability in the real world then appears as the result of wrong [italics in original] expectations (1933–34: 136) and added in a footnote to this, again referring to Hicks’ 1933 paper

Whether in any actual case anticipations will be in the right direction or not will depend partly on the nature of the change and partly on the efficiency of the institutions of the market whose function it is to anticipate future price movements. Given the forecasting ability of a speculative market, anticipations of future price-changes are as a general rule much more likely to prove correct when they are due to localised causes than when they are of a more general “monetary” character.

Now, the passages cited above have been referred to by a number of economists, as has Morgenstern’s (1928) earlier critique of “predictability in economics”. What is much less known, however, is the severe criticism leveled by Morgenstern (1935, [1963]) at the position advocated by Hicks (1933) regarding equilibrium, foresight, and expectations. For, in an article entitled “Perfect foresight and economic equilibrium” originally published in 1935 in the same journal (ZFN – in German) Morgenstern not only took issue with the position Hicks advocated, but asserted that it was completely mistaken!

In his critique, Morgenstern not only asserted that “full” and “perfect” foresight were synonymous, but he employed both terms, as he put it “in the essentially more exact sense of limitlessness” (1935, [1963]: 45). According to him, “full”, “perfect”, or as he defined the case “unlimited” foresight involved an “insoluble paradox” characterized by “an endless chain of reciprocally conjectural reactions and counter-reactions” (Keynes’ beauty contest metaphor) and claimed that “this chain can never be broken by an act of knowledge but always only through an arbitrary act – a resolution”. Morgenstern concluded therefore, that “Unlimited foresight and economic equilibrium are thus irreconcilable with one another” [Morgenstern’s emphasis] (1935, [1963]: 47). Morgenstern went on to say in regard, as he put it to “the famous Walrasian formulation that the equilibrium takes place ‘par tatonnement’ ”, i.e. “the determination of prices by Walras through the ‘prix crie’ and its successive improvements through the differing bids of buyers” that “successive adjustments are likewise irreconcilable with perfect foresight” (52) [Morgenstern’s emphasis]. He then dealt with the issue of “rational economic behavior” and what was involved if individuals acted rationally. In this regard, Morgenstern noted that “rationality” in this context posited “that the economic subjects themselves perceive the connections and dependencies – that they really see through the relationship to a certain degree” (53–4). Once again, in Morgenstern’s view, individually perfect foresight in this context would “assume that all individuals in the case have perfect knowledge – indeed uniformly perfect knowledge”, once again leading to the “completely insoluble paradox” in regard to “perfect foresight” and equilibrium (54).

In order to get around the problem, Morgenstern distinguished between perfect foresight and what he called perfect “purely theoretical knowledge of relationships” that is “perfect knowledge of a completed theory of equilibrium” assuming “that the theory of equilibrium already exists in complete form” and that “this complete science would be recognized uniformly by all economic subjects and understood equally well by all” (54–5). In other words, according to Morgenstern “a group of economic subjects can, consequently, have a perfect knowledge of the science, but they need not have greatly different knowledge of the future than men have today. These individuals are distinguished only by deeper insight into the relationships which arise from the arrangement of the data. But they may err in their assumptions about the data; optimism and pessimism can be expressed” (55).

Morgenstern then proceeded to state what can be said to be the strong form of the rational expectations hypothesis (REH). As he put it “with perfect foresight . . . there is identity between foresight and the expectation of the future” [Morgenstern’s emphasis]. He went on to explain this as follows (58):

If I know quite clearly that in three days a specified price will be at a specified level, then my knowledge is precisely the same as my expectation of the occurrence of this event. Had I expected another price, I should not have had certain, perfect foresight. In such an economy too, all factors of sentiment etc. would be eliminated. In the case of imperfect foresight, some other price is conceivable, for I cannot eliminate factors of disturbance from my expectation.

Having rejected the notion of perfect foresight, Morgenstern concluded that “Expectation depends, thus, only to a limited degree on foresight” (59) and that “it follows that the assumption of ‘perfect’ foresight is to be cut out from economic theory” (64).

However, while Morgenstern rejected the notion of perfect foresight and its link with equilibrium, he still maintained, as cited above, that there was a linkage between expectation and foresight. Thus, at the end of his 1935 paper, when talking about areas for “broader investigation”, Morgenstern proposed that it “proceed in a direction such that there are always . . . expectations about the future and that these . . . are bound up with a certain degree of foresight” which also “assumes a certain minimum amount of insight into economic relationships” (65). To obtain what he called “some picture of the relevance of the element of expectation” therefore, Morgenstern said that this would “require a new technique” and cited “a fruitful example of the introduction of the element of expectation” as being “illustrated by the special theory of duopoly” (66). In other words, as early as 1935 Morgenstern was advocating the introduction of expectations into economic analysis via a framework similar to the game – theoretic one he was to propose and publish with von Neumann a decade later, but more about this below.

Hicks, Hawtrey, Pigou, and Keynes

The early reactions to Hicks’ V&C (1939a) in general and Hawtrey’s 1939 review of it have already been dealt with (Young 1991) and this material will not be repeated. What has not been dealt with in detail up to now, however, is the unpublished exchange between Hicks and Hawtrey over the nature of foresight and expectations, and what can be seen as Pigou’s “early expectations augmented Phillips curve”. In addition, while the relationship between Keynes’ treatment of uncertainty and expectations in the Treatise on Probability, as against the General Theory, has been dealt with by many authors, the main issues deserve to be recalled here.

Hawtrey had taken issue with the V&C approach to “detailed forecasts” and expectations and challenged what Hicks called “perhaps the most important proposition in economic dynamics” that is his notion of the “elasticity of expectations” (Hawtrey 1939a: 310–11; Young 1991: 300–1). Attached to a letter from Hicks to Hawtrey dated 15 August 1939 can be found his “Notes on Hawtrey’s review of ‘Value and Capital’ ”. With regard to Hawtrey’s critique of Hicks’ assumed need for detailed “forecasts of input, output, prices, and rates of interest” by both “traders and consumers” in order to “regulate their actions” (Hawtrey 1939a: 310). Hicks replied:

This is of course a very crucial matter. I believe that the objections raised here at some length are largely answered on pp. 125–6 of the book [V&C] and by the reviewer himself on p. 310. There are two issues: the assumption of price-expectations rather than expectations of the “state of the market”; and the assumption of detailed expectations at all. On the first issue, I quite agree that it would have been better to assume “state of the market” expectations, but this meant assuming imperfect competition throughout, and I couldn’t see any way of getting to grips with my main problems if I assumed imperfect competition. But here I quite admit that my solution is a pis aller; still I hope the deficiency may be rectified some day to some extent.

On the other issue I feel on stronger ground. I do not of course suppose that a person who sets up a boot factory has to have some particular expectation of the state of the market in (say) three years’ time. Still this is a thing which will affect the profitability of his enterprise; so he has to make (implicitly) some assumption about it, even if that assumption is nothing more than a vague expectation of the continuance of something like his present conditions. However, he won’t always assume that; and even if he does, he will hold to these stationary expectations with more or less confidence in different cases, and these differences in confidence will make a difference to his policy. The whole point of my analysis is to get something general enough to include all these cases; and of course to include (as it does include) the ordinary cases as well.

The reader is perfectly at liberty to assume expected prices equal to current prices [our emphasis]. If he does so, he gets a special case of my more general construction. But it is not actually a very much simpler special case, because it is only in special circumstances that stationary expectations mean stationary plans (inputs and outputs constant over time) . . . We really have to be more general in order to get to the unity underlying this diversity.

Interestingly enough, this is quite similar to what Keynes wrote on conventions utilized in the face of untractable uncertainty and on the undue weight given to day-to-day fluctuations in business profits (1936: 148). Keynes went on to say (1936: 154) that he did not think undue weight should be given to day-to-day fluctuations in business profits “which are obviously of an ephemeral and nonsignificant character”.

According to his “Reply to notes on the review of ‘Value and Capital’ ” found in his papers (Hawtrey 1939b), Hawtrey replied as follows:

You say that to assume “state of the market” expectations would have meant “assuming imperfect competition throughout”. I do not think this is so, unless you regard the existence of goodwill or of the selling power of the individual trader as itself implying imperfect competition. But to my mind the trader’s selling power is such an essential part of the economic mechanism that it cannot be disregarded. The primary motive of enterprise is the expectation that the product contemplated can be sold at a remunerative price. That expectation may be based on the actual expansion of demand felt by an existing concern, or a new concern may be started to serve a new or expanding community or to put a new product on the market. The reward expected is demand at a remunerative price, but apart from being remunerative the price itself does not enter explicitly into the trader’s calculations. He is quite prepared to assume that, if his costs change, those of his competitors will change similarly, and that no very violent change in the volume of demand will result. There is of course a risk of a big change of costs (e.g. a serious scarcity of raw material) or of a collapse of demand, which would upset all his calculations. But traders are not deterred by these hazards. If you want to form a picture of the trader’s expectations, you will put in the foreground his hope of a steady stream of sales at a price which will yield a normal margin over the cost of producing in the most up-to-date manner prevailing in the industry. This hope will be modified by the possibility of various contingencies, favourable and unfavourable, but few if any of these will have even an approximate date suggested for them. By attributing dates to a series of contingencies extended into the remote future, you are not giving the theory greater generality, but only divorcing it completely from the facts.

Finally, Hawtrey concluded that he did not think “traders to make detailed estimates for every week a thousand weeks ahead”.

Pigou’s early expectations augmented Phillips curve

Pigou sought to bring business cycle theories to “the test of fact” (1927: 23–4, 34–5, 120, 192–3) and while warning against inferring causation from simple correlation, presented a curve (Curve 11) displaying British unemployment and prices for an almost identical period (and sub-periods) from which Phillips (2000 [1958]) derived his curve. Chart 16 (Pigou 1927: 194) showed a “very close” correspondence between unemployment (inverted) and the rate of change of prices; in “close accord” with Fisher’s results for the United States.

Inflationary expectations and animal spirits, i...