Expectations, Uncertainty, and Decision-Making

When Robert Lekachman put together in 1964 a Report of Three Decades of Keynes’s General Theory, he reproduced reviews and obituaries of Keynes originally published in the 1930s and 1940s, together with new contributions in which authors were asked to give their present evaluations of the Keynesian revolution. Champernowne (1964) chose the theme of expectations, which he saw as central to Keynes’s research programme and hitherto (i.e. in the early 1960s) relatively unexplored. As observed by Don Patinkin (1990), interpretations of Keynes’s central message as focused on uncertainty and expectations started to appear in the early 1960s only and in particular in books and essays by Shackle. Champernowne’s (ibid.: 175) ‘revisiting’ of The General Theory meant a suggestion to read the text with an eye to the role of expectations as a link between the economic future and the present, with no effort to take into account ‘more recent attempts’ to interpret and improve upon Keynes’s ideas in that regard (see Champernowne’s 1963 review of Shackle, which is critical of the latter’s theory of ‘degree of beliefs’ as a basis for the theory of decision-making).

Champernowne (1964) may be seen as one of the first comprehensive accounts of Keynes’s discussion of the effects of expectations about the future on present economic behaviour. As pointed out by Champernowne, it is through the concepts of marginal user cost, marginal efficiency of capital, and liquidity preference for money that opinions about the future affect current economic decisions in Keynes’s framework. The Marshallian concept of marginal user cost allowed Keynes to bring in the influence of long-term expectations on firms’ pricing policy. The marginal user cost includes all extra cost connected with the firms’ long-term prospects, as it corresponds to the consequent loss of capital equipment at the end of the period plus extra expenditure during the period on the product of other businesses. The value of capital equipment is related not to its replacement cost but to its prospective yields. Keynes’s definition applied not only to the effect of extra output on the dates in which the equipment would need replacement but also to any effects on the prospective markets for the firm’s goods. Champernowne (ibid.: 178, fn. 4) recalls that such notions were part of Cambridge’s oral tradition, as illustrated by Shove’s lectures on the theory of value at King’s College in 1934-1935. Expectations also play an important role in Keynes’s treatment of the aggregate demand function as part of the formulation of the principle of effective demand. Champernowne (ibid.: 180), however, finds problematic Keynes’s shift, in the middle of Book II of The General Theory, to actual ‘ex post’ values, as it suggests a greater stability to the model than can in fact be claimed. Under the definition of the propensity to consume as composed of the sum of a set of separate functions determined by the firms’ expectations of expenditures on their products (as a function of the known level of employment), it is liable to oscillation due to changes in each firm’s prospects.

According to Champernowne, Keynes’s concept of marginal efficiency of capital contained at least two novelties as compared to the traditional approach to investment. The relevant expectations are not quite those of employers planning capital investment, but private investors dealing on the stock exchange. On the other hand, Keynes’s assumption of diminishing returns in the capital goods industry reduced the amplitude of fluctuations in the rate of investment and ensured that small temporary changes in the economic climate do not set off a cumulative process of ‘rising’ expectations and expanding activity. Expectations involved in Keynes’s notion of liquidity preference cannot be described in the form of a probability distribution, but as a range of possibilities associated with such psychological states as bullishness, bearishness, and uncertainty. Moreover, as observed by Champernowne, Keynes’s demand for money function clearly distinguished between variety amongst individuals in their expectations and uncertainty within each individual’s own personal expectations. Chapter 17 of The General Theory, entitled ‘The Essential Properties of Interest and Money’, attracted Champernowne’s attention. He regarded it as the most perplexing and theoretically deep chapter in the book. Keynes attempted in that chapter to explain why the volume of output and employment is more intimately bound up with the money rate of interest than with own rates of interest of other assets. After a detailed discussion of Keynes’s involved argument, Champernowne (ibid.: 198-199) concluded that Keynes’s case rested on the assumption that wages would necessarily be and were expected to be more sticky in terms of money than in terms of other assets, which he found wanting. He agreed with Keynes’s point that instability would follow if money wages changed in response to movements in employment, but contended that a constant expectation of a gradual drift upwards of money wages and money prices may not cause instability. On the contrary, it may allow the commodity rates of interest to fall very low (due to price expectations) even when the money rate of interest remains above that level due to liquidity preference.

Champernowne (ibid.: 200-201) endorsed Keynes’s (1936 [1973]: 293) suggested division of economics between the theory of stationary equilibrium and the theory of shifting equilibrium where changing expectations are capable of affecting the current situation. The importance of money as a ‘link between the present and the future’ is only apparent when one is dealing with the theory of output and employment as a whole in shifting equilibrium. Champernowne had not always been positive about Keynes’s distinction. He was ‘sorry to find that in 1936 I added at this point in my copy [of the General Theory] the marginal note “cranky”’ (Champernowne 1964: 201, fn. 30). Indeed, when he wrote his 1936 article about Keynes and the classics, Champernowne focused on the labour market, not on the inter-temporal coordination problem caused by the absence of futures markets. This changed in his 1964 revisiting of The General Theory:

Keynes made it abundantly clear that there was no efficient link between the decisions of employers how much to invest and the decisions of individuals how to divide their incomes between consumption and savings ... But although he did not give such a clear demonstration, he was greatly concerned also with the lack of any proper link between employers’ plans regarding capacity for production of consumption goods in the future and the future decisions of individuals to spend on consumption goods. He realized that the lack of such a link would lead to violent swings in expectations and activity (ibid.: 201).

Keynes had much to say about how expectations about the future affect present economic behaviour, but not so much about the converse problem, that is, the effect of the present on expectations about the future. Champernowne (ibid.: 192-194), like others before and after him (see e.g. Hart 1947), was of the opinion that Keynes lacked a theory of the formation of expectations. Instead of setting up a detailed theory of the causation of expectations, Keynes discussed the ‘conventions and habits of thought’ which are behind these expectations. This remains a controversial aspect of the interpretation of Keynes’s framework. It has been argued by Bateman (1996: 130-131), for instance, that Keynes assumed that agents form their expectations (or beliefs) intersubjectively by social conventions, via an extension of Ramsey’s subjective probabilities.

The nature and influence of uncertainty in many areas of the economic system, as well as the methods available for proceeding in the face of it, formed the subject matter of Champernowne’s 1969 trilogy, probably the first of its kind in the economics literature. He aimed at discussing the techniques for interpreting imperfect or incomplete statistical information and obtaining statistical estimates, or choosing decisions in the light of evidence (Champernowne 1969: volume 1, page v). The book dealt in detail with the general problem of how imperfect information may be used for estimation, decision-making, and statistical inference in economics. It was part of the series ‘Mathematical Economics Texts’ published between 1969 and 1972, edited by Arrow, Solow, Hahn, and others. Amongst other well-known titles, the series included Sen’s Collective Choice and Social Welfare and Arrow and Hahn’s General Competitive Analysis. The trilogy was suggested to Champernowne by Hahn, and benefited from discussions with other Cambridge economists including Reddaway, Deane, and Mirrlees, and from comments by Arrow (who visited Cambridge when Champernowne was working on volume 1). It combined theoretical, applied, and philosophical discussions of uncertainty and estimation, in typical Cambridge fashion.

Champernowne’s (1969: volume 1, page 5) starting point is that the more important economic problems involving probability arise ‘not directly from the “stochastic” or “chance” element in economic change, but from the ignorance and uncertainty influencing the economic decisions of individuals’. This is why the ‘objective’ theories of probability, so convenient in natural sciences, are ill-suited to economics, where ‘personal’ (or ‘subjective’) theories of probability—which deal with the effects of ignorance and uncertainty on the behaviour of individuals— should be deployed instead. What sets Champernowne (1969) apart from most other books in the field, as pointed out by Stone (1972) in his extended review, is the comprehensive coverage of the foundations and varieties of probability theory. Champernowne (1969: volume 1, Chapter 9) was especially careful in his account of Bayesian techniques for estimation, supposed to be the most general procedures available for revising probability models in the light of statistical evidence. Bayesian procedures are ‘extremely powerful’ (ibid.: 7) provided sufficient initial information on probabilities is available, which is often not the case. It is mainly for this reason that ‘cruder and less ambitious special purposes statistical techniques’ (ibid.) are often used, as discussed in volume 2. The last volume is concerned with the microeconomic and macroeconomic effects of uncertainty and risk aversion, including portfolio selection, and decision-making policy to prevent disasters. Interestingly enough, Champernowne (1969: volume 3, pages 2-3) provides some illustrations, taken from the Old Testament, of the ability to protect from natural disasters which can be anticipated: Noah’s Ark, Joseph’s buffer stock scheme, and Catastrophes associated with Moses, who ‘took advantage of a foreknowledge of very unusual tidal movements to lead his people over a stretch of dry land which was at all other times submerged’. Surely, those are not passages usually to be found in econometrics textbooks. But, then again, Champernowne was one of the ‘most colourful’ economists of the last century.