The Point of Effective Demand

Abstract

Keynes's principle of effective demand conceives competitive equilibrium in terms of the choices of entrepreneurs, investors and consumers, rather than of the optimal allocation of factors of production. In The General Theory, effective demand is distinguished from aggregate demand and from income, expected or realised, and there is no suggestion that equilibrium means the convergence of expectations. Reconsideration of Keynes's use of time and equilibrium periods leads to the conclusion that he treats employment as in continuous equilibrium, at the point of effective demand, determined by the state of expectation, the correctness of which is strictly irrelevant. The nature of the equilibrium represented by the point of effective demand is here described, not in terms of the multiplier, but in terms of the continuous equilibrium of supply and demand in short-term forward markets. This reading is faithful to Keynes's conception of aggregate demand as dependent upon the expectations of entrepreneurs, and it resolves the meaning of his ‘long-period employment.’ Formal appendices identify the differences between Keynes and Walras and the nature of the multiplier. The paper concludes that the Keynesian cross and ‘Swedish’ analysis should be abandoned, and the Walrasian conception recognised as only the limiting case of general competitive equilibrium in a monetary economy.

Acknowledgments

An early version of this paper was read to a meeting of the Post Keynesian Economics Study Group at Cambridge, UK, in May 2004. I am grateful to Victoria Chick, Paul Davidson, Giuseppe Fontana, Geoff Harcourt, Jochen Hartwig, Mark Roberts and two anonymous referees for their comments. The usual disclaimer applies.

Notes

¹For a comprehensive account of early treatments from 1947 to 1964 see King (1994). On the literature between 1964 and 1989 see Section 2 below.

²The unqualified term ‘Keynesian’ will be used in this paper to refer to the neoclassical synthesis of Hicks, Samuelson, Tobin and Modigliani rather than the ideas of Keynes himself.

³The term ‘involuntary’ unemployment intrinsically requires a non-Walrasian concept of equilibrium with non-clearing factor markets, to some extent found in Pigou as well as in Keynes (see Hayes, 2006). Keynes should not be confused with the French school of ‘general disequilibrium’ models based on quantity rather than price adjustment, in which disequilibrium remains defined with reference to the Walrasian equilibrium, since Keynes maintains the assumption of market clearing in all markets other than factor markets. Arrow & Hahn (1971, p. 366) refer to Keynes's concept of general equilibrium as a ‘quasi-equilibrium.’

⁴See Kregel (1976), Patinkin (1976), Roberts (1978), Davidson (1978), Wells (1978), Parinello (1980), Casarosa (1981, 1984), Chick (1983), Torr (1984), Dutt (1987), Asimakopulos (1989) and Amadeo (1989).

⁵Roberts (1978, p. 371), in denying that ‘any dynamic process is required … to bring an economy to a position of unemployment equilibrium’, comes within a whisker of making the same claim, but draws back in his footnote 10.

⁶The consensus about Keynes's use of time periods, from which this paper departs, is that Keynes's day and production periods coincide, and correspond to a Hicksian week (Chick, 1983, p. 20; Hansson, 1985, p. 334; Amadeo, 1989, pp. 12–16; Fontana, 2004, p. 80; Hartwig, 2004, p. 312), an equation which tacitly assumes a uniform production period for all goods. Daily employment thus differs from the short-period employment equilibrium in which expectations are fulfilled (Casarosa, 1984, p. 942n). Hicks takes the short period to be Marshall's ‘few months or a year’ (Hicks, 1980, p. 141; Marshall, 1920, p. 314).

⁷Keynes's ‘day’ relates to calendar time since it reflects employment practices; it is not simply the indefinite interval between changes in the opinion of an entrepreneur, but the period since the last employment decision after which any such change or sequence of changes can be acted upon. Bradford & Harcourt (1997, pp. 120–121) suggest otherwise, and so question the precision of Keynes's definition of the period of production as n days (GT, p. 287). Keynes's day need not be a terrestrial day, but it does no harm to think of it as such.

⁸The production period of an individual good in isolation is not explicitly defined in The General Theory but appears in a 1933 draft (CW XXIX, p. 88) as the time after which a decision to employ labour realises finished output, ready for use in production or sale, as at p. 46 of The General Theory. Keynes uses the terms ‘accounting’ and ‘employment’ period to mean this production period in early drafts (CW XXIX, p. 74). See also Marshall (1920, p. 315n) on the distinction between the macro and micro production period.

⁹It is the quantum nature of Keynes's short period that justifies treatment of the state of long-term expectation as strictly independent, and answers Kalecki's (1936) critique.

¹⁰These production periods represent the optimal production plan (in the sense of Hicks, 1939, p. 193) in a given state of expectation, if more than one technical process, each with different delivery dates, is available to a given industry (GT, pp. 215–216).

¹¹By contrast, Kregel (1976, p. 210) associates ‘short-period’ with particular individual expectations and ‘long-period’ with the general state of expectation, rather than with production and investment decisions respectively (GT, p. 47). This association leads to a quite different reading, including the likelihood that employment is in disequilibrium (Kregel, 1976, p. 217), although the main conclusion remains that the disappointment of expectations is not the central issue.

¹²Yet more confusion is caused by the distinction often made in Keynesian economics between the short term, in which factor and product prices are sticky, and the long term in which they are flexible in the absence of permanent rigidities. The underlying concept of equilibrium is Walrasian (i.e. all markets clear in the long term), so that flexible prices would deliver full employment.

¹³In contrast with the present argument that Keynes's long-period employment is an integral part of his theory of effective demand, Carvalho (1990) and Asimakopulos (1984, 1989) consider ‘Keynes's discussion of [long-period employment] has the nature of an aside, without any consequence for the model presented in his book’ (Carvalho, 1990, p. 286). For Carvalho, the long-period equilibrium may never be reached, even in principle. Nevertheless, these authors correctly identify that ‘a long-period model “in the old sense” [of Marshall] … will not do’ (Carvalho, 1990, p. 289). Hansson (1985, p. 336) argues that Keynes's long-period employment requires, as a theoretical fiction, the fulfilment of long-term expectations. Yet Keynes's long period is short enough that a constant state of expectation, in which short-term expectations are fulfilled, is at least conceivable in practice. From this perspective, Keynes did in fact consider the dynamic consequences of short-period investment for future productive capacity (Asimakopulos, 1989, p. 26), without which the theory of effective demand is incomplete. See also CW XXIX, pp. 221–222 on the distinction between long term (run) and long period. Hicks (1939, p. 206–212) analyses convergence from short- to long-period equilibrium consequent upon production time lags in similar fashion to Keynes, and both, in certain respects, anticipate ‘real business cycle’ analysis.

¹⁴Amadeo (1989) interprets this switch as Keynes almost dropping ‘supply’ in moving from the Treatise on Money (Keynes, 1930) to a final ‘expenditure’ version of the principle of effective demand. On the present reading, the principle of effective demand relates exclusively to ‘supply’, meaning the production decisions of entrepreneurs in short-period equilibrium, but after Chapter 5 of The General Theory, Keynes assumes that the short-term expectations behind those production decisions are based mainly on realised results, as determined by the ‘expenditure’ decisions of consumers and investors in market-period equilibrium.

¹⁵This suggestion develops Keynes's explanation of effective demand to Hawtrey: ‘The main point seems to me to affect the whole supply and demand theory and not my version of it in particular. I have the impression that you restrict the supply and demand method to market prices only, that is to say, they relate to the higgling of the market in respect of stocks which already exist. But that is not what Marshall or Pigou or most modern economists do. The demand which determines the decision as to how much plant to employ must necessarily concern itself with expectations. And I am in this respect simply trying to put more precisely what is implicit in most contemporary economics’ (CW XIII, p. 602). Also, a few pages further on: ‘The only thing that really matters is that the given state of expectation, whatever it is, does produce by its effect on the minds of entrepreneurs and dealers a determinate level of employment. But I should find it difficult to do without my schematism as a convenient method of quantifying the state of expectation’ (CW XIII, p. 616; emphasis added).

¹⁶Under perfect competition, it is of no consequence if the employers are also dealers in their own or other employers' products; the analytical division corresponds merely to the two types of decision. It is also easier, but not essential, to think of the dealers as self-employed or as employing a fixed workforce.

¹⁷These forward markets extend only so far into the future as required to govern output over Keynes's long period; they do not represent a complete set of futures and insurance markets in the sense of Arrow & Hahn (1971). Keynes's ‘correct foresight’ (CW XIII, p. 603) or ‘rational expectation’ is limited to the short term and is contingent on the state of expectation; such foresight will be disappointed if the state of expectation changes.

¹⁸See Keynes's response to Robertson: ‘I do not remember attributing the disappointment of entrepreneurs “to a divergence between aggregate demand and aggregate supply price”. I attribute their failure to produce more to this; but their disappointment, if any, I attribute (like you) to a divergence between aggregate demand price and income’ (CW XIV, p. 89; cf. Robertson, 1936, p. 169).