Investment and Accumulation

Abstract

The standard view of accumulation goes something like this. In the short period, fraught with frictions and maladjustments, the demand for investment interacts with the supply of saving, more or less à la Keynes, to determine the growth of the capital stock. Keynesian policies may have fallen into disrepute, but for the short period the representative economist continues to use the tool box developed in the General Theory and its wake: accumulation falls out from the determination of national income.

The standard view of accumulation goes something like this. In the short period, fraught with frictions and maladjustments, the demand for investment interacts with the supply of saving, more or less à la Keynes, to determine the growth of the capital stock. Keynesian policies may have fallen into disrepute, but for the short period the representative economist continues to use the tool box developed in the General Theory and its wake: accumulation falls out from the determination of national income.

In the longer period the same economist falls back on very different arguments: the mainstream of the profession takes accumulation to be determined by saving propensities, with nary a side glance at investment demand. That Japan has over the last quarter century devoted 30 per cent of gross output to fixed capital formation and Great Britain 20 per cent is conventionally explained in terms of higher Japanese saving propensities, not in terms of a greater propensity to invest.

In a still longer time frame, even saving propensities become irrelevant. In the asymptotic future beyond all future, accumulation is determined solely by population growth and technical change. Saving propensities may affect the steady-state capital: output ratio if the technology admits of substitution between labour and capital, but that is the limit of their influence.

Economists of a Marxian bent share the mainstream view, up to the asymptotic future, which they rightly dismiss as an irrelevant construct. The terminology may differ: difficulties of ‘realization’ is favoured for describing a shortfall of aggregate demand relative to aggregate supply, and hence (abstracting from foreign trade and government surpluses or deficits) for a shortfall of investment relative to saving. But for Marxians realization problems are generally confined to the short period; in the long run it is once again the saving propensities of capitalists, along with the rate of profit determined by class struggle, which determine the rate of accumulation. To be sure, neoclassical and Marxian theories of the determination of saving propensities differ, but for present purposes this is a secondary issue; the short run apart, the two theories agree that investment propensities are irrelevant to accumulation.

Investment in the General Theory

Against these views stands the Keynesian view, which, applied to the long run, tells a very different story of the accumulation process. In the General Theory Keynes formalized his view of investment demand in terms of a ‘marginal efficiency of capital’ schedule which showed the amounts of investment that would be forthcoming at different rates of interest. The basic idea behind this schedule can be captured by supposing there to be a set of projects indexed by i = 1, …, n, each requiring a unit of investment and returning, respectively, a cash flow of r₁,…, r_n in perpetuity. If the n projects are arrayed in descending order of r_i, then with the simplifying assumption that each investment costs one dollar, r_i is the marginal rate of return on the investment of i dollars. In Keynes’s language, r_i is the marginal efficiency of capital.

Suppose now that the interest rate d, which represents the cost of capital to the investor, is also expected to be constant in perpetuity. Then the present value of the ith project’s return is r_i/d. The profit-maximizing investor will go down the list until he reaches the project at which r_i = d, that is, the point at which the marginal rate of return just equals the cost of capital. In Keynesian terms, investment is determined by equating the marginal efficiency of capital to the cost of capital. More precisely, the array of projects being discrete, the profit-maximizing rule is to undertake all projects for which r_i > d and to reject those for which r_i < d. If there is a project for which the relationship between r_i and d is one of exact equality, it is a matter of indifference whether the project is undertaken or not. The main point is that the discounted present values of returns, r_i/d − d acts as the discount rate – exceed the assumed unit cost of the investment provided r_i > d.

So far there is nothing novel in this theory. Knut Wicksell would have had no problem making the argument his own, and indeed the schedule of the marginal efficiency of capital bears a close resemblance to the marginal productivity of capital schedule of mainstream theories of accumulation. Even the overall theoretical structure which Keynes builds by joining this schedule to schedules of consumption and liquidity preference would not have been uncongenial to Wicksell, particularly if its application is confined to the short period. This is presumably why Wicksell’s Swedish followers chided Keynes for indulging ‘the attractive Anglo-Saxon kind of unnecessary originality’ (Myrdal 1939, p. 8; the comment naturally refers to the Treatise, not to the General Theory).

Probably because of the affinity to a Wicksellian version of neoclassical economics, the mainstream of American economists has been able, as was suggested at the outset of this essay, to accept a version of the Keynesian analysis for the short period – much to the annoyance of Keynes’s Cambridge disciples, like Joan Robinson and Nicholas Kaldor, who all their lives insisted that Keynes’s main message was being lost in the translation. In the standard American view, the main point of Keynes was the limit of monetary policy, conceived of in terms of its affect on d, to affect investment demand, either because of a low elasticity of the marginal efficiency schedule, or, in the limit, because of the impossibility of reducing the interest rate (the famous ‘liquidity trap’, of which Keynes said in the General Theory that it was, as yet (1935), a theoretical possibility of which there had been no actual instances). The bottom line was the need not only for state intervention in the form of an activist monetary policy – this was fully present in Wicksell’s analysis – but also in the form of fiscal policy. Indeed it does no disrespect to Keynes to accept that his influence owed as much to the intellectual justification he provided for an activist, interventionist state – for the end to laissez faire – as to the intellectual power of his ideas, certainly as these ideas were, reflected through the prism of the mainstream of the American economics profession.

An Alternative Reading

But there is another reading of Keynes. The real departure of the Keynesian theory of investment from the orthodox one, in my judgment, starts from a recognition that the formalism of his theory of investment demand obscures its real content. The starting point is the recognition that the returns of any project, lying in the future, are inherently uncertain. The r_i’s are not objectively given reality but a subjective construction of the investor. It has become fashionable to blur the old Knightian distinction between risk, objective and quantifiable, and uncertainty, subjective and qualitative, by means of the theory of subjective or personal probability. Even if the axioms underlying this theory are neither compelling (particularly the assumption of a complete ordering and the assumption of ‘independence’, what Leonard J. Savage called the ‘sure thing’ principle), nor borne out empirically in the behaviour of untutored individuals, subjective probability theory still has some heuristic value in modelling investment decisions, particularly in its emphasis on the psychology of the decision maker.

Subjective probability allows us to go behind the r_i’s of present value calculations like the simple perpetuity formula r_i/d to more complex sums of the form

$$ {r}_i^h={p}_1^h{u}_1^h{r}_{i1}^h+{p}_2^h{u}_2^h{r}_{i2}^h+\cdots +{p}_m^h{u}_m^h{r}_{im}^h, $$

in which the generic term p^h_ju^h_jr^h_ij is composed of these elements: p^h_j is Mr h’s subjective probability of the occurrence of a particular complex of events (a ‘state’) in which his marginal utility of income (normalized) is u^h_j and his estimated return from project i is r^h_ij. The central point is that in the Keynesian view, each of the constituents – the probability p^h_j, the marginal utility of income u^h_j, and the state-specific return r^h_ij – owes as much to the imagination of the investor as it does to an objective reality. The more optimistic are investors, the higher the probabilities they will attach to states in which the returns and the marginal utilities of these returns are high, and the consequence will be higher values of r^h_i. Conversely, the more pessimistic are investors, the lower the r^h_i’s that will be attached to the same projects. Thus the ‘animal spirits’ of investors play a crucial role in investment demand.

The recognition of a crucial role for animal spirits directs one’s attention away from movements along the marginal efficiency schedule. The question becomes, what determines the position of this schedule? Evidently, according to what has just been said, it is, in the last instance, investors’ evaluations of probabilities of various states, of the relative utility of income in different states, and of returns in different states.

It is equally evident that this makes a cumbersome theory. A more tractable model can be constructed by making the prospective returns the r^h_i’s, depend on the general anticipations of capitalists. The higher is the general expectation of profits, the higher will be the anticipated rate of return on specific projects – a rising tide will be anticipated rate of return in specific projects – a rising tide will presumably lift all boats. In this view, movements of the entire marginal efficiency schedule, triggered by changes in expectations of profitability, not movements along a given schedule induced by changes in the interest rate, are the key to understanding the ups and downs of investment and output.

Such reasoning – this is of course a ‘rational reconstruction’ – permitted Keynes’s heirs, Roy Harrod in Oxford and Robinson and Kaldor in Cambridge, who re-situated the General Theory in a long-period context, to recast the marginal efficiency schedule in terms of variables related to expected profits. Robinson’s investment demand function, for example, and the argument of Keynes’s own formulation, the rate of interest, disappears into the background of ceteris paribus.

The rationale for ignoring the cost of capital is the assumption of a highly insensitive responsiveness of investment and saving to plausible interest rate changes. Clearly, there are strong assumptions at work here about the relevant range of interest rate variation. As long as the anticipated returns are finite, there must be some level of the interest rate at which even the most attractive projects appear to all and sundry as uneconomic. Thus, unless saving is negatively related to the rate of interest and highly elastic, sufficient variation in interest rates could, with enough time, adjust the demand for investment to the supply of saving.

For most of the postwar period, however, the range of variation of interest rates has been too modest to test this possibility, at least if we identify the interest rate with the difference between the nominal rate on government or high grade corporate bonds and the rate of inflation. Indeed, from 1945 until 1980 this ‘real’ rate of interest never moved very far from zero in the United States. In the post-1980 disinflation and recession, real interest rates rose to levels which must give pause to even the most devoted neo-Keynesian. It is certainly too early at this writing to determine whether in the sweep of history this was a momentary aberration or the dawn of a new era; my own leaning is towards aberration, but that may reveal my neo-Keynesian predilections rather than a reasoned guess about the future.

The rate of profit generally expected on the capital stock as a whole (re) is itself no more observable than any other variable that lies in the future. In the neo-Keynesian literature re’s customarily are taken to be a function of a small number of variables which summarize the relevant information available to investors. The standard version of the theory simply extrapolates the current or immediate past rate of profit.

With simplifying assumptions like constant returns to scale, homogeneous capital, and a uniform, exogenously given rate of capacity utilization, it must be true ex post that the average rate of profit on new investment (which is the marginal rate of profit on the capital stock) turns out to equal the average rate of profit on all capital. Under these assumptions it might appear reasonable for the expected average rate of profit on the entire capital stock re to equal the expected rate of profit on new investment, which would impose, in the spirit of ‘rational expectations’, a consistency requirement for each investor

$$ {r}^e\equiv \sum_i{r}_i^h\equiv \sum_i\sum_j{p}_j^h{u}_j^h{r}_{ij}^h. $$

But no such inference is warranted. Neo-Keynesian theory in my view is compatible with rational expectations – insofar as this notion makes any sense at all under conditions of subjective uncertainty. But an important measure of realism would be lost by constraining the theory in this way, for even under the stringent assumptions that lead to a uniform realized profit rate’, there is no good reason to believe in rational expectations. The general expectation of a 4 per cent return is perfectly consistent with individual expectations that special gifts, opportunities, or kismet will result in a 10, 20 or even 50 per cent return on one’s own projects. As P.T. Barnum would have it, a sucker is born every minute.

Self-Fulfilling Prophecy

Keynes made Barnum’s sucker into a kind of hero: in Keynes’s view it was doubtful that capitalism could function without such a ‘spontaneous urge to action’. Neo-Keynesian theory makes an even stronger assertion, which has its roots in Keynes’s Treatise on Money. As long as there are enough of them, the ‘suckers’ can turn the tables on the rest of us. Capitalists, as a class, have the power to shape conditions so that their expectations come true, at least in large enough part to maintain their confidence.

Unique among economic actors, this class has the power of self-fulfilling prophecy. Not only do actual profit rates affect capitalists’ beliefs about future profits, capitalists’ beliefs also have an impact on actual profits. This is not the same thing as the ability Joan Robinson imputed to capitalists to make the profit rate anything they liked, but it is a formidable power nonetheless.

There are two mechanisms by which capitalists’ prophesies about profits become self-fulfilling, the distribution of income between capital and labour and the level of capacity utilization and employment. The first is the less familiar one, except to readers of Keynes’s Treatise on Money. Suppose a closed economy with no government spending or taxation in which the starting point is a long run equilibrium characterized by equality both between desired saving and investment and between expected and actual (average) rates of profit. Imagine a change in ‘animal spirits’ which makes capitalists willing to undertake more investment at the going rate of profit than earlier was the case. In other words, imagine an outward shift in the investment demand function. In the simplest neo-Keynesian story, this addition to aggregate demand increases spending relative to income – remember income has remained unchanged – and drives the price level upward. Assuming money wages are fixed or at least sluggish, this reduces the real wage and shifts the income distribution in favour of profits. The process continues since higher realized profits lead to further expectations of higher profits and still more investment demand. But it does not continue indefinitely, because capitalists are assumed to save a higher proportion of their incomes than do workers. The upward spiral of prices and investment continues only until the extra saving induced by higher profits absorbs the extra investment, at which point the economy comes to a new equilibrium where desired saving and investment and actual and anticipated profits are agains equal, albeit at a higher level. In addition to the existence of an investment demand function that is distinct from the saving function, the sluggishness of money and wages and the difference in saving propensities between classes are crucial to this result.

Several observations are in order here. First, although the fons et origo of this theory is Keynes’s Treatise, the theory has much in common with the model outlined in Josef Schumpeter’s Theory of Economic Development. There are two important conceptual differences however. First, the neo-Keynesian equilibrium is formulated as a steady growth rather than stationary (zero growth) state that dominated in Schumpeter’s time. Thus shifts from one equilibrium to another involve changes in the rate of growth rather than changes in the level of output.

A second difference is more fundamental. In both the Schumpeterian and the neo-Keynesian view, an outward shift in the investment demand function plausibly involves an expansion of the array of projects yielding returns in excess of the cost of capital. And in both views, the psychology of the capitalist class is crucial. Schumpeter no less than the neo-Keynesians recognized the subjective element in the estimation of returns. ‘Invention’, for Schumpeter, was necessary but not sufficient for ‘innovation’. But here the resemblance ends. In the neo-Keynesian view invention is neither necessary nor sufficient for innovation. Investment can be a pure boot-strap operation; the theory requires nothing more than a change in business psychology to change investment demand, and a change in investment demand can lead the economy to a new equilibrium with a different rate of growth. Within the limits of saving propensities and the malleability of real wages, capitalists wishes are self-fulfilling. Prices and profit rates change to validate the changes in capitalists’ expectations!

The Crucial Assumptions

There must be a trick. In fact, there are four crucial assumptions, two of which – the flexibility of real wages and the difference in propensities to save between capitalists and workers – have already been mentioned. The assumption that wages are set in money terms plays a central role in the analysis. Money wages need not be fixed once and for all, but Robinson’s version of the neo-Keynesian story (and the Schumpeterian story for that matter) cannot be told at all without the assumption of sluggish money wages. Evidently it is the income distribution that adjusts saving and investment to each other. So if the income distribution is fixed, then it cannot do the job which neo-Keynesian theory assigns it.

A difference in saving propensities is equally important to the neo-Keynesian view of capitalism. There are various versions of the so-called Cambridge saving equation, and a fair amount of confusion about the content of alternative versions exists two decades after the most important contributions to this debate. But all versions of the theory take it for granted that capitalists’ propensity to save exceeds workers’. By contrast, the principal neoclassical theory of saving, Franco Modigliani’s life-cycle hypothesis, suggests that the propensity to save out of wages will exceed the propensity to save out of property income: wages will be disproportionately in the hands of people preparing for retirement and profits disproportionately in the hands of retirees. Theoretical dispute is of course nothing new in economics. What is more surprising is that we lack persuasive empirical evidence of one view or the other two decades after the theoretical battle was fairly joined.

The essential role that flexible real wages and differences in saving propensities play is relatively transparent, so perhaps ‘trick’ is not an appropriate description for either of these assumptions. But a third assumption is necessary to make capitalists’ investment spending into a ‘widow’s cruse’ (Keynes’s metaphor), filling up with saving as fast as it is emptied in investment. This assumption is better hidden. We can see its role more clearly by asking how the process described for moving from one equilibrium in consequence of a shift in investment demand ever gets started. How is it that an increase in desired investment gets translated into effective demand? (The same question, it may be noted, might be asked about any displacement from macroeconomic equilibrium, for instance, the textbook displacement of a short-period equilibrium by a shift in the investment function.)

One possibility is that desired saving increases in line with desired investment, but this assumption in essence requires us to abandon the idea of separate saving and investment schedules. A second possibility that can be dismissed almost as easily is the just-so story fashionable in my youth: we were told to think in terms of cash ‘hoards’, with ‘dishoarding’ as the essential mechanism for initiating the disequilibrium transition from one steady state to another. That story won’t wash because under contemporary conditions there simply aren’t cash hoards of the requisite magnitude – if there ever were.

There is a better answer, and interestingly Wicksell, as well as Schumpeter and Keynes, give it in about the same way. Schumpeter is the clearest of the three, in contrast to whom Keynes is practically incoherent, perhaps because he thought the main point of the story so obvious that it did not require elaborate explanation. The main point, in two words, is credit money. The process of expanding investment can get started with no accompanying increase in desired saving if capitalists are assumed to have access to an accommodating banking system which one way or another can create claims on scarce resources out of whole cloth. Here the psychology of financial capital joins the psychology of industrial capital, for unless the financiers share the optimism of industrialists, there is no way, absent those mythical cash hoards, by which investment can increase without a contemporaneous increase in desired saving, as the neo-Keynesian and the Schumpeterian story, with their reliance on price level and profit rate shifts to accommodate capitalists, would have it. Or for that matter, the Wicksellian story. Although ultimately it is the interest rate which adjusts desired investment and saving, Wicksellian disequilibrium is not a Walrasian virtual or hypothetical imbalance of tâtonnement with false trading, but an imbalance in real time which is sustained by credit money.

The importance of an accommodating banking system, or passive or endogenous money – these are all approximate equivalents – to the neo-Keynesian system explains why partisans of this view are necessarily hostile to the quantity of money theory of prices (the so-called ‘quantity theory of money’, but this is an obvious misnomer). The dispute is evidently not about the relationship between the quantity of money and the price level, a definitionally true relationship in its customary form, but about causality. In the neo-Keynesian view it is aggregate demand which drives both sides of the quantity equation, not MV (the product of money and velocity) which drives PX (the product of prices and quantities).

In an earlier essay on this subject (Marglin 1984a), I suggested that in the neo-Keynesian story capitalists, as a class if not individually, were approximately in the position of the present Aga Khan, who in his student days is reputed to have asked his Harvard economics instructor how the theory of consumer choice worked without the budget constraint. V. Bhaskar has persuaded me that the analogy is misleading. The point is not that capitalists, even as a class, face no budget constraint: capitalists must be assumed to repay whatever debts are contracted to finance investment. The point is rather that, as a class, capitalists are able to change what are normally regarded as parameters of the budget constraint. By increasing relative demand, capitalists drive up the prices of the goods they sell relative to costs of production. The consequent increase in profits provides the wherewithal to retire debt as it becomes due.

There is a final assumption which must be introduced in order to make the neo-Keynesian argument that the longer run growth of the capitalist economy, as well as the share of investment in output and the profit rate, is sensitive to capitalists’ animal spirits. This is the assumption of slack resources, specifically, slack labour-force growth must lead to continued capital deepening, which has finite limits both in the real world and in the textbook world of smooth substitution between capital the labour. The simplest, as well as the most realistic, justification of slack labour, is the argument that the capitalist sector of the economy is embedded in a larger entity from which it can, if needed be, draw labour. I refer here to a long run ‘reserve army’ constituted both by sectors of the national economy (the farm in an earlier day, the kitchen more recently) and by sectors of the international economy (the immigrants who provided labour for 19th- and 20th-century expansion in the United States and for the post-World War II boon in Europe).

Without this assumption, the power of capitalists to shape the history of capitalist economies would be much closer to the power any group has in a standard Arrow–Debreu model to influence relative prices through its preferences or, in a stochastic model, through its expectations. On the assumption of a limited quantity of land suitable for growing tobacco, a shift of smokers’ preferences with respect to the pleasures of nicotine and the horrors of lung cancer will affect the equilibrium price of cigarettes. In a stochastic model, the belief that sunspots or any other exogenous variable matters may be sufficient to make the variable matter, even with the assumption of rational expectations (Azariadis 1981; Cass and Shell 1983). At issue in both these cases however is the distribution of a given pie, which stands in sharp contrast to neo-Keynesian theory, where distribution bears on the size of the pie itself.

Observe that the existence of slack resources is much more problematic in the long run than in the short period which was originally the focus of Keynes’s analysis. For the short period, although controversy is not lacking even here, mainstream economists will generally accept it is the exception rather than the rule that capacity or manpower constrains output. It is in the long run that the true proportions of the neo-Keynesian departure from orthodoxy reveal themselves.

The Level of Real Wages: Effect or Cause of Accumulation?

In a sense, this version of neo-Keynesian theory, like Keynes’s own General Theory, proves too much. One cannot ask many of the more interesting questions about how changes in investment demand change the economy. For most of these questions turn out to be about movements along the demand curve rather than changes in its position, which mean that they are necessarily transitory if the initial position was one of equilibrium. For instance, one cannot ask what the effect of lower (or higher) real wages might be because the real wage, determined ultimately by the price level, is a consequence rather than a thermostat. Thus to find a thought experiment which illustrates the neo-Keynesian theory requires some care; it would not have done, for example, to take as the premise of a shift in the investment demand function that Mrs Thatcher and Mr Reagan had abolished collective bargaining. For while this might plausibly lead to the expectation of a higher profit rate, this expectation would translate into a movement along the existing investment demand schedule rather than a shift of the schedule. And supposing the economy was at equilibrium in the first place, a movement along the investment demand schedule is not sustainable unless the saving schedule shifts simultaneously.

In fact questions about the effect of changes in real wages on equilibrium come out of a very different approach to capitalism than that embodied in the General Theory. There Keynes calls this approach ‘classical’, even though the economists Keynes apparently had in mind as its exemplars were more neoclassical than classical. Classical may still be the right label since the Marxian strand of the classical theory, as typified by Michal Kalecki, even more than the neoclassical strand, makes the level of real wages a central determinant in the theory of accumulation, a thermostat rather than a thermometer. But if real wages are given exogeneously, whether as a rate or a share, something else has to adjust if both investment and saving propensities are to continue to play a determining role in the accumulation process.

There are at least two ways out, besides the possibility of partitioning outcomes by ‘regimes’ in which one or another consideration – wages, investment, or saving – is a nonbinding constraint. One possibility is to allow each of the three determinants to operate with diminished force; this is the tack followed in Marglin (1984a, b).

Capacity Utility as an Adjustment Mechanism

Another possibility is to allow capacity utilization to enter the discussion more fully both as cause and as effect. This is not entirely unproblematic since it can be argued that the appropriate assumption for the long run (as distinct from Keynes’s short period) is that capacity utilization settles down at some ‘normal’ rate, in other words, that it is not a variable. But it is hard to regard this issue as resolved; both the view that capacity utilization is endogenous and the view that it is exogeneous in the long run can claim some empirical support.

The endogenous view, which Amit Bhaduri and the present author are currently developing, builds indirectly on the work of Kalecki and more directly on the work of Bob Rowthorn (1982). This view entails a significant reformulation of the investment demand function: investment demand is no longer a function of the expected profit rate but rather is a function of two of its constituents, the expected profit margin qe and the expected rate of capacity utilization ze. Definitionally re = qezey, where y represents the output:capital ratio at full capacity utilization, but with capacity utilization variable, there is no compelling reason why qe and ze should affect investment demand symmetrically, as they do in this formula. If, for example, the expected rate of profit is high because the profit margin is high and the expected rate of capacity utilization is low, the impact on investment demand may be very different from what a high expected rate of profit based on lower profit margins and higher capacity utilization would induce. In the second case there would be relatively little need for investment in order to have sufficient capacity to meet expected demand. Investment demand would be limited to new products, new processes, or the substitution of capital for labour. Thus the formulation of the expected rate of profit from a project as

$$ {r}_i^h={p}_1^h{r}_{i1}\left({q}^e,{z}^e\right)+\cdots +{p}_m^h{u}_m^h{r}_{im}\left({q}^e,{z}^e\right) $$

is at once more general and more plausible than the formulation in which r^h_i turns on re alone.

In this model, unlike the model in which aggregate investment turns on the expected rate of profit, one can ask questions about exogenous shifts in real wages. Profit margins are determined in large part by struggles over real wages, so an increase in the expected wage lowers qe and leads to a reduction in investment demand, that is, to a movement down the investment demand schedule. But the movement may be permanent since there is now another degree of freedom, capacity utilization, to accommodate the change.

A word of caution: the outcome of a change in real wages cannot be predicted solely on the basis of the qualitative structure of the model. Whether higher wages and lower profit margins will increase capacity utilization and accumulation depends on the relative strength of two opposing tendencies. On the one hand, from the capitalists’ point of view, a decrease in profit margins makes investment less desirable at a given rate of capacity utilization. On the other hand, a shift in the distribution of income from capital to labour may be expected to increase consumption demand and thus to stimulate capacity utilization. Since investment is by assumption sensitive both to profit margins and to capacity utilization, the relative strength of these two influences becomes crucial. This framework is thus broad enough to accommodate both the ‘stagnationist’ interpretation of Keynes with its emphasis on high real wages as an engine of accumulation and an ‘exhilirationist’ interpretation, which emphasizes high profit margins.

Evidently a variety of theories, and an even wider variety of models, is possible within the broad Keynesian vision that aggregate demand matters. The essence of that vision is the role of investor psychology, the strength, in Keynes’s words, of ‘animal spirits’ that ‘urge to action rather than inaction’. Perhaps a paraphrase of Marx puts the main point best: capitalists make the history of capitalist economy, but not in circumstances of their own choosing. Whatever the mechanism of adjustment, whether income distribution or capacity utilization or some combination of the two, capitalists occupy a position of singular privilege in the neo-Keynesian conception, possessing the ability to impress their subjective construction of the future on the present functioning of the economy.

See Also

• Accumulation of Capital

• Classical Growth Models

• Neoclassical Growth Theory