You are on page 1of 6

Cambridge Journal of Economics 1985,9, 173-178

NOTES AND COMMENTS Jean Baptiste Marglin: a comment on 'Growth, Distribution and Inflation'
Edward Nell*

It is not mere politeness to describe this article by Stephen Marglin as an extraordinary work, a synthesis, both clear and profound, of the central elements of two great traditions of economic thinking. It brings together and contrasts, then synthesises, basic ideas from the Marxist and the neo-Keynesian traditions. It does so in an exceptionally clear and simple analytical framework, admirably suited for the purpose. Yet, its very clarity makes it possible to see its shortcomings all the more plainly. The first question that springs to mind is whether the project is, in fact, possible. Can it be done at all, at least in the way Marglin proposes? This is not an idle question. Of course, in one sense, it can be done, and very well, too. Marglin has done it. But the question is whether what he has done is legitimate. To compare neo-Marxist and neo-Keynesian theory, they have to be made comparable. The Marxist tradition pays little attention to problems of aggregate demand, which are central to the Keynesian. Marglin adopts the Marxist position, and re-interprets the neo-Keynesian approach, setting it in a framework in which problems of effective demand are eliminated by assumption. Whatever is produced will be sold, and the neo-Keynesian principles come into play in determining the relative shares of workers and capitalists. But for Keynes, the point of The General Theory was to change the question from one of distribution to the determination of the level of output as a whole. Keynes without effective demand is Hamlet without the Prince.
Marglin's approach

Downloaded from cje.oxfordjournals.org by guest on August 18, 2011

Marglin begins by defining a one-sector framework of capitalist production. Corn is produced by workers planting seed-corn, in accordance with fixed coefficients. The output after subtracting the necessary seed-corn is divided between consumption and investment. Growth is denned as the ratio between next year's output and this year's. The price of corn is defined in relation to labour cost, and the profit rate is defined in terms of the current price. Both prices and wages are defined in terms of money, although no account of the monetary system is ever provided. Output equals consumption plus investment and also equals wages plus profits. But output isfixedin size, which is taken as unity in any given period. Next Marglin turns to savings behaviour, which he takes to be fundamentally a matter of class. Workers do not save, capitalists do, but also
*New School for Social Research, New York.

174

E.Nell

consume. Hence the growth rate (since all savings are invested) will be given by the product of the capitalist savings propensity and the rate of profit. But this does not close the model; there are two unknowns, g and r, but only one equation connecting them. A second is needed, and Marglin canvasses two candidates. Neo-Marxian theory completes the model by determining the real wage through the reserve army, while neo-Keynesian theory closes it by determining the level of investment through the interaction of the rate of interest, the rate of profit and the price level. Each of these ways of closing the model appears to capture a significant insight into the working of modern capitalism, but together they would overdetermine the system. Yet this need not be a defect; it might, on the contrary be an important insight in itself (Nell, 1970), for it could provide the basis for an account of inflation, and the last part of the paper is devoted to this. Fixed output and the Quantity Theory But the assumption of fixed output is crucial to the entire argument, and seriously distorts his account of the neo-Keynesian approach, although it echoes a famous stance once adopted by Kaldorand mocked by Samuelson (Kaldor, 1956; Samuelson, 1964). This shows up very dramatically in the derivation of the basic neo-Keynesian behavioural functions. Both the short-run investment function, for example, and longperiod growth formula are derived by setting aggregate demand equal to aggregate supply, where the latter is assumed to befixedby technology (Marglin, 1984, Equations 16 and 18, and Fig. 6). Fixing supply in this way, however, leads to some difficulties. Consider the short-run analysis. By assumption, workers consume their wages, which are assumed fixed in money terms in the short run, while capitalist investment demand is also fixed in money terms, in the short run, by decisions to borrow taken on the basis of last period's price level and rate of profit. Current real investment demand therefore varies inversely with the current rate of profit and price level. The level of real investment must equal that of real savings, so this is determined together with the price level. That is, given total demand in money terms, with total supply fixed, the money price level follows immediately, and the rate of profit is calculated by finding the real wage. For once the price level is known, since the money wage is given, the real wage follows, and so the rate of profit is determined. It is simple. It is also precisely the approach of the old Quantity Theory of Money. According to the Quantity Theory, the money supply determined the level of money spending, given the velocity, so that with total output fixed, the price level followed. Similarly Marglin has fixed both the level of money spending, corresponding to MV, and total output. But he gives no reason why firms cannot adjust their levels of borrowing for investment purposes in the light of the difference between this period's price level and last period's. If they could, then firms could be supposed to make their real investment plans on the basis of the current price level and rate of profit. This would result in the normal neo-Keynesian argument: when (real) investment demand is below/above savings, the price level will fall/rise, so the real wage will rise/fall, on the assumption that money wages are lessflexiblethan money prices. The real wage adjusts to clear the product market, rather than the labour market. But this no longer treats the real wage as a residual; far from it. The level of the real wage sets one component of effective demand, and investment the other. This adds a dimension to wages not present in the Marxian schemeor in Marglin's version of neo-Keynesian thinking. Moreover, there is no reason to suppose that investment and savings will be equated only at full capacity. At less than full capacity, adjustments will primarily take the form of income changes rather

Downloaded from cje.oxfordjournals.org by guest on August 18, 2011

Comment on 'Growth, distribution and inflation'

175

than of movements in the price level. The assumption offixedoutput, Equation 16, and the entire subsequent discussion, runs counter to the Keynesian approach.
Credit and the rate of interest

There is another respect in which Marglin's argument parallels the Quantity Theory: both neglect the relationships between money, credit and interest. Marglin claims to capture the central neo-Keynesian insight that because of the credit system, capitalists, unlike workers, operate without a budget constraint, at least in the usual sense. That is, given attractive projects a firm's ability to raise money will not be constrained by its current earnings; in general, business's investment spending is not constrained by current earnings in the way household consumption spending is by current incomecontrary to the way business decisions are portrayed, for example, in much contemporary 'disequilibrium' literature. This is, indeed, a crucial element in the neo-Keynesian vision. But Marglin tries to present it in a model which includes money borrowed and lent, but has no specification of the money supply and no determination of the rate of interest. Yet he himself points out that investment depends not simply on the (expected future) rate of profit, but on the ratio of this rate to the current (and expected future) rate of interest. Consider the effects of this point on his short-run investment demand function. Suppose the system is in disequilibrium, with the rate of profit too high. Savings will exceed investment, and the price level must fall, so that real investment can increase. But the investible funds will have been borrowed at the current rate of interest, equal to or proportional to the initial rate of profit. If the price level falls, the rate of profit comes down, and it is going to be difficult to service debt contracted at the earlier, higher rate of interest. But if the debt can be rolled over, why are firms supposed to be constrained, in the short period, to a certain level of money expenditure? Of course, if they are not so constrained then Marglin's short-period story collapses, and we are back with the more traditional neo-Keynesian system, in which investment and saving need not balance at full capacity.1 But, to repeat a point made above, if output and employment vary with effective demand, we are outside the Marxian framework. The two systems cannot be mixed, if one of them assumes that output isfixed,while the other treats it as variable. Of course, the Marxian framework could be used to examine the long-period path of accumulation, while the Keynesian dealt with fluctuations about that path. But that is not Marglin's approach; nor is it adequate, since there are long-period problems of demand. Say's Law It might be replied that since the problem being studied is long run, it is reasonable to neglect the (essentially short-run) variations in sales due to the vagaries of effective demand, in order to concentrate on the true problems of accumulation and distribution. From a neo-Keynesian perspective this is nonsense. Effective demand is not a matter of short-run 'vagaries'; it is a matter of investment, i.e. of capital accumulation. The growth of capacity must keep in step with the growth of demand. This is the essense of the multiplier-accelerator process, more generally expressed as the capital stock adjustment principle. There are problems in this approach, and it may well be that some of them have never been adequately dealt with, but the idea that capital accumulation must
'Notice that if investment is below the capacity level, then employment, and so with a given money wage, the wage bill, must also be below capacity. But total money expenditure will fall short of the full capacity level by mart than the shortfall in investment itself. This is the multiplier effect here.

Downloaded from cje.oxfordjournals.org by guest on August 18, 2011

176

E.Nell

proceed part passu with the growth of demand is surely not subject to doubt. Indeed, Say's Law is nothing but the statement that this happens automatically because supply always creates adequate demand. By contrast, the fundamental proposition of the Keynesian tradition is that neither in the short run nor in the long is there any reason to suppose that this balance will be achieved at full capacity by the unregulated market, except by accident, or as a temporary phase in a cyclical movement. This point can be illustrated by examining a remark of Marglin's in connection with the pattern of causality in the neo-Marxian model: ' . . . change the conventional wage and both the distribution of income and the growth rate change: change the propensity to save and only the rate of growth changes' (1984, p. 119). The second part of this will be true only if Say's Law holds, that is if all and only what is saved is automatically invested. Otherwise, suppose at full employment of capacity, with a given real wage, and a given level of investment, the propensity to save changed. Suppose it fell; at the initial rate of profit, investment would then exceed savings, so prices would be driven up, the real wage would fall, and the rate of profit increase. The process would stop when savings once again equalled investmentunless the higher rate of profit stimulated further investment, in which case the process might repeat itself, perhaps indefinitely. But if the propensity to save rose, the changes would not be symmetrical, for when savings exceed investment, income and employment will be adjusted downwards, rather than prices. Nevertheless, in either case a change in the propensity to save changes the rate of profit; when the propensity to save falls, this follows from a change in distribution, but when it rises, from a change in income and employment.
Real wages and growth

Downloaded from cje.oxfordjournals.org by guest on August 18, 2011

There is an even deeper problem. Is it, in fact, the case that a rise in the growth rate at full capacity can only come about through a fall in real wages? This is certainly the neo-Keynesian tradition, and is also in the spirit of Ricardo and Marx, but is it really true? Of course, if we assume no technical progress, no change in labour intensity, rigid coefficients and strict capacity limits, etc., we can make it true tautologicallyno help to anyone. The serious question is whether this is a good way to theorise about increasing the growth rate when the economy is operating at what is regarded as the normal level of capacity utilisation, let alone during a period of stagnation. One of the largest sustained expansions ever recorded under capitalism took place in the USA in the early 1940s, yet real wages rose faster than productivity throughout this period, so that the share of wages and the rate of growth both increased (Nell, 1983). Of course a wartime economy is special; nevertheless, high employment together with high and rising wages led to a burst of productivity increases, due partly to increased effort on the part of labour, and partly to modernisation. High and rising wages, after all, can be expected to stimulate labour-saving innovation. Nor does it make much sense to think of capacity as rigidly fixed. Firms normally carry excess capacity for the same reasons they carry inventories and precautionary money balances. Of course, these are comments, not about Marglin's argument, but rather about his assumptions. But they turn out to be crucially important to his policy conclusions. Inflation When we come to his account of inflation, however, some questions about the argument itself arise. The neo-Marxian determination of the real wage gives rise, according to Marglin, to a simple adjustment process, in which the current real wage adapts by

Comment on 'Growth, distribution and inflation'

177

moving towards the equilibrium level. But when the price level is determined by the short-run neo-Keynesian mechanism, the neo-Marxian pressures will govern the movements of money wages. Which of these sets of pressures is stronger, and why? Under what circumstances? Marglin does not inquire. Instead he assumes a steady-state condition, that the rate of money-wage inflation exactly equals the rate of price-level inflation. When this holds the real wage is steady, and so therefore are the profit and growth rates. Starting from some initial position, the long-run equilibrium is determined, with a permanent inequality between savings and investment, disequilibrium real wage, and steady inflation.' But why do the pressures of aggregate demand on aggregate supply just exactly balance the pressures that workers can exert over money wages? Marglin's steady state is possible, but why is it a more likely outcome than any other? What economic forces, with what strength and speed of operation, tend to pull the two rates of inflation into line with one another? And which of the two tends to adapt the most to the other? These questions are never raised, much less answered. Further, the answers are likely to depend on the way investment and effective demand are treated. If the accelerator principle were introduced, the story might look quite different (Nell, 1982). And it might be important to provide an account of the monetary and financial system. Worse, there are difficulties in each of the adjustment equations. The wageadjustment equation (Marglin, 1984, Equation 20 or 31) implies that if the initial real wage is above the equilibrium, money wages will fall, even if the economy is at full employment, with prices rising due to an excess of investment demand over savings. This is surely not plausible. Similarly, if savings exceed investment, prices will fall even if money wages are rising (Equation 22). Marglin can reply that he only intends to draw on these adjustment equations in conditions of steady, balanced inflation, but this means that he cannot use his model to analyse the forces that would tend to establish such a state. In other words, he could not, in principle, answer the questions of the previous paragraph. At this point, too, the one-sector assumption becomes problematical. Suppose two distinct kinds of goods had been assumed, produced in different sectors: investment goods and consumer goods. An excess of investment demand over savings would then imply an increase in the price of investment goods, but the shift in the distribution of income required to increase savings would need a fall in the real wage, i.e. a rise in the price of consumer goods. The price increase, in other words, takes place in the wrong sector, and moreover, changes the relative price ratio as well. Clearly the pattern of adjustment will have to be substantially more complicated, and it cannot be assumed that it will always work out (Nell, 1982).
Policy implications and conclusions

Downloaded from cje.oxfordjournals.org by guest on August 18, 2011

As so often, these theoretical questions have policy significance, but the implications are not as exact as well-organised and tidy minds might desire. Most of what Marglin says would be supported by the ideas I have drawn on in criticising him. But there are a few crucial areas where a difference might arise. For example, Marglin accepts the necessity for the Left to ' . . . respect the logic of the economic situation. Productivity does place
'In Marglin's inflation the realised real wage is always in disequilibrium; so it is hard to understand how the equilibrium level can be 'conventional' or 'normal'. How can something which never happens be the norm?

178

E. Nell

limits on wages, and not just physical limits. As long as profitability remains the mainspring of investment, there are economic limits that constrain the wage share. Under capitalism, profits are indeed the geese that lay the golden eggs' (p. 47). This rests squarely on his assumption of Say's Law; when output is determined by effective demand, investment governs current profitability, not the other way around. Marglin wants the Left to demand worker participation in management in return for agreeing to wage restraint. Reasonable enough, but is a generalised policy of wage restraint necessary or even sensible in modern conditions? First, modern economies have clearly not been operating at anything like full capacity during the past decade and a half. Raising wages would increase demand in consumer goods markets. But, secondly, it would also increase the pressures on backward firms to modernise or go out of business, an effect that does not depend on the economy operating at less than capacity. This raises productivity, but it also leads to investment. In short, under suitable conditions raising wages may stimulate growth. Of course, wages could be raised too much, or in the wrong spheres, leading to bankruptcies and collapse. Likewise, an investment strike could be provoked, or a flight of capital abroad. It is easy to imagine scenarios in which high wages could be fatal (although controls on prices and capital movement might prevent or mitigate these effects). But it is hard to avoid the conclusion that the phenomenal growth of US capital had something to do with the high wages imposed by the existence of the frontier, high wages which generated a huge market, on the one hand, and strong pressures to innovate on the other. In any event Marglin's framework, since it assumes Say's Law, cannot deal with these questions, and is forced to accept the position that austerity policies are nothing but simple 'economic logic', which any rational agent must accept. Conservatives could ask no better defence. Bibliography
Kaldor, N. 1955-56. Alternative theories of distribution, Review of Economic Studies, vol. XXIII Marglin, S. A. 1984. Growth, distribution and inflation: a centennial synthesis, Cambridge Journal of Economics, vol. 8, no. 2, June Nell, E. 1970. A Note on the Cambridge Controversies, Journal of Economic Literature, vol. VIII, no. 1, March Nell, E. 1982. Growth, Distribution and Inflation, Journal of Post Keynesian Economics, vol. 5, no. Nell, E. 1983. Beyond austerity: economic growth and economic policy, unpublished ms, New School For Social Research, New York Samuclson, A. 1964. A brief survey of post-Keynesian developments, in R. Lelachman (ed.), Keynes' General Theory. Reports of Three Decades, London, Macmillan

Downloaded from cje.oxfordjournals.org by guest on August 18, 2011