But, then, there is no objective
yardstick
- that is, other than the laws, habits and customs of bourgeois society - to tell us which method to use.
Nitzan Bichler - 2012 - Capital as Power
It turns out that, as we move from labour values to prices of production, the results do not 'add up': the outputs of the price system generally differ from its inputs.
12 According to Bortkiewicz, the inconsistency occurs because Marx's transformation is incomplete.
It converts surplus value counted in labour time into profits counted in prices, but it does not do the same for constant and variable capital.
The resulting price system therefore is half-baked - partly price denominated, partly value denominated.
Capitalists and workers, though, buy and sell commodities in prices rather than values, so it makes sense to specify two distinct systems - one
11 Bortkiewicz's arguments were summarized in Paul Sweezy's Theory of Capitalist Development (1942) and later included in his translated collection, Karl Marx and the Close of his System (1949).
12 Save for special assumptions, this inequality means that in simple reproduction the overall price of the wage goods produced by the system differs from the total value of wages (so workers consume less or more than capitalism makes available to them); that the overall price of investment goods differs from the total value of the used up constant capital (so there is net investment or divestment); and that the overall price of luxuries is different from the total profits of capitalists (so they must save or dissave).
? 100 The enigma of capital
denominated in values, the other in prices of production. And, indeed, with a dual-system specification, Bortkiewicz showed the transformation of values into prices of production to be mathematically consistent.
But there was a hefty price to pay for this consistency: the dual system no longer satisfies Marx's aggregate conservation principle. With separate systems for prices and values, there is no longer a single ? that would satisfy Equation (3) for all commodities. The result is that total prices need not be equal to total labour values and total profit does not have to be the same as total surplus value. And given these inequalities, the rate of profit denomi- nated in price terms is no longer the same as the rate of profit calculated from labour values.
This was no longer a secondary problem. Marx claimed his theory to be superior to the bourgeois alternatives, partly because it did something they couldn't: it objectively derived the rate of profit from the material conditions of the labour process. 13 Bortkiewicz turned this asset into a liability: he showed that Marx's original framework was logically inconsistent and that it could be fixed only by making the rate of profit independent of the value system.
The complicating detour
Adding insult to injury, in 1957 Paul Samuelson demonstrated that, mathe- matically, the first transformation was a pointless 'complicating detour'. Whether consistent or not, the theory was largely irrelevant. Marx stipulated a two-stage analytical process, moving from the conditions of production to labour values, and from labour values to prices of production. In fact, argued Samuelson, the process requires only one step - from the conditions of production to prices of production - without any intermediate resort to labour values. 'Marxolaters, to use Shaw's term', he suggested triumphantly, 'should heed the basic economic precept, valid in all societies, cut your losses' and dump the labour theory of value (1957: 892).
Many Marxists, although silent on Samuelson's mathematics, rejected his conclusion as typical bourgeois misunderstanding. The crux of value analysis, they argued, was never to explain prices per se, but to emphasize the under- lying social essence of their formation - the class conflict in production, the social imperative of technical change and the workings of the capitalist mode of production as whole. According to Sweezy (1942: 129), value analysis, despite its many problems, is necessary in order to 'look beneath the surface phenomena of money and commodities to the underlying relations between people and classes'. 14
13 Prices of production, writes Marx, 'are conditioned on the existence of an average rate of profit' which itself 'must be deduced out of the values of commodities. . . . Without such a deduction, an average rate of profit (and consequently a price of production of commodi- ties), remains a vague and senseless conception' (1909, Vol. 3: 185-86, emphasis added).
14 For reiterations of this view, see also Mattick (1972), Baumol (1974), Fine and Harris (1979: Ch. 2) and Harvey (1982: 35-38).
? The Marxist entanglement I 101
What remains unclear, though, is why only the analysis of labour time can do the job. Why should we assume that the logic of class conflict, technical change and the interaction between production, circulation and distribution must all be anchored in the quantitative combination of labour inputs - partic- ularly when this combination is entangled in logical contradictions and sheer impossibilities? Surely, there could be other explanations - or are labour values necessary here, by definition?
Joint production
The picture was clouded further in the 1970s. Building on Sraffa's Production of Commodities by Means of Commodities (1960), Ian Steedman (1975; 1977) and others claimed that the labour theory of value is not only inconsistent and redundant, but also impossible. The basic problem comes from 'joint' production processes, where multiple inputs jointly produce multiple outputs. Such jointness makes it difficult to identify which inputs are 'responsible' for which output, leading to strange mathematical results, with labour values becoming indeterminate, nil, or even negative!
This predicament, to which we return in Chapters 8 and 12, arises because, quantitatively, production is forever a black box. Even if we could identify its inputs and outputs (itself a questionable proposition), we could never examine the inner process by which the former quantitatively 'generate' the latter. The only way to attribute outputs to inputs is indirectly - first by finding a unique mathematical relationship between them, and then by assuming that this relationship represents the productive transformation occurring inside the black box. This superficial solution works well when there is only one output and therefore a single mathematical equation. But it tends to fail with joint production. The latter requires simultaneous equa- tions, and there is nothing inherent in joint production to guarantee that these equations could be solved with unique and positive labour values.
The existence of joint production is potentially devastating for the labour theory of value. Since everything in that theory depends on labour values being positive, the possibility that they could be undefined, nil, or negative undermines Marx's 'laws of motion' and, indeed, the very meaning of exploi- tation. These implications are all the more serious since joint production is the rule rather than the exception. In fact, even when there is only one sellable output, the process is still joint insofar as it also generates a (somewhat) depreciated constant capital.
Some Marxists have tried to minimize the significance of jointness by arguing that modified (i. e. depreciated) constant capital is not produced for the market and therefore should not be counted as a commodity (for instance, Fine and Harris 1979: 41-42). 15 The basis of this claim, though, is difficult to
15 The question of whether or not commodities are 'purposely' produced for the market and the relevance of such purposiveness for political economy are discussed in footnote 1 above.
? 102 The enigma of capital
comprehend. Entire firms are constantly bought and sold on the market for real dollars; indeed, some firms, particularly in research and development, are formed with the sole purpose of being sold to the highest bidder. And with that being the case, why should we consider second-hand machinery and structures as non-commodities? Alternatively, if we are to accept Fine and Harris' position, should we not, just to be consistent, classify all intermediate products as non-commodities?
Another escape route is to argue that capitalist specialization tends to reduce jointness, so that eventually, if not immediately, theory will prevail over reality (ibid. ). Specialization, however, is not the same thing as jointness. The former is about division, the latter about interaction, and more of one does not imply less of the other. To illustrate this point, consider the fact that General Electric has hundreds or thousands of different job descriptions, that its different corporate segments are formally associated with different types of output, and that these outputs are not the same as those of Ford and Exxon. These divisions attest the heightened specialization of contemporary capitalist production, but what do they reveal about the interaction of its different processes?
Do these divisions tell us what portion of the labour of a General Electric accountant, engineer, or driver goes into a GE pump? And when General Electric sells this pump to Ford, can we know - based on these divisions and without resorting to prices - how much of that labour is being 'transferred' to a typical Ford car? How will our answer differ if the plant using the pump jointly produces two types of automobiles rather than one? And what if the entire process were to be 'internalized' by Ford taking over GE's pump division? Would such a takeover make the process less or more joint? Labour- value theorists cannot answer these questions; and as long as the issue of jointness hangs, the input-output structure stays unspecified and the theory remains stuck.
Capitalism sans values?
According to Cornelius Castoriadis (1988, Vol. 1: 20), Marx's acceptance of wage inequality during a socialist transition to communism - 'to each according to his work' - shows how much he had bought into the 'self-evident facts' of bourgeois common sense; that is, into the presumption that we can somehow impute a product to its 'producer'. As a 1958 AFL-CIO document on Automation and Technological Change puts it, this assumption can no longer be made, even in a socialist world: 'Automation in its largest sense means, in effect, the end of measurement of work. . . . With automation, you can't measure output of a single man', and there is no longer 'any reason at all to pay a man by the piece or pay him by the hour' (cited in Tsuru 1993: 7, original emphasis). 16
? 16 As we shall see later in the book, according to Veblen the issue is not automation but the
The Marxist entanglement I 103
Paradoxically, Marx was remarkably prophetic in anticipating the demise of his own labour theory of value, and for this very reason. His insight is worth quoting at some length:
As large-scale industry advances, the creation of real wealth depends less on the labour time and quantity of labour expended than on the power of the instrumentalities set in motion during the labour time. . . . Human labour then no longer appears enclosed in the process of production - man rather relates himself to the process of production as supervisor and regulator. . . . He stands outside of the process of production instead of being the principal agent in the process of production. In this transfor- mation, the great pillar of production and wealth is no longer the imme- diate labour performed by man himself, nor his labour time, but the appropriation of his own universal productivity, i. e. his knowledge and his mastery of nature through his societal existence - in one word, the devel- opment of the societal individual. . . . As soon as human labour, in its immediate form, has ceased to be the great source of wealth, labour time will cease, and must of necessity cease to be the measure of wealth, and the exchange value must of necessity cease to be the measure of use value. . . . The mode of production which rests on the exchange value thus collapses.
(Grundrisse der Kritik der politischen Oekonomie: 592f, translated from the German by Marcuse 1964: 35-36, emphases added)17
And, from there,
. . . with the abolition of the basis of private property, with the commu- nistic regulation of production. . . the power of the relation of supply and demand is dissolved into nothing, and men get exchange, production, the mode of their mutual relation, under their own control again. . . .
(Marx and Engels 1970: 55)
This intriguing idea is typical of Marx's search for inherent contradictions: the very development of the forces of production set against the relations of production is certain to undermine the quantitative logic of capitalism. In a complex socio-technological setting, he argues, the direct relationship between labour inputs and final prices is bound to break down. When that happens, price setting becomes increasingly arbitrary; capitalists lose their moral conviction and business compass; and with their sense of hegemony thus undermined, their system becomes vulnerable and eventually unsustainable.
hologramic nature of production. The fact that production is social means that we cannot
impute output to 'its' producer even in the simplest of social settings.
17 We much prefer Marcuse's translation to the official one given in Marx (1857: 704-5).
? 104 The enigma of capital
Marx of course was proven wrong in believing that the demise of his own theory would bring capitalism down. Perhaps, contrary to his conviction, labour values are not a prerequisite for a functioning capitalism in the first place. However, his insight into the societal nature of production and into the insurmountable problems it creates for the theory of political economy was prescient.
The transformation so far
As our brief journey so far suggests, labour values cannot easily be trans- formed into prices. To recap, again, the transformation involves two parts: a first, logical step from labour values to prices of production and a second, logical/empirical step from prices of production to market prices. Beginning from the end and assuming that the first step is feasible, we saw that labour values still shed little light on the actual trajectory of capitalism. The reason is that the second step is inherently incomplete: Marxist value theorists readily agree that there is more to market prices than prices of production, yet delib- erately refuse to theorize anything other than those prices. And since the theory itself gives no indication of how important these additional factors may be, its significance cannot be assessed.
Some Marxists have tried to settle the issue with statistical analysis. Their results seem very robust - but only because the dice in their tests are loaded and the tests themselves are circular. These studies tend to measure the value- price correlation not of individual commodities, but of total sectoral outputs. Worse still, given that labour values cannot be observed, the researchers commonly 'derive' them from market prices. The result is to have us go in a circle. We are shown that market prices (presented as values) are tightly corre- lated with market prices, but it isn't clear how this revelation vindicates the labour theory of value.
The problem, however, begins much earlier, with the logical conversion from labour values to prices of production. First, this conversion is logically inconsistent - and can be made consistent only at the cost of casting out many of Marx's key conclusions. Second, even if perfectly consistent, the conversion is mathematically redundant in that it adds little to what we know to start with. And finally, once we enter the realm of joint production, it becomes difficult if not impossible to even specify the labour values to be converted.
New solutions, new interpretations
Value theorists have responded to these challenges with a voluminous litera- ture of new solutions. Most have accepted that Marx's own transformation was problematic and that the subsequent critiques needed to be addressed. But they've also insisted that the difficulties can be overcome - by modifying one or more of Marx's assumptions.
Changing the assumptions
Michio Morishima, for example, examined the so-called Fundamental Marxian Theorem, according to which a positive rate of profit both implies and is implied by a positive rate of exploitation. He argued that the Theorem is valid - but only if we valuate commodities differently than Marx. Whereas for Marx value is based on average labour time, Morishima claimed that the Theorem would hold only if value is counted in minimum labour time (Morishima 1973; Morishima and Catephores 1978). Another approach was offered by David Laibman (1973-74). Laibman correctly observed that the class struggle rages not over the rate of profit, but over the distribution of income; and he therefore concluded that it is the rate of exploitation rather than the rate of profit that tends to equalize across industries.
A very different and highly original solution was developed and empiri- cally examined by Emannuel Farjoun and Moshe? Machover (1983). Conventional Marxist models, they argued, specify their variables as discrete, equilibrium magnitudes. But the competitive capitalist reality is inherently stochastic. Competition, they noted, tends to disperse rates of profit as well as equalize them. Of these two processes, the former is stronger, both within and across industries, and rates of profit, as a result, do not converge to equality but rather spread across a range. This tendency means that we should not expect the correspondence between market prices and labour values to be deterministic but probabilistic. And with determinism gone, the logical consistency of the system obviously is no longer an issue.
Complexity
Over time, the proliferation of different solutions, the restrictive nature of their assumptions and the heightened disagreements among their advocates have served to complicate and fracture Marx's original framework. Most value theorists, though, have remained unmoved by this increasing com- plexity.
According to Marx, the deviation of prices from values is inherent in the price-form itself. 'This is no defect', he insists, 'but, on the contrary, admi- rably adapts the price-form to a mode of production whose inherent laws impose themselves only as the mean of apparently lawless irregularity that compensate one another' (Marx 1909, Vol. 1: 115). From this viewpoint, the transformation problem undermines neither the value principle nor the insight it offers into capitalist relations. 'On the contrary', argues Laibman (2000: 314), the complicated process of converting values into the prices 'is a further development of a core insight: capitalist market relations mystify and obscure - and thereby enable - the class exploitation at the heart of the system' (original emphasis). Indeed, in this sense, says Laibman, the rejection of Marx's pulling and redistributing process merely serves to show that 'exploi- tation, far from taking place in mutually isolated sectors, is systemic and
The Marxist entanglement I 105
106 The enigma of capital
inseparable from the entire web of interconnections in the structure of production and exchange' (ibid. ).
The problem with this approach is twofold. First, it is hard to evaluate a theory whose purpose is both to explain and obscure reality (since a weaker explication implies greater mystification - and vice versa). Market prices may indeed seem chaotic and bourgeois theory certainly serves to mystify them. But it is doubtful that Marx wanted his own value theory to serve the same end. Second, all things being equal, tells us Occam's razor, the best solution is the simplest. Complicated explanations tend be intellectually unattractive and politically ineffective - a suicidal combination for a theory whose ulti- mate purpose is praxis.
Changing the definitions
And so there has emerged since the 1980s a different approach altogether, one that offers not new solutions, but new interpretations. 18 According to the new interpretationists, the theory that Marxists have laboured to fix for over a century is indeed logically inconsistent and plagued by insoluble problems - but that theory isn't Marx's. Interpreted correctly, they argue, Marx's texts yield a very different labour theory of value - one that is both internally consistent and fully congruent with all of his theoretical results. 19
The new-interpretation literature has different variants, all sharing one thing in common: they redefine the meaning of labour value. In what follows, we focus on the version that departs most radically from received Marxism - the Temporal Single-System Interpretation, or TSSI. 20 According to the TSSI, Marx's 'true' value theory differs from the conventional interpretation on two main counts, which we consider in turn.
Recounting costs
First, the system is not simultaneous but temporal. To illustrate the differ- ence, consider the production of a personal computer by Dell. The company purchases inputs, such as motherboards from Intel, hard drives from Toshiba, sound cards from Creative Labs, etc. ; it assembles these components with its own workers; and then, three weeks later, when the process is com- plete, it sells the computer to the final consumer. According to the standard,
18 Key collections and reviews of this literature include Freeman and Carchedi (1996), Foley (2000), Freeman, Kliman and Wells (2000) and Kliman (2007).
19 Andrew Kliman (2004: 23, Table 2. 1) compares the different interpretations on how well they replicate Marx's theoretical claims.
20 For a concise exposition of the TSSI, see Kliman and McGlone (1999). Other versions include the particular 'New Interpretation' attributed to Foley (1982) and Dume? nil (1983) and the 'Simultaneous Single System Interpretation' associated with Wolff, Roberts and Callari (1982) and Moseley (1993), among others.
? The Marxist entanglement I 107
simultaneous approach, the value of these inputs should be recorded at current rates. They should be measured based on what they would have taken to produce now, at the end of the process, regardless of what they actually took to produce earlier in the process. This assumption guarantees that commodities that act both as inputs and outputs have a single value.
The TSSI disagrees. This logic makes no business sense, and certainly it isn't Marx's. The dollar costs are recorded when Dell purchases its inputs, not when it sells its output. To be consistent, argue TSSI advocates, the same temporal sequence must also hold for labour values. We need to record them when the inputs enter the process, not when the output leaves.
There is really no way to decide which of these two methods is 'valid'. According to Michael Perelman (1990), Marx himself left the issue open. He used antecedent (past) labour at the micro level of the firm and coexisting (current) labour at the macro level of capitalism as a whole - though without indicating why. Conventional accounting practices make a similar distinction between the LIFO and FIFO methods - and here, too, there is no 'right' or 'wrong'. With LIFO (meaning 'Last In, First Out'), Dell would calculate its costs currently, based on the price of its most recently purchased inputs. With FIFO (or First In, First Out), it would compute the cost historically, based on the price of the most dated inputs in its inventory. The two methods can have dramatically different financial consequences - depending on the rate of infla- tion, tax laws, transfer pricing, currency movements and other assorted considerations, including the duration of the production run and the path of technical change.
But, then, there is no objective yardstick - that is, other than the laws, habits and customs of bourgeois society - to tell us which method to use. Now, unlike the accountants, who deal with observable prices, Marxists do not even know what abstract labour looks like. So it seems that their quest for the Holy Grail here is a bit presumptuous.
Theoretically, however, historical valuation is far more 'flexible' than a current one. Current valuation as used by the conventional approach requires all instances of a given commodity to have one and only one value, so that a barrel of oil has the same value whether it appears as an input or an output. By contrast, historical valuation as used by the TSSI allows each barrel of oil to have its own value - depending on its particular temporal position in the production process. This difference allows the TSSI to appear more theoreti- cally 'robust' than its conventional alternative - but that appearance is misleading. Obviously, if the same commodity can have multiple values, the likelihood of the valuation system as a whole being logically inconsistent is much reduced.
Prices as values
Labelling these considerations 'theoretical', though, is somewhat misleading - for the simple reason that TSSI really is more of an accounting device than a scientific proposition. The definitional nature of the TSSI is evident from its
108 The enigma of capital
second assumption. According to this assumption, Marx had in mind not two systems, but one. Recall that, in the standard dual-system view as formalized by Bortkiewicz, labour values and prices of production are determined sepa- rately, each in their own distinct system, and that the question is whether or not the former can be mathematically 'transformed' into the latter. This ques- tion does not even arise in the TSSI. Here, there is only one system for both prices and values and therefore nothing to transform in the first place.
The language of the last sentence is a bit deceptive. In fact, the single system articulated by the TSSI does not have prices and values. It has prices as values. The distinction is crucial. The conventional Marxist approach argues that labour values are the cause of prices. This causal link is mean- ingful because the definitions of the two magnitudes are different. Prices are counted in money, whereas values are counted in labour time. The two magnitudes could be used interchangeably - but only if the theory is correct.
The setup of the TSSI is completely different. Here, there is no point in asking whether or not prices are equal to values, simply because values are defined by market prices. What the empirical studies do out of necessity and with much unease, the TSSI marshals with fanfare and as a vindication of Marx. The value of constant capital used by Dell is defined here as the market price the company paid for its (depreciated) capital, raw materials and semi- finished goods. Similarly, the value of the labour power it employs is defined as the money wages paid to its workers. Finally the company's surplus value is defined by subtracting from its current dollar revenues the constant and variable capital, as well as an unknown discrepancy that may arise from unequal exchange. 21 Labour is still held responsible, by definition, for the creation of all value in the aggregate. But it is no longer necessary for any of the underlying computations. Individual values are defined as the sum of market prices, with the result that the single-system equations no longer contain embedded labour coefficients.
New interpretationists do not see a problem with this circularity. Duncan Foley, for instance, argues that when Marxists define values by prices they do nothing different than Newton, who defined force by mass and acceleration (F = m * a). In each instance the definition serves a purpose insofar as it 'decisively disciplines and directs scientific investigation in conceptual and fruitful ways' (2000: 28). And, of course, there is nothing wrong with defini- tions per se.
But there is an important difference between the two cases: whereas Newton offers one definition for force, the new interpretationists have two for value: one based on price, the other on labour time. And since value is made proportionate to both price and labour time, it follows that prices are proportionate to labour time and that the labour theory of value is true before
21 This discrepancy, although unknowable for individual commodities, disappears in the aggregate since, by assumption, unequal exchange cannot create value, only redistribute it.
? The Marxist entanglement I 109
we even begin. Kliman and McGlone (1999: 44) claim that this is a 'circular, Hegelian type' of explanation, although it isn't clear how this supposedly dialectical technique differs from the positive neoclassical habit of defining the quantity of capital by its price or the marginal product of labour by the wage rate. Given these considerations, it is perhaps little wonder that, while the definition of force enabled Newton to accurately describe, explain and predict the movement of actual bodies, the concepts of utils and labour values enable their theorists to say pretty much nothing about the actual movement of prices.
Proponents of the TSSI argue that this is what Marx had in mind. And maybe they are right. But if that was Marx's intention, it seems unclear why he would bother to call the result a theory of value, let alone a labour theory of value. The TSSI says nothing about labour, save for asserting that labour creates all values. And it is not a scientific theory in the sense of cause X (value) explaining consequence Y (price). It merely states the truism that current money prices are the sum of past money prices, plus a deviation.
Writers who advocate this approach emphasize that they do not claim Marx's value theory to be empirically valid. Their only purpose is to show that his framework is logically consistent and fully in agreement with his analytical claims. But in the process of achieving this purpose, they seem to have shifted into reverse. Whereas Marx's labour theory of value claims to reason phenomena according to their 'essence', the TSSI advocates move in the very opposite direction. Not only have they turned his theory into an irre- futable tautology, but by defining labour values in price terms, they have made it practically impossible to transcend the very appearance they wish to explain. They have ended up with a dogma.
7 The Marxist entanglement II
Who is productive, who is not?
Their husbands, you know, are the two principal stock holders in the Mills. Like all the rest of humanity, those two women are tied to the machine, but they are so tied that they sit on top of it.
--Jack London, The Iron Heel
So far, our discussion of Marxism has examined how production determines the quantitative architecture of capitalism - while taking production itself for granted. But, then, there is nothing very simple about production, certainly not in the way that Marx analyses it. Recall that his value equations concern productive labour only. They therefore presuppose that we can objectively distinguish labour that is productive from labour that is not. Yet, as we shall now see, differentiating the two types of labour is no simpler than trans- forming values into prices. And even if a line could somehow be drawn between them, its meaning and significance would be anything but clear.
Productive and unproductive labour
The question of 'productivity' dates back to the eighteenth-century conflict between the nobility and the bourgeoisie. The issue was who contributes more to society. Franc? ois Quesnay's Tableau e? conomique argued that it was the tillers of the land. The peasants were the sole producers, while the bourgeoisie merely circulated their product. The opposite position was marshalled by Adam Smith, who prioritized the industrial entrepreneur as the engine of economic growth and the wealth of nations. Marx leveraged the debate for his own purpose, arguing that only the industrial working class produced value.
In all three cases, the reasoning was based on Descartes' notion that cause cannot act at a distance. Leaves move when touched by wind, and the same must be true for productivity: Quesnay's peasants were closest to the land, Smith's capitalists pulled the levers of technical change and Marx's industrial workers were tied to the machines.
The Marxist entanglement II 111
The problem in Marx is that not all workers are industrial, which means that the labour process itself needs to be bifurcated. Marxists classify produc- tion according to the type of wage labour employed. Many types of labour produce use value and many generate exchange value, but for Marx only labour that produces use value, exchange value and surplus value is produc- tive in the capitalist sense. 'That labourer alone is productive, who produces surplus-value for the capitalist, and thus works for the self-expansion of capital', he writes (Marx 1909, Vol. 1: 158). Only labour that is 'directly consumed in the course of production for the valorization of capital' can generate surplus value and produce capital (Marx 1864: 1038). Other types of labour - such as those employed in domestic services (servants for example), in circulation (to convert one form of value to another) and in state organs (as public officials and service providers) - are all unproductive from the stand- point of accumulation. Although they may be important for the reproduction of capitalism as a whole, as well as for the realization of surplus value, they themselves do not create surplus value; in fact, their wages are paid from - and therefore eat into - that very surplus value.
Clearly, distinguishing between the two types of labour is crucial. Since productive labour boosts accumulation whereas unproductive labour hinders it, knowing which is which must be a first step in any investigation of capi- talism. And yet this first step is not easy to take. 1
For a start, Neo-Ricardian Marxists deny the very productive-unproduc- tive distinction. Orthodox Marxists love to dismiss this denial as heresy, on the ground that Neo-Ricardians reject value analysis altogether and therefore are not 'true' Marxists. But even if we ignore the dissenters and assume for the moment that indeed there are two types of labour, we are still faced with the question of how to tell them apart.
According to Fine and Harris (1979: 56) the answer is simple. Paraphrasing Marx, they state that 'If labour directly produces surplus value it is produc- tive; if not, it is unproductive'. This criterion looks straightforward, but in fact it puts the cart before the horse. If our final purpose in separating produc- tive from unproductive labour is to identify surplus value - how can we begin by assuming we already know what surplus value is? Recall that surplus value, like utility and factor productivity, is a theoretical construct, not a directly observable quantum. So how can we know when it is being produced and when it is not?
Recognizing the problem, Fine and Harris, still in the spirit of Marx, offer an indirect litmus test: 'only labour which is performed under the control of capital (on the basis of the sale of labour-power from workers to capitalists),
1 For various analyses and summaries of the issues involved in this distinction, including Marx's own inconsistencies, see Baran (1957: Chs 2-3), Morris (1958), Gough (1972), Harrison (1973), Hunt (1979), Laibman (1992) and Savran and Tonak (1999). Our discus- sion draws in part on these writings.
? 112 The enigma of capital
and in the sphere of production, is productive' (ibid. ). In other words, to iden- tify productive labour we must back-step and decide (1) what constitutes 'production', and (2) which aspects of production come under the 'direct control' of capital. Unfortunately, answering these two questions is not much easier than observing values.
Production versus circulation
Financial intermediation, advertising and insurance
The standard Marxist view is that production mediates the 'relationship of society to nature' - in contrast to circulation and to the reproduction of the social order, which affect the historically specific 'relations among human beings' (Savran and Tonak 1999: 122). Based on this nature-society distinc- tion, the employees of companies such as Deutche Bank (diversified finance), InterRepublic Group (advertising), ING (insurance), Fannie Mae (mort- gages) and CIGNA (real estate) are for the most part unproductive. Their labour, insofar as it is devoted to circulation, does not mediate the relation- ship between society and nature and therefore cannot, by definition, create surplus value.
But is it really that simple? Begin with the sphere of financial intermedia- tion. Over the past century, credit has become the most important mechanism for directing social reproduction (or 'allocating resources' in neoclassical parlance). Despite this fact, Marxists do not consider financial intermedia- tion as 'mediating society's relationship to nature', and therefore do not see it as productive. To be consistent, though, they should apply the very same criteria to R&D, cost accounting, industrial relations and strategic planning. After all, these activities - just like financial intermediation - help allocate resources in 'productive' corporations such as Toyota, ExxonMobil and Microsoft, and if credit is deemed unproductive these corporate activities should be unproductive as well. It is true, of course, that credit is intimately related to the redistribution of ownership and therefore to power; but could this aspect of credit be at all separated, even conceptually, from its other features that help guide reproduction?
Or take advertising. Undoubtedly, this activity is designed to promote sales. But what about the incessant remodelling of automobiles, clothing, detergents, cosmetics, architecture, news media and what not - remodelling that according to some estimates accounts for over 25 per cent of the cost of production? 2 Given that the main purpose here, much like in advertising, is to enhance circulation, shouldn't we consider the labour put into such
2 The 25 per cent estimate is taken from Fisher, Griliches and Kaysen (1962), the first to study the cost of automobile remodelling. Pashigian, Bowen and Gould (1995) claim that such remodelling has since become even costlier - to the point of forcing US-based firms to cut down the frequency of model changes.
? The Marxist entanglement II 113
remodelling to be unproductive as well? Paradoxically, even a positive answer would not solve the problem here. After all, any new product characteristic can persuade people to buy, so how do we distinguish between the adver- tising-like aspect of remodelling that merely circulates existing values and its productive aspect that by definition creates new values?
Finally, consider insurance, which Marxists commonly but erroneously classify as purely redistributional and therefore unproductive. Every produc- tion system requires some means of reducing uncertainty. 3 Historically, this reduction has been achieved through kinship commitments, community support organizations, state programmes like business and farm subsidies, welfare payments, unemployment insurance and pensions and, of course, capitalist insurance. Although all these methods are based on redistribution, they also serve to provide stability for production.
Similarly with business conglomeration and diversification. This strategy, which was first practised by the large trading companies of the seventeenth century and is presently exercised by most large corporations, tends to reduce uncertainty and therefore serves as a form of insurance. If we consider the labour involved in initiating, planning and managing corporate diversifi- cation as productive, we must treat the labour of private insurance employees in a similar manner. If the former creates value and surplus value, so does the latter.
Disaggregates in the aggregate
Marxists commonly bypass these ambiguities by conceding that circulation activities, like financial intermediation, advertising and insurance, do have an impact on production: they affect the reproduction of the social order as a whole, and therefore the overall magnitude of value and surplus value. But they also insist that such activities, because of their general character, do not bear on the relative magnitudes of specific values and surplus values. It is in this latter sense that circulation is deemed unproductive.
This claim is difficult to sustain for two basic reasons. To start with, finan- cial intermediation, advertising and insurance - like every social activity - are invariably differential: they affect some aspects of production more than others. Changes to old-age security programmes or the shifting stance of monetary policy - just like the purchase of life insurance by a single employee or the extension of credit to a specific corporation - have different impacts on different sectors, firms and workers. The same with marketing: the overall
3 In fact, the reduction of uncertainty has had an important, if understudied, impact on the very direction of social development. The pre-historic Australian population, for instance, became nomadic partly because of the continent's infertile soil and the dearth of domesti- cable plants and animals. But there was another crucial reason: El Nin? o made agricultural production intolerably unpredictable, in contrast to the relative security offered by the diver- sity and flexibility of nomadic hunting and gathering (Diamond 1999: Ch. 15).
? 114 The enigma of capital
sales effort creates what John Kenneth Galbraith (1967) called the 'propa- ganda on behalf of goods in general'. But the consequences of this general propaganda vary along the production chains, with some processes being affected differently and more extensively than others. Thus, if we are ready to accept that a certain social activity affects the production of value and surplus value in general, we also have to accept that this impact is distributed unevenly and therefore affects the relative structure of values and surplus values.
Unfortunately, though - and here we get to the second point - value and surplus value cannot be observed, detected or examined, so there is no way to ascertain the existence of these differential impacts, let alone their size. We cannot know if monetary tightening by the central bank alters the relative value and surplus value of microchips vs crude oil - and, if so, by how much. We don't know whether the effect of tax breaks to advertisers raises, reduces or leaves unchanged the value and surplus value of brand names relative to brandless commodities. We have no clue as to the impact of insurance legisla- tion on the value and surplus value in the production of mobile homes rela- tive to the production of university graduates.
But, then, if we are unable to say whether these so-called circulation activi- ties are productive or not, let alone to what extent, how can we ever separate them from production 'proper'?
Objective exchange values?
Eating the cake and having it too
In the final analysis, the difficulty of identifying the so-called sphere of production is rooted in a basic theoretical schizophrenia: the attempt to deduce objective exchange values from subjective use values - while denying this very deduction in the first place. 4
Marx believed that the boundaries of production were objectively delin- eated. They were given, he said, by 'the definite social form, the social rela- tions of production, within which the labor is realized' (Marx 1863, Part I: 152, emphasis added). In practice, though, the categories that Marxists consider definitive and objective almost always rest on subjective criteria.
A recent taxonomy of productive and unproductive labour by Savran and Tonak (1999) illustrates this uneasy derivation. Following Marx, they empha- size the need to distinguish clearly between two types of circulation: the circulation of use value typical to all advanced socio-economic structures, and the specifically capitalist circulation of exchange value embedded in commodities, money and capital (p. 142). However, when they come to distinguish the circulation of exchange value from its creation, their criteria are based entirely on use value:
4 This schizophrenia, of course, is hardly unique to Marxism. As we will see in Chapter 8, it underlies both the util of the neoclassicists and the abstract labour of Marx.
? The Marxist entanglement II 115
[A] society can only increase its wealth through the purposeful transfor- mation of nature and only that amount that has thus been produced can be distributed among the individual members or social classes of society. No amount of exchanging parts of the social product already produced can increase this product itself. . . . This means that any activity which is not directly necessary for humanity's intercourse with nature in order to transform aspects of it in accordance with human needs cannot be regarded as productive labour in general, nor therefore, as productive labour under capitalism.
(143-44)
To see the difficulty, note that, according to this statement, any under- standing of objective exchange value requires prior agreement on no less than five different questions: (1) the definition of the 'product itself'; (2) the 'amount of wealth' this product represents; (3) whether or not the production of this 'wealth' involves a 'purposeful transformation of nature'; (4) whether or not the transformation is 'directly necessary' for humanity's intercourse with nature; and (5) the 'human needs' the product is made to fulfil. Now, since the answers to these questions are all subjective to a lesser or greater extent, and given that they are nonetheless necessary for separating the production from the circulation of exchange value, it follows that exchange values cannot be derived objectively from production.
Capitalist answers, pre-capitalist questions
The difficulty here is one of historical mismatch. Simply put, we are trying to give capitalist answers to pre-capitalist questions. In the feudal-agricultural order, the separation between productive and unproductive activity seemed fairly clear cut. On the one hand, there were the peasants who tilled the land, along with the artisans who crafted with their tools and serviced with their skills. One could reasonably argue that their 'direct and purposeful inter- course with nature produced wealth in accordance with human needs'. It seemed rather evident - certainly to their coevals - that they were productive.
On the other hand, there were the nobility and the clergy who merely appropriated, along with the merchants, money lenders, usurers and taxmen who mostly 'exchanged and redistributed the already produced social product'. Since neither group participated in production as such, it wasn't too hard to conclude that they were unproductive. Even the Church found it diffi- cult to deny what everyone could see. 5
5 A 1020 poem by Bishop Adalbero of Laon provides a succinct summary of the feudal division of labour: 'The community of the faithful is a single body, but . . . human law distin- guishes two classes. Nobles and serfs, indeed, are not governed by the same ordinance. . . . The former are the warriors and the protectors of churches. They are the defenders of the people. . . . The other class is that of the serfs. This luckless breed possesses nothing except
? 116 The enigma of capital
There was relatively little ambiguity: the division between those who laboured and those who didn't was fairly trenchant; needs and subsistence were practically synonymous and wealth a self-evident residual; product diversity was miniscule and the items on offer fairly well-defined; joint production was limited and technical knowledge diffused. In this pre-capi- talist context, use value seemed pretty straightforward - indeed, almost 'objective'. It was an obvious yardstick for separating the productive from the unproductive.
But these clear ecological, occupational and legal distinctions began to melt with the arrival of capitalism. Nowadays, use value is no longer evident, let alone agreed upon. It cannot be an 'objective' yardstick, or even a rough rule of thumb, for delineating what Marxists call the sphere of capitalist production. The remainder of this section illustrates the impasse.
The product itself and the amount of wealth
Take our first question regarding the notion of the 'product itself'. Even in the simplest cases, products are no longer well-defined. When a New York resident buys a loved one flowers flown in from Mexico, does she purchase just 'flowers' - or specifically 'Mexican flowers'? This distinction is crucial since the latter entails the additional costs of air transport, insurance and hedging against currency changes - expenses without which Mexican flowers would be unavailable in New York. If the 'product itself' is merely flowers, these additional costs are socially unnecessary and therefore count as unpro- ductive circulation expenses.
Capitalists and workers, though, buy and sell commodities in prices rather than values, so it makes sense to specify two distinct systems - one
11 Bortkiewicz's arguments were summarized in Paul Sweezy's Theory of Capitalist Development (1942) and later included in his translated collection, Karl Marx and the Close of his System (1949).
12 Save for special assumptions, this inequality means that in simple reproduction the overall price of the wage goods produced by the system differs from the total value of wages (so workers consume less or more than capitalism makes available to them); that the overall price of investment goods differs from the total value of the used up constant capital (so there is net investment or divestment); and that the overall price of luxuries is different from the total profits of capitalists (so they must save or dissave).
? 100 The enigma of capital
denominated in values, the other in prices of production. And, indeed, with a dual-system specification, Bortkiewicz showed the transformation of values into prices of production to be mathematically consistent.
But there was a hefty price to pay for this consistency: the dual system no longer satisfies Marx's aggregate conservation principle. With separate systems for prices and values, there is no longer a single ? that would satisfy Equation (3) for all commodities. The result is that total prices need not be equal to total labour values and total profit does not have to be the same as total surplus value. And given these inequalities, the rate of profit denomi- nated in price terms is no longer the same as the rate of profit calculated from labour values.
This was no longer a secondary problem. Marx claimed his theory to be superior to the bourgeois alternatives, partly because it did something they couldn't: it objectively derived the rate of profit from the material conditions of the labour process. 13 Bortkiewicz turned this asset into a liability: he showed that Marx's original framework was logically inconsistent and that it could be fixed only by making the rate of profit independent of the value system.
The complicating detour
Adding insult to injury, in 1957 Paul Samuelson demonstrated that, mathe- matically, the first transformation was a pointless 'complicating detour'. Whether consistent or not, the theory was largely irrelevant. Marx stipulated a two-stage analytical process, moving from the conditions of production to labour values, and from labour values to prices of production. In fact, argued Samuelson, the process requires only one step - from the conditions of production to prices of production - without any intermediate resort to labour values. 'Marxolaters, to use Shaw's term', he suggested triumphantly, 'should heed the basic economic precept, valid in all societies, cut your losses' and dump the labour theory of value (1957: 892).
Many Marxists, although silent on Samuelson's mathematics, rejected his conclusion as typical bourgeois misunderstanding. The crux of value analysis, they argued, was never to explain prices per se, but to emphasize the under- lying social essence of their formation - the class conflict in production, the social imperative of technical change and the workings of the capitalist mode of production as whole. According to Sweezy (1942: 129), value analysis, despite its many problems, is necessary in order to 'look beneath the surface phenomena of money and commodities to the underlying relations between people and classes'. 14
13 Prices of production, writes Marx, 'are conditioned on the existence of an average rate of profit' which itself 'must be deduced out of the values of commodities. . . . Without such a deduction, an average rate of profit (and consequently a price of production of commodi- ties), remains a vague and senseless conception' (1909, Vol. 3: 185-86, emphasis added).
14 For reiterations of this view, see also Mattick (1972), Baumol (1974), Fine and Harris (1979: Ch. 2) and Harvey (1982: 35-38).
? The Marxist entanglement I 101
What remains unclear, though, is why only the analysis of labour time can do the job. Why should we assume that the logic of class conflict, technical change and the interaction between production, circulation and distribution must all be anchored in the quantitative combination of labour inputs - partic- ularly when this combination is entangled in logical contradictions and sheer impossibilities? Surely, there could be other explanations - or are labour values necessary here, by definition?
Joint production
The picture was clouded further in the 1970s. Building on Sraffa's Production of Commodities by Means of Commodities (1960), Ian Steedman (1975; 1977) and others claimed that the labour theory of value is not only inconsistent and redundant, but also impossible. The basic problem comes from 'joint' production processes, where multiple inputs jointly produce multiple outputs. Such jointness makes it difficult to identify which inputs are 'responsible' for which output, leading to strange mathematical results, with labour values becoming indeterminate, nil, or even negative!
This predicament, to which we return in Chapters 8 and 12, arises because, quantitatively, production is forever a black box. Even if we could identify its inputs and outputs (itself a questionable proposition), we could never examine the inner process by which the former quantitatively 'generate' the latter. The only way to attribute outputs to inputs is indirectly - first by finding a unique mathematical relationship between them, and then by assuming that this relationship represents the productive transformation occurring inside the black box. This superficial solution works well when there is only one output and therefore a single mathematical equation. But it tends to fail with joint production. The latter requires simultaneous equa- tions, and there is nothing inherent in joint production to guarantee that these equations could be solved with unique and positive labour values.
The existence of joint production is potentially devastating for the labour theory of value. Since everything in that theory depends on labour values being positive, the possibility that they could be undefined, nil, or negative undermines Marx's 'laws of motion' and, indeed, the very meaning of exploi- tation. These implications are all the more serious since joint production is the rule rather than the exception. In fact, even when there is only one sellable output, the process is still joint insofar as it also generates a (somewhat) depreciated constant capital.
Some Marxists have tried to minimize the significance of jointness by arguing that modified (i. e. depreciated) constant capital is not produced for the market and therefore should not be counted as a commodity (for instance, Fine and Harris 1979: 41-42). 15 The basis of this claim, though, is difficult to
15 The question of whether or not commodities are 'purposely' produced for the market and the relevance of such purposiveness for political economy are discussed in footnote 1 above.
? 102 The enigma of capital
comprehend. Entire firms are constantly bought and sold on the market for real dollars; indeed, some firms, particularly in research and development, are formed with the sole purpose of being sold to the highest bidder. And with that being the case, why should we consider second-hand machinery and structures as non-commodities? Alternatively, if we are to accept Fine and Harris' position, should we not, just to be consistent, classify all intermediate products as non-commodities?
Another escape route is to argue that capitalist specialization tends to reduce jointness, so that eventually, if not immediately, theory will prevail over reality (ibid. ). Specialization, however, is not the same thing as jointness. The former is about division, the latter about interaction, and more of one does not imply less of the other. To illustrate this point, consider the fact that General Electric has hundreds or thousands of different job descriptions, that its different corporate segments are formally associated with different types of output, and that these outputs are not the same as those of Ford and Exxon. These divisions attest the heightened specialization of contemporary capitalist production, but what do they reveal about the interaction of its different processes?
Do these divisions tell us what portion of the labour of a General Electric accountant, engineer, or driver goes into a GE pump? And when General Electric sells this pump to Ford, can we know - based on these divisions and without resorting to prices - how much of that labour is being 'transferred' to a typical Ford car? How will our answer differ if the plant using the pump jointly produces two types of automobiles rather than one? And what if the entire process were to be 'internalized' by Ford taking over GE's pump division? Would such a takeover make the process less or more joint? Labour- value theorists cannot answer these questions; and as long as the issue of jointness hangs, the input-output structure stays unspecified and the theory remains stuck.
Capitalism sans values?
According to Cornelius Castoriadis (1988, Vol. 1: 20), Marx's acceptance of wage inequality during a socialist transition to communism - 'to each according to his work' - shows how much he had bought into the 'self-evident facts' of bourgeois common sense; that is, into the presumption that we can somehow impute a product to its 'producer'. As a 1958 AFL-CIO document on Automation and Technological Change puts it, this assumption can no longer be made, even in a socialist world: 'Automation in its largest sense means, in effect, the end of measurement of work. . . . With automation, you can't measure output of a single man', and there is no longer 'any reason at all to pay a man by the piece or pay him by the hour' (cited in Tsuru 1993: 7, original emphasis). 16
? 16 As we shall see later in the book, according to Veblen the issue is not automation but the
The Marxist entanglement I 103
Paradoxically, Marx was remarkably prophetic in anticipating the demise of his own labour theory of value, and for this very reason. His insight is worth quoting at some length:
As large-scale industry advances, the creation of real wealth depends less on the labour time and quantity of labour expended than on the power of the instrumentalities set in motion during the labour time. . . . Human labour then no longer appears enclosed in the process of production - man rather relates himself to the process of production as supervisor and regulator. . . . He stands outside of the process of production instead of being the principal agent in the process of production. In this transfor- mation, the great pillar of production and wealth is no longer the imme- diate labour performed by man himself, nor his labour time, but the appropriation of his own universal productivity, i. e. his knowledge and his mastery of nature through his societal existence - in one word, the devel- opment of the societal individual. . . . As soon as human labour, in its immediate form, has ceased to be the great source of wealth, labour time will cease, and must of necessity cease to be the measure of wealth, and the exchange value must of necessity cease to be the measure of use value. . . . The mode of production which rests on the exchange value thus collapses.
(Grundrisse der Kritik der politischen Oekonomie: 592f, translated from the German by Marcuse 1964: 35-36, emphases added)17
And, from there,
. . . with the abolition of the basis of private property, with the commu- nistic regulation of production. . . the power of the relation of supply and demand is dissolved into nothing, and men get exchange, production, the mode of their mutual relation, under their own control again. . . .
(Marx and Engels 1970: 55)
This intriguing idea is typical of Marx's search for inherent contradictions: the very development of the forces of production set against the relations of production is certain to undermine the quantitative logic of capitalism. In a complex socio-technological setting, he argues, the direct relationship between labour inputs and final prices is bound to break down. When that happens, price setting becomes increasingly arbitrary; capitalists lose their moral conviction and business compass; and with their sense of hegemony thus undermined, their system becomes vulnerable and eventually unsustainable.
hologramic nature of production. The fact that production is social means that we cannot
impute output to 'its' producer even in the simplest of social settings.
17 We much prefer Marcuse's translation to the official one given in Marx (1857: 704-5).
? 104 The enigma of capital
Marx of course was proven wrong in believing that the demise of his own theory would bring capitalism down. Perhaps, contrary to his conviction, labour values are not a prerequisite for a functioning capitalism in the first place. However, his insight into the societal nature of production and into the insurmountable problems it creates for the theory of political economy was prescient.
The transformation so far
As our brief journey so far suggests, labour values cannot easily be trans- formed into prices. To recap, again, the transformation involves two parts: a first, logical step from labour values to prices of production and a second, logical/empirical step from prices of production to market prices. Beginning from the end and assuming that the first step is feasible, we saw that labour values still shed little light on the actual trajectory of capitalism. The reason is that the second step is inherently incomplete: Marxist value theorists readily agree that there is more to market prices than prices of production, yet delib- erately refuse to theorize anything other than those prices. And since the theory itself gives no indication of how important these additional factors may be, its significance cannot be assessed.
Some Marxists have tried to settle the issue with statistical analysis. Their results seem very robust - but only because the dice in their tests are loaded and the tests themselves are circular. These studies tend to measure the value- price correlation not of individual commodities, but of total sectoral outputs. Worse still, given that labour values cannot be observed, the researchers commonly 'derive' them from market prices. The result is to have us go in a circle. We are shown that market prices (presented as values) are tightly corre- lated with market prices, but it isn't clear how this revelation vindicates the labour theory of value.
The problem, however, begins much earlier, with the logical conversion from labour values to prices of production. First, this conversion is logically inconsistent - and can be made consistent only at the cost of casting out many of Marx's key conclusions. Second, even if perfectly consistent, the conversion is mathematically redundant in that it adds little to what we know to start with. And finally, once we enter the realm of joint production, it becomes difficult if not impossible to even specify the labour values to be converted.
New solutions, new interpretations
Value theorists have responded to these challenges with a voluminous litera- ture of new solutions. Most have accepted that Marx's own transformation was problematic and that the subsequent critiques needed to be addressed. But they've also insisted that the difficulties can be overcome - by modifying one or more of Marx's assumptions.
Changing the assumptions
Michio Morishima, for example, examined the so-called Fundamental Marxian Theorem, according to which a positive rate of profit both implies and is implied by a positive rate of exploitation. He argued that the Theorem is valid - but only if we valuate commodities differently than Marx. Whereas for Marx value is based on average labour time, Morishima claimed that the Theorem would hold only if value is counted in minimum labour time (Morishima 1973; Morishima and Catephores 1978). Another approach was offered by David Laibman (1973-74). Laibman correctly observed that the class struggle rages not over the rate of profit, but over the distribution of income; and he therefore concluded that it is the rate of exploitation rather than the rate of profit that tends to equalize across industries.
A very different and highly original solution was developed and empiri- cally examined by Emannuel Farjoun and Moshe? Machover (1983). Conventional Marxist models, they argued, specify their variables as discrete, equilibrium magnitudes. But the competitive capitalist reality is inherently stochastic. Competition, they noted, tends to disperse rates of profit as well as equalize them. Of these two processes, the former is stronger, both within and across industries, and rates of profit, as a result, do not converge to equality but rather spread across a range. This tendency means that we should not expect the correspondence between market prices and labour values to be deterministic but probabilistic. And with determinism gone, the logical consistency of the system obviously is no longer an issue.
Complexity
Over time, the proliferation of different solutions, the restrictive nature of their assumptions and the heightened disagreements among their advocates have served to complicate and fracture Marx's original framework. Most value theorists, though, have remained unmoved by this increasing com- plexity.
According to Marx, the deviation of prices from values is inherent in the price-form itself. 'This is no defect', he insists, 'but, on the contrary, admi- rably adapts the price-form to a mode of production whose inherent laws impose themselves only as the mean of apparently lawless irregularity that compensate one another' (Marx 1909, Vol. 1: 115). From this viewpoint, the transformation problem undermines neither the value principle nor the insight it offers into capitalist relations. 'On the contrary', argues Laibman (2000: 314), the complicated process of converting values into the prices 'is a further development of a core insight: capitalist market relations mystify and obscure - and thereby enable - the class exploitation at the heart of the system' (original emphasis). Indeed, in this sense, says Laibman, the rejection of Marx's pulling and redistributing process merely serves to show that 'exploi- tation, far from taking place in mutually isolated sectors, is systemic and
The Marxist entanglement I 105
106 The enigma of capital
inseparable from the entire web of interconnections in the structure of production and exchange' (ibid. ).
The problem with this approach is twofold. First, it is hard to evaluate a theory whose purpose is both to explain and obscure reality (since a weaker explication implies greater mystification - and vice versa). Market prices may indeed seem chaotic and bourgeois theory certainly serves to mystify them. But it is doubtful that Marx wanted his own value theory to serve the same end. Second, all things being equal, tells us Occam's razor, the best solution is the simplest. Complicated explanations tend be intellectually unattractive and politically ineffective - a suicidal combination for a theory whose ulti- mate purpose is praxis.
Changing the definitions
And so there has emerged since the 1980s a different approach altogether, one that offers not new solutions, but new interpretations. 18 According to the new interpretationists, the theory that Marxists have laboured to fix for over a century is indeed logically inconsistent and plagued by insoluble problems - but that theory isn't Marx's. Interpreted correctly, they argue, Marx's texts yield a very different labour theory of value - one that is both internally consistent and fully congruent with all of his theoretical results. 19
The new-interpretation literature has different variants, all sharing one thing in common: they redefine the meaning of labour value. In what follows, we focus on the version that departs most radically from received Marxism - the Temporal Single-System Interpretation, or TSSI. 20 According to the TSSI, Marx's 'true' value theory differs from the conventional interpretation on two main counts, which we consider in turn.
Recounting costs
First, the system is not simultaneous but temporal. To illustrate the differ- ence, consider the production of a personal computer by Dell. The company purchases inputs, such as motherboards from Intel, hard drives from Toshiba, sound cards from Creative Labs, etc. ; it assembles these components with its own workers; and then, three weeks later, when the process is com- plete, it sells the computer to the final consumer. According to the standard,
18 Key collections and reviews of this literature include Freeman and Carchedi (1996), Foley (2000), Freeman, Kliman and Wells (2000) and Kliman (2007).
19 Andrew Kliman (2004: 23, Table 2. 1) compares the different interpretations on how well they replicate Marx's theoretical claims.
20 For a concise exposition of the TSSI, see Kliman and McGlone (1999). Other versions include the particular 'New Interpretation' attributed to Foley (1982) and Dume? nil (1983) and the 'Simultaneous Single System Interpretation' associated with Wolff, Roberts and Callari (1982) and Moseley (1993), among others.
? The Marxist entanglement I 107
simultaneous approach, the value of these inputs should be recorded at current rates. They should be measured based on what they would have taken to produce now, at the end of the process, regardless of what they actually took to produce earlier in the process. This assumption guarantees that commodities that act both as inputs and outputs have a single value.
The TSSI disagrees. This logic makes no business sense, and certainly it isn't Marx's. The dollar costs are recorded when Dell purchases its inputs, not when it sells its output. To be consistent, argue TSSI advocates, the same temporal sequence must also hold for labour values. We need to record them when the inputs enter the process, not when the output leaves.
There is really no way to decide which of these two methods is 'valid'. According to Michael Perelman (1990), Marx himself left the issue open. He used antecedent (past) labour at the micro level of the firm and coexisting (current) labour at the macro level of capitalism as a whole - though without indicating why. Conventional accounting practices make a similar distinction between the LIFO and FIFO methods - and here, too, there is no 'right' or 'wrong'. With LIFO (meaning 'Last In, First Out'), Dell would calculate its costs currently, based on the price of its most recently purchased inputs. With FIFO (or First In, First Out), it would compute the cost historically, based on the price of the most dated inputs in its inventory. The two methods can have dramatically different financial consequences - depending on the rate of infla- tion, tax laws, transfer pricing, currency movements and other assorted considerations, including the duration of the production run and the path of technical change.
But, then, there is no objective yardstick - that is, other than the laws, habits and customs of bourgeois society - to tell us which method to use. Now, unlike the accountants, who deal with observable prices, Marxists do not even know what abstract labour looks like. So it seems that their quest for the Holy Grail here is a bit presumptuous.
Theoretically, however, historical valuation is far more 'flexible' than a current one. Current valuation as used by the conventional approach requires all instances of a given commodity to have one and only one value, so that a barrel of oil has the same value whether it appears as an input or an output. By contrast, historical valuation as used by the TSSI allows each barrel of oil to have its own value - depending on its particular temporal position in the production process. This difference allows the TSSI to appear more theoreti- cally 'robust' than its conventional alternative - but that appearance is misleading. Obviously, if the same commodity can have multiple values, the likelihood of the valuation system as a whole being logically inconsistent is much reduced.
Prices as values
Labelling these considerations 'theoretical', though, is somewhat misleading - for the simple reason that TSSI really is more of an accounting device than a scientific proposition. The definitional nature of the TSSI is evident from its
108 The enigma of capital
second assumption. According to this assumption, Marx had in mind not two systems, but one. Recall that, in the standard dual-system view as formalized by Bortkiewicz, labour values and prices of production are determined sepa- rately, each in their own distinct system, and that the question is whether or not the former can be mathematically 'transformed' into the latter. This ques- tion does not even arise in the TSSI. Here, there is only one system for both prices and values and therefore nothing to transform in the first place.
The language of the last sentence is a bit deceptive. In fact, the single system articulated by the TSSI does not have prices and values. It has prices as values. The distinction is crucial. The conventional Marxist approach argues that labour values are the cause of prices. This causal link is mean- ingful because the definitions of the two magnitudes are different. Prices are counted in money, whereas values are counted in labour time. The two magnitudes could be used interchangeably - but only if the theory is correct.
The setup of the TSSI is completely different. Here, there is no point in asking whether or not prices are equal to values, simply because values are defined by market prices. What the empirical studies do out of necessity and with much unease, the TSSI marshals with fanfare and as a vindication of Marx. The value of constant capital used by Dell is defined here as the market price the company paid for its (depreciated) capital, raw materials and semi- finished goods. Similarly, the value of the labour power it employs is defined as the money wages paid to its workers. Finally the company's surplus value is defined by subtracting from its current dollar revenues the constant and variable capital, as well as an unknown discrepancy that may arise from unequal exchange. 21 Labour is still held responsible, by definition, for the creation of all value in the aggregate. But it is no longer necessary for any of the underlying computations. Individual values are defined as the sum of market prices, with the result that the single-system equations no longer contain embedded labour coefficients.
New interpretationists do not see a problem with this circularity. Duncan Foley, for instance, argues that when Marxists define values by prices they do nothing different than Newton, who defined force by mass and acceleration (F = m * a). In each instance the definition serves a purpose insofar as it 'decisively disciplines and directs scientific investigation in conceptual and fruitful ways' (2000: 28). And, of course, there is nothing wrong with defini- tions per se.
But there is an important difference between the two cases: whereas Newton offers one definition for force, the new interpretationists have two for value: one based on price, the other on labour time. And since value is made proportionate to both price and labour time, it follows that prices are proportionate to labour time and that the labour theory of value is true before
21 This discrepancy, although unknowable for individual commodities, disappears in the aggregate since, by assumption, unequal exchange cannot create value, only redistribute it.
? The Marxist entanglement I 109
we even begin. Kliman and McGlone (1999: 44) claim that this is a 'circular, Hegelian type' of explanation, although it isn't clear how this supposedly dialectical technique differs from the positive neoclassical habit of defining the quantity of capital by its price or the marginal product of labour by the wage rate. Given these considerations, it is perhaps little wonder that, while the definition of force enabled Newton to accurately describe, explain and predict the movement of actual bodies, the concepts of utils and labour values enable their theorists to say pretty much nothing about the actual movement of prices.
Proponents of the TSSI argue that this is what Marx had in mind. And maybe they are right. But if that was Marx's intention, it seems unclear why he would bother to call the result a theory of value, let alone a labour theory of value. The TSSI says nothing about labour, save for asserting that labour creates all values. And it is not a scientific theory in the sense of cause X (value) explaining consequence Y (price). It merely states the truism that current money prices are the sum of past money prices, plus a deviation.
Writers who advocate this approach emphasize that they do not claim Marx's value theory to be empirically valid. Their only purpose is to show that his framework is logically consistent and fully in agreement with his analytical claims. But in the process of achieving this purpose, they seem to have shifted into reverse. Whereas Marx's labour theory of value claims to reason phenomena according to their 'essence', the TSSI advocates move in the very opposite direction. Not only have they turned his theory into an irre- futable tautology, but by defining labour values in price terms, they have made it practically impossible to transcend the very appearance they wish to explain. They have ended up with a dogma.
7 The Marxist entanglement II
Who is productive, who is not?
Their husbands, you know, are the two principal stock holders in the Mills. Like all the rest of humanity, those two women are tied to the machine, but they are so tied that they sit on top of it.
--Jack London, The Iron Heel
So far, our discussion of Marxism has examined how production determines the quantitative architecture of capitalism - while taking production itself for granted. But, then, there is nothing very simple about production, certainly not in the way that Marx analyses it. Recall that his value equations concern productive labour only. They therefore presuppose that we can objectively distinguish labour that is productive from labour that is not. Yet, as we shall now see, differentiating the two types of labour is no simpler than trans- forming values into prices. And even if a line could somehow be drawn between them, its meaning and significance would be anything but clear.
Productive and unproductive labour
The question of 'productivity' dates back to the eighteenth-century conflict between the nobility and the bourgeoisie. The issue was who contributes more to society. Franc? ois Quesnay's Tableau e? conomique argued that it was the tillers of the land. The peasants were the sole producers, while the bourgeoisie merely circulated their product. The opposite position was marshalled by Adam Smith, who prioritized the industrial entrepreneur as the engine of economic growth and the wealth of nations. Marx leveraged the debate for his own purpose, arguing that only the industrial working class produced value.
In all three cases, the reasoning was based on Descartes' notion that cause cannot act at a distance. Leaves move when touched by wind, and the same must be true for productivity: Quesnay's peasants were closest to the land, Smith's capitalists pulled the levers of technical change and Marx's industrial workers were tied to the machines.
The Marxist entanglement II 111
The problem in Marx is that not all workers are industrial, which means that the labour process itself needs to be bifurcated. Marxists classify produc- tion according to the type of wage labour employed. Many types of labour produce use value and many generate exchange value, but for Marx only labour that produces use value, exchange value and surplus value is produc- tive in the capitalist sense. 'That labourer alone is productive, who produces surplus-value for the capitalist, and thus works for the self-expansion of capital', he writes (Marx 1909, Vol. 1: 158). Only labour that is 'directly consumed in the course of production for the valorization of capital' can generate surplus value and produce capital (Marx 1864: 1038). Other types of labour - such as those employed in domestic services (servants for example), in circulation (to convert one form of value to another) and in state organs (as public officials and service providers) - are all unproductive from the stand- point of accumulation. Although they may be important for the reproduction of capitalism as a whole, as well as for the realization of surplus value, they themselves do not create surplus value; in fact, their wages are paid from - and therefore eat into - that very surplus value.
Clearly, distinguishing between the two types of labour is crucial. Since productive labour boosts accumulation whereas unproductive labour hinders it, knowing which is which must be a first step in any investigation of capi- talism. And yet this first step is not easy to take. 1
For a start, Neo-Ricardian Marxists deny the very productive-unproduc- tive distinction. Orthodox Marxists love to dismiss this denial as heresy, on the ground that Neo-Ricardians reject value analysis altogether and therefore are not 'true' Marxists. But even if we ignore the dissenters and assume for the moment that indeed there are two types of labour, we are still faced with the question of how to tell them apart.
According to Fine and Harris (1979: 56) the answer is simple. Paraphrasing Marx, they state that 'If labour directly produces surplus value it is produc- tive; if not, it is unproductive'. This criterion looks straightforward, but in fact it puts the cart before the horse. If our final purpose in separating produc- tive from unproductive labour is to identify surplus value - how can we begin by assuming we already know what surplus value is? Recall that surplus value, like utility and factor productivity, is a theoretical construct, not a directly observable quantum. So how can we know when it is being produced and when it is not?
Recognizing the problem, Fine and Harris, still in the spirit of Marx, offer an indirect litmus test: 'only labour which is performed under the control of capital (on the basis of the sale of labour-power from workers to capitalists),
1 For various analyses and summaries of the issues involved in this distinction, including Marx's own inconsistencies, see Baran (1957: Chs 2-3), Morris (1958), Gough (1972), Harrison (1973), Hunt (1979), Laibman (1992) and Savran and Tonak (1999). Our discus- sion draws in part on these writings.
? 112 The enigma of capital
and in the sphere of production, is productive' (ibid. ). In other words, to iden- tify productive labour we must back-step and decide (1) what constitutes 'production', and (2) which aspects of production come under the 'direct control' of capital. Unfortunately, answering these two questions is not much easier than observing values.
Production versus circulation
Financial intermediation, advertising and insurance
The standard Marxist view is that production mediates the 'relationship of society to nature' - in contrast to circulation and to the reproduction of the social order, which affect the historically specific 'relations among human beings' (Savran and Tonak 1999: 122). Based on this nature-society distinc- tion, the employees of companies such as Deutche Bank (diversified finance), InterRepublic Group (advertising), ING (insurance), Fannie Mae (mort- gages) and CIGNA (real estate) are for the most part unproductive. Their labour, insofar as it is devoted to circulation, does not mediate the relation- ship between society and nature and therefore cannot, by definition, create surplus value.
But is it really that simple? Begin with the sphere of financial intermedia- tion. Over the past century, credit has become the most important mechanism for directing social reproduction (or 'allocating resources' in neoclassical parlance). Despite this fact, Marxists do not consider financial intermedia- tion as 'mediating society's relationship to nature', and therefore do not see it as productive. To be consistent, though, they should apply the very same criteria to R&D, cost accounting, industrial relations and strategic planning. After all, these activities - just like financial intermediation - help allocate resources in 'productive' corporations such as Toyota, ExxonMobil and Microsoft, and if credit is deemed unproductive these corporate activities should be unproductive as well. It is true, of course, that credit is intimately related to the redistribution of ownership and therefore to power; but could this aspect of credit be at all separated, even conceptually, from its other features that help guide reproduction?
Or take advertising. Undoubtedly, this activity is designed to promote sales. But what about the incessant remodelling of automobiles, clothing, detergents, cosmetics, architecture, news media and what not - remodelling that according to some estimates accounts for over 25 per cent of the cost of production? 2 Given that the main purpose here, much like in advertising, is to enhance circulation, shouldn't we consider the labour put into such
2 The 25 per cent estimate is taken from Fisher, Griliches and Kaysen (1962), the first to study the cost of automobile remodelling. Pashigian, Bowen and Gould (1995) claim that such remodelling has since become even costlier - to the point of forcing US-based firms to cut down the frequency of model changes.
? The Marxist entanglement II 113
remodelling to be unproductive as well? Paradoxically, even a positive answer would not solve the problem here. After all, any new product characteristic can persuade people to buy, so how do we distinguish between the adver- tising-like aspect of remodelling that merely circulates existing values and its productive aspect that by definition creates new values?
Finally, consider insurance, which Marxists commonly but erroneously classify as purely redistributional and therefore unproductive. Every produc- tion system requires some means of reducing uncertainty. 3 Historically, this reduction has been achieved through kinship commitments, community support organizations, state programmes like business and farm subsidies, welfare payments, unemployment insurance and pensions and, of course, capitalist insurance. Although all these methods are based on redistribution, they also serve to provide stability for production.
Similarly with business conglomeration and diversification. This strategy, which was first practised by the large trading companies of the seventeenth century and is presently exercised by most large corporations, tends to reduce uncertainty and therefore serves as a form of insurance. If we consider the labour involved in initiating, planning and managing corporate diversifi- cation as productive, we must treat the labour of private insurance employees in a similar manner. If the former creates value and surplus value, so does the latter.
Disaggregates in the aggregate
Marxists commonly bypass these ambiguities by conceding that circulation activities, like financial intermediation, advertising and insurance, do have an impact on production: they affect the reproduction of the social order as a whole, and therefore the overall magnitude of value and surplus value. But they also insist that such activities, because of their general character, do not bear on the relative magnitudes of specific values and surplus values. It is in this latter sense that circulation is deemed unproductive.
This claim is difficult to sustain for two basic reasons. To start with, finan- cial intermediation, advertising and insurance - like every social activity - are invariably differential: they affect some aspects of production more than others. Changes to old-age security programmes or the shifting stance of monetary policy - just like the purchase of life insurance by a single employee or the extension of credit to a specific corporation - have different impacts on different sectors, firms and workers. The same with marketing: the overall
3 In fact, the reduction of uncertainty has had an important, if understudied, impact on the very direction of social development. The pre-historic Australian population, for instance, became nomadic partly because of the continent's infertile soil and the dearth of domesti- cable plants and animals. But there was another crucial reason: El Nin? o made agricultural production intolerably unpredictable, in contrast to the relative security offered by the diver- sity and flexibility of nomadic hunting and gathering (Diamond 1999: Ch. 15).
? 114 The enigma of capital
sales effort creates what John Kenneth Galbraith (1967) called the 'propa- ganda on behalf of goods in general'. But the consequences of this general propaganda vary along the production chains, with some processes being affected differently and more extensively than others. Thus, if we are ready to accept that a certain social activity affects the production of value and surplus value in general, we also have to accept that this impact is distributed unevenly and therefore affects the relative structure of values and surplus values.
Unfortunately, though - and here we get to the second point - value and surplus value cannot be observed, detected or examined, so there is no way to ascertain the existence of these differential impacts, let alone their size. We cannot know if monetary tightening by the central bank alters the relative value and surplus value of microchips vs crude oil - and, if so, by how much. We don't know whether the effect of tax breaks to advertisers raises, reduces or leaves unchanged the value and surplus value of brand names relative to brandless commodities. We have no clue as to the impact of insurance legisla- tion on the value and surplus value in the production of mobile homes rela- tive to the production of university graduates.
But, then, if we are unable to say whether these so-called circulation activi- ties are productive or not, let alone to what extent, how can we ever separate them from production 'proper'?
Objective exchange values?
Eating the cake and having it too
In the final analysis, the difficulty of identifying the so-called sphere of production is rooted in a basic theoretical schizophrenia: the attempt to deduce objective exchange values from subjective use values - while denying this very deduction in the first place. 4
Marx believed that the boundaries of production were objectively delin- eated. They were given, he said, by 'the definite social form, the social rela- tions of production, within which the labor is realized' (Marx 1863, Part I: 152, emphasis added). In practice, though, the categories that Marxists consider definitive and objective almost always rest on subjective criteria.
A recent taxonomy of productive and unproductive labour by Savran and Tonak (1999) illustrates this uneasy derivation. Following Marx, they empha- size the need to distinguish clearly between two types of circulation: the circulation of use value typical to all advanced socio-economic structures, and the specifically capitalist circulation of exchange value embedded in commodities, money and capital (p. 142). However, when they come to distinguish the circulation of exchange value from its creation, their criteria are based entirely on use value:
4 This schizophrenia, of course, is hardly unique to Marxism. As we will see in Chapter 8, it underlies both the util of the neoclassicists and the abstract labour of Marx.
? The Marxist entanglement II 115
[A] society can only increase its wealth through the purposeful transfor- mation of nature and only that amount that has thus been produced can be distributed among the individual members or social classes of society. No amount of exchanging parts of the social product already produced can increase this product itself. . . . This means that any activity which is not directly necessary for humanity's intercourse with nature in order to transform aspects of it in accordance with human needs cannot be regarded as productive labour in general, nor therefore, as productive labour under capitalism.
(143-44)
To see the difficulty, note that, according to this statement, any under- standing of objective exchange value requires prior agreement on no less than five different questions: (1) the definition of the 'product itself'; (2) the 'amount of wealth' this product represents; (3) whether or not the production of this 'wealth' involves a 'purposeful transformation of nature'; (4) whether or not the transformation is 'directly necessary' for humanity's intercourse with nature; and (5) the 'human needs' the product is made to fulfil. Now, since the answers to these questions are all subjective to a lesser or greater extent, and given that they are nonetheless necessary for separating the production from the circulation of exchange value, it follows that exchange values cannot be derived objectively from production.
Capitalist answers, pre-capitalist questions
The difficulty here is one of historical mismatch. Simply put, we are trying to give capitalist answers to pre-capitalist questions. In the feudal-agricultural order, the separation between productive and unproductive activity seemed fairly clear cut. On the one hand, there were the peasants who tilled the land, along with the artisans who crafted with their tools and serviced with their skills. One could reasonably argue that their 'direct and purposeful inter- course with nature produced wealth in accordance with human needs'. It seemed rather evident - certainly to their coevals - that they were productive.
On the other hand, there were the nobility and the clergy who merely appropriated, along with the merchants, money lenders, usurers and taxmen who mostly 'exchanged and redistributed the already produced social product'. Since neither group participated in production as such, it wasn't too hard to conclude that they were unproductive. Even the Church found it diffi- cult to deny what everyone could see. 5
5 A 1020 poem by Bishop Adalbero of Laon provides a succinct summary of the feudal division of labour: 'The community of the faithful is a single body, but . . . human law distin- guishes two classes. Nobles and serfs, indeed, are not governed by the same ordinance. . . . The former are the warriors and the protectors of churches. They are the defenders of the people. . . . The other class is that of the serfs. This luckless breed possesses nothing except
? 116 The enigma of capital
There was relatively little ambiguity: the division between those who laboured and those who didn't was fairly trenchant; needs and subsistence were practically synonymous and wealth a self-evident residual; product diversity was miniscule and the items on offer fairly well-defined; joint production was limited and technical knowledge diffused. In this pre-capi- talist context, use value seemed pretty straightforward - indeed, almost 'objective'. It was an obvious yardstick for separating the productive from the unproductive.
But these clear ecological, occupational and legal distinctions began to melt with the arrival of capitalism. Nowadays, use value is no longer evident, let alone agreed upon. It cannot be an 'objective' yardstick, or even a rough rule of thumb, for delineating what Marxists call the sphere of capitalist production. The remainder of this section illustrates the impasse.
The product itself and the amount of wealth
Take our first question regarding the notion of the 'product itself'. Even in the simplest cases, products are no longer well-defined. When a New York resident buys a loved one flowers flown in from Mexico, does she purchase just 'flowers' - or specifically 'Mexican flowers'? This distinction is crucial since the latter entails the additional costs of air transport, insurance and hedging against currency changes - expenses without which Mexican flowers would be unavailable in New York. If the 'product itself' is merely flowers, these additional costs are socially unnecessary and therefore count as unpro- ductive circulation expenses.
