A value-oriented distinction between productive and unproductive labour.

I. Introduction

The distinction between productive and unproductive labour is one of the cornerstones of the labour theory of value. Savran and Tonak (1999: 115-120) provide a brief enumeration of the relevant theoretical aspects: the analysis of capital accumulation, the determination of economic variables, the rate of profit, state intervention, the growth of the service sector, financial and consumer services, privatisation, etc. Particularly in its fundamental importance to the empirical analysis of capitalist economies, the concept of productive labour is essential for the conversion of economic variables of the conventional national account systems into categories coherent with the labour theory of value. Nevertheless, this importance is paralleled by a continuing controversy over the definition and delimitation of the concept of productive labour in Marxian literature. This can be traced back to the heterogeneous treatment of the subject in Marx's work, principally, due to his main writings on productive labour being inconclusive. As a consequence, any exegetic approach to the reading of Marx must be abandoned. Instead, the search for the greatest explanatory power guides the development of the concept of productive labour in this paper. Controversy over the concept has recently become evident in the debate held both on theoretical and empirical grounds by significant proponents of the labour theory of value. On the one hand, some authors (led by Shaikh) have defended the need to maintain the distinction, and have emphasised its importance in empirical analysis. On the other hand, other authors (led by Laibman) have proposed abandoning the distinction, arguing that it is insignificant for the labour theory of value.

The lack of agreement about the concept of productive labour is really harmful for the labour theory of value. In fact, it hinders any breakthrough in empirical analysis. Does the rate of profit fall? Does it fall as a consequence of a rising proportion of unproductive labour? Is the accumulation of capital affected by an expansion of unproductive activities? What are the consequences of the growth of services, finance, etc. for capital accumulation? What is the role of the state in capital accumulation? These questions have completely different answers depending on the attitude taken towards the concept of productive labour.

Using the first approach, the evolution of the ratio of unproductive to productive labour becomes the key factor in explaining the main economic aspects of the present times. With the second approach, this ratio is irrelevant, and alternative explanations are sought.

In this situation, the labour theory of value does not represent a useful basis for empirical analysis. As a consequence, no valid conclusion for economic policy can be extracted from economic theory (the labour theory of value), and economic agents lose the material basis for their economic decisions: workers can no longer rely on workers' theory. For these reasons, the theoretical analysis of the concept of productive labour is a requirement for the successful development of the labour theory of value and, consequently, for the development of the progressive project.

Marx's most precise definition of productive labour is found in extracts from Chapter VI (unpublished) and Theories of Surplus Value, in which productive labour is defined as labour that produces surplus value. In the former text, Marx establishes that

   since the direct purpose and the actual product of capitalist
   production is surplus value, only such labour is productive
   ... as directly produces surplus value. Hence only such
   labour is productive as is consumed directly in the
   production process for the purpose of valorising capital'.
   (p. 77)

More concretely, in Theories of Surplus Value he writes that

   productive labour, in its meaning for capitalist production,
   is the wage-labour which, exchanged against the variable
   part of the capital (the part of capital that is spent on
   wages) reproduces not only this part of the capital ... but
   in addition produces surplus value for the capitalist. (p.
   152)

Its essential feature lies in its specifically capitalist content, as Marx mentions in a crystal-clear passage:

   Productive labour is only an abbreviated expression for
   the whole relation, and the manner in which labour
   capacity and labour figure in the capitalist production
   process. Hence if we speak of productive labour, we speak
   of socially determined labour. (Chapter VI: 83)

Nowadays, this simple definition is widely accepted among Marxist economists, and it is the basis for the rejection of wrong definitions existing in the literature. (1) However, this consensus soon vanishes when the definition is applied to the classification of different forms of concrete labour. The main point of this paper is to show that the apparent adherence of Marxist economists to Marx's basic definition of productive labour is deceptive. In particular, most interpretations do not entirely consider the social determination of the concept but, on the contrary, are founded on a use-value criterion, which relies heavily on features non-specific to the capitalist mode of production. In contrast, a distinction between productive and unproductive labour is proposed here that is based on specifically capitalist criteria.

Section 2 introduces Shaikh's defence of the distinction based on the concept of productive labour in general, and his support of an 'extensive' classification of unproductive labour. Section 3 deals with Laibman's proposal that the distinction between productive and unproductive labour be abandoned. In both sections, I show that their arguments are not coherent with the social and historical content of the concept of productive labour. In section 4 I offer an alternative approach, in which relevance is placed on the production of value rather than on the production of use value. At the end of this section, two essential aspects of the distinction are addressed: the articulation of production and circulation spheres, and the relation between value and use-value creation. Finally, by way of conclusion, section 5 draws out the implications of this approach for empirical research in capitalist economies.

2. Productive labour in general

Shaikh and Tonak's (1994: 20) analysis of productive labour rests on the existence of a 'prior and more general distinction between production and non-production activities' that sheds light on 'more concrete distinctions between labours which are and are not productive of capital'. They argue, moreover, that the neglect of this general concept has caused the lack of understanding of the concept of productive labour in the literature (Savran & Tonak, 1999: 120). The application of this general concept leads to the distinction of four basic activities of social reproduction: production, distribution, social maintenance, and personal consumption. Personal consumption does not require any expenditure of labour. From the remaining labour activities, only production involves productive labour in general (the term 'productive labour' is used in Savran and Tonak [1999], while Shaikh & Tonak [1994] refer to 'production' and 'non-production' labour. The terms are equivalent). Therefore, 'labour is not synonymous with production' (ibid: 22). (2)

The concept of productive labour in general is only the starting point of their definition of productive labour for capital (Savran & Tonak, 1999: 123). They proceed to incorporate into the analysis the specifics of the capitalist mode of production. (3) In Shaikh and Tonak's words (1994: 29), the concepts of productive and unproductive labour 'take on additional content when ... considered in relation to specific social relations under which they might be conducted ... Labour might be conducted for direct use, for sale for income, and for sale for profit'; but only in the last case 'it represents capitalist commodity production that produces not only use values and values but also surplus value'. Therefore, they arrive at the accepted definition of productive labour: labour that creates surplus value.

To sum up, in addition to being productive in general, productive labour for capital must be 'wage labour which is first exchanged against capital (i.e. it is capitalistically employed)' (ibid: 30). According to this approach, productive labour for capital is a sub-set of productive labour in general, since 'surplus value can only be produced in the immediate process of production' and 'only labour which is productive in general ... can produce surplus value' (Savran & Tonak, 1999: 124).

It should be noted that Shaikh and Tonak's definition of productive labour in general rests on a use-value criterion. For them, 'the process of production involves the creation or transformation of objects of social use by means of purposeful human activity' (Shaikh & Tonak, 1994: 22). Accordingly, 'in the case of production activities, the labour involved is production labour, which utilises certain use values in the creation of new use values' (ibid: 24). Conversely, non-production labour does not create new wealth, as 'certain types of labour share a common property with the activity of consumption--namely, that in their performance they use up a portion of the existing wealth without directly resulting in the creation of new wealth' (ibid: 25). This type of labour is related to distribution and social maintenance activities. Thus, 'although distribution activity does transform the use values it circulates, this transformation relates to their properties as objects of possession and appropriation, not to the properties which define them as objects of social use' (ibid: 26). Likewise, in the activities of social maintenance, 'use values enter as material inputs into activities designed to protect, maintain, administer, and reproduce the social order, and as such they are quite distinct from production labour' (ibid: 27).

The distinction between productive and non-productive labour, therefore, implies the existence of labour that does not produce use values. In other words, Shaikh and Tonak's approach severs the link between the execution of concrete labour and the creation of use values--a link I believe to be generally valid. If the link is retained, all labour is production labour (of use values) and there is no room for the concept of non-production labour. Therefore, my critique of Shaikh and Tonak is centred around a critique of their concept of productive labour in general. For this reason, the definition of use value that they employ is worth some close examination.

Guerrero (1997: 1) rightly points out, in a comment about their definition of productive labour, that Shaikh and Tonak define the concept of use value according to Lancaster's classification of the various characteristics of the objects of social use:

   Shaikh and Tonak (1994) use Lancaster's well-known
   approach of the many 'characteristics' of every commodity
   in order to distinguish between the different properties
   that encompass the material thing or effect produced. So,
   in any commodity it is possible to find some material
   (objective) properties, some social properties and other
   properties that are not relevant to our purposes. Only the
   material or objective properties (such as colour, shape or
   location) enter in the definition of use-value. (Guerrero,
   1997: 1)

But in spite of this narrow definition of use value, I do not think it is possible to consider activities such as distribution and the maintenance of social order as performed by non-production labour, as Guerrero himself argues. 'The crucial point is that if one just looks at the definition of use-value, it is absolutely impossible to leave activities like distribution, sales, advertising, social maintenance, etc., outside the borders of the definition given by Shaikh and Tonak themselves' (Guerrero, 1997: 1).

Instead, I would argue that both distribution and social-maintenance labour create objective or material properties in their products on the same grounds as production labour does. For instance, a cashier in a supermarket creates an object of social use (a service sale) characterised by its objective properties (and in fact, this is the only way to obtain goods from a supermarket without the risk of being stopped by the security service). Precisely, a security guard arresting a shoplifter is creating a use value (an arrest) whose objective properties are undeniable. Therefore, the separation of production and non-production labour by means of a use-value criterion seems unfounded.

The debate about the role played by the concept of use value in the distinction between productive and unproductive labour has recently been enriched by the controversy between Mohun and Laibman. Mohun (1996: 44) holds that the assertion that circulation labour creates use values is only possible if we employ a neoclassical concept of use value: 'If it is argued that all workers employed by capital are productive, irrespective of their location with respect to the circuit of capital, then there must be a slippage in the category of "use-value" to its neo-classical sense as a subjective property, qua psychological characteristic, of the purchaser'.

Laibman (1999: 68), surprisingly, accepts the subjective content of the use-value concept, since he believes that it does not contradict the labour theory of value. He writes, 'Use-values do have a subjective component, if they are appropriated by conscious human beings; the subjective aspect is of course socially and historically conditioned, and shaped by class location and production relations. Does Mohun wish to argue that use-values are physical artefacts, outside of social and cultural determination?' In my view, only objective properties enter into the definition of use value, thus I reject Laibman's position. This is not to say, of course, that I do not recognise that conscious human beings have different subjective appreciation of the objective properties of the use value of commodities. For instance, the objective properties created in the production of a car advertisement are independent of the different subjective responses that the advertisement may cause.

On the other hand, I cannot agree with Mohun's statement that circulation labour only adds subjective properties to the use values. As a matter of fact, the subjective perceptions of conscious human beings cannot arise from nothing, but must arise from the new objective properties created. I agree with Guerrero's (1997: 1) statement that 'we do not need to remove the neo-classical identification between production and labour if we understand production as production of use-values', and that 'maintaining this identification is not the same thing as maintaining the neo-classical approach'.

Consequently, the only way to rescue the distinction between production and non-production labour is by classifying the different objective properties of use values following some previously established evaluative criterion. Hence, the claim that the 'classical distinction between production and non-production labour is essentially analytical' (Shaikh & Tonak, 1994: 25) is unfounded. (4) In conclusion, I believe that it is not possible to identify the production of use value (under capitalist social relations) with the creation of value and surplus value.

That is to say, it is not possible to sever the link between concrete labour and use value. As a consequence, all kinds of labour must be regarded as production labour (of use values), and the concept of non-production labour is simply misconceived. To sum up, Shaikh and Tonak's intention to provide an 'analytical' definition of productive labour is faulty because of their utilisation of a use-value criterion. In this sense, it is very illuminating to analyse the authors' reasons for this use:

   The distinction between productive and unproductive
   labour is necessary, but not sufficient, for the analysis of
   reproduction. We need also to know the specific
   components of unproductive labour and their interaction
   with the circuits of capital and revenue. This is precisely
   why we began our analysis with the general distinction
   between production, distribution, social maintenance, and
   personal consumption activities, rather than merely
   beginning with Marx's definition of productive labour.
  (ibid: 31-32)

Hence, it is the emphasis on social reproduction that is misleading in their definition of productive labour. For the social reproduction of capitalist society, two requisites must be met: 1) the quantity of surplus value created and transformed into capital must be enough to guarantee the extended reproduction of capital; and 2) the composition of the output must have the adequate proportion of means of production, means of consumption, etc. Of these requisites, the first, social one is specific to capitalist society, while the second, technical one is necessary in all modes of production.

However important the technical aspect of the social reproduction may be, it undoubtedly has nothing to do with the definition of productive labour. Productive labour is not defined as labour that meets both the above requisites--the concept of 'reproductive labour' would fit that description. On the contrary, the concept of productive labour is limited to the analysis of value and surplus-value creation; that is, to the first aspect of social reproduction. Despite Marx's heterogeneous treatment of productive labour, he is very clear on this point:

   A large part of the annual product ... consists of extremely
   paltry products (use values), serving to satisfy the most
   miserable appetites, fancies, etc. But content is entirely
   irrelevant to whether the labour is determined to be
   productive or not (although the development of wealth
   would of course be checked if a disproportionate part
   were reproduced in this way, instead of being reconverted
   into means of production and means of subsistence, which
   enter anew into reproduction, whether that of the
   commodities or that of the labour capacities themselves--which
   are, in short, consumed productively)'. (Chapter VI:
   85)

In conclusion, I find no support for the use of the distinction between production and non-production labour for the delimitation of the concept of productive labour for capital. Hence I consider all labour activities to be productive of use values. (5)

3. Laibmans' rejection of the distinction

Among Marxist economists, the loudest voice raised against the distinction between productive and unproductive labour is Laibman's:

   The productive/unproductive labour distinction, in the
   strong uses that most proponents (including, it must be
   said, Marx) claim for it, is unsound and should be
   discarded as a residue of bourgeois classical economics.
   (Laibman, 1992:71)

Laibman identifies seven different definitions of productive labour, each of which, in turn, he refuses--although in fact, he does not reject the socioeconomic definition, which distinguishes between the capitalist and non-capitalist spheres of production. He simply asserts that it 'does nothing to resolve the thorny issues surrounding the presumably central categories of labour working within capitalist enterprises, especially in the areas of circulation and supervision, that are thought to be unproductive' (Laibman, 1999: 63). As stated below, this problem can be solved using two different levels in the distinction between productive and unproductive labour. For the purposes of this paper, I limit the analysis to the analytical definition of productive labour, which applies

   the unproductive label to some of the labour that is
   organised under the control of capitalists for the purpose
   of producing and realising surplus value.... The labour
   associated with the circulation of commodities, as opposed
   to their production, is often singled out as a major
   component of unproductive labour. (Laibman, 1992: 76)

Although Laibman believes this definition to be compatible with the generally accepted definition, he rejects it since 'this definition, as stated, is empty or "non-operational"' (Laibman, 1999: 62). Concretely, he asserts that it is not possible to maintain this distinction without making use of a non-analytical criterion:

   The problem is finding an operational criterion that will
   identify workers who do not create value or surplus value.
   In looking for such a criterion, we have to be careful not to
   fall back into alternative, and unsatisfactory, definitions such
   as the physicalist or the evaluative. (Laibman, 1992: 76)

This last warning seems most appropriate; and in fact, I have criticised modern versions of the distinction on the same grounds. For example, Laibman rightly observes that the classification of disciplinary and coercive workers into the unproductive group implies the abandonment of an analytical criterion. According to him, 'workers who perform disciplinary and coercive functions within the workplace are deemed unproductive, because they exercise functions of capitalist control rather than actually "contributing to production"'. And he points out that the logic behind this argument is that 'given more advanced social relations, the coercive apparatus in capitalist workplaces are unnecessary and wasteful'. Hence the argument falls into 'the presumably surpassed and unacceptable evaluative position' (Laibman, 1999: 63). Laibman, here, is answering Mohun's (1996: 37) position: 'Labour which enforces hierarchy and discipline arises out of the need for capital to retain coercive control over the class antagonisms inherent in the capital relation. The labour which performs this function merely personifies the coercive power of the latter, and in so doing does not create value'. Surprisingly, when Mohun considers these critiques, he asserts that 'the necessity or otherwise of a particular type of labour is a different criterion for distinguishing types of labour from that with respect to their location in the circuit of capital' (ibid: 42). As Laibman (1999: 65) evidences, his reasoning 'does not fit easily into the circuitry framework'.

Laibman carries on with his critique of the analytical definition, tackling the distinction between production and circulation labour. He recognises that this distinction 'is at least superficially more convincing, in that circulation labour can presumably be identified with distinct phases of the capital circuit'. However, he insists that no convincing criterion has been given to support it 'other than the imagery of the circuit with its phases or "metamorphoses"' (ibid: 66). Without any doubt, the hub of the matter lies in what Laibman considers to be a valid analytical criterion: what would be, in his opinion, an operative criterion with which to distinguish between production and circulation labour? His answer cannot come as a surprise: an analytical criterion must identify the moments in which labour activity creates new use values from the moments in which it does not:

   The critical question is whether an aspect of buying, selling,
   insuring, legal, accounting labour can be identified, that is
   not reducible in further analysis to some aspect of the
   transformation and processing of use values: the concrete
   labour activity that provides the bodily form (to use the
   well-known Volume I metaphor) for abstract labour and
   the creation of value.... At what point does the production
   of use value stop? (Laibman, 1992: 77-78)

Laibman, therefore, is of the same opinion as that expressed in the greater part of the literature, including in Shaikh and Tonak: for him, the distinction between production and circulation must be grounded on a use-value criterion. The only difference is that he believes this criterion to be non-operative: either it is not feasible to delimit these moments, or all the labour activity must be considered as production labour. As a consequence, Laibman (1999: 63) thinks that 'the attempt to ground the analytical distinction on something solid had failed, and the analytical use of the productive/unproductive distinction should therefore be abandoned'. Therefore, 'the clear positive implication of this conclusion is that all waged labour employed by capitalists creates value' (ibid: 64). However, as we will see in section 4, the failure to differentiate between production and circulation in employing a use-value criterion does not impede the grounding of this distinction on different, capitalist criteria.

4. A value-oriented definition of productive labour

In this section, I provide a value-theoretic definition of productive labour, in which the specificity of the capitalist mode of production is grasped at all times. This definition takes the form of two consecutive steps, as shown in Table 1. In the first step, I distinguish between capitalist production and non-capitalist production. In the second, I differentiate between production and circulation activities within capitalist production.

[TABLE 1 OMITTED]

In the capitalist mode of production, forms of capitalist and non-capitalist production coexist. The goal of the first step is to separate them through the identification of the distinctive features of capitalist labour. Concretely, capitalist labour must be subsumed under the circuit of capital: M--C ... P ... C'--M'; its characteristics are: 1) that it is wage labour; and 2) that its product is sold in the market. This distinction is made by Marx in a concise and precise manner in Chapter VI. (Note that since Chapter VI is an unpublished appendix to the first volume of Capital, which is devoted to the process of capitalist production, Marx's analysis deals exclusively with the first level of the distinction). He acknowledges that 'the capitalist labour process does not abolish the general determinations of the labour process. It produces products and commodities'. But he also points out that 'the labour process is only a means to capital's valorisation process' (Chapter VI: p.8). In other words, 'the owner of labour capacity ... is a wage labourer', whose labour 'not only preserves in part and reproduces in part the capital values that have been advanced, but at the same time increases them, and therefore converts them into self-valorising value, into capital' (ibid: 79-80). To sum up, 'in capitalist production, the production of products as commodities, on the one hand, and the form of labour as wage labour, on the other, become absolute' (ibid: 81).

Following this definition, non-capitalist labour can be classified according to the requisite(s) it fails to meet. For example, 'every productive worker is a wage labourer; but this does not mean that every wage labourer is a productive worker'. This is the case of the wage labour hired 'in order to be consumed as use value', because it is not employed 'in order to replace the value of the variable capital as a living factor and to be incorporated into the capitalist production process' (ibid: 80). This kind of labour, though waged, is not intended to make profits through the sale of its products. (6) Only wage labour exchanged for capital is capitalist labour, while wage labour exchanged for revenue is non-capitalist. Clear examples of non-market wage labour are domestic labour and public employees. Also, labour can be directed to the market but performed by non-waged labourers. Again, we cannot talk about capitalist labour, far less about productive labour. This situation occurs as a consequence of the coexistence of pre-capitalist forms of production within the capitalist social formation; independent workers belong in this category. As Marx points out, 'Some of the labour which produces commodities in capitalist production is performed in a manner which belongs to earlier modes of production, where the relation of capital and wage labour does not yet exist in practice, and therefore the category of productive and unproductive labour, which corresponds to the capitalist standpoint, is entirely inapplicable' (ibid: 82). Finally, there is labour that is neither waged nor directed to the market, such as household labour, which Gouverneur (1990: 2) calls 'autonomous labour'. (See his Table 1 for a similar classification of labour in the capitalist mode of production.)

The next step in the definition of productive labour consists in differentiating production and circulation labour. (7) In this task, the general formula of capital is again a useful starting point. Along its circuit, M--C ... P ... C'--M', capital performs different functions--as money capital, productive capital and commodity capital--and suffers several metamorphoses, from money form to commodity form and vice versa. Some transformations--those represented by a hyphen in the formula--take place in the circulation sphere, while those represented by dots occur in the production sphere.

It is precisely this distinction between the production and the circulation spheres in which the foundation of the distinction between productive and unproductive labour within capitalist production lies. Specifically, it is grounded on Marx's statement that value (and surplus value) is only created in the production process, since in the circulation process, value solely undergoes changes in its form which do not affect its magnitude:

   There is in an exchange nothing (if we except the replacing
   of one use-value by another) but a metamorphosis, a mere
   change in the form of the commodity. The same exchange
   value, i.e., the same quantity of incorporated social labour,
   remains throughout in the hands of the owner of the
   commodity first in the shape of his own commodity, then
   in the form of the money for which he exchanged it, and
   lastly, in the shape of the commodity he buys with the
   money. This change of form does not imply a change in
   the magnitude of value.... So far therefore as the
   circulation of commodities effects a change in the form
   alone of their values, and is free from disturbing influences,
   it must be exchange of equivalents. (Marx, 1906: 176)

Nevertheless, both production and circulation are necessary instances of the global reproduction of capital. Hence, capital reproduction is constituted by the sum of production and circulation time, though these periods affect the valorisation of capital--the very aim of capitalist production--differently. In Marx's words:

   The time of circulation and the time of production
   mutually exclude one another. During its time of
   circulation, capital does not perform the functions of
   productive capital and therefore produces neither
   commodities nor surplus value. (Marx, 1909: 142)

Subsequently, the activities belonging to the circulation sphere imply the performance of circulation labour, which is unproductive of value and surplus value. Moreover, the labour and the means of production employed in circulation activities 'consume' part of the already-created value. In other words, unproductive circulation activities are financed out of value-creating production activities. For this reason, we must distinguish clearly between capital invested in the production sphere--productive capital--and capital invested in the circulation sphere--unproductive capital. Both capitals claim an aliquot part of total surplus value, but only productive capital creates it.

The unproductive functions of capital related to the circulation sphere and the agents performing them are analysed by Marx in volume III of Capital, in which section 4 deals with commercial capital, section 5 with financial capital, and section 6 with ground rent. The emphasis is placed on the mechanism by means of which unproductive capitals appropriate an aliquot part of total profit: these capitals obtain the general rate of profit through the redistribution of value which occurs in the circulation sphere due to the deviations between materialised labour time in the production of commodities and realised labour time in their sale.

In summary, I have presented a distinction between productive and unproductive labour in two steps. The first step differentiates between capitalist and non-capitalist production, and its content is widely accepted in the literature. Shaikh and Tonak consider it to be a second step in the definition of productive labour when it is applied only to productive labour in general. Laibman identifies it with the socioeconomic definition, which he considers correct in its content, though limited in its application. Consequently, the modern controversy over the definition of productive labour lies in the subtleties of the second step: the distinction between production and circulation activities. The two opposing approaches analysed here base this distinction on a use-value criterion, i.e. on the identification of activities unproductive of use value, yet with different conclusions. In contrast, I have rejected this criterion and have argued instead for a criterion founded on the production of value. In the remaining part of this section, I again focus on this second step, setting the patterns for the identification of production and circulation activities according to my value criterion. In this task, the focus must be directed onto two essential matters: the division between production and circulation, and the relation between value and use-value creation.

Production and circulation

Let us return to Laibman's rejection of the definition of productive labour. Laibman clearly acknowledges that it is not possible to distinguish between labouring activities productive and unproductive of use value, so he concludes that the distinction between productive and unproductive labour is ill-founded. While I agree with his acknowledgement, the consequence he extracts is, in my view, wrong. It is easy to see that the differences lie in the interpretation of the distinction between the production and circulation spheres. Laibman (1992: 77) maintains that it is 'essential to distinguish between production and circulation of commodities.... It is quite another thing, however, to speak of production and circulation as distinct "places" and to attempt to visualise an instant in which a commodity "leaves" the sphere of production and "enters" the sphere of circulation'. In other words, 'production and circulation are not "sectors". They are simultaneous-yet-distinct moments of the same social process' (Laibman, 1999: 68).

Therefore, Laibman defends a unity of the process of capital reproduction, as formed by the production and circulation processes, in which it is not possible to 'physically' distinguish between these two moments; instead, both moments take place simultaneously during capitalist reproduction. In other words, there is no division between the labour deployed in production and circulation activities: the distinction between the production and the circulation spheres is purely formal, rather than material. Hence, capital turnover time is both and simultaneously production and circulation time. This interpretation clearly opposes Marx's statement that 'the time of circulation and the time of production mutually exclude one another' (Marx, 1909: 142). Laibman considers 'the circuitry formulas [to] have been misinterpreted in a mechanistic fashion to imply an artificial separation of circulation and production.... These conceptual devices should not, however, be used to imply that use-values pass "from" a distinct sphere of circulation "to" a sphere of production, and then out again' (ibid.). Actually, he suggests, 'the metamorphoses in the circuit of capital can be treated more metaphorically than literally' (Laibman, 1999: 71).

I agree entirely with Laibman that production and circulation are not 'sectors', as we will see in section 5. However, I cannot agree that the distinction between production and circulation is founded on purely formal criteria with no material content, since I do not share his contention that the labour theory of value is not undermined by the abandonment of the productive-unproductive distinction: 'Neither the proposition that value and surplus value are created in production (not circulation), nor the distinction between production and (re)distribution of value, requires that labour be rigidly separated into production labour and circulation labour' (Laibman, 1999: 71). Laibman intends to endorse a distinction of the production and circulation spheres valid for Marx's Capital Volume I, but not for Volumes II and III. On the contrary, I believe that the theoretical power of the analysis in Volume I is dubious if we do not separate production and circulation labour. Also, the explanatory power of the analyses carried out in sections 4, 5 and 6 of Capital Volume III is neglected.

First, Laibman (1992: 77) acknowledges the usefulness of the separation of production and circulation in Volume I: 'Marx lays the groundwork for the theory of surplus value in volume I of Capital by developing both the distinctness and the close interdependence of production and circulation, and his generating insight is that surplus value arises only in production that is also a moment of circulation, that is, both within and without the process of circulation'.

He also praises the use of the comparison between pure commodity circulation and capital circulation: 'Marx's C--M--C vs. M--C--M' distinction is a brilliant insight into two forms of market relations, and his questions surrounding the origin of surplus value were a path-breaking means of investigating the connection between the surface structure of formal equality in exchange and a deep structure of exploitation and coercion' (1999: 68). However, he judges that the comparison can be maintained even though he rejects the concept of unproductive labour. In my opinion, Marx's aim in Volume I is not to show labour as the only source of value (this constitutes his starting point), but to demonstrate that value is only created in the act of production. Therefore, Marx's effort is senseless if we consider exchange as a purely formal act, with no labour activity attached to it. To sum up, the separation of production labour and circulation labour is essential for the proposition that labour is solely created in production. (8)

Second, Laibman is against the distinction between the production and the circulation sphere used in Volume III, and he rejects the theory of commercial capital. In my opinion, this implies that many phenomena occurring in capitalism remain unexplained. For instance, the generalised difficulties of commodity realisation associated with crises cannot entail an absolute increment in circulation labour (devaluation) and a relative decrement in production time (valorisation); instead, they produce a general increase in the value of commodities.

In contrast, Laibman (1999: 70) asserts that 'other questions in the theory of redistribution of surplus value, such as the interest rate and the return to interest-bearing capital' remain. However, this is only reconcilable with his rejection of the existence of circulation labour if he argues that although interest-bearing capital is not productive of value or surplus value, the labour activities carried out within its circuit (as represented in the simple formula M--M') are productive. In other words, that unproductive capital employs productive labour! Alternatively, one may think that no labour is necessary for interest-bearing capital to obtain its share of total surplus value.

In any case, no convincing reconciliation is provided. Therefore, the existence of circulation labour is a necessary condition for the existence of unproductive capital and unproductive activities. Of course, there is productive labour activity related to interest-bearing capital--I am not defending a 'sectoral' classification of circulation labour, as argued in section 5. But it does not negate the necessary existence of unproductive labour due to the existence of interest-bearing capital.

To summarise, I think a complete distinction between the production and circulation spheres is needed if we are to preserve the essentials of the labour theory of value. This distinction, nonetheless, cannot be founded on a use-value criterion; it must be built on a pure value criterion.

Value and use-value creation

At this point, my distinction between production and circulation labour seems to be indeterminate, as I have not offered an alternative concrete criterion for the practical separation of forms of labour that is different from the use-value criterion. What is the operational criterion for this distinction? Paradoxically, some defenders of the use-value criterion give us clues as to how we might confer content on the value criterion. Mohun (1996: 43) refers to the moments in the process of valorisation of capital:

   The reproduction of capitalist social relations is
   inseparable from the reproduction of circuits of capital,
   and from the way in which values, manifested in particular
   use-values appear in successively different forms. That is,
   concrete labours are understood as the bearers of particular
   moments in the valorisation of value, and hence
   consequences, not causes, of the latter. The issue is not
   what is happening to value as a result of this or that
   concrete labour, but rather what concrete labour is
   required by this or that moment in the valorisation of
   value; not the manner in which some particular concrete
   labour determines how valorisation occurs, but rather the
   manner in which valorisation determines what concrete
   labours are necessary.

Here, Mohun rightly points out the core of the matter: the determinant criterion is the production of value, not of use value; nevertheless, in capitalist production, value production is inevitably connected to use-value production. As a consequence, the delimitation of production and circulation labour must entail the identification of certain concrete forms of labour with the corresponding label of productive or unproductive in the different particular cases. It is not the labour content that operates this classification, but the value creation. Actually, the same concrete labour may be regarded as productive or unproductive depending on its relation to the valorisation process. Conceptually, the analytical distinction between productive and unproductive labour does not require a use-value criterion. In practice, it is the use values and the concrete forms of labour that must be classified in the particular cases.

In the same vein, Savran and Tonak (1999: 42) address the critique that since circulation labour is necessary in all modes of production, it must be productive, noting that 'this type of criticism confuses and conflates in an unjustifiable manner the circulation of use-values and the circulation of commodities, money and capital'. Hence, it is the circulation of commodities as values (not as use values) and the need to transform the money capital into commodity capital and vice versa that determines the productive character or not of the labour employed. Therefore, the task consists in investigating the processes of reproduction of the different capitals, identifying the moments of valorisation and the moments of circulation (and the concrete labours related to them).

Finally, it is useful to address Laibman's (1999: 68) critique: 'To avoid arbitrary reliance on concrete labours', he writes, 'we will have to explain why the concrete labour of title transferring is treated differently from other concrete labours. It is this imputed difference that accounts for the assignment of such labour in the circuit of capital in the first place'. Now, it is possible to show why this critique is wrong. Simply, there are no concrete 'labours' generally associated with the title-transferring activity (if we identify it with the circulation activity).

Therefore, we cannot assign any concrete labour to the unproductive label. On the contrary, it is the title-transferring character of some concrete forms of labours that place them in the unproductive group. Of course, this consideration requires a thorough investigation of particular cases and of particular branches of production.

That is to say, the classification of productive and unproductive labour needs microeconomic foundations, based on the analysis of the valorisation process of every branch of production. However, this need does not preclude the real content of the distinction being founded on an analytical value criterion.

5. Unproductive labour and empirical research

The alternative definition provided above implies an alternative approach to empirical research. In this section, I briefly elucidate the important differences between the orthodox approach and mine, drawing the conclusions of this paper.

The main differences lie in the delimitation of unproductive circulation activities. I agree with Laibman that circulation activities are not sectors, nor can they be identified with branches of production of conventional national account systems. On the contrary, I think this identification is a consequence of a pervasive utilisation of the use-value criterion, with perverse effects for the empirical research of capitalist economies. Actually, the conventional classification of branches of production in national accounts is based on the different use values produced. Therefore, it is only if we use a use-value criterion that we can identify circulation activities with some branches of production. In conclusion, commercial capital, financial capital and ground rent do not have any correspondence with the branches of production of conventional national accounts: trade, finance or other sectors.

In spite of this, I regard these sectors as productive ones, inserted in the general circuit of capital M--C ... P ... C'--M'. Correspondingly, inside them we find both production and circulation labour, as in any other sector. For instance, the trade sector involves the production activity of a sale service (which includes concrete operations such as transport, storing, packing and advertising, among others) that creates both value and surplus value. Of course, circulation activity is also to be found within this sector. The same holds true for the financial sector and production activities such as money transfers, security-box renting, the creation of bank accounts, etc. In conclusion, the presence of production and circulation activities in these branches of production is qualitatively comparable to that existing in any other branch. The quantitative presence may vary from one branch to another; and there might exist evidence supporting a greater participation of circulation labour in these branches. However, I explicitly reject any direct identification of trade, finance, etc. with the unproductive circulation spheres.

Notwithstanding, this methodological error is committed in a vast majority of empirical investigations. For instance, Shaikh and Tonak (1994: 252) consider the totality of the trade, finance and real-estate sectors as non-production activities; hence, they classify all the labour employed in these sectors as unproductive. What is more, they include in this category the non-production labour belonging to production sectors (ibid: 295). Similarly, Moseley (1991: 176-177) assigns the financial, insurance and rent sectors to the unproductive category. In terms of the trade sector, he assumes that unproductive labour accounts for the greater part of the sector. Other empirical works assume identical hypotheses. As a consequence, their estimation of the ratio of unproductive labour to total labour has steadily increased during the last decades, due to the increasing participation of trade and finance in total output. Hence, estimates of economic categories like the rate of surplus value or profit are strongly affected by this evolution.

In my opinion, a microeconomic classification of productive and unproductive labour based on a value criterion would not show these trends, and the proportion of unproductive to total labour would not play any significant role in the explanation of the long-term evolution of main economic categories. In this respect, my view is closer to that of those who reject the concept of unproductive labour within capitalist production: 'There are many reasons why profits may rise or fall, but the proportion of "productive" and "unproductive" workers has nothing to do with this rise or fall' (Houston, 1997: 138). Actually, in the absence of an adequate microeconomic separation of production and circulation, an empirical analysis of capitalist economies may perfectly assume that the ratio of circulation to total labour is constant over the period, as in Camara (2002, 2003). It is worth noting that this unorthodox alternative approach to the distinction between productive and unproductive labour implies a much more orthodox interpretation of the accumulation trends than the traditional orthodox approach.

However, there are still many points about the influence of unproductive labour in capitalist economies to be investigated. Although I believe that the proportion of unproductive labour does not play any role in accumulation trends in the long term, it is not so in the short term. Thus, recurrent difficulties in the realisation of commodities, the corresponding increase in unpaid loans and mortgages and so on, which are associated with capitalist production, may cause cyclical behaviour in the ratio of unproductive to total labour. In other words, circulation costs may become excessive in periods of crisis, as happens with production costs. It is essential to investigate the general behaviour of 'circulation costs' and the question of whether there exists a tendency of these costs to decrease in a competitive environment, as happens with production costs. These investigations could provide a real breakthrough in the understanding of capitalist economies within the labour theory of value, instead of the devastating effects of the increasing importance being placed on unproductive labour.

Acknowledgments 

This paper was written while I was a postgraduate student in the Department of Applied Economics at the Universidad Complutense de Madrid. I wish to thank Diego Guerrero and my student colleagues for their helpful comments during our discussion of the subject. Any remaining errors, of course, are mine. A previous version of this paper was presented at the fourth annual conference oft he Association of Heterodox Economists, held in Dublin from 10-11 July 2002.

References

Camara, Sergio (2002) 'La rentabilidad del capital y sus componentes: El caso de Espana (1964-2000)' in Economia: Teoria y Practica, no. 17, pp. 45-74.

-- (2003) 'Tendencias de la rentabilidad y de la acumulacion de capital en Espana, 1954-2001', Ph.D. thesis, Universidad Complutense de Madrid.

Carchedi, Guglielmo (1991) Frontiers of Political Economy (Verso).

Gouverneur, Jacques (1990) 'Productive labour, price/value ratio and the rate of surplus value: Theoretical viewpoints and empirical evidence' in Cambridge Journal of Economics, vol. 14, no. 1, pp. 1-27.

Guerrero, Diego (1989) 'Acumulacion de capital, distribucion de la renta y crisis de rentabilidad en Espana (1954-1987)', Ph.D. thesis, Universidad Complutense de Madrid.

--(1997) 'Unproductive labour', unpublished paper, New School for Social Research.

Houston, David (1997) 'Productive-unproductive labour: Rest in peace' in Review of Radical Political Economics, vol. 29, no. 1, pp. 131-139.

Laibman, David (1992) Value, Technical Change and Crisis: Explorations in Marxist Economic Theory (M. E. Sharpe).

--(1999) 'Productive and unproductive labour: A comment' in Review of Radical Political Economics, vol. 31, no. 2, pp. 61-73.

Leadbeater, David (1985) 'The consistency of Marx's categories of productive and unproductive labour' in History of Political Economy, vol. 17, no. 4, pp. 591-618.

Marx, Karl (1969 [1861]-3) Theories of Surplus Value, Part I (Lawrence and Wishart).

--(1864) Capitulo VI [inedito]: Resultados Del Proceso Inmediato de Produccion (Siglo XXI editores), English translation taken from <www.marxist.org> [Ch. VI]

--(1906 [1867]) Capital, Volume I: The Process of Capitalist Production (Charles H. Kerr & Company).

--(1909 [1885]) Capital, Volume II: The Process of Capitalist Circulation (Charles H. Kerr & Company).

Mohun, Simon (1996) 'Productive and unproductive labour in the labour theory of value' in Review of Radical Political Economics, vol. 28, no. 4, PP. 30-54.

--(2000) 'Productive and unproductive labour in the us economy: Does the distinction matter? A reply to Houston and Laibman', paper presented to the mini-conference of the International Working Group on Value Theory (IWGVT), Washington, 24-26 March 2000.

Moseley, Fred (1991) The Falling Rate of Profit in the Post-War United States Economy (Macmillan Press).

Savran, Sungur & Ahmet Tonak (1999) 'Productive and unproductive labour: An attempt at clarification and classification' in Capital & Class, no. 68, Summer, pp. 113-152.

Shaikh, Anwar & Ahmet Tonak (1994) Measuring the Wealth of Nations: The Political Economy of National Accounts (Cambridge University Press).

Notes

(1.) It is possible to find several classifications of wrong definitions of productive labour. The physicalist definition considers productive that labour which creates tangible commodities. This is Adam Smith's second definition and, though practically banished from the literature nowadays, it was dominant among the Soviet theorists and present in the work of some Western economists such as Poulantzas (Savran & Tonak, 1999: 150n, 136; Guerrero, 1989: 252-255, 255-258). The evaluative definition evaluates labour and its products according to a standard of social usefulness, and its main proponents are Gillman, and Baran and Sweezy (Laibman, 1992: 75; Shaikh & Tonak, 1994: 20; Guerrero, 1989: 276). Finally, the reproductive definition classifies into the unproductive group that labour whose products are not used in the reproduction of capital. Morris, Blake and Gough consider the production of luxury goods as unproductive (Guerrero, 1989: 277-282). The Sraffian and 'global' versions also fit into this definition (Laibman, 1992: 72-73). Finally, it also encompasses the claim that labour which increments the use value of the workforce is productive, as found in Gough, O'Connor, or in Yaffe and Bullock (Guerrero, 1989:289-298).

(2.) Savran and Tonak (1999: 121-123) provide a slightly different classification, with five social activities: production, circulation, distribution of the product, personal and social consumption, and reproduction of the social order. Consumption and distribution do not involve expenditure of labour, while circulation and reproduction of the social order imply the performance of non-production labour.

(3.) In this vein, Savran and Tonak (1999: 124) assert that 'a definition of productive labour based on the concrete character of the labour spent in the production process is manifestly insufficient within the context of capitalism.

(4.) Note that it is not the usefulness or inner coherence of the distinction that is under critique here. Even if this evaluative distinction might be operative and coherent, it is not useful for the definition of productive labour for capital, since the 'general rule' of the social and historical content of the concept is broken. In fact, Savran and Tonak (1999: 144-145) themselves admit to the limited validity that they assign to this 'rule':

When Marx emphasises that distinction between productive and unproductive labour is independent of and indifferent to the type of use-value produced, his statement is ... restricted to that set of use-values which correspond to the set of productive labour in general.... The type of use-value that results from a certain activity is immaterial to the distinction between productive and unproductive labour only when the activity in question is part of production.

(5.) Although not explicitly identified with the concept of productive labour in general, the use-value criterion is very common in the literature. Carchedi (1991: 28) writes that 'any labour process which does not affect the use value of a material object, such as the purchase and sale, banking and insurance, etc., should be regarded as a formal transformation and thus unproductive. This is a formal material labour process which can produce neither value nor surplus value'. In the same vein, Leadbeater (1985: 597) considers that 'the decisive point Marx recognised is that labour expended in genuine cost [of circulation] does not affect physically the process of production or use values of commodities'. Leadbeater admits, too, that Marx's 'remarks to the effect that use value does not matter refer specifically to the sphere of production, not circulation' (ibid: 599).

A flagrant instance is to be found in Mohun. After stating that his distinction is 'analytic' and that 'in the valorisation of value, it does not matter what use-value is produced; instead, what is required is a specification of the moments of that valorisation', he goes on to say that 'the commodity being traded undergoes no transformation whatsoever, other than a change of ownership, and neither use-value nor value is produced' (Mohun, 1996: 44, 43-44). More recently, he reiterates this proposition:

   Does an activity produce a new use-value or alter an
   existing one in some way? ... For those who assert
   that all wage labour is productive, then all such labour
   does produce a new use-value, or does alter an existing
   one, and it does produce surplus value.... Conversely,
   for those who assert that some wage labour is
   unproductive, then that labour does not produce a
   new use-value, or does not alter an existing one, and
   it does not produce surplus value, but rather consumes
   it'. (Mohun, 2000: 5)

(6.) The distinction between market and non-market wage labour is analogous to the distinction between capital and revenue: it plays an important but limited role in the definition of productive labour. In spite of this, it has often been used to encompass its whole complexity.

(7.) This level of the distinction is only introduced into the analysis in Volumes II and III of Capital:

   The capitalist, as representative of capital engaged in
   its valorisation process--productive capital--performs
   a productive function, which consists precisely in
   directing and exploiting productive labour. The
   capitalist class, in contrast to the other consumers of
   surplus value, who do not stand in a direct and active
   relation to its production, is the productive class par
   excellence.... As yet, we are only acquainted with
   capital within the direct production process. The
   situation with the other functions of capital--and with
   the agents used by capital to perform these functions
   can only be examined later'. (Chapter VI: 89)

(8.) Houston (1997: 132) believes that the conclusions of Volume I cannot be made extensible to capitalism: 'Simple commodity production and exchange are part of a system of social relations which are not capitalist, and abstract conclusions drawn from the C--M--C circuit are not necessarily applicable to a different concrete, capitalism. Whatever the role of exchange and circulation in petty commodity production, it is hardly comparable to that of an advanced capitalism system'. He commits two errors: 1) he believes that Volume I deals with non-capitalist social relations, since he considers value to be a category non-specific to capitalism; and 2) he assumes the unacceptable 'sectoral' determination of circulation labour.

(9.) Moseley makes two sets of estimates. First, he obtains an average proportion of unproductive labour of 55.46 per cent (see tables A.7 and A.8 on pp. 168-169), and then he raises it to 60.49 per cent (see tables A.9 and A.10 on p. 170).