1.1. Capitalist relations of production

On an influential Marxist view, the relations of production are ‘relations of effective power over persons and productive forces, not relations of legal ownership’ (Cohen Reference Cohen2000: 63). An effective power is a nonnormative ability to effect outcomes. A taxonomy of effective powers à la Cohen permits a characterization of capitalist relations of production in terms of the subordinate producer’s control over her labour power and means of production, as follows:

For any pair {r, s}, where r and s take their values from the set {All, Some, None}, r represents the direct producer’s degree of control over her own labour power, and s over means of production. That the producer controls ‘all’ of her labour power entails that she enjoys a formal right of self-ownership, that is, is not liable to extra-economic coercion. On Cohen’s orthodox view, the {All, None} set of powers is both necessary and sufficient for capitalist relations of production.

The orthodox view faces well-known objections. Take, for example, antebellum slavery in the American South. Cohen himself suggests that ‘[a] producer could be coerced into working for a wage and producing surplus value’ (Cohen Reference Cohen2000: 83 refers to Marx Reference Marx1976: 925). So antebellum slavery involves slaves performing wage-labour, but lacking full control over their labour power. That is, antebellum slavery provides a presumptive counterexample to the orthodox view (for a wealth of similar cases, see Banaji Reference Banaji2010). Consider, further, merchant and usury capital, the forms dubbed ‘antediluvian’ by Marx (Reference Marx1976: 266; Reference Marx1981: 728). Under putting-out, for example, the merchant makes a profit by providing raw material to the peasant, buying the final product below its market value. Putting-out seems to be an instance of {All, Some} relations of production, since the peasant must own some of the conditions of production.Footnote ¹ If putting-out betokens a capitalist relation of production – as I will argue below – then it constitutes another counterexample to the orthodox view.

This paper argues that, in light of these difficulties, it is preferable to define capitalist relations of production functionally. A relation R is a capitalist relation of production if and only if R involves relations of effective control over labour power and means of production that serve the subsumption of labour by capital. This definition makes R compatible with any combination of the r, s variables in the production relations set (whether or not it is represented in Table 1). Capitalist relations of production are then constituted by that subset of reproducible economic relations in which the extraction of surplus labour is subjected to the maximization of exchange value – or what Marx calls the ‘law of value’. The connections between these concepts will, I hope, become more transparent in the discussion that follows.

Table 1.

Relations of production (Cohen Reference Cohen2000: 65).

1.2. The capitalist mode of production

This paper defends a definition of capitalist social relations that draws upon Marx’s distinction between the relations and the mode of production. Most discussions of Marx’s theory of production note that he uses the term ‘mode of production’ in at least two senses, one narrow and one broad. In the narrow sense, the mode of production refers to the form of technology employed in production, including the technical division of labour. The narrow sense of the mode of production therefore excludes power relations. Here is an instance of this usage:

Handicrafts and manufacturing therefore share the same mode of production. Cohen (Reference Cohen2000: 79-80) calls this the material mode. The broad sense of the mode of production, on the other hand, includes facts about the purpose of production and the form of social labour. Cohen calls this the social mode. For example:

In what follows, I shall refer to the mode of production exclusively in the narrow, or material, sense. Marx usually relegates claims about the social mode, including claims about the power of capital over labour, to the mode of subsumption or subordination.Footnote ² This paper argues that Marx’s mature works underwrite at least four modes of subsumption: hybrid, formal, real and abstract. Hybrid subsumption, on the one hand, represents the way capital dominates workers under usury, putting-out, and the so-called ‘domestic system’. It is, in that sense, a ‘transitional form’ from feudalism to capitalism. Formal and real subsumption are explicitly discussed by Marx in Capital, volume I, and the Results of the Immediate Process of Production. Finally, abstract subsumption refers to the ways capital dominates workers under the cooperative factory. It is, in this sense, a transitional form from capitalism to communism.

1.3. Exploitation and subsumption

Consider, first, the subsumption of labour by capital. I assume that Marx’s claims about subsumption entail the possession, by capitalists, of dominating power over workers. Domination obtains when a powerful agent, A, leads another agent, B, to act for reasons that are not independent of A’s power. Suppose B is hungry. A uses her power, not in order to help B act for hunger-based reasons (e.g. to eat because hungry), but rather in order to get B to act for reasons based on A’s power (e.g. to eat because otherwise A will beat B). Here A dominates B. The power of capital over labour is a form of dominating power, in the sense that it is a power compelling workers to act for power-dependent reasons. There are, after all, no reasons that workers should work for capitalists, other than those grounded on the power of the latter over the former. That power, in turn, obtains in virtue of capitalist control over the means of production.Footnote ³

Now consider the exploitation of labour by capital. As a matter of conceptual necessity, A exploits B if and only if A presumptively benefits at B’s expense by taking advantage of B.Footnote ⁴ Marxists believe that the relevant metric of ‘expense’ is labour time.Footnote ⁵ A extracts surplus labour from B if there is an unequal exchange of labour time between A and B, such that A pockets the difference. That is, A extracts surplus labour from B if: A’s consumption bundle embodies X hours of labour time, B’s consumption bundle embodies Y hours of labour time, whereas A’s product embodies Y hours, B’s product embodies X hours, and X > Y ⩾ 0.Footnote ⁶ A necessary condition for specifically capitalist exploitation is that surplus labour assumes the value-form, that is, constitutes surplus value. I shall assume, throughout this paper, that what enables capitalist exploitation is the power of capital over labour. In a word, capitalist exploitation presupposes capitalist domination.

2.1. Marx’s attempt to defend (1)

In chapter 4 of Capital, volume I, Marx defines capital as ‘self-valorizing value’. Capital is represented by the formula M − C − M′, where M and M′ are money magnitudes, C is a commodity, and M′ > M, such that M′ involves newly created value (and is not a mere redistribution of value). Surplus value is defined as: M′ − M = ΔM. According to Marx, capital assumes three main forms: merchant’s capital, usurer’s capital and industrial capital. Merchant’s and usurer’s capital are ‘antediluvian’ (Marx Reference Marx1976: 266), in the sense that they ‘appear before the modern primary form of capital’ (Marx Reference Marx1976: 267) Marx then argues that wage-labour is the only source of value. He does this by assuming an equivalence between prices and values. If that equivalence holds, then merchant and usury capital do not produce surplus value. Therefore only industrial capital produces surplus value. Chapter 6 of volume I completes Marx’s argument, by showing that, on these assumptions, the only source of surplus value can be ‘a commodity whose use-value possesses the peculiar property of being a source of value, whose actual consumption is therefore itself an objectification of labour, hence a creation of value’ (Marx Reference Marx1976: 270). That commodity is labour-power. Hence (1) holds. I now show that this argument fails on its own terms.

Marx’s exposition has two parts. The first goes from a premiss about the definition of surplus value to a conclusion about merchant and usury capital, whereas the second goes from a premiss about value to a conclusion about the source of surplus value. The first part is as follows:

1. (3) The exchange of nonequivalents is never the source of newly-created surplus value, and always a mere redistribution of existing value. (Lemma)

2. (4) Merchant and usury capital presuppose exchange of nonequivalents. (Definition)

3. ∴ (5) Merchant and usury capital ‘appear impossible’, that is, are mere redistributions of existing value.

Marx asserts (4) in Marx (Reference Marx1976: 266–7).Footnote ⁸ Lemma (3) is the crucial move. It is formulated as follows: ‘If equivalents are exchanged, no surplus-value results, and if non-equivalents are exchanged, we still have no surplus-value’ (Marx Reference Marx1976: 266). What needs to be shown is that both sides of this biconditional hold: price-value nonequivalence must not only be necessary for mere redistribution of value, but also sufficient for it. But this is precisely what Marx does not show. That is, by Marx’s own lights, some forms of merchant capital, such as putting-out, have a foothold in production.Footnote ⁹ They do not, therefore, constitute mere redistribution of existing value. In effect, Marx moves illicitly from the true claim that value redistribution is possible only if someone buys cheap and sells dear – that is, only if nonequivalence holds – to the false claim that nonequivalence is always value redistribution. This is important because, if (3) were true, then assuming equivalence would be tantamount to purging exchange of false positives: it would exclude merely redistributive exchanges. But if (3) is false, then assuming equivalence – as Marx proceeds to do in the second part of the argument – yields false negatives, that is, it excludes nonequivalent exchanges that do involve creation of surplus value. The false negatives are precisely those forms of merchant and usury capital that create surplus value. If Marx is taken to be offering a deductive argument, then he is not entitled to assume (3).

Having assumed the truth of (3), Marx asserts price-value equivalence. The second part of his argument is as follows:

1. (6) Equivalents are exchanged.

2. (7) If equivalents are exchanged, then the only source of surplus value, that is, the basis for the reproduction of capitalist relations of production, must be a commodity whose consumption creates value.

3. (8) That commodity is labour-power, and labour-power alone.

   ∴ (1) Capitalist relations of production presuppose wage-labour.

Suppose (8) is granted.Footnote ¹⁰ The argument still begs the question. That is, (6) covertly assumes the truth of (1). For consider: by (3) and (4), merchant and usury capital do not produce surplus value. But then to affirm (6) is to assume that only industrial capital produces surplus value. And since industrial capital and wage-labour are extensionally equivalent, (1) follows without the help of auxiliary premiss (7). The second part of the argument is redundant. Chapters 4–6 of volume I therefore amount to little more than further clarification of Marx’s assumptions – as opposed to a nontrivial deduction of (1) from independent axioms.Footnote ¹¹ Indeed, Marx’s dropping the assumption of equivalence in volume III of Capital makes (6) false and the putative derivation of (1) invalid. No deduction of (1) is ever provided.

2.2. Cohen’s attempt to defend (1)

It is worth considering another influential, but equally invalid, defence of (1). Consider the following famous passage from the Grundrisse:

Cohen claims that the last sentence of this passage ‘is, in fact, a compressed statement of thesis [(1)]’ (Cohen Reference Cohen2000: 184, n. 2). The idea here is that, since capital implies the ‘personification’ of the conditions of labour in the capitalist, the existence of the capitalist implies the existence of the wage-labourer. This inference is invalid: the objective conditions of labour can acquire a ‘personality as against labour’ if capitalists merely hire them out to workers (through lending, for example), as opposed to hiring workers in to use them. Whether capital hires labour or labour hires capital, it is still the ‘objective conditions of labour’ that dominate workers.Footnote ¹² The same invalid inference, from Marx’s assertions about the alienated form of social relations to the existence of wage-labour, pervades the works of Sweezy (Reference Sweezy1942), Mandel (Reference Mandel1974), Fine and Saad-Filho (Reference Fine and Saad-Filho2004), Harvey (Reference Harvey2010), Callinicos (Reference Callinicos2014), among many others.Footnote ¹³

Again, no valid deduction of (1) has been provided.

2.3. The locus of the inconsistency

I now locate the apparent inconsistency in Marx’s writings. In chapter 36 of volume III of Capital, Marx distinguishes between two forms of usury: lending money to ‘extravagant magnates’ for the consumption of luxuries and lending to ‘small producers who possess their own conditions of labour’ (Marx Reference Marx1981: 729). The latter form of usurer’s interest, he suggests, represents surplus value. Marx then distinguishes usury (or interest-bearing) capital under the pre-capitalist mode from the form it assumes under the capitalist mode of production (Marx Reference Marx1981: 730). In a famous passage, he argues that:

Several things follow. First, capitalist relations of production pre-date the capitalist mode of production. Second, those direct producers subsumed under usury capital are subject to {All, Some} relations of production (see Table 1). They differ, in that respect, from the industrial proletarian. Third, it is false that usury capital does not subordinate labour: Marx asserts that it merely fails to ‘directly subordinate labour’ (my emphasis). Indeed, he later adds that ‘[u]surer’s capital has capital’s mode of exploitation without its mode of production’ (Marx Reference Marx1981: 732). In the next section, I show that this ‘indirect’ subordination of labour by usury capital is tantamount to capitalist domination of the producer as such, only outside the factory.

Marx therefore thinks that usury capital, insofar as it appropriates newly created value, is a capitalist relation of production; it entails a relationship of dominating power geared towards the production of surplus value, not its mere redistribution.Footnote ¹⁴ Similar considerations apply to merchant capital: putting-out, for instance, has a similar exploitative structure.

Now consider the following triad of claims:

1. (1) Capitalist relations of production presuppose wage-labour.

2. (9) Merchant and usury capital constitute capitalist relations of production.

3. (10) Merchant and usury capital do not presuppose wage labour.

The triad is inconsistent; no more than two of its claims are simultaneously assertable. In volume I of Capital, Marx seems to be deducing (1). In volume III he asserts the conjunction of (9)–(10). The next section makes the case for keeping the conjunction. If that case is sound, then (1) must be false. The upshot is that capital dominates workers in a variety of ways, wage-labour being only one of them.

3.1. Skillman on subsumption

It is well-known that Marx posited an important distinction between manufacturing and mechanized industry, discussed with consummate brilliance in chapters 14 and 15 of Capital, volume I. The distinction is between the formal and real subsumption of labour by capital. Under the former, often identified with manufacturing, the technical basis of the mode of production has not yet ‘become adequately realized – it [has] not become indispensable, and that also means technologically indispensable’ to ‘producing for the sake of production’ (Marx Reference Marx1976: 1037–38). According to Marx:

The ‘less developed’ stage of ‘capitalist production’ is formal subsumption; the ‘more developed’ real subsumption. To these different modes of subsumption there correspond different forms of surplus value (what Marx calls ‘absolute’ and ‘relative’ surplus value) and different degrees of worker independence. Crucially, however, both formal and real subsumption presuppose a complete separation of the worker from the means of production. That is, they are characterized by {All, None} relations of production, the set of property relations uniquely coextensive with wage-labour. The obverse of this separation is that dark archetype of modernity, the factory.Footnote ¹⁵

This taxonomy is incomplete. For, as the discussion in chapter 36 of volume III shows, Marx manifestly believes that capitalist relations of production precede the factory. Indeed, it is incoherent to think that capitalist relations of production can ‘transform’ or ‘penetrate’ a mode of production without predating its transformed successor. Marx writes, moreover, that the ‘reproduction of the handicraft system on the basis of machinery only forms a transition to the factory system’ (Marx Reference Marx1976: 589) and refers to the ‘medley of transitional forms’ from handicrafts to manufacturing (Marx Reference Marx1976: 602). He adds:

It follows that ‘transitional’ and ‘hybrid’ forms represent ‘indirect’ modes of subsumption of labour to capital. That is, they constitute capitalist relations of production transitional to the capitalist mode of production. Skillman (Reference Skillman2007) offers a helpful taxonomy of these hybrid forms. He uses Marx’s expanded schema of the capital circuit: M − C. . .P. . .C′ − M′, where . . .P. . . represents the production of new value, with two modifications. The first modification uses arrows to represent the transformation of value-adding inputs into outputs, such that Marx’s schema now appears as M − C → P → C′ − M′. The second modification uses brackets to characterize the portion of the labour process that remains in the control of the worker. Skillman’s schema is depicted in Table 2.

Table 2.

Skillman’s taxonomy.

The first row represents usury capital: the usurer advances money capital M to the direct producer, who uses it to buy commodities worth C. The producer then adds value, turning C to C′, which she sells for M′. She then uses the proceeds to pay interest to the usurer. Crucially, the producer retains some degree of control over her product, over the labour process, and over the process of circulation. A typical complement to this autonomy is the producer’s possession of certain means of production, tools and so on, some of which are used as collateral for the loan. The second row represents merchant capital: the merchant uses money capital to purchase raw materials worth C. The merchant then gives these raw materials to the producer, who valorizes C to C′, which is then appropriated by the merchant and sold for M′. This is, for example, how the capitalist dominates the producer under putting-out.Footnote ¹⁷ Skillman (Reference Skillman2007) calls the subsumption obtaining under merchant capital primitive subsumption. I here follow Marx in calling it hybrid.Footnote ¹⁸

The third and fourth row represent formal and real subsumption, respectively. Under formal subsumption, the worker still retains a modicum of autonomy because the labour process has not been completely objectified: the manufacturing worker is still ‘a fragment of his own body’ (Marx Reference Marx1976: 482), but not yet a mere ‘living appendage’ of the machine (Marx Reference Marx1976: 548). This modicum of autonomy is represented in the third row by the worker’s control over the labour process, which is completely forsaken under mechanized industry. The move downwards along this table, from usury to industry, schematically represents the thinning out of producer autonomy, together with the corresponding extension of capital’s domination of the producer. Skillman’s elegant taxonomy thus vindicates the developmental continuity between hybrid, formal and real subsumption as capitalist relations of production. It thereby vindicates (9), the view that merchant and usury capital constitute capitalist relations of production.

I turn now to (10), which says that one can have hybrid subsumption of labour to capital without wage-labour, indeed without a labour market. Suppose that (10) is false. That is, suppose that putting-out presupposes the existence of a labour market. Now the putter-out optimizes by hiring a proletarian. This reduces monitoring, waste and pilfering costs, assuming that the cost of monitoring effort is not excessively high. But then putting-out is unprofitable; labour markets imply the euthanasia of the putter-out. This contradicts the anti-(10) assumption that being a putter-out is profitable when wage-labour is available.Footnote ¹⁹ This simple argument shows why labour markets presuppose an {All, None} set of relations of production, rendering them incompatible with putting-out. As Skillman points out, the historical survival of antediluvian forms of subsumption largely depended on producers’ nonmarket access to nonlabour collateral, mainly land. The gradual erosion of that collateral led to the eventual demise of hybrid subsumption and ushered in the era of the factory (see Marx Reference Marx1981: 730–1). Hence hybrid subsumption does not entail wage-labour. Therefore (10) holds. Given the truth of (9), it follows that (1) is false.

3.2. Banaji on subsumption

Like Skillman, Jairus Banaji has recently argued that antediluvian capital constitutes capitalist relations of production. Banaji offers numerous penetrating historical studies of these relations, from early Byzantium to contemporary India. Banaji’s studies also vindicate (9). In discussion of the relationship between big, middle and proletarianized peasantry in the mid-19th century Deccan districts of India, Banaji writes:

Banaji’s ‘pre-formal subordination’ is equivalent to Skillman’s ‘primitive’ subsumption, both of which are equivalent to what Marx calls hybrid subsumption.Footnote ²⁰ It follows that Banaji affirms (9): merchant and usury capital sometimes constitute capitalist relations of production.Footnote ²¹

But unlike Skillman, Banaji baulks at (10), preferring variations on (1) instead. In response to the charge that antediluvian forms of capital do not involve wage-labour, Banaji replies that they entail ‘disguised wage-labour’. He writes, for example:

This attempt to defend (1) is incoherent.Footnote ²² For if usury and merchant capital entail wage-labour, disguised or not, then they (trivially) presuppose industrial production. The point of distinguishing sharply between hybrid and formal subsumption, however, is to show how, in the manner of Skillman, the factory constitutes a turning point in the economic articulation of modernity. The factory presupposes an all {All, None} set of production relations, that is, the ‘complete expropriation’ of the producer. This is not the predicament of the producer under putting-out or usury, as Banaji’s own studies show. In other words, if ‘pre-formal’ subsumption entails ‘disguised’ wage-labour, then Banaji’s distinction between formal and preformal subsumption – indeed, the whole edifice of distinctions between a ‘dependent middle peasantry’ and a ‘semi-wage-labour peasantry’ (Banaji Reference Banaji2010: 317–323) – collapses.Footnote ²³ Banaji should, in all consistency, embrace (10) and reject (1). Capitalist relations of production do not presuppose commodified labour-power, or even workplace authority relations.

3.3. Brenner and Wood on subsumption

I now consider an influential account of the transition from the pre-capitalist to the capitalist mode of production. This account posits the extraction of surplus labour in the value form as a necessary but not sufficient condition for the subsumption of labour by capital. On this view, the set of sufficient conditions is completed by the existence of a specific form of relationship between markets and producers.

Robert Brenner has argued that capitalist relations of production presuppose market dependence on the part of both capitalists and producers. It is the asymmetric nature of that dependence that explains the rise of English agrarian capitalism. Indeed, market dependence is Brenner’s explanation for the secular growth of agricultural productivity in early modern England, largely accounting for its early industrialization (Brenner Reference Brenner1976, Reference Brenner1977). E.M. Wood elaborates:

Since, according to Brenner and Wood, these forms of market dependence constitute capitalist relations of production, the Brenner/Wood account contradicts the orthodox definition of capitalist social relations. Indeed, it entails the falsehood of (1). But what exactly is ‘market-dependence’? Here is a typical formulation, due to Charles Post:

By ‘law of value’ Post means the imperative to produce commodities at ‘socially necessary labour time’ – roughly cost of production – thereby maximizing exchange value.Footnote ²⁵ There are three different, and progressively narrower, senses of market-dependence in Post’s definition: (a) The broadest sense refers to lack of nonmarket access to the means of subsistence and production. (b) A narrower sense refers to subjection to market competition and the law of value. (c) A yet narrower sense refers to subjection to market competition that leads to ‘specialization, technical innovation, accumulation’.Footnote ²⁶

For our purposes, (b) offers the best specification of market-dependence. On the one hand, (a) is too broad: on this definition, the monopsonistic feudal lord, or Marx’s ‘extravagant magnate’, who lacks nonmarket access to grain, is market-dependent. (c), on the other hand, is too narrow. For on this definition, the pre-revolution agriculture of Russia (Lenin’s agrarian capitalism), the Deccan capitalism of Banaji, the decadent industries of ‘actually-existing socialism’, and many other instances of – presumptively capital-positing – backwardness involve noncapitalist relations of production. Market-dependence is therefore best defined in terms of subjection to the law of value, but not the possible effects of that law.

In light of these amendments, subsumption – and by extension capitalist relations of production – can now be defined as the production of surplus labour, subjected to the imperative of cost minimization, and extracted by exploiters in the value form. This definition has important implications. It implies, for example, that all forms of petty commodity production – self-employment, family farms, agricultural cooperatives, etc. – that are both exploited and subjected to the imperatives of profit maximization, involve subsumption of labour to capital. It follows, again, that capitalist relations of production do not presuppose a labour market; capital-subsumed self-employment is not a kind of wage-labour. By the same token, the petty commodity production of 17th-century middle English peasants (Byres Reference Byres2006), or of 19th-century American family farms (Post Reference Post2011), betoken capitalist relations of production.

I now show that this definition of subsumption implies that market socialism represents yet another, distinctive, form of subsumption of labour by capital.

4.1. Cooperatives as a transitional form

In discussion of cooperative factories, Marx writes:

The putative connections between the ‘cooperative enterprise’ and the ‘capitalist joint stock company’ are opaque unless juxtaposed with Marx’s critique of the ideological conflation of ‘profits of management’ with ‘wages of supervision’:

This contrast is further illuminated by Marx’s famous orchestra analogy:

Through a characteristically Marxian dialectic of self-abolition, the capitalist system generates a credit system, which makes possible the separation of ownership from control, which, if realized, paves the way for worker-controlled factories under a system of social ownership of the means of production.Footnote ²⁷ Suppose all factories were turned into cooperatives. Surely that would suffice to abolish the domination of labour by capital. But then worker-controlled, socially owned, market-operating cooperatives are not a mere transitional form towards socialism; they are socialism. Marx rejects this conclusion. He writes:

Like hybrid subsumption, abstract subsumption is a transitional form. There is much debate as to exactly what Marx means by ‘workers’ becoming ‘their own capitalist’.Footnote ²⁸ But the conjunction of this passage with passages where Marx claims that both workers and capitalists are dominated under capitalism (see, for example, Marx Reference Marx1976: 990), seems to entail that the reciprocal pressures of competition cause workers-turned-capitalists to suffer some form of domination, or ‘enslavement’ even where all factories are profit-maximizing cooperatives. Is this claim defensible? It is, I believe, defensible, but not for the widely held reason that market-operating cooperatives ‘self-exploit’.Footnote ²⁹ The next section refutes the ‘self-exploitation’ claim and suggests a reading of what ‘valorize their own labour’ means, one based on the abstract subsumption of labour to capital.

5.1. Self-exploitation

Market socialism has been proposed as a half-way institutional remedy to exploitation of all forms that is neither based on central planning, nor is exclusively laissez-faire. In models of market socialism the basic idea is captured by a decentralised form of worker management based on profit maximization, in which: (a) the means of production are jointly owned (either by workers, or by the state), (b) firms are democratically controlled by the workers in the firm, allowing them to share in the profit, and (c) production is for a market, in which values are determined by average or marginal cost. Economic democracy of the market socialist variety achieves inter-firm equilibrium through maximization of profit: the individual firm’s maximand is revenue per worker. Thus in long-run competitive equilibrium the firms that survive are those, and only those, that make an economic profit greater than, or equal to, zero.Footnote ³⁰

A common complaint levelled against market socialism is that it encourages the ‘self-exploitation’ of workers. The idea seems to be that, since capitalists exploit workers, and all workers are capitalists, capitalist-workers exploit capitalist-workers, i.e. capitalist-workers exploit themselves.Footnote ³¹ This complaint is incoherent, for two reasons. As a matter of conceptual necessity, to exploit is to benefit at someone else’s expense. It follows that A cannot exploit A, and that A cannot exploit B if A does not extract a presumptive net benefit from some transaction with B. So even if it did make sense to say that workers somehow harm or degrade themselves under market socialism, it would still be false that they self-exploit.

5.2. The real case against market socialism

Suppose there is a system of commodity production without a labour or a credit market. Under such a market system, different firms will come to own differential assets, depending on how successful they are at accumulating capital. The whole point of allowing the existence of profit-seeking firms, after all, is to take advantage of the coordinating informational signals putatively furnished by competitive markets. Now, some firms will do badly and will have to lay people off. Others will do well and will need to hire people. More successful firms will accumulate more wealth. It is a short step from this inequality to the claim that the member of firm A can exploit the member of B. Here is one way this can happen, in the absence of A hiring B, or indeed of any kind of labour market:

There are two producers, Friday and Robinson. Friday is wealth-poor, Robinson is rich. If they do not trade, Robinson will work 8 hours and Friday will work 16, so that each can satisfy his needs (assuming they’re both rational, the fact that they decide to trade shows that they both benefit from trading). They only trade in final goods, and there are no labour or credit markets. In equilibrium under free trade, Friday works 12 hours, Robinson works 4, and both attain subsistence.Footnote ³²

Roemer claims, and Marx agrees,Footnote ³³ that there is exploitation here. Indeed, as long as Friday appropriates the surplus in value-form, there is extraction of surplus value: having worked for 4 hours, Robinson can relax for the rest of the day, while Friday toils to produce what Robinson would otherwise have produced only with an extra 4 hours of work. Hence:

Note that this exchange, insofar as it is subject to the law of value, is a capitalist relation of production. I now want to argue that such exploitative transactions, obtaining through mere trade in goods and in the absence of labour and credit markets, constitute abstract subsumption. Subsumption, I said at the outset, is a freedom term: it describes the way capital dominates labour. In the present example, Robinson clearly possesses a power over Friday, indeed more power than Friday possesses over Robinson (notice Roemer’s claim that Robinson can ‘get Friday to work’). The evidence for this is the discrepancy in the number of hours worked. It is, moreover, in virtue of that power that Friday has a reason to work for Robinson: in equilibrium Friday acts as Robinson’s unhired servant. So their relationship is one of domination. This form of domination falls under abstract subsumption in the sense that, unlike formal and real subsumption, it does not depend on the existence of a labour market and, unlike hybrid subsumption, it does not depend on the existence of a credit or money market. And yet the relationship engenders extraction of surplus labour – not mere value-redistribution – subjected to the law of value. This capitalist relation of production is, for all present purposes, abstract.

It is, now, but a short step to show that market socialism involves abstract subsumption. For cooperatives, even the most successful ones, will need access to raw materials, machinery and other objective conditions of labour. Under market socialism, by assumption, these can only be purchased from other cooperatives. As the process of accumulation proceeds, more successful cooperatives will accumulate wealth, which will eventually manifest itself in unequal exchanges of the Friday-Robinson variety.Footnote ³⁴ Such exchanges are capital-positing, insofar as they involve the extraction of surplus value subject to the law of value. It follows that such exchanges involve abstract subsumption. I now argue that this form of subsumption finds corroboration in Marx’s mature published writings.

5.3. Abstract subsumption in Marx

In volume III of Capital Marx introduces the famous ‘transformation problem’ of dealing with the transformation of values into prices. Although this problem is beyond the scope of this paper, Marx’s solution has some bearing on whether he thought abstract subsumption possible. I now argue that Marx’s proposed solution – whether sound or not – provides a Robinson-Friday-type mechanism of exploitation of cooperatives by cooperatives.

The transformation problem arises because different capitalist firms or industries have different organic compositions of capital.Footnote ³⁵ If prices and values are assumed equivalent and if the ratio between surplus value and the wage bill is assumed constant – Marx calls this ratio the rate of exploitation – then firms with higher than average organic composition (call them F _high) will tend to earn lower than average profits, whereas firms with lower than average organic composition (call them F _low) higher than average profits. But if there is a uniform rate of profit, which Marx (Reference Marx1981: 254–301) assumes, then the excess surplus value extracted in F _low will be redistributed from F _low to F _high until profit rates are equal. The only way this transfer can be effected is if relative prices satisfy: p _Fhigh/p _{Flow > 1}. This, in turn, implies that relative prices will not be equivalent to relative values. The set of prices formed out of the equalization of the rates of profit Marx calls prices of production.

Marx’s argument implies that F _high extracts surplus value from F _low à la Roemer: capital-intensive cooperatives exploit labour-intensive cooperatives. This is the Friday-Robinson-type mechanism we were looking for, which corroborates the abstract subsumption view.

However, some of Marx’s writings seem to contradict this conclusion. In his discussion of Wakefield’s theory of colonization, for example, Marx writes:

Since cooperatives are the property of the immediate producers, it seems to follow from this passage that cooperative worker-owners cannot be ‘their own capitalist’. Defenders of market socialism, such as DiQuattro (Reference DiQuattro2007) take passages such as this to corroborate their view that market socialism cannot involve exploitation. This inference is too quick. For Marx’s discussion, in this context, manifestly relies on a premiss about the relative scarcity of labour-power relative to land:

What advocates of market socialism need to show is that, under a system of competing cooperatives, the ‘bulk’ of capital is ‘public property’, in the sense that one’s appropriation of the means of production does not ‘subjugate the labour of others by means of such appropriation’ (Marx and Engels Reference Marx, Engels and Stedman-Jones2002: 238). And this is precisely what market socialists need to deny, at least insofar as they want to make markets responsible for the allocation of capital and aggregate investment. That is, suppose the allocation of investment were made to depend on the market-mediated private decisions of worker-owned cooperatives. It would follow that some firms accumulate more capital than others, either due to the technical nature of their production process, or due to success in competition. This implies differences in organic composition between F _high and F _low, generating transfers of surplus value from the latter to the former. Workers in labour-intensive cooperatives thus become the unhired servants of workers in capital-intensive cooperatives. This is how cooperatives exploit cooperatives through mere exchange. In the absence of democratic control over aggregate investment, abstract subsumption obtains.

5.4. How abstract subsumption is distinctive

Capitalist power characterized by abstract subsumption does not presuppose credit or labour markets. I now take up the differences between abstract subsumption and the other modes of subsumption discussed so far: hybrid, formal, real. Take hybrid subsumption first. There are some relationships between cooperatives that satisfy the usury circuit (first row of Table 2). Vanek’s (Reference Vanek1970) labour-managed firms, for example, have this form. Vanekian cooperatives are funded by loans from profit-seeking cooperative banks. They are therefore subject not to abstract, but to hybrid subsumption: bank and cooperative occupy the same positions in the capitalist structure as usurer and peasant, respectively. Of course Vanekian labour-managed firms might be an excellent idea, as far as the medium-term freedom and autonomy of producers is concerned.Footnote ³⁶ But that does not implicate these structures any less in hybrid subsumption. By implication, abstract subsumption does not obtain in Vanek-type cooperative relations; it only obtains where a cooperative dominates another in the absence of a credit or labour market.

Now take formal and real subsumption. As I argued above, both of these entail wage-labour; that is, both presuppose a labour market. Abstract subsumption, like hybrid subsumption, does not presuppose a labour market. Suppose there is a system of profit-seeking, worker-managed cooperatives that can freely hire and fire workers. Whenever there is a layoff, the state acts as employer of last resort à la Schweickart (Reference Schweickart2011). In this institutional setting, all unemployment is frictional and wage-labour is abolished, in the sense that wages and employment are no longer determined by the cost of reproducing the worker (however that cost is determined). Suppose, further, that there is a publicly owned bank that provides loans to cooperatives without extracting surplus value from them. In this institutional setting, there is no usury-type relation between banks and Vanekian cooperatives. For all practical purposes, no capital-positing credit or labour markets exist. There is, by implication, no hybrid, formal or real subsumption of the worker. Yet Robinson-Friday relations of domination manifestly survive; cooperatives now ‘valorize their own labour’ in the sense that they exploit each other’s surplus labour, simultaneously subjecting one another to the law of value. These relations constitute abstract subsumption.

It is worth pointing out, finally, that abstract subsumption, unlike formal and real subsumption, does not presuppose bossing, the domination of producers by bosses.Footnote ³⁷ Abolishing capitalist relations of production abolishes bossing, in the sense that it replaces an intra-firm relation of servitude with an intra-firm relation of equality. In that sense, abolishing bossing might suffice to abolish intra-firm servitude. It does not, however, suffice to abolish inter-firm servitude; that is, it does not suffice to abolish capitalist relations of production.

I conclude that the case for abstract subsumption is defensible. On this view, Marx’s account of cooperative workers as ‘valorizing their own labour’ has critical force. I briefly conclude by discussing some corollaries of the views defended in this paper.