[This is a longer version of my chapter in Inequality: Radical Institutionalist Views on Race, Gender, Class, and Nation, William M. Dugger, ed. Greenwood Press, 1996: pp. 65-86. I edited it slightly in May, 2002. By coincidence, many of my points were also made by the late John Elliott in the chapter before it.]


Taxation without Representation:

A Reconstruction of Marx's Theory of Capitalist Exploitation*




James Devine



Economics Department

Loyola Marymount University

One LMU Drive

Los Angeles, CA 90045-2699 USA

phone: 310/338-2948; FAX: 338-1950

e-mail: jdevine@lmu.edu




This chapter presents a robust reconstruction of Karl Marx's theory of capitalist exploitation in a way that is coherent to modern readers. Neither his “labor theory of value” nor dialectical jargon is employed, while Marxology is largely avoided. The aim is to develop the general framework for stating the necessary conditions for the existence and persistence of the kind of capitalist exploitation described by Marx.


While John Roemer [1982] tries to fit Marxian conclusions onto a Procrustean bed of mainstream theory, this chapter does not presume that vision to be correct. Instead, it allows for the possibility that concepts and theories developed using Marx's heuristics can be “rationally reconstructed” in terms developed by another paradigm as long as points of disagreement are noted.


Any theory that uses the word “exploitation” necessarily has a normative content, so that moral dimensions must be explored. But since this chapter's goals are primarily positive, no conclusion about the word's normative meaning is reached. However, I believe that Arjun Makhijani's [1992] phrase “taxation without representation” summarizes the Marxian theory as a first approximation: capitalist exploitation involves both macro-level capitalist supremacy (the worker's proletarianization) and micro-level subjection of labor by capital. Together, these are termed “institutional coercion” and make exploitation akin to taxation. Allowing this taxation to persist is workers' conscious submission.


As with all analogies, we cannot go too far. Capitalist exploitation is not the same as taxation: the coercion that allows it to persist has a different form than the armed force used by the state and can work in a decentralized way. Further, unlike most forms of coercion, capitalist “taxation” can encourage abundant growth of production.


This paper has seven parts, starting with (A) a quick survey of mainstream views of exploitation. Then, some of Marx's (B) non-mainstream methodology and (C) ethical dimensions are discussed. Next, (D) Roemer's effort is criticized and (E) my vision of Marx's theory, based in part on that critique, is introduced. Using Dymski and Elliott's [1989] terminology, this theory centers on (F) primary exploitation, the creation of a surplus-product. Finally, (G) the theory of secondary, merely redistributive, exploitation is explained and (H) this chapter's conclusions are summarized.


A. Mainstream Views.


“Exploitation” usually does not include simple theft, since the latter is not a persistent economic relationship. Rather, exploitation involves some “normal” aspect of the system, an institution. To Webster's Ninth New Collegiate Dictionary (1991), exploitation is “an unjust or improper use of another person for one's own profit.” Mainstream economists see exploitation as “improper” in liberal terms, i.e., serving individual or interest-group goals against universal goals (the public interest).


There are three main perspectives here. First, neoclassical economists do not follow the dictionary stress on selfishness since avarice drives markets, the apotheosis of their vision. Exploitation is instead a kind of market failure, a deviation from an ideal vision of capitalism (these days, the Walrasian system). The most common neoclassical exploiter is a monopsony [cf. Robinson, 1969] or a monopoly. Another exploiter is the agent who takes advantage of the principal who hires her, under conditions of asymmetric information [cf. Stiglitz, 1975]. A third exploiter is the free rider who takes advantage of others in the production of public goods.


In the second view, exploitation coexists with “perfect” markets: given a special position in society, an interest group can shift the distribution of income in its direction, impoverishing the rest, even though their role serves no reasonable purpose. While Henry George [1879] pointed to landowners, John Maynard Keynes [1936] saw rentiers as fitting this picture.


Third, combining the neoclassical emphasis on market imperfections with the second school's stress on the exploiters forming a segment of society that gains income without contributing, Milton and Rose Friedman [1980] see the government as exploitative. It is a monopoly run by a special interest group, regularly interfering with markets.


B. Methodological Issues.


As we shall see, Marx's theory combines aspects of the mainstream visions, mixed in a unique way. Before turning to the specifics, an overview of his vision is needed, since it differs from that of most economists, in terms of the surplus problematic, institutionalism, and political economy.


First, neoclassicals follow a scarcity problematic, seeing economics as being about the allocation of scarce goods among competing ends. As Ajit Sinha [1994] points out, Marx combined the scarcity approach with an alternative, the surplus approach: aggregate output typically exceeds aggregate costs of production, implying a surplus-product (for short, “surplus”). (See also Obrinsky [1983].) Scarcity and choice play roles in this view, but the emphasis is on the production and utilization of surplus in a growth-oriented context.


Unlike many of this school such as the followers of Sraffa [1960], Marx's surplus corresponds to surplus-labor done by workers to produce it. Within those institutions described as capitalism, the surplus-labor shows up in money form, as surplus-value (property income), which is divided among industrial and commercial profits, interest, rent, much of taxes, and the salaries of top managers of capitalist enterprises.


Neoclassicals see almost all income as “property income,” thus seeing wage income as the return on “human capital.” But here we are referring only to property that can be sold. Taxes are included as part of property income to the extent that the state is an integral part of the property system; its power is needed in order for such income to be received.


The reference to “institutions” points to Marx's second relevant deviation from the mainstream, institutionalism. He rejects methodological individualism, though individuals are very important: the societal process cannot be reduced to nothing but atomistic agents' tastes and decisions under the natural conditions they face. Many constraints are not natural but artificial and benefit groups that fight to preserve them. Further, institutions feed back to shape the nature of individual consciousness, preferences, and actions. Though originally created by humans and continually recreated by them, institutions take on a life of their own, alienated from individual wills, and indeed shaping those wills.


Nowadays, capitalism is a central institution of this sort: it is not natural but human-made and is defended mightily by those who benefit most from exploitation that is its hallmark. Whereas the mainstream sees exploitation as a disease from which capitalism can be cured (though perhaps with difficulty) or as a natural part of the human condition, Marx saw exploitation as being essential to that institution's operations and to be abolished with it: “the production of surplus-value, or the extraction of surplus-labor, is the specific end and aim, the sum and substance of capitalist production . . .” [1867: 298]. The mainstream types of exploitation are relatively secondary.


A third difference between Marx and most economists is that his method was one of political economy, seeing the politics/economics distinction as artificial. The distinction between the “state” and the “economy” (civil society) arose only with capitalism. Though this distinction has become fuzzy since the 19th century, with the rise of welfare state and fascism, social scientists still employ the distinction, inventing departmental boundaries which no longer correspond exactly to objective societal divisions.


More fundamentally, Marx saw “politics” – including conflictual relations among people –as not existing only in the state sphere, but also inside other institutions, including the capitalist firm. In Bowles and Gintis' terms [1986: ch. 4], economic sites involve political practice, just as the political site (the state) includes economic activity.


Following from this, Marx's theory of exploitation goes beyond the normal purview of economics: not only capitalism but other economic systems such as feudalism and slavery are “exploitative.” But our concern is exploitation only under capitalism. Non-capitalist exploitation is assumed to have been abolished.


C. Ethics.


Inevitably, Marx's use of the word “exploitation” has normative content. In the broad Marxian tradition, there are several ethical perspectives that should be noted. Three are associated with Marx himself, Makhijani, and Roemer.


Marx has been accused of being relativist or even amoral – even though his moral commitment is obvious. This perplexity arises from his distinctive approach. Writing in an era overloaded with often-hypocritical slogans of liberty, fraternity, equality, and so forth, he never developed an ethical system and tended to disdain moralistic incantations. As with the neoclassical unease with conceptions of “fairness,” he rejected the task of discovering fundamental ethical principles. But unlike neoclassicals, he never set up ideal models of the world (e.g., the Walrasian system) as a standard of comparison for judging real-world phenomena; he left utopian designs for others and described socialism and communism in the sketchiest of terms [cf. Marx, 1891].


Instead, he used the dominant ethics, with a twist: as Cornell West argues, Marx focused on “the discrepancy between the rhetoric of universal interests [i.e., liberal ethics] and the reality of particular class interests within the limits circumscribed by particular systems of production . . .” [1991: 92; emphasis suppressed]. This emphasis on the contradiction between theory and practice produces a contrast between the dominant moral standards, e.g., “the very Eden of the innate rights of man . . . Freedom, Equality, Property, and Bentham” [1867: 176], and Marx's subsequent description of the class conflict within capitalist workplaces.


Similarly, we see his value = price assumption in volumes I and II of Capital: value is used as a standard of “equal exchange,” the trading of equivalent for equivalent that liberals could accept. Marx saw such exchange as prevailing under “simple commodity production” (in which all workers are self-employed) when there are no monopolies or similar imperfections. Thus, “exploitation” can be defined in terms of the non-exchange of equivalents or “unequal exchange” (value price).


As developed below, Marx's “primary” exploitation refers to a very special kind of exchange. This is the “equal” purchase of labor-power (the workers' ability to work) at its value, which hides a non-exchange relationship in production. Capitalists receive the results of some of workers' actual labor without payment, since they pay for only the workers' labor-power.


Since, as developed below, this exploitation is based on institutional coercion (supremacy and subjection), it ends up being akin to “taxation without representation” (to employ Makhijani's [1992] phrase), where the mechanisms of taxation are largely hidden from view. As noted, the analogy with taxation is not exact but is instead a heuristic for clarifying Marx's theory. Since it is generally agreed that much or most coercion is bad, it gives a second ethical perspective.


This perspective helps deal with an important issue of exploitation, that of the use of the revenue earned by the exploiter for good purposes, such as on the improvement of technology or other potentially good projects. However, good use does not absolve the taxers from being subject to the strictures of the consent of the governed. While most would agree that a benevolent despot is superior to a malevolent one, that does not make the principle of popular sovereignty irrelevant. After all, power corrupts: the technology and “good” projects will be biased to serve the interests of the powerful, subjectively benevolent or not.


A neoclassical might question the phrase “without representation”: does not the capitalist system represent consumers, who vote with their dollars? But the actual result of the exploitation, surplus-value, is spent by only the capitalists or their agents, so that it is only their “dollar votes” that the taxers heed. Further, Marx argues that the system does not follow consumption at all: rather, the driving force is the hunger for more surplus-value that propels accumulation.


Going further beyond Marx, Roemer presents a third view, in terms of alternatives available to the exploited: capitalism is called “exploitative” because in theory there exists a situation where the workers would be better off and capitalists worse off [1982: 194], say, a worker-controlled economy or Roemer's own “market socialist” ideal [cf. Roemer, 1994, part IV]. This exploitation persists because of the “dominance” of the workers by the capitalists.


Stretching Roemer's usage a bit, exploitation persists because of collective action problems: such an alternative to capitalism is not superior from the point of view most workers as individuals or even as members of a large minority, if they go alone. However, workers might form a united coalition and break capitalist dominance, allowing a transition to a collectively-preferable situation benefiting each individual worker. A non-exploitative situation is thus seen as a collective good for workers while it is a collective bad for the capitalists [cf. Devine 1989A, 1994B].


Given the three ethical visions, some might not see Dymski and Elliott's “secondary exploitation,” the redistribution of surplus-value, as exploitation at all. In criticizing the beneficiaries of this redistribution, such as landowners and rentiers, Marx (unlike George or Keynes) never appealed to the “public interest” or issues of fairness. Besides, landowners and rentiers do not use force to get their incomes and might be seen as advantageous to capitalism's operations compared to the alternatives available to industrial capitalists. But to Marx, they are seen as gaining from primary exploitation, so these groups might be seen as recipients of stolen goods, where the industrial capitalists have done the stealing. Secondary exploitation, like primary exploitation, is defined relative to the exploited group, the direct producers.


Finally, despite these normative considerations, Marx's theory is mainly positive in its aim. Marx's purpose in Capital was not only to rally the workers against capitalism but to help us understand the “laws of motion of modern society.” His exploitation theory is central to that aim: surplus-value is the basis for capitalist accumulation, which is not only the increase in the stocks of means of production but also the extension of capitalist social relations of exploitation on a wider scale [1867: ch. 24]. More generally, in volume III of Capital, he stressed the mode of exploitation, i.e., the “specific . . . form in which unpaid surplus-labor is pumped out of the direct producers,” as revealing “the innermost secret, the hidden basis of the entire social structure and . . . the corresponding specific form of the state” [1894: 791-2]. Despite all of the qualifications that Marx immediately piles on this assertion, his theory of exploitation is central to his socioeconomic theory of history.


D. Roemer's Theory.


Before discussing that positive theory, recall that Roemer [1982] also attempted a modern restatement. He elucidated Marxian ideas about exploitation totally in terms of the hegemonic paradigm, applying Walrasian theory, jettisoning the surplus problematic, institutionalism, and almost all of Marx's political-economic vision. He concludes that we “should not search for the cause of exploitation in the Marxian sense in the labor process or at the point of production [as Marx did], but in the distribution of property” [1994: 3].


This total emphasis on property ownership is true only if Roemer has developed a coherent and robust theory of capitalist exploitation. It also misses the point that both the labor process and the distribution of asset ownership can be complementary parts of a coherent exploitation theory, along with such factors as accumulation. But turn to both the contributions and limitations of his models [see also Devine and Dymski, 1989; 1991; 1992].


1. The core of Roemer's vision of exploitation shares much with Marx. The capitalist minority's ownership of society's scarce means of production (capital goods) leads to their exploiting workers even though there exists no overt use of force, monopoly, monopsony, or the like. Being “exploited” can be defined either as working more hours than is needed to pay for one's livelihood, as with Marx, or as that the workers' average product, the APL, exceeds their real wage rate, RW.


But unlike for Marx, Roemer's workers are not forced by circumstances to work for capitalists: they are “free to lose,” voluntarily submitting despite the existence of equally lucrative alternatives. His simplest model, which captures the essence of his theory, is a version of W. Arthur Lewis's [1958] model of the modern sector of an underdeveloped economy facing unlimited supplies of labor coming from the backward “farm” sector. A labor-demand curve intersects a horizontal labor-supply curve: the upper area between the curves represents modern-sector profits, and for Roemer, exploitation by the capitalists, which determines the profit rate obtained by owning of the capital goods.


For Roemer, the different forms of property income (industrial profit, land-rent, etc.) are scarcity rents, so that his theory is quite similar to George's or Keynes's vision. For Marx, on the other hand, property income arises originally with production of a surplus by wage labor subjected by industrial capitalists, with the scarcity of capital goods only one necessary part of their origin.


This difference does not prove Marx right and Roemer wrong. But because Roemer's models do not capture the full extent of Marx's theory, they do not present a robust theory of exploitation. Since the capital goods are products of labor rather than gifts of nature, their scarcity is inherently temporary. Endogenous market forces should drive their price to their production cost, abolishing profits: the accumulation of capital goods raises the demand for labor, eventually abolishing its abundance, so that wages rise to annihilate profits. Since capitalists are getting “something for nothing,” they strive to get as much as possible until there is none left. Collective action problems prevent them from voluntarily preventing this over-accumulation that drives profit-rates to zero in equilibrium.


Accelerating the attainment of zero-profit equilibrium is the entrance of workers into the capitalist class by saving. Further, under Roemer's constant returns to scale assumption, any worker who goes into business for herself would suffer no cost disadvantage vis-à-vis existing capitalists, no matter how minuscule her business.


Even if we can avoid these problems, a new one arises since Roemer's models are arbitrary with hardly any of their parameters justified in a theoretical way. Most importantly, there is no explanation of the key question of why the APL exceeds RW, i.e., why a surplus is produced. Production and technology are unexplained in Roemer's theory. This is not surprising, since Walrasian models in general lack production: the scarcity problematic deals not with the actual creation of products but the distribution of a fixed amount of products amongst individuals through markets. For Marx, on the other hand, production is central [1867: chs. 7, 10, 13-15].


Relatedly, social relations of work are unexplained: we are never told about how the capitalists get workers to work rather than “shirk”; the Principle/Agent (P/A) problem is never addressed. The relations between capitalists and workers are simply a matter of exchange (hiring). Crucially, unlike for Marx, there is no unemployment to push workers to work: Roemer's reserve army is all-volunteer, since workers can earn the same income by going to the farm.


2. Skillman [1995] responded to the various critiques of Roemer, bringing the debate to new territory in one way. To him, Roemer's “isomorphism” theorem represents a major blow against Marx. He argues that it proves that capitalist subjection (subsumption) of labor is unnecessary to exploitation.


In this theorem, Roemer [1982: ch. 3] compares two imaginary “islands”: on the labor market island (LMI), capitalists hire workers to work with capital goods while on the credit market island (CMI), rentiers lend them to workers. In both cases, workers are “exploited,” in that APL > RW. Since the labor market does not exist on the CMI, it is not necessary to exploitation. Second, because unemployment is absent, it is also not required. Third, since a rentier can lend and then collect interest (and exploit) without any kind of supervision over workers' (debtors') activity, Skillman argues, no supervision of labor is required.


The conclusion that the labor market, unemployment, and subjection are unnecessary to exploitation is fallacious. Roemer assumes that capitalist exploitation in production is akin to the LMI, and that lending and receiving interest are akin to his model of the CMI. Then he shows that the two islands are almost the same in their results, so that capitalist exploitation in production is almost the same as lending. But the conclusion applies only if the basic assumptions are true. Neither author presents a sufficient argument to prove the validity of their assumption, i.e., the similarity between the “islands” and the real world.


More important, however, is not the patent unreality of the CMI and LMI models, but their similarity. Since the CMI's mathematical structure is so similar to that of the LMI, one can easily interpret it as a model of hiring labor. Alternatively, one might interpret the LMI as being a model of capitalists lending to workers: after all, there is no unemployment and also no explicit supervision of a labor-process. Roemer proves the “isomorphism” between models that are substantially the same. It would be much more valid and useful if Roemer had proved isomorphism between (1) an “island” with a labor process that is under direct supervision of capitalists and their underlings and subject to constant threat by the reserve army of unemployed labor (Marx's case); and (2) an “island” where the workers are not supervised, there is no involuntary unemployment, and workers can freely choose to produce their subsistence by working on “the farm” (Roemer's case). But neither Roemer nor Skillman attempts this task.


E. Marx's Positive Theory.


To Marx, as Dymski and Elliott argue, there is a clear distinction between the industrial capitalist's exploitation and that by the rentier. The former is “primary” exploitation, while the latter is merely “secondary” (corresponding to Wolff and Resnick's [1987: 146-51] distinction between “fundamental” and “subsumed” class processes). As suggested by the names, we need to understand the former (in section F) before we can understand the latter (in G). As usual with theory, reality is more fuzzy; however, an understanding of the basic difference helps us with the ambiguous (mixed) cases.


Marx described the issue of exploitation as one in which a representative capitalist lays out money (M) to buy commodities (C) which are then used to get more money (M'). The key issue is how and why M' > M, not only on the micro but also on the macro level. Profits due to “buying low and selling high,” i.e., unequal exchange amongst capitalists (in which price value) are seen as mere redistributions (secondary exploitation) that cancel out on the macro level and are so ignored [1867: ch. 5].


To understand what is at stake, examine the theory of profits in light of neoclassical distribution theory. (Marx's exploitation theory is his profit theory.) As Obrinsky [1983: chs. 4, 11] shows, this theory has largely involved the unexplained or undertheorized assumption that profits are positive. The existence of profits – and of property income in general – has usually been a non-problem. For many years, neoclassicals explained the existence of property income, profits, or interest (concepts that were almost always conflated) with an aggregate production function: the profit (interest) rate equaled the marginal product of capital goods. That this was an “explanation” indicates the low priority put on the problem: it is equivalent to simply assuming that profits exist. This is especially true since there is a clear distinction between the social role of owning capital goods (which evokes a flow of property income) and the goods themselves (which do not), so that the latter does not justify the former. This theory fell apart when Sraffa [1960] demolished the aggregate production function. Among neoclassicals, it has been replaced by Walrasian general equilibrium theory (except in textbooks and the sloppy work on “total factor productivity” and the “new growth theory”).


Another response has been to develop a theory of profits on the micro level. For the individual capitalist, “normal” profits are a given, and income needed to justify staying in the industry (with super-normal, economic, profits canceling out on the macro level). Still needed is an theory of why such profits exceed zero. One answer is to equate normal profits with the interest that could be earned on alternative uses of a capitalist's assets, since he could decide to leave “real” activity.


But this brings up the issue of why the interest rate (i) is positive. Here a graph is drawn showing an individual's intertemporal choice, trading off current consumption against next-period income. The budget line slopes down, with slope = -(1 + i). The individual decides how much to consume, as shown by this line's tangency with the highest indifference curve feasible. In equilibrium, the budget line is also tangent to the intertemporal production possibilities frontier below it. Often, the graph is assumed to apply equally at the micro and macro levels – or this issue is not addressed.


The reason for this digression into neoclassical economics is that Marx's theory of primary exploitation was trying, in modern lingo, to explain why the intertemporal p.p.f. for the economy as a whole has a slope so that i can exceed zero. If this is true for the economy as a whole, then it can also be so for an individual without that gain simply being at the expense of others.


Marx went beyond explaining why i > 0, to help us understand all other types of property income, including dividends, retained earnings, and the like. The issue was the rate of profit (r, total property income divided by invested capital), similar to what Keynesians might call the “average efficiency of capital” or the rate of return. Then, if we can explain why r > 0, we can also explain why i > 0.


Marx's surplus problematic explains how capitalist institutions create and reproduce over time a steep p.p.f. that constrains choice, leaving the issue of the actual choice made as secondary. In Marx's [1867: ch. 24, s. 3] discussion of the abstinence theory, for example, the choice of how much surplus-value to accumulate is subordinate to the actual existence of surplus-value; abstinence alone does not explain why r > 0. He looked at the societal mode of production in order to explain the creation of surplus by workers, an issue that neoclassicals and Roemer never address.


On one level, the etiology of surplus might be extremely easy to understand, but that simply brings up a second issue, the form in which surplus appears under capitalism. We can imagine that an independent producer working hard now, taking advantage of possibilities given by nature, producing surplus in the future. Whether or not surplus is produced is problematic, depending on definitions, since “hard work” raises costs and may mean a zero surplus. Even assuming a surplus, there is as yet no exploitation or profit in the capitalist sense. Marx's exploitation excludes such concepts as the self-employed producer “exploiting herself”: as in mainstream theory, exploitation is a regular association between people. To understand it, we have to discover how someone can capture the proprietor's surplus and also motivate her to continue to produce it (so that this is not simply theft). If we can explain this, then the intertemporal p.p.f. will be such that r > i > 0 for capitalists or money-lenders rather than simply for the proprietor.


The “Austrian” school, among others, has a completely different answer: a surplus arises and i > 0 (usually equated with r > 0) due to “roundaboutness” as when wine improves with age, so that “waiting” or “abstinence” is rewarded in the form of profit. Other mainstream authors suggest that “risk-taking,” “entrepreneurship,” or “capital goods” are produce profits. A full-scale critique of these theories is far beyond the scope of this paper (but see Obrinsky [1983]). One example will suffice to show that the mainstream factors explaining profit are necessary but not sufficient at the macro level.


A professional poker-player uses equipment, takes risks, delays gratification, engages in strategic behavior, tries new tricks and tactics (innovates), owns tools (e.g. marked cards), cheats, and earns large winnings and can even do so repeatedly. But no surplus results from such behavior; the gambler's winnings are simply redistributions from others with no new production occurring. Thus, the mainstream factors might be necessary for an individual to receive profits but are far from sufficient for these winnings not to be pure redistributions. We need to understand the societal conditions which allow the risk-taker (etc.) to get a profit without being a mere thief or bunko artist.


F. Primary Exploitation.


To explain the production of surplus which shows up as profit received by non-workers, Marx's theory adds involuntary elements to Roemer's purely voluntary exploitation. Since this coercion is part of the capitalist social structure, it fits with institutionalism. But not all of the story involves coercion: an individual can make a profit (secondary exploitation) without participating directly in the production of surplus-value (primary exploitation), just as a rentier can earn interest by holding treasury bonds without personally joining in the government's forcible extraction of taxes.


Institutional coercion's role can be understood if we examine capital's macro-supremacy and then its micro-subjection of labor. This reversal of Marx's order of presentation clarifies matters because it fits with his emphasis on the importance of the societal totality.


1. Supremacy. Marx emphasized “the so-called primitive accumulation” [1867: chs. 27-33], the historical and violent creation of the “fundamental conditions of capitalist production.” Under these conditions, workers are “free in a double sense” in that “neither they themselves form part . . . of the means of production, as in the case of slaves, bondsmen, &c., nor do the means of production belong to them, as in the case of peasant-proprietors” [1867: 714]. This capitalist supremacy (workers' proletarianization) goes beyond Roemer's “dominance” in that not only do workers lack control over the capital goods, but their own income-producing assets are insufficient to provide them with livelihood for any extended period. They are thus forced by their social circumstance to work for the capitalist class, no matter how unpleasant the job.


a) The two aspects of workers' freedom help produce the reserve army of unemployed labor. First, freedom from bondage allows this army to exist: there is no joblessness of the capitalist sort under slavery (for example) since the slave owner strives to use his property as intensively as possible. Second, workers lack access not only to the means of production but also to the means of subsistence (consumption goods), which they can get only by working for capitalists. Since they cannot go to the Roemerian “farm” to work for their livelihoods, the only alternative is to be unemployed (though this conclusion will be moderated in part (b)). Just as George [1879: book 7 ch. 2] saw the exclusion of workers from the ownership of land as putting them at a power disadvantage, it is central to Marx's theory of unemployment and capitalist supremacy. He cites Wakefield's theory that in land-rich colonies, it is necessary to forcibly keep workers from the frontier if profit are to be garnered by hiring them [1867: ch. 33].


The reserve army's existence represents a form of structural coercion that complements and reinforces the instrumental coercion (threats of firing, etc.) practiced by capitalist managers discussed below. Given the existence of the reserve army, “the laborer purchases the right to work for his own livelihood only by paying for it in surplus-labor . . .” [Marx, 1867: 515].


The ability of workers to quit or to reject job offers (part of the first kind of freedom) does not typically equate the marginal disutility of time spent at the workplace to the wage rate, as for neoclassicals. (By a Keynesian reckoning, therefore, the reserve army therefore involves involuntary unemployment.) Instead, the marginal utility of work-time is equated to the marginal disutility of becoming jobless, roughly the cost of job loss (COJL) [cf. Weisskopf, Bowles, and Gordon, 1983]. Labor's mobility tends to equalize wages and working conditions between sectors, subject to the usual qualifications such as the effects of unemployment. However, absent the frontier (Roemer’s “farm”), this equalization does not abolish the capitalist supremacy that allows exploitation.


b) There exist other ways in which workers can avoid working for the capitalists besides going to the frontier, but these turn out to be quite limited. Going into business for one's self is very difficult without sufficient assets and the ability to diversify one's investments. Workers lack that ability by definition [cf. Bowles and Edwards, 1993: 130]. It is true that there is a constant flux of workers into the self-employed petty bourgeoisie (and a trickle of these that become full-scale capitalists), but there is also a constant reflux into the working class as enterprises fail. Though some individuals move between class positions, the class structure itself remains, with the vast majority where they started.


The difficulty of a worker going into business persists in part because of the competition from capitalist firms, which can benefit more easily from economies of scale and scope. At the same time, the reserve army keeps the wages low, making it extremely difficult for workers to have enough income to save enough to have the “down payment” for going into business. (Usually saving for retirement or consumer durables such as houses is difficult enough.) Since this blocks their ability to earn income for saving from their own businesses, they can be stuck in a vicious circle.


Another place to get subsistence is from the state. But unemployment insurance benefits and other welfare-state programs are paid for by redistributions from employed workers, as serious analyses of the incidence of payroll taxes indicate. The actual benefits are historically contingent, based on struggles: politicians and social scientists work to make sure that welfare-state programs do not undermine the “incentive to work” for capitalists. In the absence of a strong counter-pressure from working people, as in Western Europe after World War II, it is unlikely that these programs will represent a viable alternative to selling labor-power to the capitalists. For the U.S., Miller [1989] finds that government programs that benefit workers were more than paid for by taxes on wages for 1952 to 1985 under three different sets of definitions [cf. Tonak, 1987].


One option is to live off one's family or community. Even though this may encourage conflict within the group, extended families, communities, and labor unions can and do provide (privatized) unemployment insurance. But this does not change the redistributive nature of that insurance.


Criminal activity is another possible source of unemployed workers' livelihood, but mostly represents a redistribution from other workers (whose incomes are limited by the reserve army): as is often noted the main victims of “street crime” are the poor. Its existence is also actively opposed by the state and by non-criminals.


c) Open unemployment (as measured in the U.S.) is less important than the more general phenomenon of the COJL. In many underdeveloped countries, there is almost no open unemployment: almost all property-less people who can work have jobs, no matter how pathetic. However, the large gap between wages in the capitalist sector and the income earned from petty trading, street-hustling, crime, and the like (and the non-existence of unemployment insurance and the like) imposes a large COJL on workers in the capitalist sector [cf. Schor, 1987: 175]. Even without open unemployment, as Joan Robinson once noted, the only thing worse than being exploited by a capitalist is not being exploited by one.


Before going further, a formal description helps us summarize:


COJL = C(U, Wgap) where C(0, 0) = 0                     (1)


where C is a positive function of both the open unemployment rate (U) and Wgap, the gap between the wage earned working for the capitalists and alternatives such as street-hustling and the dole (the workers' fall-back position). Now suppose that there exists a minimum COJL (Cmin) necessary to produce an adequate profit rate (from a capitalist viewpoint). At this Cmin, U and Wgap can substitute for each other in providing an adequate COJL. In the “first world,” the Wgap is minimal but U is positive, while in the “third world,” U is minimal and Wgap is large. Note the difference with Roemer's models, in which it is possible for U and Wgap to both equal zero, so that the COJL is zero.


Bowles and Edwards [1993: ch. 8] suggest two substitutes for the COJL in ensuring that a surplus is produced: social democracy and fascism. To simplify, these alternatives will be ignored until part (e) below.


d) We are far from concluding our argument in favor of Marx's theory. What preserves the COJL over time? We do not want to leave the COJL unexplained or simply assumed to equal or exceed Cmin. We want to avoid functionalism, the view that a phenomenon exists simply because it helps the capitalist class. If we can explain the COJL's persistence, we can also understand the difficulties of workers going into business or providing privatized unemployment insurance, since a high COJL makes these difficult. Our story also gives us an understanding of the limits on welfare-state measures.


Recall that in Roemer's model, the abundance of workers relative to capital goods is abolished automatically (and with it, profits and exploitation) if capitalists follow market incentives, investing their profits in scarce capital goods. Marx's solution is simple here, because his theory of exploitation is so different. For Roemer, profits are quasi-rents. For Marx, while the scarcity of these goods is not denied, the receipt of profits is not simply based on this condition alone. Instead, profits are akin to a tax that capitalists can impose because they control the economy's growth process, including the pace and nature of investment and technical change.


Suppose that C < Cmin or that this situation is anticipated soon, threatening a profit squeeze as workers are more able to push for high wages, cut back on work intensity, or even go into business for themselves. The capitalists' control over investment decisions means that they can and will cut investment spending (and the economy's growth rate) when profits are so threatened; this “capital strike” restores the profit-boosting situation of high unemployment by depressing the economy [cf. Marx, 1867: 619f]. Richard Goodwin [1967] formalized this process, likening it to the symbiotic and cyclical relationship between predators and prey. No conspiracy is needed: profit squeezes hit all capitalists to varying degrees, encouraging a certain unity of purpose without them having to collude in any way.


The Marx-Goodwin theory presumes that capitalists cannot simply add a mark-up to money wage costs (as in some Keynesian models). In Marx's time, the gold standard prevented the passing-on of money wage costs. Today, competition faced by individual capitalists and by nations on the world market prevent complete passing-on, at least in the short run. But in a modern economy with fiat and credit money, the short-term profit squeeze could encourage accelerating inflation instead of recession [cf. Bowles and Boyer, 1990]. Central banks' vehement opposition to inflation (reflecting the power of property-owners) encourages recession, reproducing Marx's classic scenario in a politicized form. Even without this opposition, accelerating inflation eventually encourages recession that reestablishes a COJL adequate to preserve profits.


This theory differs from that of Samuel Bowles and Herbert Gintis [1990A,B; 1993] and Devine and Dymski [1991], who stressed the origins of the COJL from purely microeconomic behavior, in essence, the payment of efficiency wages [cf. Akerlof and Yellen, eds., 1986]. While the idea that managers manipulate wage payments in order to motivate workers by raising the COJL is mostly reasonable and can cause unemployment by interaction with other non-Walrasian elements of the economy [cf. Devine, 1993A], the emphasis below is on the macroeconomically-created COJL. The role of other imperfections is largely ignored.


This cycle theory also explains why the normal situation of positive unemployment does not cause wages to fall to zero or enough to “clear” the market for labor-power, abolishing involuntary unemployment (as in many textbook stories) without bringing in any microeconomic explanation such as sticky wages: when wages fall sufficiently relative to the APL, profits are boosted and with them, accumulation and the demand for labor-power. (Microeconomic theories of wage stickiness are not wrong or irrelevant; they are just not necessary to the story.)


Threatened by high wages or low work effort, capitalists can also institute labor-saving technological change [Marx, 1867: ch. 25, s. 2], while seeking new labor-power supplies at home and abroad. Indeed, they are driven to seek ways to replace labor-power by the battle of competition amongst capitalists and the possibility that wages will rise in the future. Further, they may use the state, as when capitalists push the state to change immigration laws to import “guest workers” or braceros when labor-power is scarce.


Governments are sites for intraclass competition and inter-class conflict and need not always serve the capitalist class interests. But a government that attempts to lower the COJL has to face the limits created by the profit squeeze/capital strike and capitalist control over technological change. If a government promotes true full employment or institutes programs seen as undermining profitability (even if they are not truly so), eventually investment stagnates or capital flees to more profitable shores. This induces recession or increased inflation, falling tax revenues, foreign exchange problems, and/or long-term productivity-growth stagnation. These undermine popular support for the government and push it out of office or to change its policies [cf. Kalecki, 1972; Block, 1977; Lindblom, 1982]. Examples of such “automatic destabilization” include those against Allende in Chile and Mitterand in France.


e) The receipt of profits is not guaranteed by capitalist control of investment and technological change, since capitalists can get themselves into serious crises (such as the Great Depression, in which the reserve army became “too large,” hurting the realization of profits [cf. Devine, 1994A]). Further, workers can revolt or pressure the state in a way that lowers this tax. But in “normal” (non-crisis) times, capitalists will receive positive profits, because crises are typically not as severe as the Depression and workers compete against each other. On the latter, the competition amongst workers creates the collective action problem referred to in section C, which is reinforced on a cultural and ideological level by the illusions created by competition: workers often side with “their” employers in the battle of competition, or with “their” nations in international competition, in hopes of getting benefits. This divides and rules them as a class, legitimating the system and preserving capital's supremacy over time.


Since it is based on the fact just highlighted, i.e., that the existence of exploitation is partly based on workers' conscious submission, turn to the possibility of the social-democratic contract that is one substitute for the COJL: workers might accept exploitation in exchange for a relatively good deal in terms of the distribution of benefits. Capitalists put up with this situation because of the security that the contract provides.


The possibility of such a contract is based on interpretations of the history of Sweden and some other countries; to some extent this kind of contract has existed in the “primary labor markets” of the rest of the capitalist world. In these cases, the type of consent based on the COJL and divide-and-rule stressed above is replaced in part by a more democratic and popular form of legitimation. “Taxation without representation” begins to edge toward “taxation with representation,” to stop being exploitation in a normative sense. In the Swedish case, for example, workers had some control over the way in which the surplus they had produced was utilized.


However, such a “social contract” is partial, conditional, and temporary. It is partial because it excludes many workers, either those who produce imported inputs or the “guest workers” who do menial tasks within the nation (and are exported during bad times). Even if democratic legitimation applied in Sweden or the primary labor markets, the divide-and-rule type of legitimation still applies for capitalism as a whole.


It is conditional because as soon as the social-democratic contract stops being profitable to the capitalists, it leads to a capital strike or to capital flight to areas that do not constrain profit-seeking as much. This implies that the contract is temporary: the normal globalization of capitalism (seen dramatically in the last 15 years) means that greater opportunities and inducements for capital flight arise, so that the contract becomes more likely to break down.


Alternatively, the social democratic system could be transformed into socialism through the abolition of capitalism. Not only has this never happened, but it seems unlikely. The social-democratic contract was won as a compromise because of the strong organization and socialist consciousness of workers pushing for something even better. But social democracy itself is built on the subordination of this organization and consciousness to compromise and so slowly undermines the possibility of future working-class victories.


In the current global economy, most of the old social-democratic programs are under attack and are falling apart, while primary labor markets are shrinking. This suggests a return to the strategy of relying on the COJL and divide-and-rule to preserve exploitation. However, we cannot deny the possibility of social democracy, at least a temporary one, in the far or intermediate future.


Fascism is the second alternative within capitalism to the COJL: state violence can be wielded to directly force people to work [cf. Kalecki, 1943]. This type of exploitation is relatively transparent and needs no profound analysis. But just as with slavery, fascist rule seems inadequate for promoting high-quality work, productivity growth, and the like, since forced labor is usually poor labor (a side-effect of relying so little on conscious consent). A fascist economy would not do very well in market competition with non-fascist capitalist powers and so tends toward autarchy and military competition. The societal imposition of a COJL on workers seems a more “efficient” way of promoting capitalist accumulation. The fascist solution has been imposed only in situations of severe economic and social crisis, in which the reproduction of capitalist social and property relations were threatened.


In sum, exploitation's existence requires societal impetus on workers to “pay their taxes” to the capitalist class, where the COJL is a substitute for the direct application of force in extorting these revenues. With either of these, a certain amount of “consent of the governed” is required for the system to be perpetuated over time, though fascism requires less of that. In fact, such consent can substitute for the COJL and force, at least temporarily as under social democracy. Now we turn to the impact of this type of societal pressure, on the actual production of surplus on the micro level.


2. Subjection. The societal pressure to “pay taxes” need not cause the actual production of surplus, just as capital's supremacy is only part of the story, necessary but not sufficient. For Marx, the “anarchy in the social division of labor,” i.e., the unplanned and relatively harmonious co-operation of different industries and firms through the market (in the absence of fascism or social-democratic planning), is complemented by “despotism in that of the workshop” [1867: 356]. This subjection of labor by capital (both formal and real) is what was earlier termed instrumental coercion. It refers to the direct capitalist rule in production, as opposed to voluntary exchange (the Walras/Roemer story): in this microeconomic situation, the coercer can impose a significant cost on the coercee for not complying with his wishes. The existence of a COJL > Cmin makes employers' threats credible, unlike in a Walrasian model.


a) Capital's supremacy cannot be sufficient, since it does not explain the existence of surplus: capitalists could use their macro-societal power to induce a pure redistribution that reduces the standard of living and/or net worth of workers. This would make capitalists akin to a kleptocratic elite such as that led by Zaire's Mobutu or seen in the more extreme libertarian statements about the U.S. government: the state extorts taxes from the people in order to waste the revenues on luxury, war, and further tax collection, steadily depressing peoples' standard of living and/or net worth. This is similar to Marx's vision of pre-capitalist usury which


“impoverishes the mode of production, paralyzes the productive forces instead of developing them, and at the same time perpetuates these miserable conditions in which the social productivity of labor is not developed . . .” [1894: 596].


This hurts the ability of the system to grow, so that it degenerates into a stationary state. It thus differs qualitatively from the capitalist exploitation that Marx explained, which was part of a dynamic and expanding system, one that can produce new real assets rather than simply redistributing them from others.


To rule out this purely redistributive and thus self-limiting exploitation theory, to define and thus understand the production of surplus, some standard of comparison is needed. Marx assumed that the workers are paid enough to reproduce themselves as a class over time, i.e., according to the culturally-based and historically-determined subsistence level. This notion captures the fact that workers' needs are more than just a matter of subjective wants but a matter of survival, not just physically but as human beings in society. If paid at subsistence, they need not accumulate debt or dip into savings in order to survive.


Marx stated the micro-etiology of exploitation as follows: surplus is produced if the length of the working-day exceeds the value of labor power (VLP), i.e., the socially necessary abstract work-time required to produce the daily wage determined by the subsistence level. Payment at the VLP is a key part of the equal exchange (price = value) assumption discussed in section C: exploitation of workers is not defined as payment less than competitive free-market prices for their labor-power, but rather as occurring despite their being paid the value of their labor-power. This complements the assumption that capitalists do not simply profit at each others' expense, i.e., the focus on the macro creation of surplus rather than mere redistribution of a given product.


As seems appropriate in our century, use an hour rather than a day as the unit for which workers are paid: the value of an hour of labor-power equals the amount of socially-necessary abstract labor time needed to produce the hourly real wage. So the VLP is the real wage (RW) for an hour of labor-power divided by the average output produced by the labor done during an hour of labor-power hired (output per hour hired, the APL).


To use modern language, avoid further reference to the VLP and allude to RW and APL alone. Marx's issue can be restated as whether or not APL > RW. In turn, the APL is explained by the following tautology:


APL = (output/labor done)(labor done/labor-power sold) = q e                        (2)


where q is the effectiveness of labor done and e is the intensity of labor or the “degree of effort” [cf. Devine and Reich, 1981]. The ability of the capitalists to extract a surplus and to exploit workers thus depends on technology and management systems, plus the ability of capitalists to depress wages.


Initially, changes in q (reflecting current technology and management systems) will be largely ignored. Given this, the production of surplus depends on capitalist managers' ability to evoke sufficient effort from their workers. The issue of technical change will be addressed under point (e), when “real subjection” is discussed.


Many have criticized Marx's assumption of the exogenously-given RW. Instead, Andrew Friedman [1977: 267-9], Michael Lebowitz [1992: 119-20], and others argue that the level of the RW is the product of an endogenous class struggle. To assure generality, therefore, we should go beyond Marx to consider the case in which wages are set by steadily changing social needs and class struggle in the context set by the supply of and demand for labor-power – and can change in response to exploitation.


Marx's micro-theory of exploitation is not dependent on the given RW. In the theory of the profit-squeeze/capital-strike cycle discussed above, wages are the “dependent variable” [1867: 620] rather than the independent variable that they were earlier in the book. The normal existence of unemployment “confines the field of action of this law [of the supply of and demand for labor-power] within the limits absolutely convenient to the activity of exploitation and to the domination of capital” [1867: 639]. That is, the cycle, labor-saving technical change, and the like insure that low-COJL situations that squeeze profits are abolished.


b) Now assume RW constant, with its current value determined by what was received in a previous period (roughly, their customary standard of living, as in Bowles and Edwards [1993: ch. 4]). Given this, return to the issue of how the capitalist or his hired managers can succeed in getting more out than they put in to hire workers during this period. Marx's answer is simple: workers are not paid for the actual labor they do but instead for the cost of bringing themselves (their labor-power) to work; their actual labor has no price on the market because it is not marketed. Further, they labor more intensively than necessary to exceed the labor needed to produce the RW, while the capitalists can capture the benefits of management systems and technology that raises labor's effectiveness.


This theory is not contradicted by marginal productivity theory (assuming that such theory is valid). A profit-maximizing capitalist equates the marginal productivity of hiring labor-power to the RW, the price of labor-power. But the wage is not determined by the MPL. Rather, as far as a competitive capitalist is concerned, labor-power market conditions (the COJL) determine the wage; then, the capitalist determines employment given that wage. Even if the RW is constant, the COJL helps to determine the degree of effort, the amount of actual labor done, and thus the marginal productivity of an hour of labor-power hired.


c) Given the coercion allowed by the COJL, a central question is why it is that workers get paid for providing labor-power rather than labor. Workers sell their time on the market and it is that time (labor-power) which has a price; once they have struck a deal, they are under the capitalist's authority and their product is the latter's property. The social relations within the workplace are distinctly non-market in character, involving much more than individual exchanges.


Recent research fits with Marx's vision: as Williamson [1975], Herbert Gintis [1976], and others explain, most jobs involve more than hiring an individual to do a single and simple task. In the former's terms, complete contingent futures contracts between employer and employee are impossible. This fact's general acceptance has encouraged economists to increasingly distinguish between “labor sold” and “services rendered” and to worry about the P/A problem. With the workers as agents and the capitalist manager as principal, the principal must discover how to get the agent to work hard in the real world of asymmetric information.


Otherwise, embezzlers steal from their bosses while slackers and petty bureaucrats gum up the works, making operations unprofitable. Without dictatorial supervision, in the Marxian view, no profits will be received since inadequate amounts of work will be done to do more than pay for the wages.


Skillman [1994] argues that the P/A problem does not add anything significant to exploitation theory; rather than explaining why profits are positive, dictatorial supervision explains only why profits are higher than with self-supervised workers organized through markets. But this argument that subjection is merely epiphenomenal misses the limitations of the neoclassical P/A literature. The usual P/A article starts with a totally hypothetical “full information” Pareto-optimal economy and then introduces asymmetric information. But Marx's standard for comparison in his exploitation theory is not the Paretian ideal, but rather a non-exploitative ideal of simple commodity production (posited by economists of his time). Marx, like other economists of his period, would have considered the notion of perfect information to be silly and irrelevant. More importantly, the neoclassical view simply avoids the theoretical issue of the origins of surplus (see section E). If anything, its existence in the full-information model is implicitly assumed, with P/A problems leading to the imperfect realization of these profits. Though on the technical level P/A models are unobjectionable, if we want to explain positive profits we cannot start by assuming what we are trying to explain.


Skillman's [1994: 34-9] effort follows exactly this pattern, except that he explains profits' existence with Roemer's “full information” theory. But as noted in section D, Roemer's theory simply assumes that APL > RW, while the scarcity of capital goods is insufficient to explain the actual production of surplus as opposed to mere redistribution of an already-produced surplus. We need the more complete picture of institutional coercion to explain the existence of profits.


Indeed, the P/A problem undermines Roemer's models. Both his Credit Market and Labor Market Islands lack Marines to hit the beaches to enforce contracts. Though the state is implicitly assumed to do the job, it lacks the resources (including information) needed to enforce each and every contract, even if it were possible to fully specify contracts ahead of time. State enforcement is thus at most complementary to the efforts of the contract-enforcers that capitalists hire or mechanisms that they use (including the COJL) to ensure contract compliance. Without the subjection of labor, contract compliance breaks down.


In sum, if we are trying to explain rather than presume the existence of a surplus, the P/A problem stops being an example of “inefficiency” defined relative to an imagined ideal, but is rather defined as a conflict over the amount of production, a variable: in the simplest story, workers (ceteris paribus) want to produce only what is enough to cover their wages, while the capitalist wants them to produce as much beyond their wages as is possible, of course balancing the cost of evoking that effort.


d) We must now turn to an important practical concern that goes beyond the P/A problem and limits the effectiveness of such individual solutions to the problem as “bonding” (as happens with individual contractors) and the paying of efficiency wages to individuals. Work occurs in groups, with the individual worker being part of the “collective worker.” This means that there are so many interdependencies – externalities within the workplace – amongst the various workers and their work that it is extremely difficult if not impossible to separate out the contributions of individual workers in the “team.”


Alchian and Demsetz [1972] see the capitalist as a supervisor, “monitoring” the teams' inputs and outputs to avoid “shirking.” The capitalist receives the lion's share of the “residual” gained through such monitoring. However, these authors do not note the fact that this residual accrues to the capitalist not simply because of any supervision but also because of ownership rights [cf. Mandel, 1977: 53-4]. With different ownership rights, the residual might be received by (for example) a workers' co-operative, in which workers supervise each other. Even given capitalist property rights, Alchian and Demsetz miss the way in which the reserve army of labor (and the COJL) allow the employer to threaten workers to get them to work to produce a surplus.


The collective nature of work suggests that workers have group interests (the production of collective goods, such as a leisurely pace of work) which conflict with the goal the capitalist manager. This conception differs from Williamson [1975] or Stiglitz [1975], for whom the capitalist/manager and workers share a single collective good. But these authors' vision is not totally wrong: while there exists a basic conflict of interests between capitalists and workers, in many cases workers (fearing unemployment and hoping for raises) and their employers ally against the employers' competitors.


In view of the common conflict between profits and the workers' collective good, however, the capitalist tries to encourage the free rider problem amongst workers, in order to divide and rule his employees [Marglin, 1974; Devine and Reich, 1981]. Friedman [1977] and Edwards [1979] emphasize the need for organized management systems to control workers, usually incorporating the divide-and-rule principle. In this vision, the capitalist control of production is more than simply a matter of setting up an incentive system and enforcing it; rather, it is a (micro) political problem, involving potential and actual conflicts of interest. In this management relationship, the submission that was necessary to the perpetuation of supremacy is also needed to reproduce subjection over time. A rebellious workforce willing to throw spanners in the works or to go on wildcat strikes will not allow subjection or exploitation.


In response to all of this, the case of piece rates is often mentioned, as a counter-example. But as noted in Mathewson [1931] and Devine and Reich [1981], payment of piece wages are very limited in application and further, do not prevent political problems. Besides the problem of actually isolating an individual's contribution to the total product, one problem with market-like piece-rates is that the quality of the pieces is impossible to contract ahead of time. The other is that workers band together to prevent the downward adjustment of piece-rates that so often happens when they work harder. Unlike competitive farmers who face a somewhat similar situation, workers find it relatively easy to do so because they often work together in a single workplace and thus find it easier to communicate and co-operate with each other, even if that co-operation is largely tacit while on the job. Both of these considerations indicate that in addition to the piece-rate, workers must submit to the power and authority of the employer. Piece rates and capitalist management/control systems are complements rather than substitutes.


e) As Marx realized, real wages may not be low enough (given technology and thus the effectiveness of labor) to actually allow for exploitation to occur. Given the absolute limits set by physical subsistence needs, it may be impossible to reduce the RW. Increases in the intensity of labor may simply induce a higher cost of reproducing labor-power, preventing the creation of a surplus. With undeveloped forces of production, therefore, it is quite possible that the degree of exploitation would be extremely low or even zero [Marx, 1867: 511].


However, once capitalism takes hold of the production process – what Marx called the real subjection of labor by capital – it promotes labor productivity growth, helping to assure the reproduction of exploitation over time. This real subjection is more stringent than mere supervision (formal subjection): it involves the capitalist control over technology and its introduction on the micro-level. Capitalists are no longer dealing with workers who bring their own tools and knowledge to work (as with many craft-workers in even in today's construction industry); instead, they are replacing their own machinery with new and different capitalist-owned machinery and are monopolizing the technical knowledge developed by hired scientists and engineers.


Under real subjection, the capitalists can install the technology they want, one which raises the effectiveness of labor (q) without directly causing increases in real wages. Such labor-saving technological change also helps create a reserve army of labor (technological unemployment) and the economies of scale and scope that block workers' entrance into business.


Once established, the normal production of surplus can then be the basis for a virtuous circle (from the capitalist viewpoint) of accumulation and the creation of further profits, even in the face of rising wages. After the initial stage, in which more overt force may be needed, the somewhat automatic profit squeeze/capital strike cycle and labor-saving technical change can make sure that wages do not rise too much to allow the production of profits, since the real subjection of labor implies that the capitalist controls investment and technology. Such control is not present if production simply involves workers deploying their own tools.


At the same time, automation and similar techniques are used to lower the skill content of the work process, taking control away from craft-workers [cf. Braverman, 1974]. In the end, workers may be relatively skilled, but both the machines and the knowledge are supplied by the capitalists or their institutions (such as technical schools) so that workers have no special advantage arising from their individual attributes. Unlike many craft workers in the building trades, for whom the ownership of skills and tools conveys certain advantages even in the absence of a union, under the real subjection of labor by capital, workers must band together to resist capital or to bargain collectively.


Despite the introduction of machinery and the deskilling of craft-based labor, exploitation remains a political problem at the micro-level. Even “semi-skilled” and “unskilled” workers can devise ways to block the introduction of new machinery or to undermine its effectiveness if the capitalist has not solved the political problem. This can be done either via collective action (unions, wild-cat strikes, etc.) or if some subset of workers can find a strategic location in the management structure (such as a bottleneck in the assembly line) or even by sabotage. Submission is still needed for exploitation to be reproduced over time.


f) Subjection is necessary but not sufficient; it is complementary to capitalist supremacy. First, subjection depends on supremacy, the lack of workers' options besides working for capitalists; the COJL gives subjection its force. In turn, capital's supremacy involves capitalist control over both investment and the introduction of new technology.


Completing the circle, these types of control require subjection. In an imaginary economy totally organized by self-employed craft-workers using their own tools and knowledge, capitalists would have no direct control over real investment or technological change. If there were no COJL, the craft-workers would face no disadvantage vis-à-vis contractors and bankers, so the latter could not force their decisions concerning investment and technology. With no direct control by capitalists over the workplace, there would be no way to get the craft-workers to produce a surplus or to hand it over to the capitalists.


G. Secondary Exploitation.


The six mainstream kinds of exploitation mentioned in section A fit with the Marxian theory. As pure cases, they represent “secondary exploitation.” However, there exist important “impure” cases, which are (a) mixed with and/or (b) contribute to primary exploitation. So, the simple case of “all exploitation arises out of industrial production” needs to be modified. Even so, primary exploitation sets the context in which secondary exploitation occurs so that our discussion above helps us understand that below.


1. For monopsony and monopoly, market prices deviate from competitive prices for long periods. Some capitalists have special positions that prevent the entry of competitors, so that their profit rate exceeds the average for the economy as a whole; they receive “super-profits.” Others have work forces with a limited ability to seek jobs elsewhere and are able to pay lower wages for given working conditions than on average, allowing the capitalists to “super-exploit,” to participate in exploitation more than others.


In the latter case, if capital mobility is unblocked, the monopsonist will not receive an above-average profit rate in notional equilibrium, as other capitalists rush in to take advantage of below-average wages. But it is possible that monopolistic and monopsonistic privilege are combined (as with the company town), in which case the company will receive super-profits.


Unless the monopoly or monopsony is extremely important in the economy as a whole (which seems unlikely given the magnitude of world capitalism), it does not affect the over-all degree of exploitation. If this is unchanged, the monopolist or monopsonist gains super-profits only from other capitalists, who receive sub-normal profits.


If monopsony and monopoly are extremely important to the operations of the economy, the capitalist economy has a significant admixture of feudalism or some other form of overt servitude; alternatively, fascism exists. If these phenomena are the normal way in which business is operated, then they would raise the overall degree of exploitation, though (as mentioned in part F.1.e) it may hurt long-term growth of labor productivity.


2. The P/A theory of exploitation has three levels. First, capitalist agents such as the top managers of firms are able to capture part of societal surplus-value in the form of exalted salaries, perks, stock options, and/or embezzlements, reducing the incomes of stock-owners and creditors and getting more than justified from these principals' perspective. Their ability to do so depends on the primary exploitation of the direct producers. Though they may use their power in a way that limits the size of the surplus produced, that would undermine their income in the long run.


Second, from the point of view of these managers, the members of the workforce are the “agents.” The latter can refuse to produce a surplus-product. Further, if we use the customary standard of living to define the RW and, given the APL, the surplus-product, workers can capture part of the surplus-product by pushing to earn more than their customary level. As discussed above, this represents a central part of subjection, i.e., the constant political problems that it involves.


Workers are sometimes seen as earning “employment rents,” which some see as exploitation of capitalists by workers. But this is looking at matters backwards. Workers fear the COJL and thus produce a surplus. Then they may be able to win back some of the surplus that they created. They are not the exploiters.


Third, individual workers can gain at each others' expense. While this “opportunism” [Williamson, 1975] can benefit individual workers in an “exploitative” way, this competition among workers (if kept within bounds) also divides and rules the work-force and encourages the production of profit: the back-stabbers in “office politics” can help the managers and owners by motivating workers to work hard to rise in the hierarchy rather than fight to attain collective goals. To management, such behavior must be kept within limits; doing so is part of the political problem that capitalist exploiters must face.


3. The free-rider problem is best explained in terms of a specific case, the destruction of nature and the exploitation of future generations. It is part of a Marxian theory of competition. In their efforts to survive the competitive battle, capitalists actively seek to internalize external benefits and externalize internal costs, often in innovative and highly creative ways that demonstrate an ability to plan ahead and delay gratification in the name of profit [cf. Hunt, 1980; Devine, 1993B]. Successful development of new ways to dump costs on others can give a company a competitive advantage and temporary super-profits that can be capitalized, allowing for further growth. Of course, on the macro-level, this behavior can eventually lead to the destruction of the natural conditions allowing capitalist exploitation to exist. But a competitive capitalist cannot act on such knowledge. In other words, the free rider problem is part and parcel of the way in which freely-competitive capitalism works.


4. Just as for George, Marx saw the land-owning class as crucial. For both, land-owners capture some surplus because they control scarce gifts from nature. The major difference is that George, having a different theory of profits, stressed a basic harmony between the urban classes. For Marx, on the other hand, land-owners are able to grab a part of the aggregate surplus-value, representing a redistribution from industrial capital, where surplus-value is produced.


The landowner may also be an industrial capitalist (hiring farm workers to harvest the crop, etc.) In this “impure” case, the landowner is contributing to the aggregate surplus-value. Further, as noted, the extent to which workers are landowners themselves (rather than being pure proletarians) limits the degree to which they can be exploited by normal capitalist means.


5. Marx's attitude toward rentiers was similar to that of Keynes, without sharing the latter's admiration of the entrepreneurs (industrial capitalists). The rentiers gain a share of the societal surplus-value in the form of interest from their control of loanable money-capital. But industrial capitalists are the ones who induce the workers to produce the surplus in the first place. For Marx, unlike for Roemer, the rentier's pure money-lending does not create the basis for the receipt of property income.


However, there exist cases of impure money-lending, as when a loan-shark forces a worker to work hard and long. But the loan-shark is not a rentier but instead shares some characteristics of an industrial capitalist, i.e., the direct participation in the subjection of labor. As the P/A literature on banking points out, much money-lending shares this characteristic (to varying degrees): the loaners ration credit, snoop into prospective debtors' personal lives, and impose conditions concerning collateral and the like [cf. Mishkin, 1992: 171-8].


Given capitalist supremacy, i.e., the threat of the COJL and the absence of alternatives, this type of behavior might induce an “I owe, I owe, so it's off to work I go” response on the part of workers, complementing the industrial capitalist's subjection, which (if common enough) raises the aggregate degree of exploitation. Without supremacy, the dependence of workers on industrial capitalists for their livelihood, however, the money-lender's hold over workers would be nil. So again, the institutional coercion of labor is central to the exploitation story.


Even many or most of the independent producers living on the edges of capitalism (the petty bourgeoisie and home-workers) are subject to capital's supremacy. In the extreme, they own neither the means of production (including raw materials) nor the means of subsistence, or have very limited access to these, and must gain access to them through the agency of the capitalists. These workers may not even own the homes they work in and may be borrowing money to finance consumption. They are also threatened by the existence of the reserve army, as a merchant can “fire” home-workers and hire replacements. They may also be competing in product markets with wage-workers operating much more advanced equipment, i.e., with low-cost commodities produced under conditions of the subjection of labor by capital. Of course, some independent producers are not in this kind of bind: these are not exploited and instead are on the edge of joining the capitalist class.


6. The final theory of exploitation centers on the role of the state. Contrary to the popular image of Marx as a super-statist, he was hostile to the state, striving for its “withering away” [cf. Draper, 1977, 1978, 1986, 1990; Devine, 1994B]. In Marxian theory, the state is central to the existence of both supremacy and subjection over labor, since it helps create and preserve capitalist property rights. Under feudalism and other pre-capitalist modes of exploitation, the use of armed force (a characteristic role of the modern state) is inseparable from the “economic” relations between the lord and the serf: the former simultaneously collects rent and taxes, two incomes that cannot be separated. But under 19th century capitalism there was a societal division of labor between those sectors “monopolizing the legitimate use of force” [Weber, 1918] and those that engage in purely economic activity. Taxes on capital may be seen as partly a matter of redistribution of surplus-value to the state in return for services rendered to the capitalist class.


The state can also use its force in order to raise the degree of exploitation, as under fascism, or may emulate capitalists by directly engaging in production, as with state capitalism (e.g., the U.S. Tennessee Valley Authority). To some extent the modern welfare state (like social democracy) contributes to the production of surplus-value by legitimating capitalism and its work relations. In these cases, the societal division of labor is fuzzy compared to classical liberal capitalism, so that taxes are not simply a deduction from (redistribution of) surplus-value.


On the other hand, the state is “relatively autonomous”: that is, it is not under complete capitalist control and does not always serve the interests of the capitalist class. It might serve only a fraction of that class, or there may be a P/A problem, this time with the state bureaucracy as the agent. As Friedman would be the first to point out, the state may go “too far” in the collection of taxes, feathering the government bureaucrats' nests with no obvious benefit to the capitalists or anyone else besides these parasites.


Third, and finally, the working class's political activity can (and sometimes does) succeed in winning benefits from the state in excess of the taxes they pay. While this also contributes to the system's legitimation and, as noted above, can thus raise the amount of surplus produced, to some extent workers can reduce the surplus left over for capitalists. In fact, these two effects may work together (as in Sweden for decades), where the state works to both increase the surplus through legitimation and decrease the surplus via distribution to workers, without hurting and perhaps actually helping profits. But in the end, as argued, social democracy cannot escape the laws of motion of capitalism.


H. Conclusion.


The point of this chapter is to reconstruct Marx's theory of exploitation, improving on previous efforts in this vein, applying his surplus problematic, institutionalist perspective, and political-economic vision to restate his theory in modern terms. The most profound type of exploitation, the one explained by Marx, involved not simply the redistribution of existing assets but the creation of new ones. The basic theory is both macro- and micro-institutional, involving coercion. At the same time, I gave some idea of the ethical basis of using the word “exploitation”: it is a matter of a conflict between the stated ideals of freedom and equality and the actuality of subjection, between the stated ideal of equal exchange and the actuality of unpaid surplus-labor. It might be likened to taxation without representation.


Whereas Roemer's effort to summarize Marx's positive theory of the capitalist exploitation of labor in a Walrasian framework severely limits the role of politics, the above summary defense of Marx's theory is profoundly politicized on the complementary macro and micro levels. Non-armed coercion based on the threat of costly job loss and the dictatorial rule of the workplace by the capitalist, including the real subjection of labor, the ability to revolutionize production by introducing new technology, form the micro-economic basis for exploitation. On the macro-level, the exclusion of workers from producing for themselves using the gifts of nature and the products of their labor (the means of production) creates the cost of job loss and the conditions allowing microeconomic subjection. Finally, there must be ideological submission by workers: capital's institutional coercion of labor must be legitimated in the eyes of the workers if it is to continue. Further, severe economic crises must be avoided. As fits with Marx's emphasis on the nature of capitalism as historically limited, the persistence of exploitation is not guaranteed.


The discussion of secondary exploitation leads to the conclusion that the institutions allowing purely redistributive exploitation can be mixed up with and can sometimes even reinforce the production of surplus-value.


Though it goes far beyond Marx's methodological framework, Roemer's normative definition cannot be avoided. Any kind of criticism of “capitalist exploitation” seems somewhat futile if there exists no alternative, not only non-capitalist but non-exploitative, way of organizing production beyond reversions to pre-capitalist modes of production or to Stalinist economics. Given the importance of conscious consent, this is not just an academic matter: the absence of an alternative is clearly one of the factors that legitimates capitalist exploitation and allows its perpetuation.



Akerlof, George A., and Janet L. Yellen, eds. 1986. eds. Efficiency Wage Models of the Labor Market. Cambridge, UK: University Press.


Alchian, Armen, and Harold Demsetz. 1972. Production, Information Costs and Economic Organization. American Economic Review. 62(5), Dec.: 777-95.


Block, Fred. 1977. The Ruling Class Does Not Rule: Notes on the Marxist Theory of the State. In Thomas Ferguson and Joel Rogers, eds., The Political Economy: Readings in the Politics and Economics of American Public Policy, Armonk, NY: M.E. Sharpe, 1984: 32-46.


Bowles, Samuel and Robert Boyer. 1990. Labor Discipline and Aggregate Demand: A Macroeconomic Model. American Economics Review. 78(2), May: 395-400.


Bowles, Samuel and Richard Edwards. 1993. Understanding Capitalism. 2nd ed. New York: Harper Collins.


Bowles, Samuel and Herbert Gintis. 1986. Democracy and Capitalism: Property, Community, and the Contradictions of Modern Social Thought. New York: Basic Books.


_____ 1990A. Contested Exchange: New Microfoundations for the Political Economy of Capitalism. Politics and Society 18(2): 165-222.


_____. 1990B. Reply to Our Critics. Politics and Society 18(2): 293-315.


_____. 1993. The Revenge of Homo Economicus: Contested Exchange and the Revival of Political Economy. Journal of Economic Perspectives. 7(1), Winter: 83-102.


Braverman, Harry. 1974. Labor and Monopoly Capital. New York: Monthly Review Press.


Devine, James. 1989A. Paradigms as Ideologies: Liberal vs. Marxian Economics. Review of Social Economy. 47(3) Fall: 293-312.


_____. 1993A. Microfoundations and Methodology in Modeling Capitalism. Review of Radical Political Economics, 25(3): 51-59.


_____. 1993B. The Law of Value and Marxian Political Ecology. In Jesse Vorst, Ross Dobson, and Ron Fletcher, eds. Red on Green: Evolving Ecological Socialism, Canada: Society for Socialist Studies.


_____. 1994A. The Causes of the 1929-33 Great Collapse: A Marxian Interpretation. Research in Political Economy vol. 14: 119-94.


_____. 1994B. Paradigms as Political Philosophies: Liberalism vs. Marxism. Loyola Marymount University: Unpubl. ms.


_____ and Michael Reich. 1981. The Microeconomics of Conflict and Hierarchy in Capitalist Production. Review of Radical Political Economics. 12(4): 27-45.


_____ and Gary Dymski. 1989. Roemer's Theory of Capitalist Exploitation: The Contradictions of Walrasian Marxism. Review of Radical Political Economics. 21(3): 13-17.


____. 1991. Roemer's General Theory is a Special Case: The Limits of Walrasian Marxism. Economics and Philosophy, 7: 235-275.


____. 1992. Walrasian Marxism Once Again. Economics and Philosophy. 8: 157-162.


Draper, Hal. 1977, 1978, 1986, 1990. Karl Marx's Theory of Revolution. 4 vols. New York: Monthly Review.


Dymski, Gary A., and John E. Elliott. 1989. Should Anyone Be Interested in Exploitation? Canadian Journal of Philosophy. Supp. Vol. 15: 333-74.


Edwards, Richard. 1979. Contested Terrain: The Transformation of the Workplace in the Twentieth Century. New York: Basic Books.


Friedman, Andrew L. 1977. Industry and Labor: Class Struggle at Work and Monopoly Capitalism. London: Macmillan.


Friedman, Milton and Rose Friedman. 1980. Free to Choose: A Personal Statement. New York: Harcourt, Brace, Jovanovich.


George, Henry. 1879. Progress and Poverty. New York: Robert Schalkenbach Foundation, 1942.


Gintis, Herbert. 1976. The Nature of Labor Exchange and the Theory of Capitalist Production. Review of Radical Political Economics. Summer.


Goodwin, Richard M. 1967. A Growth Cycle. In E.K. Hunt and Jesse Schwartz, eds. A Critique of Economic Theory. Harmondsworth, UK: Penguin, 1972: 442-9.


Hunt, E.K. 1980. A Radical Critique of Welfare Economics. In Ed Nell, ed. Growth, Profits, and Property. Cambridge: Cambridge U.P.


Kalecki, Michal. 1943. Political Aspects of Full Employment. In his Selected Essays on the Dynamics of the Capitalist Economy. Cambridge: University Press, 1971.


Keynes, John M. 1936. The General Theory of Employment Interest and Money. New York: Harcourt, Brace and Co.


Lebowitz, Michael, 1992. Beyond Capital: Marx's Political Economy of the Working Class. New York: St. Martin's.


Lindblom, Charles E. 1982. The Market as Prison. In Thomas Ferguson and Joel Rogers, eds., The Political Economy: Readings in the Politics and Economics of American Public Policy, Armonk, NY: M.E. Sharpe, 1984: 3-11.


Makhijani, Arjun. 1992. From Global Capitalism to Economic Justice. New York: Apex Press (the Council of International and Public Affairs).


Marglin, Steven. 1974. What Do Bosses Do? The Origins and Functions of Hierarchy in Capitalist Production. Review of Radical Political Economics. 6(2): 33-60.


Marx, Karl. 1867, 1885, 1894. Capital. volumes I-III. Paperback edition. Frederick Engels, ed. New York: International Publ., 1967.


____. 1891. Critique of the Gotha Program. In Robert Tucker, ed. The Marx-Engels Reader. 2nd ed. New York: Norton, 1992: pp. 525-41.


Mathewson, Stanley. 1931. Restriction of Output among Unorganized Workers, New York: Viking.


Miller, John. 1989. Social Wage or Social Profit? The Net Social Wage and the Welfare State. Review of Radical Political Economics. Fall 21(3): 82-90.


Mishkin, Frederic S. 1992. The Economics of Money, Banking, and Financial Markets. 3rd ed. New York: HarperCollins.


Obrinsky, Mark. 1983. Profit Theory and Capitalism. Philadelphia: University of Pennsylvania Press.


Robinson, Joan. 1969. The Economics of Imperfect Competition 2nd ed. London: MacMillan, St. Martin's Press.


Roemer, John. 1982. A General Theory of Exploitation and Class. Cambridge, MA: Harvard U.P.


_____, 1994. Egalitarian Perspectives: Essays in Philosophical Economics. Cambridge: Cambridge U.P.


Schor, Juliet. 1987. Class Struggle and the Macroeconomy: The Cost of Job Loss. In Robert Cherry et al., eds. The Imperiled Economy, Book I: Macroeconomics from a Left Perspective. New York: URPE.


Sinha, Ajit. 1994. A Critique of Part One of Capital volume one: The Value Controversy Revisited. Unpubl. ms. North York, Ont., Canada: York Univ.


Skillman, Gilbert L. 1995. Ne Hic Saltaveris!: Marx's Positive Theory of Exploitation After Roemer. Unpubl. ms. Middletown, CT: Wesleyan Univ.


Sraffa, Piero. 1960. Production of Commodities by Means of Commodities. Cambridge: Cambridge U.P.


Stiglitz, Joseph. 1975. Incentives, Risk and Information: Notes Toward a Theory of Hierarchy. Bell Journal of Economics. 6(2) Autumn.


Tonak, E. Ahmet. 1987. The U.S. Welfare State and the Working Class, 1952-1980. Review of Radical Political Economics. 19(1) Spring: 47-72.


Weber, Max. 1918. Politics as a Vocation. In Gerth, H.H. and C. Wright Mills, eds. 1946. From Max Weber: Essays in Sociology. New York: Oxford University Press.


Weisskopf, Thomas, Samuel Bowles, and David M. Gordon. 1983. Hearts and Minds: A Social Model of U.S. Productivity Growth. Brookings Papers on Economic Activity. (2): 381-441.


West, Cornell. 1991. The Ethical Dimensions of Marxist Thought. New York: Monthly Review Press.


Williamson, Oliver E. 1975. Markets and Hierarchies: Analysis and Anti-Trust Implications. New York: The Free Press.


Wolff, Richard D., and Stephen A. Resnick. 1987. Economics: Marxian versus Neoclassical. Baltimore: Johns Hopkins Press.



* An earlier draft of this paper was presented at the University of California at Riverside on October 24, 1994 and at Loyola Marymount University on November 16, 1994. A version of the appendix was presented at the URPE sessions of the ASSA convention in January 1995. Thanks to Ed Chilcote, Steve Cullenberg, Bill Dugger, Gabe Fuentes, John Harvey, James Konow, Victor Lippitt, Michael Perelman, Bob Pollin, Lynda Rush, Howard Sherman, Kamal Shoukry, Ajit Sinha, Robert Singleton, and others for their useful comments. It also benefited from a long e-mail discussion with Gil Skillman. Of course, all high crimes and misdemeanors are mine alone. A mathematical model summarizing the theory is available.