Today I would like to talk about class analysis: how we do it, how we understand and apply the method – in particular, the persistent challenge of working-class formation as a historical and sociological phenomenon, and as a political proposition. Working-class formation – we would all like to see some of that!

It is therefore a problem how unsettled the debates remain over the nature of the phenomenon that raged from the late 1970s through the 1990s. The contestants, rather than fight to a resolution, largely vacated the field and headed in different directions.¹ I do not blame them; I imagine that I would have joined them. But this theoretical cease-fire itself seems symptomatic to me of the changing terrain on which the left fought over this period. At the outset of the debate, in the late 1970s, socialist possibilities still appeared open, and an effective strategy to face the dawning political crisis was critical.² By the late 1990s, few would have said the same. The crisis that gave rise to the debate in the first place, beginning in the 1970s, reached a kind of terminus. The political possibilities that the theoretical contestants hoped to keep open became closed, and the issue was vacated. ‘Labour has been quiescent for over twenty years now’, observed Perry Anderson in the 2000 relaunching of New Left Review. ‘The only starting-point for a realistic Left today is a lucid registration of historical defeat. Capital has comprehensively beaten back all threats to its rule’.³ How can a victory be so comprehensive in a society that undermines its own basis constantly?

A paradoxical outcome over the past decades has been that the voices of Marxist analysis most expressly committed to a more orthodox or classical defence of class analysis have also been most ready to argue that the traces and effects of working-class struggles are nowhere to be seen. We see different iterations of such arguments, from the very common position that identity politics has displaced or dissolved working-class politics; to the idea of ‘class dealignment’, which holds that political behaviour is no longer explicable in class terms; to the reintroduction of Weberian concepts of status groups to explain behaviour that seems aberrant.⁴ Without dilating on any of them in particular, we might uncontroversially first agree that each of these strategies seeks to explain how increasingly unequal class societies do not produce class outcomes and class behaviours in politics, at least not on the part of the working class.

What is paradoxical about these positions is that as a rule they claim the mantle of more traditional class analysis, yet simultaneously retreat from the idea that we can systematically explain politics and ideology by reference to class relations. Each of them carves out a territory in the zone of the political and ideological that cannot be explained by class relations, and instead is determined externally to the class relationship. Typically we then hear much more talk about and demand for class analysis in the mode of exhortation than we see comrades actually doing it, and indeed I worry that such demands for class analysis seem to crowd out the practice itself.

Before I finish my introductory polemic, let me give you two examples from the American scene. Vivek Chibber’s recent book The Class Matrix argues that culture is irrelevant to the question of why workers do not organise, which instead is sufficiently explicable in terms of individuals’ rational risk–reward calculation. But when querying why workers do occasionally surmount this barrier, Chibber proposes that the answer is a ‘culture of solidarity’.⁵ Culture is dismissed from causality on one side, only to be summoned back again in the breach on the other, with no rigorous effort made at specifying a systemic relationship between the economic and the cultural: economic rationality excludes cultures of solidarity, except when it does not. The basic Marxist insight that social action must be explained as simultaneously economically rational and ideologically meaningful, and that economic objectivity is accessed subjectively through the cultural meaning that arises from it, is almost entirely lost here. As the relationship between social structure and ideology is severed, each becomes unintelligible and incommensurable with the other. Thus, either economy or ideology is effective, but never both at once, and with no theory of their alternation.

Similarly, across their work of the last twenty years, Walter Benn Michaels and Adolph Reed Jr. have argued that the professional-managerial class deploys the politics of race to displace the politics of class. Without gainsaying the empirical occurrence of this phenomenon, which seems undeniable (if perhaps less uniform and all-encompassing than they would argue), let us note that it presents a glaring methodological dilemma: how can it be that race, a purely ideological phenomenon on their account, can override the objective reality of class? Note that this is a stronger claim than Chibber’s: Chibber advances the idea that non-solidaristic practice is individually rational and thus culture is causally irrelevant to workers’ division, only (inexplicably) to their occasional unity. Michaels and Reed go further, arguing that non-solidaristic practice is successfully imposed upon workers through cultural means, causing their division. For this to be the case, either race (and more generally ideology) must have a quality of social objectivity; or class must lack one. If race has such a quality, we are in some version of a theory of racial capitalism, which Michaels and Reed would adamantly reject. Theoretically, we could derive a coherent such account through Antonio Gramsci, Louis Althusser, Stuart Hall, or Cedric Robinson, depending on taste and emphasis; but they would choose none of these, and it is difficult to discern another pathway. On the other hand, if class lacks objectivity and is merely one competing discourse among many, we are of course landed squarely back in the terrain of the cultural turn and post-Marxism. Yet this is exactly the tendency against which Michaels and Reed and many fellow travellers have set their face. In the face of this paradox, I am drawn to recall Gramsci’s famous warning of the methodological Scylla and Charybdis of ideologism and economism.⁶ Here the analysis manages to ricochet from one obstacle into the other rather than pass safely between them: a self-contradictory account of ideology’s effectivity over the economic, grounded theoretically in economism.

I begin with this issue because it seems to me unhealthy that the loudest voices advocating class analysis on the left are also the ones that write off huge areas of political struggle. For them these are, at best, noble but unrelated to class processes. Neutrally they are seen as distractions. At worst they are sites of ruling-class infiltration and subornment. Much of the world is consumed by struggles over racism, nationalism, gender, sexuality, migration, religion: both left- and right-wing movements seek to mobilise these social forces for their purposes of liberation and hierarchy. If we decline to make these phenomena intelligible and historically specific by uncovering their determinate relation to class processes, we have abandoned our post. Rather than enjoin that we talk about class, we ought to study the topic: to take it seriously as a subject of concrete inquiry, including empirical inquiry – an effort to learn something we do not already know.

This requires us to engage with the concrete social variation in working-class history and experience. The proletariat has always been many things concretely, even as it is unitary in a more abstract sense. To hold only to this more abstract unity and refuse to engage with concrete social variation is to reduce the working class to a mere idea – indeed at worst, a fetish, a talisman, an object. Marxism has a rich storehouse for exploration of social difference among the socially subordinated, reaching all the way back to debates about peasant classes, the woman question, child labour, imperialism and the national question, and so on.⁷ These matters are not supplementary to the core issue of the formation of the working class, but are decisively part of that question. Today I will discuss how Marxist feminism endows us with a legacy to reapproach questions of class formation – a relevant topic given the recent, understandable concern about the composition of past winners of this award. I have titled this talk with a cliche – ‘the baby and the bathwater’ – that concerns the difficulty of distinguishing between transitory and essential qualities. The cliche is also a metaphor for reproductive labour which says something about the effort involved in caring for the young and the attention required not to treat people like objects.

A challenge we face in doing class analysis after deindustrialisation is how much of what we thought was general to working-class experience may prove specific to the industrial age. The arc of proletarianisation, in which peasants were made into workers, enacted a certain kind of uniformity. The socialisation of labour through industry created the possibility of wider political horizons for proletarians, by gradually producing intraclass homogeneity: the real subsumption of labour eventually somewhat suppressed differences of skill, region, religion, even race and ethnicity.⁸ As intraclass difference has reasserted itself, we ought to inquire into what its relationship is to the close of this historical arc.

To the extent we continue to debate Marxian questions of class and class formation, particularly in the anglophone world, we do so with far too little attention to the social experiences of labour markets, employment, and unemployment that mediate the concrete existence of the proletariat; labour processes, skill, and deskilling; labour-market segmentation by race and gender; family form, reproduction, and demography.⁹ Investigation of these and similar matters were central to class debates in the 1970s, at the last high point of Marxism, yet we have not sufficiently refreshed the empirical content of these lines of inquiry. Or to the extent we have done so, we have not yet closed the circle back toward theoretical and methodological issues of class analysis. How has a half century of historical defeat registered, not merely at the level of method, but in the sociology of the proletariat? We do not know, not at the level of general concepts.

The undeniable crisis of the industrial working class and its organisations generated innumerable forms of intellectual disorientation, as well as social dislocation – this much we do know. My own empirical research has been directed toward this problem, and I will summarise this a bit and try to generalise from it. Let me also say that my perspective is limited by geography and language: this will be a discussion overweighted toward the North Atlantic and especially anglophone world, the United States most of all. I will do my best at various points to speak more broadly but do not wish to overreach too far beyond what I know best, and hope that in the discussion we can widen and engage in some comparison. I do hope you will amend and correct my claims in regard to these kinds of oversights in discussion.

As Harry Braverman argued in 1974, historical and social fluctuation is a necessary and unavoidable element in class antagonism in capitalist societies. ‘The term “working class”, properly understood, never precisely delineated a specified body of people, but was rather an expression for an ongoing social process’, Braverman observed.

Nevertheless, to most people’s minds it represented for a long time a fairly well-defined part of the population of capitalist countries. But with the coming of broad occupational shifts and a growing consciousness of these shifts in recent decades, the term has lost much of its descriptive capacity. … We are dealing not with the static terms of an algebraic equation, which requires only that quantities be filled in, but with a dynamic process the mark of which is the transformation of sectors of the population.¹⁰

Implicit here is the suggestion of a recurrent contradiction: one that is at once formally necessary and static, and yet mobile and dynamic in its empirical specifics. This contradiction occurs between the process of social fluctuation that produces class positions and recruits layers of the population into them; and the slower developing political and ideological armature of working-class organisation – unions, parties, and the language and symbols in which they deal, what Braverman notes as the problem of an obsolescent ‘fairly well-defined’ image. As the cultural historian Michael Denning observes, ‘No matter what our position in the class-structuring of the population, our class images are usually a generation behind the realities of class formation, behind the forces of capitalism that are reshaping the working population’. There is no reason, Denning observes, to believe that the left would be exempt from this problem.¹¹ It is in address to this exact issue that Adam Przeworski observed that ‘the ideological struggle is a struggle about class before it is a struggle among classes’.¹²

What, then, is the empirical nature of the differences that have proliferated since the 1970s, which seem to require of us a new round of struggles about class?

Let me begin from my own research programme, in all its provinciality. The book awarded the prize that brings me here, entitled The Next Shift: The Fall of Industry and the Rise of Health Care in Rust Belt America, is a social history of labour-market transition. Focusing on the US city of Pittsburgh, the historic centre of the American steel industry, I show how today’s low-wage service economy emerged out of the limits and paradoxes of industrial working-class formation. This transition cannot be understood simply as a shift from good jobs to bad ones: residual and emergent fractions of the working class are interwoven in complex historical patterns, partly solidaristic and partly antagonistic.¹³

In the US labour market, the category called ‘health care and social assistance’ is the largest sector of employment, accounting for about one in seven jobs. In deindustrialised cities and towns of the Northeast and Midwest, the so-called Rust Belt, this number rises to one in six, even sometimes one in five. In the bottom quintile of the American wage structure, the care economy – broadly construed – accounted for 56 per cent of all job growth in the 1980s, 63 per cent in the 1990s, and 74 per cent in the 2000s. Overwhelmingly it remains women who do these jobs, especially African American and immigrant women.¹⁴ Home health aide has been for many years the fastest growing job in America.

About a decade ago in Pittsburgh, in the context of a union representation dispute, the University of Pittsburgh Medical Center claimed before the National Labor Relations Board that it had no employees. This was remarkable since this institution is widely known as the largest employer in the state of Pennsylvania – at the time with about 70,000 employees and today near 110,000. Here was a prelude of what has become a very familiar paradox since the pandemic: how can workers be both essential and disposable, everywhere yet invisible? Posed more rigorously: how can such relentless growth of a certain industry and demand for its products not issue into leverage, opportunity, and power for the workers who perform it? What are the mediations or articulations of race, gender, class, and state that structure whatever class formation may be crystallising here, and how are they organised?

The outsize proportions of the care economy in places that once were home to concentrations of American industrial capital suggests a historical linkage between one phase of class formation and another. Reconstruction of this linkage can illustrate both the origins of this new care economy, and the causes of its peculiar characteristics.

When the industrial working class consolidated itself and won the concessions that amounted to the welfare state, like all working-class formations in stable capitalist societies, it did not do so as a hegemonic class. The terms by which its activity was legalised and its economic security was negotiated in the 1930s and 1940s cemented its organisation as belonging to the ‘economic-corporate’ kind, in Gramsci’s terms. In the United States this process took the particular form of a private-sector welfare state, established primarily through firm-level collective bargaining.

The apotheosis of the economic power of American labour was thus at the same time the moment that political sectionalism again began to set in, with McCarthyism and the Cold War environment operating to erect boundaries of representation around organised labour. As an interest group in the Democratic coalition, organised workers were structurally divided from the rest of the working class.¹⁵ Industrial workers were unionised and protected by a privatised welfare state, which doled out social benefits through private-sector collective bargaining, and from which workers without access to industrial labour markets – women, racial minorities – were largely excluded. This would famously result in political division by the 1960s and 1970s, but it had a structural, economic element that is of great significance as well.

Disbursement of the social wage to the industrial working class through private-sector employment institutionalised a structural antagonism between different layers of the class, an antagonism generally demarcated and activated by race and gender. The best known form is that which occurred within the household: the employed worker’s wage packet and benefits had to be turned into food, cleaning, laundry, leisure, sex, children, by the unwaged housewife: this might manifest in actual interpersonal conflict or might not. Let me note here my debt to the finalist for the prize Silvia Federici, it is of course an honour even to be considered in her company. But I would like to suggest that this is merely a major example of a wider pattern, in which a portion of the wage essentially becomes a coupon to claim reproductive labour from an unwaged or otherwise marginal woman worker. In particular, employment-based health insurance, one of labour’s central conquests in the private-sector welfare state after the defeat of a public-sector effort in the 1940s, represented just such a voucher: what it paid for primarily was time in the hospital, which is to say not medical attention alone but also the labour of nurses and nursing assistants, laundry and kitchen workers, and technicians. Health-care workers, however, were not protected by labour law – the same labour laws that enabled steelworkers to win health insurance excluded hospital workers from the protections of formal employment. Operating outside such protections, hospitals hired from the most vulnerable strata of the labour market.

What we have here is a classic dualised labour-market structure, one mediated by race and gender, variations of which have occurred across the North Atlantic welfare states where organised working classes won major gains in the twentieth century without ever achieving hegemony. The interesting additional question is how the marginal sectors, once positioned as auxiliary to the industrial economy, became so numerically dominant, and why their economic marginality persisted through this inversion.

Here it is critical to understand the dynamics of the end of Fordism and the consequences for reproductive labour. As is argued by Nancy Fraser, the transition to neoliberalism destabilised the Fordist regime of caregiving.¹⁶ But this is not sufficient to grasp it, for the system of social reproduction was not destabilised while standing still; it also expanded rapidly, becoming distended even as its internal logic was transformed.

In particular, the process of factory job loss, which began gradually in the late 1950s and accelerated rapidly in the 1970s and 1980s, caused the population to become older, poorer, and sicker.

First, deindustrialisation brought economic immiseration. In the Pittsburgh region 150,000 factory jobs were lost between 1975 and 1985, following two decades of slower decline. The regional poverty rate more than doubled over these years; so did the homeless population. In the steel town of Aliquippa, researchers observed a 55 per cent decline in earned income from 1981 to 1984, commenting, ‘Falls of this magnitude inevitably set in motion a struggle for individual and collective survival’.¹⁷

The language of survival is apt: the process of dislocation moved from the economic to the demographic and epidemiological. Deindustrialisation made the population older, because of the centrality of seniority-based labour markets to unionised industry. Inevitably, this meant that declining rates of employment would produce a demographically aged workforce and population: the young moved away to seek opportunity, but the largest cohorts aged in place. In 1950, 38 per cent of Pittsburgh steelworkers were over the age of 45; in 1980, half were. Or put another way: in 1970, 11 per cent of the regional population was older than 65. By 1990, more than 17 per cent of the population was over 65. From 11 per cent to 17 per cent in 20 years, including not just relative but absolute growth. The large cohort hired for 1940s war production aged in place, even as the young left. By 1990, Pittsburgh would become the second oldest metropolitan area in the country – behind only Florida’s Broward County.¹⁸

Finally, a poorer population and an older population is also a sicker population. Older people are generally sicker, of course. Poorer people too are as a general rule sicker, and the economic shock became a health shock in short order. For high-seniority industrial workers laid off between 1980 and 1986 in Western Pennsylvania, mortality hazard in the immediate aftermath increased 50 to 100%.¹⁹ For each laid-off steelworker who died, many more were ill. Observable increases in domestic violence, addiction, suicide, and even phenomena such as heart disease gave deindustrialisation a distinctive epidemiology.

There is also a more complex mechanism here which we might call biopolitical. Because so much of the welfare state was privatised and employment based, industrial workers’ claim on health-care services, provided in the private sector, was much more robust than other social-welfare benefits. For displaced workers and their families seeking to claim social support, health manifestations of social distress presented one of the most promising avenues. Use of the health-care system expanded very rapidly in industrial centres in the 1970s and 1980s, rising far above the national average, especially among blue-collar workers and their families. This consisted of frequent and especially very lengthy hospital stays. In 1979, Pittsburgh generated an average 1.6 hospital inpatient days per capita, about triple the national US rate today. This obtains equally for steelworkers and family members, suggesting not workplace hazard but a socioeconomic effect, in which the health system is used for social care via the management of non-critical conditions.²⁰ As deindustrialisation drove this form of usage, the system grew enormously, with health insurance acting as a kind of countercyclical fiscal pump. As the system grew, it hired.

Here it is helpful to think of a family: a steelworker, hired during the Korean War boom in 1950 into the mill where his dad worked. Once he gets the job, he and his girlfriend get married, she quits her job as a waitress, they buy a house, have a few kids. Over the decades, he is laid off and called back and goes out on strike many times; he is injured a few times but never critically. Through these years, his wife takes care of him, as she does their kids. By 1980, it is time to retire, and both members of the couple are ready: a lifetime’s worth of the physical burdens of blue-collar life – injuries, pollution, stress, cigarettes, and alcohol – have built up.

Now in 1980, the sons cannot follow the fathers and grandfathers into the mill: the jobs are not there. So they begin to think about migrating away. But who will take care of their aging parents, their wives ask? And so the daughters of the younger generation must look for work, departing from their mothers’ lives of housework. And who is hiring? The hospitals and nursing homes. All the aging steelworkers and housewives are being taken care of by the daughters of other families like theirs. Women thus filed into the labour market from the late 1960s through the 1990s, beginning at the bottom with African American women whose husbands lost their factory jobs first, and gradually filling in the lower strata as white women joined them. In 1980, health care passed steel as the largest employer. Overall the health-care industry functioned as a shock absorber, expanding against the contraction of the industrial economy and soaking up some of its dislocations: through the fiscal structure of third-party health insurance and private health-care administration; and through the labour supply of women entering the formal labour market, institutionalising and commodifying care functions once located in families.

At one level, this appears to be a very classic form of increasing socialisation of labour, as the web of interdependency required for working-class social reproduction broadens again. Once peasants made their own food and clothing; then they came to depend on each other as workers, through the mediation of the market, for such goods. Once families produced more of their care, then they came to depend on each other through hospitals, agencies, schools, and care homes. At another level, though, this latter process has new, anomalous features.

Here is the point at which I would like to move up the ladder of abstraction from Pittsburgh. I observed earlier that much of the care economy has remained mired in the low-wage, precarious layers of the labour market, despite consistent sectoral growth and increasing demand for labour. This anomaly points our attention to what is distinctive about the care economy, and indeed about interpersonal service work more generally: the resistance that it presents to the kinds of productivity increase that we associate with capitalist industry.

This phenomenon has become widely observed in recent years and deployed in debates about global economic stagnation.²¹ I would like to add two observations to the discussion of its significance, by way of broadening my argument up until here. The first is that we have taken this empirical observation as altogether self-explanatory without expending sufficient effort to generate an explanation of service-economy stagnation from within the Marxist tradition: for this the resources are available in Marxist feminism.²²

First, let us note that Marx makes a bit of a hash of this question himself. Famously, he observes of social reproduction that, during the labour process, ‘a definite quantity [bestimmtes Quantum] of human muscle, nerve, brain, etc. is expended, and those things have to be replaced’.²³ But how can the quantity possibly be definite? On this point, as was argued in the ‘housework’ debates of the 1970s, we are certainly in the murky qualitative realm of concrete labour – not the quantitative, market-equilibrated clarity of abstract labour.²⁴ And in this concrete domain, the quantity of human muscle, nerve, brain, etc., can hardly be definite: its expenditure will necessarily vary worker to worker, day to day, according to an infinity of contingencies that are partly opaque. I do not always know why I sleep badly, or what I desire, or how to eat well, or why a part of my body hurts, or the cause of my depression, or – harder still – what would be required for remedy.²⁵ Even an expert often does not know or cannot intervene efficiently. Indeed in the same passage, Marx begins to generate an answer to this problem in his famous recognition that the standard of living, which sets the costs of labour-power’s reproduction, is historically specific and determined in part through class struggle. Abstracted, this point implies objective indeterminacy in the use-values required for the social reproduction of labour-power – just the opposite of the ‘definite quantity’ of the earlier passage.

The human body itself – the product as well as the instrument of reproductive labour – presents a limit to capital, a limit which manifests in the labour process of care. This limit, as the political theorist Alyssa Battistoni emphasises, arises from the natural cycles of time through which the body itself must move.²⁶ Although these cycles gain their specific shape and meaning from social context, they are not infinitely plastic, nor are they even wholly knowable. The body presents physical resistances and epistemic opacity to intervention, inhibiting the forms of rationalisation that have tended to homogenise production within capitalism. Indeed, it is the quality of being amenable to such rationalisation and homogenisation that makes a production process ‘industrial’. In its absence, wide diversity is imposed upon the practices of social reproduction.

By the same token, the processes of social reproduction cannot crystallise into value satisfactorily. They must pass through the body, and the body cannot be made to conform wholly to value’s abstract compulsion. Satiation of hunger and need for sleep, satisfaction of sexual desire, maintenance of hygiene and health, management of disability, negotiation of age and death, fertility and procreation, accumulation of knowledge and experience – such phenomena define living labour and its reproduction as employable labour-power; they are difficult or impossible to absorb wholly into capital. For example, while there exists a relationship between economic conditions and human fertility, throughput of children does not correspond in a linear fashion to labour-market demand. Nor can children be made to assimilate the information that education imparts in an orderly manner – so-called ‘human capital’ itself arises from a labour-intensive, temporally uneven process enacted upon an intermittently resistant object – the child.

Since living labour is reproduced by such indefinite interpersonal processes, indeterminacies are endemic to social reproduction. Precisely what is required for replenishment of labour-power is partly obscure even to workers themselves regarding their own capacities within the daily and weekly cycles, much less along the timescale of generational reproduction of labour-power. As replenishment of labour-power is mediated through institutions of kinship, market, or state, this uncertainty is multiplied: the regeneration required by the worker who needs the reproductive labour of another – something often called ‘care’ – may be opaque to the caregivers charged with the assignment; or there may be misalignment between the caregiver’s diagnosis and the recipient’s own account. Children and parents often disagree with each other, or with teachers; clients resist social workers; elders resent being managed. Such misalignment is often produced by contradiction between the abstract quantitative demands of market pressures on the one hand, as they operate on market-mediated providers of reproductive services; and on the other the concrete indeterminacy of social reproduction – the problem of the dynamic physical body in which labour-power is stored. From the perspective of service workers and clients, the problem of productivity appears as a zero-sum struggle over attention and effort. An acceptable quantity of attention and care from the perspective of an institution – a hospital, a school – and its managers, creditors, and regulators may appear wholly insufficient for a patient or a student, or exhausting for a worker.

In neoclassical economics, this pattern is well known under other names. One is William Baumol’s famous ‘cost disease’, in which service industries suffer from productivity stagnation, which then inflicts rising costs and compels price rises.²⁷ Similarly – in another insight from 1960s neoclassical health economics – Kenneth Arrow argued that uncertainty and information asymmetry systematically plague health-care provision. This uncertainty ‘has created many social institutions in which the usual assumptions of the market are to some extent contradicted. The medical profession is only one example’.²⁸ Arrow did not, of course, think himself to be making an intervention at the intersection of the law of value and the service economy. Nonetheless, it is easy to see his observation in these terms: the political economy of social reproduction tends to develop anomalous organisational forms. It cannot be bought and sold in the normal way, and often features unusual market institutions or outsize roles for the state. The family is only the chief exemplar, and the most ideologically freighted.²⁹ The conceptual connection between the family and the dysfunctions of commodified social reproduction was observable even when Arrow wrote in 1963. He continues, ‘The economic importance of personal and especially family relationships, though declining, is by no means trivial in the most advanced economies; it is based on non-market relations that create guarantees of behavior which would otherwise be afflicted with excessive uncertainty’.³⁰

In other words, some forms of social interdependence, because of the uncertainty that characterises them, cannot be systematically commodified. In fact, family relationships offer no alternative guarantee of reliability, contra Arrow. There exists no institutional configuration that eliminates the uncertainty that occurs in a relation of care, because that uncertainty arises not from the form of the relation but from its content.

Here let me draw a connection between two famous theories. One we have already discussed, the problem of productivity stagnation in service work. The other is Judith Butler’s classic theorisation of gender. In Butler’s terms, gender is a ‘constituted social temporality … instituted through acts which are internally discontinuous’, forming ‘a stylized repetition’.³¹ Slippage in this repetition is inevitable, which creates the possibility for Butler of novelty in gendered enactments.

At an abstract, formal level of analysis, the inevitability of slippage that arises through this discontinuous repetition is notably similar to the problem of productivity stagnation in service work: what each theory says is that the body cannot be relied upon to follow any exact protocol through the cycle of everyday life or across the stretch of a lifetime. In the service workplace, the abstraction of labour into labour-power glitches and stutters. Just as the subject cannot ever master gender and produce it perfectly, similarly no flight attendant can produce exactly the same smile every time or become more efficient at doing so.³² Indeed, these are nearly two ways of saying the same thing. Moreover if the flight attendant could do so, there would be no way of guaranteeing the same effect on each passenger.

Having allowed human biology some effectivity in political-economic relations, it is incumbent to take a moment to banish it once again from our account of gender. The distinctive problem of embodiment, which makes human service work labour-intensive and hence high-cost, itself has nothing intrinsically to do with gender. Rather, the problem of embodiment causes such work to be assigned, for economic reasons, to marginalised layers of the proletariat to carry out, through waged and unwaged forms.³³ It is of course true that women’s subordination is historically linked to the social forms surrounding the biological processes of gestation, birth, and breastfeeding.³⁴ Crucially, though, the bearing of this point for our discussion here is that such historical socio-biological forms enable the confinement of women within these socially marginalised layers. As Battistoni emphasises, the relationship between the gendered service economy and modern society’s historic ‘woman question’ is incidental only, connecting at the point where women are assigned the least desirable work because they have the least social power – a historical fact rooted ultimately in the social regulation of biological processes. The fact that the least desirable work often bears that quality because it too pertains to the body is, despite superficial similarity to the purported relationship between biology and gender, analytically distinct.

The irregularity of the body’s demand for care corresponds to a requirement for elasticity in its provision – hence the search for vulnerable, cheaply-available labour. Commodification is either undertaken on a small-scale and at a low margin, sometimes verging on self-exploitation (food service, waged domestic labour, informal economic activity of all kinds); or it is coaxed by public investment or public subsidy under political demand (health care, elder and social care, education, child care); or it does not occur at all (the unwaged household sector).

Such afterthought, low-productivity sectors – Battistoni names them ‘remnants’; Maya Gonzalez and Jeanne Neton call them ‘the abject’ – stand in what we might call, following Stuart Hall, a relationship of both ‘articulation’ and ‘conservation-dissolution’ to capitalist production.³⁵ Human service providers frequently, though not always, supply a product – labour-power – required by capitalism, but whose production capitalist processes can only unevenly absorb and direct. They also continue to reproduce living labour even where it is not demanded as labour-power, a further feature in their resistance to full subsumption into capital. Here let me quote Stuart Hall, himself quoting Charles Bettelheim, on the relationship between the capitalist mode of production and other exterior modes that it encounters through colonial expansion:

Charles Bettelheim, who may appear to take a more ‘classical’ view, argues that the dominant tendency is toward the dissolution of other modes by the capitalist one. But this is often combined with a secondary tendency – that of ‘conservation-dissolution’: where non-capitalist modes, ‘before they disappear are “restructured” (partly dissolved) and thus subordinated to the predominant capitalist relations (and so conserved)’. ³⁶

While social reproduction is not per se originally outside and prior to capitalism in the way that precolonial social formations have been, it is, due to its qualitative characteristics, resistant to total absorption in a similar fashion. Its subordination similarly causes conservation of its distinctive qualities, as capital continually probes for opportunities for real subsumption and finds its limits. Seeing care as internal to capitalist society but marked off from capitalist production by this limit, describes a critical difference between this view and that of social-reproduction theory, which generally holds that the distinctiveness of social reproduction is a political effect. What I am arguing here instead is that this distinctiveness is internal to the differential profit-seeking logic of capital, and that the relationship between caregiving and capital is therefore an articulation, not a separation; whose logic is economic first and political only second. A wide variety of institutional configurations thus becomes possible for the provision of care, with diverse political struggles mediating a fundamental economic constraint. Gender is in turn produced through these mediations, as that constraint’s social consequences are meted out through the division of labour. Who can be compelled to do which kinds of labour is a question that often precedes and activates gender, rather than only following it.³⁷

Having developed a more theorised account of the origins of the productivity problem in the service economy, I now wish to make a second observation about the productivity limit, which characterises the relationship between the service economy and the wider capitalist social formation – through the question of demand for reproductive services. Within mainstream social science, the standard story is to see the shift to demand for services, sometimes called ‘tertiarisation’, as a linear function of growth. The move up the developmental ladder from primary to secondary to tertiary sectors accompanies rising affluence, as in a climb up Maslow’s hierarchy of needs. But what I have argued from my empirical case is that expansion of demand for care may also be driven by economic stagnation – or more precisely in the specific case, a cumulative cycle of industrial overdevelopment followed by stagnation. The operative mechanism here is the decline of opportunities for single-earner households that industrial expansion afforded, and the subsequent fractal pattern of mutual responsibilities that emerges as more members of households and kinship units enter into the cash nexus, in turn generating expanding household care needs as labour supply faces outward increasingly. The daughter gets a job as a nurse, but now she needs to find someone to watch her kids.

On this perspective, the global phenomenon of deindustrialisation, although extremely varied in its expressions, is likely to produce a global pattern not only of the destabilisation of care provision, but also its expansion. The failure of Fordist and developmental models did not merely damage the ability of peasant and proletarian households to get their existing care needs met; it generated new care needs. Yet the nature of these needs, indeterminate as they are, corresponds to enormous elasticity in possible social arrangements for meeting them: there being no one best way, no possible rationalised mass-produced care, a unitary global process – the relative decline of industrial employment – manifests heterogeneously across the world-system. The resulting relations of care interweave forms of interdependency, even solidarity, with exploitation, since they consist of the negotiation of zero-sum economies of effort. Crises of staffing-availability in health-care facilities worldwide illustrate this phenomenon, as staff both engage in self-sacrifice to keep things running and patients living, and also enter into sometimes intense antagonism with patients: the United States is facing an epidemic of violence in health-care facilities, which host 73 per cent of nonlethal incidences of workplace violence; while in China specialised gangs, called Yi Nao, extort compensation from doctors and nurses on behalf of aggravated patients.³⁸

Globally, the faltering of capitalist growth models in almost all parts of the world has thrown proletarian populations back onto collective and interdependent modes of survival. Some of these modes are state-mediated, some market-mediated, some kinship-mediated. Some follow industrial job loss, others de-agrarianisation. All depend for their functioning in the teeth of difficult conditions first upon the exploitation of vulnerable layers of women labourers who are increasingly responsible for generating income (within or outside of formal labour markets); and second, simultaneously upon the formation of new patterns of everyday proletarian solidarity, to make zero-sum economies of effort functional.³⁹

In much of the world, the North Atlantic normative model took hold much more weakly if at all, where the family-wage industrial jobs on which it rested were never broadly forthcoming. In its stead, women have frequently taken on complex positions at the points of connection between country and city and between household and market, generating a conceptually challenging interplay between novelty in the gendered division of labour and continuity with historical forms of patriarchy. As Deborah Bryceson argues, ‘Social boundaries have been redrawn to maximize entry of market participants. The scramble for cash has caused an upheaval in age-old gender and generational divisions of labor. Types of work ascribed strictly to men, or alternatively women, have broken down’.⁴⁰

One increasingly common pattern in a destabilised world labour market is for extended kin systems to substitute unwaged childcare labour for what is sacrificed when a mother seeks waged work, domestically or internationally – ‘women’s reserve army of reproductive labor’, as Ayşe Arslan describes it.⁴¹ In the context of international migration, there often forms a ‘chain of care’, in which households generate remittance income through the export of reproductive labour, while family members care for left-behind children or elders and secure reproduction in the absence of the migrant; or an internal hinterland supplies cheap domestic labour to fill the gap.⁴²

While the ‘scramble for cash’ does not occur identically everywhere, the basic dynamic of increased pressure on women to buffer the stagnation or failed take-off of the formal labour market is broadly shared. Summing up the world pattern, Shahra Razavi and Shireen Hassim write, ‘It is becoming increasingly necessary for all household members – whether female or male, young or old – to take on paid work. … Much of this tends to be badly remunerated, with women being overwhelmingly clustered in low-entry, low-return type activities’.⁴³

The situation of stagnating or declining industrial employment and underdemand for labour has several simultaneous contradictory effects. It puts pressure on existing institutions of social reproduction, causing their functional hypertrophy (either in the form of more reproductive labourers or intensification of reproductive labour); and in so doing exacerbates the impossible situations in which reproductive labourers are inevitably enmeshed. In short, their assigned responsibilities expand while the available resources with which to discharge those responsibilities contract: this situation, which Nancy Fraser has called the ‘crisis of care’, gains its specific characteristics not only from the collision of fiscal austerity and underemployment with systems of provision organised through the family and the welfare state, but also from the distinctive additional roles families and welfare states have been forced to take on as normative employment has disintegrated. Social reproduction is not only ‘undermined’, as Fraser observes.⁴⁴ Paradoxically at the same time, through its articulated linkage to capital, social reproduction is also expanded. To see the relation between neoliberalism and social reproduction as straightforwardly atomising and privatising is to miss this contradictory expansive dimension, which generates new forms of reproductive linkages within kinship units and among strangers alike.⁴⁵

The crisis of care occurs not first for capital, which evades the problem with relative ease, but for living labour, as a deficit experienced in everyday life. This deficit has an indefinite quality: what would be needed to fill it is an open question, and there is usually more than one possible answer. Reproductive labourers, those assigned with the task of care can – and to some indeterminate extent will – intensify their own efforts in correspondence with the deficit.⁴⁶ As Sarah Mosoetsa summarises her research in KwaZulu-Natal, ‘There was a much greater reliance on the household than had been the case in previous years, when households had depended almost entirely on permanent wage labor for their survival. In the absence of formal employment and wage income, the work done in households becomes critical to the survival of its members’. Here too, intergenerational relations prove analytically critical in the management of emerging new care needs. ‘Older women, in their role as caregivers and providers for the extended household, do most of this work. The impact of AIDS and unemployment on younger women, makes older women central figures in poor households today’.⁴⁷ This expansion of interdependency has in many places – most impressively Latin America – opened up expressly political possibilities, as feminist struggles gain a kind of universality, adhering class and racial-justice demands to themselves.

Globally this expansion of reproductive labour takes a vast diversity of forms, from women who drop off children with a relative for care during their workshifts, to elders placed into growing social-care institutions subsidised by the state. Yet in few places are mothers and wives producing meals, baths, lessons, therapies, medicines, or minutes of custodial attention for immediate kin in rising proportion as compared to aunts, grandmothers, daughters, granddaughters, teachers, nurses, nannies, and domestics; more extended and complex patterns of reproductive relationships, including those with waged workers, are becoming the rule with greater frequency almost universally. This world pattern forms a fundamental element of the structure of social inequality, as it describes increasing transfer of responsibility for social reproduction onto relatively unprotected forms of labour: this unprotected labour regime might be informalised and kinship-based, as with a Filipina grandmother who watches her grandchildren while doing her own informal market work; or its informality might itself have an explicit juridical shape, appearing when we follow her daughter to her job in the United States as a home health aide, where she lacks citizenship and workplace protections.⁴⁸

Finally, and more suggestively, it seems to me that this analysis provides us a new way of understanding the global politics of gender, most notably the failure of heteropatriarchal normativity to reproduce itself in younger generations, and correspondingly the new mobilisations to attempt to enforce it. Conceiving of the performance of gender as the effort to bring the enactment of self into a stable orbit around an imaginary focal point, it then becomes apparent that the economic disintegration of the dual model must introduce disarray, as individuals face the imperative to improvise and stretch their capacities, and thereby their selves, to meet the complex, multidimensional demands of the ‘scramble for cash’. Where a seemingly viable normative life-course pathway once existed, this imagined future would reach back and shape the young in its own image – not without friction – but with a power that is no longer operative when the future presents instead as an opaque, buzzing cloud of opportunity and constraint mixed together. For instance, as more members of working-class families experience pressure to bring in cash, parents are more often obliged to entrust children to other family members, acquaintances, or strangers. To the extent they do so, the formation of the children in the image of the parents weakens. ‘The threat posed to the family by children’s gender self-determination may be that it reveals the image of the child to be a social object parents cannot alone control’, observes Max Fox (in regard to the right-wing paedophilia panic). ‘Any real, living child’s ultimate gender is a mystery which unites its material body and social existence, synthesizing uncountable contingent encounters and their traces into something that coheres and persists’.⁴⁹

The fundamental point I have argued is that the quantity of those uncountable contingent encounters exterior to the immediate family has increased – not just for children but in general. Kathi Weeks names this phenomenon as ‘the expansion of the scope of necessary labor’.⁵⁰ Social-movement challenges to normative orders of gender and sexuality of course have played a critical role here, although the material logic of their successes is to be found in the phenomenon named by Weeks.⁵¹

Another way of saying this is that individuals who cannot negotiate the passage into and across the life course following old trail markers do not have the option to enact gender as their elders did – at times generating incomprehension or even hostility. Fordism established certain rigidities of self in correspondence with routinised production. In contrast, zero-sum economies of effort in the low-productivity service economy have no fixed typologies; individuals negotiate their own positions through their relations of interdependency, a process which accelerates slippage in the performance of gender. No example illustrates this mechanism as decisively as the hikikomori phenomenon first recognised and named in Japan, although now observed globally: long-term stagnation of economic growth has been greeted with radical social withdrawal by hundreds of thousands of Japanese young people as worsening economic opportunities became de-aligned from the steps on a rigidified life-course pathway.⁵² Hikikomori also exemplify in extremis how the social damage wrought by economic stagnation expands feminised reproductive labour: the arrival of the phenomenon was met with a new industry of ‘rental sisters’ to gradually coax young men back out into public life: rental sisters are to hikikomori as nurses are to Pittsburgh steelworkers.⁵³ Even short of such extremities, breakdown in intergenerational transmission of gender norms emerges as a generative global possibility, one fiercely resisted by new fascist movements.⁵⁴

What I have argued is that the socialisation of reproductive labour by capitalist development and underdevelopment is producing systematic processes of both working-class social differentiation and social interdependency, processes which are determined by the qualitative characteristics of both supply of and demand for service labour. But the circumstances of their realisation seem unlikely to repeat those of the classic cases. There are possibilities of solidarity within this differentiation, arising not from uniformity but from this interdependency; and those possibilities are not merely latent. In the United States, the struggles of teachers and nurses have been central to the revival of the labour movement. In Latin America, the new pink tide is profoundly linked to the Ni Una Menos struggle against gendered violence.⁵⁵ Today in the Palestinian struggle, the value of sumud, steadfastness and resistance, is rooted profoundly in the everyday social practices required of women to survive the conditions of exile and confinement – not to mention the notable role of doctors and nurses.

To return to methodological reflection, it seems to me that in light of this history of recomposition, the crisis of class analysis at the end of the twentieth century had a quality of historical necessity. It swept over everyone, and could not really be avoided, only made conscious and grappled with. The dialectical challenge here is to recognise that a formal pattern of recurrence and even continuity – working classes are made and unmade and made again, according to the law of value – may nonetheless consist substantively in historical, empirical novelty. Our obligation as historical materialists is to confront that empirical novelty with the enduring powers of our method and to absorb the result of this confrontation.

In the troughs of the cycle of class formation, political and ideological efforts are required to redefine the class in the midst of its recomposition, to rediscover its specific historical features and their political affordances and limits. Such efforts, as Przeworski insisted, must precede phases of more open struggle between classes and are characterised by an open-ended dynamic in which the forces and relations of production do not dictate any specific historical configuration for the proletariat. The proletariat gains its shape through a political process, the ‘struggle about class’, that defines its boundaries. As the class recomposes, we will recognise old features, discover new ones, and perhaps make out old ones appearing in new forms: this again is the challenge of baby and bathwater. As Michael Denning observes,

The central dynamic of working class life under capitalism is neither a secular trend toward homogenization, nor an infinite proliferation of differences, but the dialectic between, to use Marx’s words, competition and association. … To the command of neo-liberal globalization – ‘workers of the world, compete’ – we must answer with that old slogan of the global justice movement: ‘workers of the world, unite’. But in doing so we need to put it in new words, new songs, new figures of that yet unimagined collectivity.⁵⁶

Such figuration will arise out of struggles, not in lecture halls. But the expansion of the scope of necessary labour – a process of both distress and cooperation, association and competition – is the social experience for which we need, and may yet find, a name, a symbol, a slogan, and a song to sing.