Migration and the invisible economies of care: Production, social reproduction and seasonal migrant labour in India

Funding information Economic and Social Research Council, Grant/Award Number: ES/K002341/1; FP7 Ideas: European Research Council, Grant/Award Number: ERC‐2012‐ RtG_20111124 This paper focuses on the processes of migrant labour exploitation which are crucial for capitalist growth and the inequalities they generate. Ethnographic research conducted in different sites across India shows how patterns of seasonal labour migration are driven by class relations marked by hierarchies of identity (caste and tribe) and the spatial geopolitics of internal colonialism (region) – differences that are mobilised for accumulation. Labour migration scholarship has mainly explored sites of production. We extend recent social reproduction theory (SRT) and an older literature on labour migration and reproduction to argue that the intimate relationship between production and social reproduction is crucial to the exploitation of migrant labour and that this means we have to place centre‐stage the analysis of invisible economies of care which take place across spatiotemporally divided households, both in the place of migration and in the home regions of migrants. Furthermore, we develop recent work on SRT and migration to argue that an analysis of kinship (gender over generations, not just gender) is crucial to these invisible economies of care. This analysis is important in showing the machinations of capitalist growth and for political alternatives.

policy discussions for the first time and online conferences, seminars, and reports emerged to try to understand internal labour migration in the country.
This paper aims to deepen our understanding of this migrant labour by highlighting the spatiotemporally divided invisible economies of care that are crucial to their exploitationrarely focused on in migrant labour analysis but essential for policies and politics of labour. The focus is on migrant labour at the very bottom of labour hierarchieslow-caste and tribal people who seasonally migrate within India, leaving their homes for a few months a year to take up low-wage precarious work in faraway brick factories, sugar-cane fields, or construction sites, for instance. The central issue examined is how such migrant labour is exploited, on which there is now some literature. But rare are studies that centre a spatially aware theory showing how appropriation of surplus value from labour occurs both at the site of production as well as through labours' social reproduction in what we call the "invisible economies of care" situated not only in the place of migration but, importantly, also in migrants' home areas, through spatiotemporally divided households. We argue that crucial to these "invisible economies of care" are kinship relations (gender and generation; not only gender) and that in the home areas they, in fact, also encompass productive activities, challenging a production-reproduction divide. In a multi-scalar analysis, we also show how these invisible economies of care are underpinned by class relations marked by hierarchies of identity (caste and tribe) and the spatial geopolitics of internal colonialism (region) which drive seasonal labour migration. Our analyses are important not only in showing the machinations of neoliberal capitalist growth but also for the question of what is to be done, explored in a concluding discussion.
The paper is based on ethnographic fieldwork conducted as part of a major India-wide research programme focusing on five different sites (in the states of Tamil Nadu, Kerala, Himachal Pradesh, Telangana, and Maharashtra) that we led. 1 This research signalled the significance of seasonal labour migration in understanding how and why India's low castes and tribes -Dalits and Adivasi communities who, at 16.6% and 8.6% of the Indian population, make up 1 in 25 people in the world remain at the bottom of the socio-economic hierarchy despite economic growth . We also draw on ethnographic research by one of us: long-term participant observation living with tribal communities in the state of Jharkhand (including moving with them to migration sites) over a total of four and a half years (Shah, 2010(Shah, , 2018. While there is a large literature on the migrants' point of viewearlier writings by one of us focused on how Adivasis saw migration to brick kilns as a space of "freedom" from constraints at home (Shah, 2006)here we pay attention to the structural dynamics that our cross-nation research allows us to observe.

| THE SPATIAL POLITICS OF LABOUR EXPLOITATION
Capitalism has long relied on migrant labour that is paid less, works longer, harder, more flexibly, and is more easily controlled than local workers (Castells, 1975;Kelly, 2002;Smith & Winders, 2008). Neoliberal restructuring which has accelerated accumulation and dispossession has directly led to the growth and exploitation of a precarious migrant workforce (Ferguson & McNally, 2014;Hanieh, 2019). ILO estimates show 164 million such international migrant workers in 2017 (ILO, 2018). But low-wage international migrants are numerically dwarfed by low-wage domestic migrants within countries (Kofman & Raghuram, 2015, p. 23); in China alone this involves 281 million people 2 and in India,100 million. This section lays out our theoretical framework for analysing exploitation processes of seasonal migrant labour. We argue that by transcending existing divides between analyses of international and domestic labour migration we can gain a better understanding of some core aspects of exploitation. We draw on feminist political economy and scholars of race and ethnicity to show how capitalism articulates with and reinforces gendered, racialised, and classed oppression asthrough labour migrantsdifference is mobilised for accumulation (Balibar, 1991;Bhattacharya, 2018;Buckley et al., 2017;Ferguson & McNally, 2014;McDowell, 2015;Werner et al., 2017). This leads to our main discussion. Scholarly attention has focused on how a low-cost labour force is created in the production process and the wider work setting. Drawing on theories from the 1970s on labour migration and reproduction in sub-Saharan Africa as well as recent social reproduction theory (SRT), we argue that this analysis should be extended across multiple scalesfrom nation to regions, into migrant households, and back out.
In our view, similar processes can influence the conditions of both international and domestic migrants, and we show this by drawing on literatures and analyses of both. One aspect of this is the "hyper-precarity" of migrant workers (Lewis et al., 2015), predicated on their non-citizen status and harassment by government institutions (see also Buckley et al., 2017). Coercive immigration regimes curtailing the social rights of migrants have been shown by scholars to be deliberate policy; it is profitable to make migrants illegal because they become vulnerable, controllable, and hence "cheap" labour (e.g., Burawoy, 1976;Castells, 1975;Ferguson & McNally, 2014;De Genova, 2002;Sassen, 1988). Most of this research focuses on cross-border migration; here we show how similar processes exist in domestic migration. There are of course differences.
Most importantly, unlike many international migrants, domestic migrants are formally still citizens. Nevertheless, as we show, on migrating they can de facto be treated as non-citizens, making them similarly vulnerable. 3 Our focus is on Adivasi and Dalit seasonal labour migrants in India. We have argued elsewhere that, similar to oppression by race and ethnicity in other parts of the world, caste and tribe-based oppression entrenches Adivasis and Dalits at the bottom of socio-economic hierarchies in India Shah et al., 2018). Here, in exploring the precarity of domestic migrants, we show that caste and tribe-based stigmasincluding those held by local workers about migrantsare underpinned by a spatial geopolitics of uneven development. The regions that supply migrant labour for the lowest rungs of the labour hierarchy in more developed parts of the country are marked by a long history of domination and subjection, namely a process of internal colonialism. Racism/casteism and the spatial politics of internal colonialism work together to disadvantage migrants: class, caste/tribe, and region all matter in the exploitation of migrant labour.
Second, "invisible economies of care" across the spatiotemporally divided households are shown to be crucial to migrant labour exploitation. Here we draw inspiration from two literatures. In the 1970s, anthropologists researching domestic seasonal migration in sub-Saharan Africa explored the relationship between production and reproduction. Meillassoux (1972Meillassoux ( , 1981 in Southern Africa highlighted how capitalism (gold mining) used migrant labour because the social security cost of that labour could be passed onto their agricultural communities at home. Burawoy (1976), analysing migrant agricultural labourers in California and migrant gold miners in South Africa, made similar points; the system allowed the state/employer to externalise reproduction costs of labour. This made labour migrants "super-exploited" (Meillassoux, 1981): capital cheapens labour from the countryside much below the standard cost of local workers. They can be paid less than required for their full maintenance (and have worse terms and conditions of work) because capital can rely on their families and kinship networks back home to care for them.
The insights from this classic literature are rarely drawn on today. 4 Inspiration to rethinking the production-reproduction relationship more often comes from non-migration related literature by Marxists and Marxist-Feminists, also from the 1970s, which highlights that reproductive activitiessuch as women's unpaid domestic and care workmust be considered part of the productive labour that generates value (Dalla Costa & James, 1972;Federici, 1975;Fortunati, 1995Fortunati, [1981; Mies, 1982Mies, , 1986. This is because such reproductive activity enables the day-to-day survival of those considered productive workers (usually male industrial workers) and reproduces the next generation of workers. This analysis has enjoyed a revival as "SRT" (Bhattacharya, 2017;Ferguson, 2020;Fraser, 2016;Katz, 2001;Mezzadri, 2019). One focus of this recent literature is international migrants, often women, in commodified care work or "care chains" (for instance, nurses, maids, nannies) especially in well-off parts of world society. Recognising the limits of a narrow focus on care work, some SRT analysis has widened to the role of migrant labour in general under capitalism, looking beyond the workplace and exploring not only the "house-work" that sustains the worker but also the wider "array of activities and relationships in maintaining people both on a daily basis and intergenerationally" (Kofman & Raghuram, 2015, p. 2; see also Ferguson & McNally, 2014;Ferguson, 2020;Kofman, 2012;Pearson & Kusakabe, 2012). Nevertheless, analytically and spatially, this research still predominantly focuses on the place of destination for international migrants. Consideration of the role of production and reproduction in the home region at the same time is rare, as pointed out by Gidwani and Ramamurthy (2018). Though the rationale for doing so is succinctly put by Ferguson and McNally (2014, p. 13): that in using migrant labour, capital and the state "regularly draw from a pool of effectively 'cost free' labour power on whose past social reproduction they have not spent a dime" (see also Cousins et al., 2018;Mezzadri, 2020). By this, the SRT literature is now reaching similar conclusions to those of the 1970s anthropology work on labour migration and reproduction in sub-Saharan Africa. Here, we deepen the line of shared inquiry between these literatures to promote an overall analysis that focuses on production and reproduction in both work destinations and home regions of migrants.
The need for an overall analysis of migration that also includes home regions undermines the analytic separation between reproduction and production. We start again with the older anthropology literature, namely O'Laughlin's critique of Meillassoux. O'Laughlin (1997, p. 21) argues that there is no analytical opposition between productive and reproductive activities as capitalist wage labour is also reproductive. This position is elaborated through Fraser's recent argument that social reproduction is an indispensable background condition for the possibility of production (2014, p. 61); it is central and necessary to the existence of waged work, the accumulation of surplus value, and the functioning of capitalism. In fact, Fraser suggests that the very separation of "economic production" from "social reproduction" is constitutive of capitalism itself (2014, p. 67) and that this separation is not fixed as the categories are also co-constituted and their separation is normatively and politically contestable (2014, pp. 69-70).
In this vein, focusing on both "receiving" sites of migration and "home" regions of migrants, our analysis will develop how productive and reproductive activities are analytically and empirically intertwined by drawing attention to the invisible economies of care across spatiotemporally divided households that sustains workers. We highlight that the "invisible economies of care" can include certain "productive" work such as maintaining assets that enable a livelihood and a safety net back at home; fields for some subsistence production, forests that give firewood etc., and livestock. At the centre of invisible economies of care are shown to be the productive and reproductive work of kinship over generationsthat is the work of kin (e.g., husband or wife, parents and children, or brothers and sisters) who are left at home to maintain the household of a seasonal migrant (both in place of migration and at home). We thus expand the focus of recent migration and social reproduction theory from gender to kinship over generations.
Conceptually, the terms "social reproduction" and "care" have overlapped but also had distinct meanings and usages. Different authors have also used one term in varied ways (Smith, 2019). We draw on Fraser, who views social reproduction as the "forms of provisioning, caregiving and interaction that produce and maintain social bonds. It is "variously called 'care', 'affective labour' or 'subjectivation'" and involves the socialisation of the young, making communities, social cooperation etc., and it takes place in households, neighbourhoods, and public institutions such as schools and child centres (2014, p. 61).
In recent years, the term "care" has been side-lined and "social reproduction" has gained ground as feminist research on globalised migration and labour has moved from a narrow focus on care chains and paid care work (Ehrenreich & Hochschild, 2002;Hoang et al., 2012;McDowell, 2015) to wider social reproduction concerns (Kofman, 2012). As pointed out by Kofman and Raghuram (2015, p. 12) and empirically shown by Pearson and Kusakabe (2012), a narrow definition of care limits the analysis of migration and excludes wider care responsibilities of migrant workers outside the care sector.
Nevertheless, the term "care" remains useful, not the least for a politics of mobilisation (see for instance Tronto, 1993Tronto, , 2005, who promotes an "ethics of care"). In this paper, we therefore move interchangeably between "social reproduction" and an expanded concept of "invisible economies of care" (beyond both unpaid/paid domestic and caring labour) which are crucial to sustain workers. By this we view care as a broad concept encompassing "everything we do to continue, repair, and maintain ourselves so that we can live in the world as well as possible" (Tronto & Fisher, 1990, p. 41). Of course, the point is that the capitalist employer never acknowledges such economies, the costs of which are borne by the worker and their kin. We call this an "invisible" economy because it is never considered in worker remuneration; making care invisible constitutes the very basis on which the overall workforce is cheapened.

| IDENTITY, REGION, AND SEASONAL LABOUR MIGRATION IN INDIA
This understanding is now developed through an analysis of seasonal labour migration within India. The numbers are hard to establish. 5 The most reliable estimate suggests 100 million people are involved (Deshingkar & Akter, 2009); nearly one quarter of the labour force. But this figure is a decade old and probably an underestimate as seasonal work is rarely visible in surveys/censuses. 6 Such migration has a long history (de Haan, 2002;Tumbe, 2018) but its significance has increased over the last 50 years (Breman, 2007;Datta, 2016;Jain & Sharma, 2019;Mosse et al., 2002). Moreover, as we will argue, all figures are underestimates because of their basis in a methodological individualism which means that none of the data measure what matters the most.
Following Breman (1996), there is now a small but strong core of research highlighting issues such as debt and dependency, labourer-contractor relations, working conditions, income and remittances, gendered migrant dynamics, the worldviews of seasonal labour migrants and the reasons why people migrate. 7 Indeed, it is increasingly being recognised as a policy concern and an issue for social mobilisation.
Seasonal migrant workers can include men and women across castes, but those at the bottom of the social hierarchy (Dalits and Adivasis especially) are often found doing the hardest work in the worst living conditions. 8 We show that there is also a strong geopolitical dimension to patterns of labour migration based on the class politics of the vastly uneven development across India. Migrant workers who are the worst off often come from the poorest regions, with the lowest Human Development Indices (HDI) and worst multidimensional poverty: Jharkhand, Bihar, Chhattisgarh, Madhya Pradesh, Odisha, and east Uttar Pradesh. These are typically places whose wealth has been extracted by dominant castes and classes for centuries, leaving little for most of its inhabitants.
In our cross-nation field research, we found that in the three field sites in states that had "high" HDI -Kerala, Tamil Nadu, and Himachal Pradeshthe cheapest and most exploited workforce was low-caste and tribal seasonal migrant labour from the central and eastern Indian forested belts, 9 doing work that even local Adivasis and Dalits in areas they migrated to would not touch or were moving away from. A whistle-stop tour of these three field sites illustrates the relationship between capital and labour through the dynamics between local low-caste labour and seasonal migrant low-caste and tribe labour, before focusing on a "home" area of that labour.
The first site, where Raj worked, is the Kerala tea plantations where, from the 1860s, Tamil Dalits were brought as indentured labourers by colonial planters. Over the years, they won some rights (some security of employment, some free health care, and the promise of pensions). But the 1990s global collapse of the tea economy, alongside Indian economic liberalisation, brought the closure of some plantations, the take over of others by Indian multinational companies, and cutbacks in jobs and labour rights. Dalit men lost permanent work in the tea factories which closed. Some older Dalit women maintained "permanent" tea-picker jobs but with severe welfare cuts and others were reduced to "temporary" work status. As they fought for their rights (Raj, 2019), the plantation owners seized the moment to introduce a new, cheaper workforce from eastern India. Where Raj worked, these were Adivasis (men and women) from Jharkhand, brought in as casual workers for just eight to nine months.
Second is the coastal belt of Cuddalore District in Tamil Nadu. From the mid-1980s, economic liberalisation turned this landlord-dominated agricultural belt into one developing chemical and pharmaceutical industries. Where Donegan worked, an export-oriented gelatine factory emerged (with a Japanese business partner and joint ventures in China and Taiwan). Locally dubbed "the bone factory," it turned cow bones into gelling agents used in food, pharmaceuticals, photography, and cosmetics. The toxic and polluted work of cleaning the bones was done by local Dalit men hired as daily contract wage workers, controlled by the erstwhile landlords whose relatives owned the factories, who got the permanent jobs and dominated the supply of labour and material to the factories. But, as in the tea plantations, when the Dalits protested their exploitation, their bargaining power was cut by employers bringing in cheaper seasonal migrant Adivasi and Dalit labour from Eastern India (Odisha, Bihar, and Jharkhand) for 8-10 months of the year.
Third, the final field site with a high HDI, where Axelby worked, is Himachal Pradesh, the homes of the Gaddis and Gujars, both tribal groups, who diversified out of nomadic agro-pastoralism. The Gujars went to Punjab to sell milk and the Gaddis moved more locally into skilled, semi-skilled, and some manual work. But here too one finds that for the hardest manual work constructing roads in the extreme cold of the high Himalayas, cheaper seasonal migrant labour from Bihar and Jharkhand was brought in. As Sabhlok (2019) has shown, this labour is mainly tribal and Muslim, comes in gangs (usually kin-related) led by a "mate" from Jharkhand, is selected for its docility and capacity to endure hard labour in high altitudes, and lives in the poor cramped conditions of roadside tents.
From these "receiving sites," let us now turn to a "home" area of this labour. We focus on Jharkhand (where one of us conducted ethnographic research), one of the states with the highest proportion of seasonal outmigration in the country, 10 the lowest HDI, and a high incidence of poverty. 11 This is despite its enormous natural resource wealth: 40% of the country's mineral reserves. Alongside the minerals, its forests were plundered for their timber from colonial times and land revenue was extracted by forcible settlement of the tribals. Outsidersespecially from Bengal and Biharwere the main beneficiaries of this wealth, taking the best jobs and resources, such that scholars and activists have argued that a process of internal colonialism is at work (Devalle, 1992). 12 Moreover, after the 1830s people were taken (many forcibly) to work in the tea plantations of the north-east (Assam, West Bengal), where some stayed for generations. The Adivasis, who originally dominated the area, experienced an outright fall in their numbers and are now among the poorest of the local populations. Today, unable to make ends meet, many migrate seasonally to work as manual labour.
In the rural areas of Jharkhand where Shah (2013) lived last, though there was an abundance of bauxite and coal nearby, almost everyone lived in mud houses with no sanitation, running water, or electricity, and 65% of the population was illiterate. Across 400 households (2,000 people) in nine villages surveyed in 2009-2010, just under half had a member who migrated seasonally (46%). Seventeen percent of OBC 13 (Sahu) households had migrant members, as had 42% of Muslim households, 45% of Adivasi (Oraon, Lohra, and Kherwar) households, and just over 70% of Dalit (mainly Ghasis and Bhuiyas) households.
The workers travelled long distances on crowded jeeps, buses, and trains (and sometimes foot), typically leaving for six to eight months, after the paddy harvest in November, returning for the July rains. It was rare that entire households departed; usually someone stayed back to look after the home. It was common for members to take turnsone person migrating one year, another the next. Contractors negotiated work with employers and navigated the difficult long journey for labourers (who were largely illiterate) and often gave advances 14 which tied them into the migration process (in "neobondage"; see Breman et al., 2009). Labourers rarely had direct connection with employers.
Reflecting developments elsewhere (Picherit, 2009(Picherit, , 2018aRaj & Axelby, 2019), the contracting system changed from contractors who were higher status (higher caste) and relatively wealthy, engaging several hundred labourers, to Adivasis and Dalits contractors taking small numbers (from a handful to 20) of their own family members, kin, and acquaintances (Shah, 2006). 15 One effect of these changes is that there were a wide variety of contractors and places to migrate to. It is quite common for migrants to go to one destination one year, and an entirely new place the next.
While some of the Adivasis and Dalits from Jharkhand ended up in manufacturing units such as a Tripura beer factory, most went to carry bricks in kilns in West Bengal, Bihar, and Uttar Pradesh (Figure 1); as agricultural labour in West Bengal, Uttar Pradesh, and sometimes even Gujarat/Punjab; and to construction sites. In destinations of work they went into the poorest of all jobs, paid at piece rates. Accommodation was temporary housing (make-shift brick slums, or plastic or hay tents), with poor access to water and no electricity. Payment was only at the end of a season. With the advance and meagre weekly subsistence allowances cut from the final payment, accounting left to the whims of the clerks of employers, and money given to contractors who may take their own share of the labourer's salary, it was not uncommon for Adivasi and Dalit migrants to return with not much more than when they left (though some reported "savings").
Some intra-caste differences must be noted. First, in line with other studies and all India figures, while labouring conditions were universally poorall went into indecent informal sector workthere was a qualitative difference between the work Adivasis and Dalits could get (in the absolute pit of the labour hierarchy) and that which OBCs and Muslims attained (marginally better work) (see also, e.g., Jain & Sharma, 2019;Srivastava, 2019). The latter considered the work and the accompanying living conditions of Adivasis and Dalits low status and, though only marginally better off themselves, were usually paid monthly rates (rather than piece rates), which in theory guaranteed some monthly income, had slightly better F I G U R E 1 Adivasi seasonal migrant labour from Jharkhand head loading cooked bricks at a factory in Uttar Pradesh.
Source: Photo by Jens Lerche, 2017 accommodation (concrete rooms with a fan), and typically were in the construction sector as brick layers, electricians, mistrys, or manual labour in manufacturing. Of course, some bucked the trend. Adivasis who had stronger assets and better social networks could sometimes join the slightly higher castes in their places of work. But it was rare for OBC and Muslims (even if they were much poorer than Adivasis) to go to the worst work sites where the Adivasis and Dalits ended up ( Figure 2).
Second, there were intra-household gender differences in patterns of migration between different caste groups. Whereas it was rare for Muslim or OBC women to seasonally migrate, Dalit and Adivasi women migrated (like their men) as the norm, not the exception. 16 Moreover, it was common for the latter to either migrate without their partners or look after assets at home alone.
Third is the difference between Adivasi and Dalits. Marking an inverse to the India-wide trend (Adivasis tend to fare worse than Dalits), Dalits in the Adivasi-dominated "home" areas (here, the forested belts of Jharkhand) fared worse than Adivasis. This was because Dalits had less land than Adivasis to rely on back at home. These qualitative differences are conjoined with strong spatial geopolitical dynamics of the political economy of seasonal labour migration. Regions that have experienced a long history of internal colonialism, with outsiders siphoning off F I G U R E 2 Adivasi seasonal migrant labour from Jharkhand carting uncooked bricks at a factory in Uttar Pradesh.
Source: Photo by Jens Lerche, 2017 SHAH AND LERCHE indigenous wealth leaving behind high levels of poverty, are now providing labour for the lowest rungs of the labour hierarchy in more developed parts of the country. This is as true of Chhattisgarh and Odisha as it is of Jharkhand. Much of the land in these belts was historically protected for Adivasis to halt land alienation and dispossession under colonialism, through the Chotanagpur Tenancy Acts and other land laws, and enshrined in the Indian Constitution's Fifth Schedule. But these laws are being undermined. Since the mid-1990s, hundreds of memorandums of understanding have been signed between multinational and national corporations and state governments promising land for mineral extraction, industries, and power plants. The result is a new wave of accumulation by dispossession and even an internal civil war, which human rights activists argue has the main purpose of dispossessing the land from the people who live there (Shah, 2018). Most of the central and eastern Indian belt is now highly militarised, in the name of clearing India of its terrorists, its greatest single internal security threat: the Maoist-inspired Naxalite rebels (Shah, 2018). Meanwhile the armies of vulnerable migrant labourers from these alleged rebel lands grow.
There are similar patterns of internal colonialism at work in other parts of the countryfor instance, in the likewise poor tribal areas of Rajasthan, Madhya Pradesh, Maharashtra, and Gujarat providing seasonal migrant labour for elsewhere in those states (Breman, 1996;Jain & Sharma, 2019;Mosse et al., 2002). No doubt, we could also extend the analysis to cross-border migration into India; Nepalis and Bangladeshis also labour at the bottom of the Indian labour hierarchies.
Of course, spatial relations of domination and subjugation which mark internal colonialism are not the only dynamics of this migration. In some parts of the countrysuch as the agricultural plains of Bihar and Uttar Pradeshescaping oppressive landlords has been a reason for low-caste migration (Gidwani & Sivaramakrishnan, 2003;Rodgers & Rodgers, 2011). The point is that the co-constitution of the spatial geopolitics of uneven development and caste/tribe trump identity alone, or region alone, as significant features of the politics of oppression and exploitation underlying patterns of seasonal labour migration.

| THE INVISIBLE CARE ECONOMIES OF MIGRANT LABOUR
The use of migrant labourin India seasonal migrant labouris particularly profitable for employers because they can be exploited more than local labour in multiple spaces: the receiving areas where migrants work and live, as well as home regions which are crucial for migrant social reproduction. The spatial dynamics of exploitation hide the relationship between production and care that unfolds across multiple sites. These concerns force us to move between different scales of analysisfrom nation to region, from migrant work sites and into migrant homesand centre the spatiotemporal division of households.
One domain of exploitation is in relation to receiving areas. Adivasi and Dalit migrant labour from central and eastern India worked harder for less pay than even local low-caste labour. In Tamil Nadu, while local Dalit contract labourers in the bone factory worked one eight-hour shift, six days a week, the eastern Indian migrants worked between one and four shifts in a row, seven days a week, taking turns to sleep in the factory corner. In Kerala, the Jharkhand Adivasis were employed only as casual workers who would never be promoted to the "temporary" or "permanent" status which Dalit workers had. Kept as contract labour meant no access to provident funds, annual leave, or medical leave provided to "permanent" and "temporary" workers. Piece rate end-of-season pay dominated for many migrants and they would often be cheated of some earnings.
Living conditions of seasonal migrant workers are more squalid than those of local low-caste workers. In the Himalayas, the eastern India Adivasi road construction workers slept in tents and huts, often under freezing conditions in the high mountains. In Tamil Nadu, the eastern Indian migrants were housed in disused crumbling buildings, 40 to 70 people squeezed in one large room, only able to sleep in shifts, while the local Dalits had mud or brick houses. In the brick factories of West Bengal, Adivasis from Jharkhand lived on the edge of the factory in rudimentary shacks with low tin roofs and no electricity, 500 people sharing water from three taps, using the Hooghly River for shitting and bathing. No local labour is found there. In Kerala, several Adivasi families from Jharkhand were forced into the small tenements that only one local Dalit family would occupy.
Seasonal migrants are also not eligible for elementary citizenship rights and welfare measures (cf. Abbas, 2016;Mander et al., 2019;Mander & Saghal, 2010). They cannot access rice, dal, and kerosene for the poor subsidised by the state's Public Distribution System. 17 Hospitals are difficult to access, schools impossible; children sometimes accompany parents to insecure work sites and begin working themselves. Language gulfs prevent migrants from making demands and keep them highly alienatedin Kerala, the Jharkhandi Adivasis did not speak Malayalam; in Tamil Nadu, the eastern Indian migrants could not converse in Tamil. They cannot vote in elections unless they return home. For the authorities, they simply do not exist and are treated as second-class citizens, if citizens at all. Moreover, casteism creates an overall hostile environment for migrants, including from the local workforce who can participate in their stigmatisation. Workers from Jharkhand and Bihar have been beaten on trains and chased away by other workers tormenting them as uncouth, jungli (wild and savage), and accusing them of polluting regional cultures and languages. In Maharashtra in 2008, more than 40,000 Bihar-Jharkhand workers fled after attacks by local Hindu right-wing Shiv Sena Marathi workers. Similarly, in 2017, tens of thousands of eastern Indian workers fled north Gujarat after being attacked by Gujaratis for allegedly raping a 14-month-old. 18 The xenophobia stigmatises them as bringing disease, violence, job insecurity, and accuses them of being BIMARU, a Hindi or Urdu word for illness, which has become an unfortunate term for the general poverty and underdevelopment of states such as Bihar, Odisha, and Jharkhand.
For tribal migrants who have access to land and forests, who have historically tried to move away from the caste oppression of the plains (Shah, 2010), migration forces them into the heart of a world of hierarchical distinctions to experience new stigmas. Tribal migrants are often regarded as "simple, ignorant and uncivilised," their girls "loose," 19 thoughunlike Dalitsthey are often seen as "clean." The stigma of filth and pollution attached to Dalits is prevalent everywhere, such that in some of our sites Dalits hid their names to rent a room. 20 The overall conditions of this migrant workforce were aptly summarised by Breman (1990) in relation to Adivasi and low-caste migrant cane-cutters in Gujarat, "Even Dogs Are Better Off"! A second domain of exploitation is in relation to the domestic economies of care that sustain the whole system of migrant exploitation. The links between the migrants' work and home regions have been explored by some work on seasonal migration in India (Breman, 2007;Mosse et al., 2005;Pattenden, 2016Pattenden, , 2018Picherit, 2012Picherit, , 2018aRogaly, 2003;Rogaly & Rafique, 2003;Rogaly & Thieme, 2012). But these studies have not analytically addressed the relationship between production and the invisible economies of care at the heart of the appropriation of value from labour.
How exactly are the domestic economies of care important to the value generated from migrant labour? The example of the migrant home areas of Jharkhand shows that the economies of care can include the land that provides grain for some of the year for the migrant household, the forests that provide firewood and timber for building houses, and the cows, buffalos, and goats that provide a livelihood and a safety net back at home.
So, the economies of care for household reproduction encompass production too, of use-value (as in subsistence farming or hunting and gathering), but also simple commodity production which itself exploits labour (e.g., picking kendu leaves for making bidis to sell to traders for cash to buy everyday goods necessary to reproduce households). Of course, not all migrant households will have access to such "safety nets" back at home. But, as argued earlier, the point is not to demarcate production and reproduction, but to show how production and remuneration of migrant labour in the migrant destination hide the value produced by an entire economy of care of that labour back home (which itself can include household production and direct exploitation).
Furthermore, the economies of care include the kin who enable the migrant to work for a part of the year in a certain period of their working life. Kin tend the fields, keep the household functioning, look after livestock, and help ensure that the migrant has something to come back to, is looked after in old age when he/she gets no pension, and becomes part of generating the next generation of the labour force. 21 In some cases, the worker at the production coalface is a woman and it is her partner who sustains the household. In other cases, it is both the partner but also older generationsin particular grandparents and uncles and auntswho do the care work enabling a worker to migrate. Sometimes, it is even children who are crucial to the everyday reproduction of a migrant household as they do the agricultural work, cook, and clean, bring fire-wood, and look after other children. Some of these kin may have migrated themselves in their youth, and others will become the next generation's migrant labourers caring for the current migrant when he or she is old.
Indeed, an analysis of men and women, and different generations, is more important than just gender in the invisible economies of care which sustain a migrant worker in production. If Marxist-Feminists highlighted the significance of a gendered analysis and of domestic work done by women to be counted as work, the arguments presented here highlight the need to centre an analysis of kinship over generations and across spatiotemporally divided households in our analysis of the exploitation that enables capital to accumulate from migrant workers.
There is one more point to note before concluding. Taking seriously the relationship between production and the economies of care forces us to rethink the premises of migration statistics. All current estimates focus on individuals, but this ignores that it is the entire household that participates in enabling one person to migrate. It is not just that people take turns to go, but also that, as we have argued here, production cannot be separated from reproduction. More appropriate estimates would therefore include all people across the household affected by and enabling labour migration (even if at any one moment in time all are not physically moving) as the very point of such migration is that the migrant is inextricably linked to other members of the household. Seen this way, several hundred million people -400 million would not be too large an estimateare intimately involved in seasonal casual labour migration in India.

| THE LONG MARCH OF WHAT IS TO BE DONE
We have attempted here to create a framework for understanding the exploitation of migrant labour, arguing for the analysis of international and domestic migration in the same theoretical frame, empirically focusing on seasonal casual labour migration within India. If by the summer of 2020 anything positive has come out of the COVID-19 crisis in relation to India, it is that scholars, activists, and some policymakers have been reinvigorated in their pursuit of a new approach to providing social security to migrant workers. This would certainly be extremely important, but it would not be enough to account for the multiple and often hidden ways in which migrant labourers are exploited. We have argued for the need to explore the invisible spatial processes of migrant labour exploitation which feature the geopolitics of internal colonialism (alongside caste and tribe oppression), and the extraction of surplus value from labour taking place both at the site of production and through the invisible economies of care spanning spatiotemporally divided households. We have done so by critically resuscitating early anthropological understandings of how employers use migrant labour to externalise the cost of maintaining the labour force, and building on contemporary analyses of migration and SRT, and studies of labour, race/ethnicity, and class.
The analyses of our ethographic field research fleshed out the theoretical argument: that the relationship between the migrant workers' involvement in production and the invisible economies of care which sustain them, and which are thus crucial for surplus value extraction from them, are central for the analyses of labour. Rather than drawing boundaries between production and reproduction, or fretting over the use of care versus social reproduction, we have argued for an expanded notion of the invisible economies of care. We have shown the significance of paying attention to how the care of the migrant labour force is spatiotemporally divided. There is the day-to-day reproduction at the work place. But there is also the care of the migrants at home which allows him/her to migrate in an annual cycle for a part (the working years) of their entire lifecycle. This process involves assets back in the village and the people, especially kin, who maintain them. The migrant's role in production and the analyses of this are fundamentally tied in with the entire lifecycle of care across several sites, through kinship and over time across generations.
There is much more work to be done. Exciting are the new directions opened by SRT and others drawing on feminist theory to propose new directions for the study of capitalism (Bear et al., 2015). In relation to exploitation of migration labour, one key arena is in trying to understand more deeply how exactly intra-and extra-household dynamics of care are being reconfigured because of migrationfor, as some of the best anthropological work has pointed out (see Harris, 1984; for a review, Moore, 1988), we cannot assume what a household is but must study the transforming dynamics of family and kinship, gender and generation. 22 Another key issue is the practical-political implications of our analyses. We argue that the informal economies of care and the value generated beyond the workplace must be as central to any question of what is to be done as workplace concerns. This we see not only in relation to improving the conditions of seasonal migrant households but also in relation to keeping alive alternative values from which an anti-capitalist politics may emerge. We address these implications in relation to India but their relevance is of course wider.
Starting with the standard focus on workplace issues, the mobilisation of seasonal migrant workers in India is rare, although there are exceptions. Unions have historically only focused on the working classes in formal jobs (Breman, 1999;Parry, 2009) even though 93% of the Indian workforce (National Commission for Enterprises in the Unorganised Sector, 2009)which includes seasonal migrantsis informal (can be fired any moment, has no welfare benefits, no medical benefits, no insurance, no pensions). Labour legislation includes an Inter-State Migrant Workmen (Regulation of Employment and Conditions of Service) Act, 1979. But workers usually don't know their rights (Government of India, 2017a) and the state does little to implement them. Labour inspectors rarely operate. Agitation for law implementation is one field of struggle. But it is unclear if more legislation would be beneficial. At any rate, the move is in the opposite direction, as the government is "reforming" labour legislation, legalising parts of existing illegal employment practices.
Obstacles to mobilising are many. Language barriers, treatment of migrants as second-class citizens, and permanent transience lead to isolation. Employers actively clamp down on attempts to organise through private security firms and by placing labour from different states and groups at the same worksite, making it hard for them to unite (Srivastava & Jha, 2016). Circulation in work with no protection can encourage low-caste labourers to be dependent on political patronage and persistent debt-bondage, preventing collective action (Picherit, 2018b). And how can people organise if one year they are in one site and in another year in an entirely different place?
State repression is another obstacle. This is particularly so for the most radical attempts at organising; those rare organisations working with contract workers in the informal economy. The Maoist or "urban Naxal" label today is used to discredit the most radical of labour organising, and allows for the state to use draconian anti-terror laws and incarcerate alleged suspects for years without even providing a charge-sheet, let alone bringing them up for trial. We have seen such labelling in the case of the Maruti workers in Manesar organised in protest (Teltumbde, 2012) and those like Anand Teltumbde who wrote about them (who is now in prison). Or for those representing informalised workers at Reliance Energy (now Adani Electricity) in Mumbai. Or those who carried on the legacy of the assassinated Shankar Guha Nyogi and his Chhattisgarh Mukti Morchaone of those rare organisations that worked with contract workers in the informal economy (Parry, 2009)such as Sudha Bhardwaj, also imprisoned.
The working-class movement itself can also be a barrier. It has neglected migrant workers and sometimes reproduced racist attitudes towards them, such that workers are used to break each other's strikes (see Burawoy, 1976, p. 1,062;Anitha et al., 2018). Although there are now unions in other countries (e.g., Brazil) with anti-racism units, implementing affirmative action, and tackling race-based wage disparities, in India the discrimination of workers because of their caste, tribe, or religion has rarely been fought against by working-class movements .
Crucially, as indicated, the co-constitution of production and care points to mobilisation beyond the workplace. Feminist politics highlighted the limits of production-based organisation. A range of campaigns resulted, the most famous of which, "Wages for Housework" (Dalla Costa & James, 1972;Federici, 1975), argued for domestic work of women to be given a wage. Notwithstanding critiques, 23 these struggles highlighted that (male) workers in production could only be workers because of the work that women did in the home.
Similarly, for migrant labour households there is a need to consider centrally the spatiotemporally divided economies of care of migrant labour as an arena for raising demands. Labour scholars have recognised the need to move away from single location struggles to their spatial and temporal fragmentation, allowing for fluid, episodic, and multi-sited forms (Pattenden, 2018;Picherit, 2012). However, rarely do they keep the economies of care central. Focusing on the economies of care would need to include both the wider conditions of the workforce in the destination and their circumstances in their home region. In the destination place, it would be important to remove migrant illegality/criminality and language barriers; provide access to welfare measures lost on migrating (e.g., PDS card); and provide decent housing and childcare. In the home regions, the main issue would be to recognise capital's appropriation of the value production of the entire household when any one worker is employed for a part of the year and over the full lifecycle, including after workers are too worn out to keep working. Taking a partial leaf from the "wages for housework" campaigns, the aim could, for example, be to ensure either that the migrant (male or female) wage includes the cost of running the workers' household (which includes men and women, children and elders) all year round or that the household was otherwise additionally and directly remunerated. Social security not only for the migrants but also for those staying back in the villages could be another demand.
By keeping central the spatiotemporal division of the migrant household, there are other important battles which emerge. In the Adivasi-dominated areas of central and eastern India such as Jharkhand, household reproduction of migrant labour is dependent on access to land and forests and the ability to sustainably farm these. For such regions, struggles against internal colonialism as manifest in the undermining of the protection of land and forests for Adivasis would be crucial, as would further support for Adivasi livelihoods in home areas. In other areas, where for example landless low-caste Dalits dominate the out-migrating labour force, demands for redistributive land reforms may be significant.
Centring the invisible economies of care also has potential implications for an anti-capitalist politics. The Adivasi-dominated forested heartlands of India, home of seasonal migrants at the bottom of the labour hierarchy, have long been spaces of what Guha (1998) once called "dominance without hegemony." Reminiscent of Rancière's (2012) "Proletarian Nights," they have allowed for spaces where workers can "withdraw themselves, intellectually and materially" from the domination of their bodies and actions, in a struggle to free and to engage oneself on one's own terms, to liberate one's own life from the bonds of one's situation (Rancière, 2012, p. ix). In a country marked by hierarchy and inequality, access to land and forests has enabled Adivasis to keep alive alternative relatively egalitarian values which have manifested themselves in kinship and internal political organisation, the ethic of production and consumption (for instance, the sharing of labour without payment, consuming only for one's needs), and in gender relations (women are not subject to the same degree of patriarchal controls as they are in the rest of India). In comparison with the Dalits of the plains, who suffer from more direct domination and oppression, and internal stratification, and who reproduce caste hierarchies between their communities, a "counter-politics" (Shah, 2019) of the values of Adivasi egalitarianism, shaped by the negotiation of their integration into the state and with the processes of capitalism, could form the basis of an alternative anti-capitalist politics (albeit one that the Maoist guerrillas, and other Marxist in India and elsewhere, would call utopian).
However, the material basis for Adivasi counter-politics is fast eroding with the onslaught of the state and corporate capital nexus. Indeed, today the counterinsurgency forces of the Indian state surrounding the hilly forests mark an impending social death of the Adivasis. This may turn Adivasis into pauperised tribal castes with nothing but their labour power to sell, stripped of their counter-politics. Whether this happens or not remains to be seen.
While there are particularities to the Indian situation, the underlying issues and processes of uneven development are of global relevance. A first step is to include centrally the spatially divided invisible economies of care crucial to the exploitation of migrant labour in scholarship and for an alternative politics.

ACKNOWLEDGEMENTS
This paper draws on the joint work of the authors and on the 28th Wertheim Lecture delivered by Alpa Shah in Amsterdam in 2019 on "The long march of labour: radical change and the social reproduction of migrant labour" and her Silver Jubilee Lecture given in Patna in 2017 at the Asian Development Research Institute International Conference on Bihar and Jharkhand. Thanks to Kathinka Sinha-Kherkoff, Shaibal Gupta, Luisa Steur and the Moving Matters Programme, and the Wertheim Family for these invitations which propelled this piece. Also thanks to organisers and participants for comments on Jens Lerche's presentation at the Centre for Migration and Diaspora Studies seminar, SOAS, 2019. The research on which this lecture is based was funded by a starting grant from the ERC, an ESRC major research grant, as well as an ESRC Research Fellowship. We are grateful to comments from Gavin Smith on an earlier draft and to Patricia Noxolo and Phil Emmerson at Transactions of the Institute of British Geographers for the care they took in publishing this piece and to the comments of two excellent anonymous reviewers.

DATA AVAILABILITY STATEMENT
Data are unable to be shared.

Alpa Shah
https://orcid.org/0000-0003-1233-6516 Jens Lerche https://orcid.org/0000-0003-3817-7058 ENDNOTES 1 Between 2014 and 2017 we worked closely with a team of five postdoctoral anthropologists (Richard Axelby in Himachal Pradesh, Dalel Benbabaali in Telengana, Brendan Donegan in Tamil Nadu, Jayaseelan Raj in Kerala, and Vikramaditya Thakur in Maharashtra) to explore how and why Adivasis and Dalits remained at the bottom of the Indian economic and social hierarchy. We undertook extensive research training and developed comparative research questions, research methods, and writing to generate explicitly comparative work, the result of which is our co-authored book, Ground down by growth . The fieldwork itself lasted a year. Each postdoctoral researcher lived with Adivasi and Dalit communities in their respective sites but comparative research questions were developed in each site with constant communication and consultation with others, including extended visits by the authors of this paper and at least one other postdoc. Here, we draw on ethnographic insights of that work as presented in Ground down by growth  but combine that with Shah's long-term ethnographic fieldwork in Jharkhand (between 1999 and 2010), and migration data she collected over the course of 18 months living in an Adivasidominated area on the borders of Lohardaga, Gumla, and Latehar districts. 2 See https://blogs.wsj.com/chinarealtime/2017/04/30/journey-to-the-rest-chinas-migrant-workers-top-280-million/ (Accessed June 2020). 3 The Chinese "hukou," a domestic passport system, is for instance infamous for denying rural migrants the same rights and benefits as urban workers (Alexander & Chan, 2004). 4 One exception is Cross (2013). 5 See Government of India (2017a) and Breman (2009) for data problems. 6 Government of India (2017b) estimates 140 million seasonal migrants, but the basis of this number is unclear. 7 Breman (1985;1994;1996;2007;), Datta (2016, Deshingkar & Farrington (2009), de Haan (2002, Mosse et al. (2002), Mosse et al. (2005), Picherit (2009Picherit ( , 2018a, Rodgers et al. (2016), Rodgers (2016), Rogaly (2003Rogaly ( , 2009, Rogaly & Rafique (2003), Rogaly & Coppard (2003), Rogaly & Thieme (2012), Shah (2006). 8 See Government of India (2017a, pp. 15-16) and case studies from brick factories (Breman, 1996;Committee of Concerned Citizens constituted by Jagrut Kashtakari Sanghatana, 2018;Guérin, 2013;Prakash, 2009, p. 204;Shah, 2006); cane cutting (Guérin, 2013;Breman, 1990); and construction (Jain & Sharma, 2019;Srivastava & Jha, 2016). 9 Of the other studies, one (by Vikramaditya Thakur) was in Nandurbar, Maharashtra and included hill Bhils, Bhils resettled from the hills to the plains because of the damming of the Naramada, and still others who lived in the Gujjar-dominated plains. Seasonal labour migration to brick kilns and sugar cane fields is prevalent in the hills and in the Gujjar-dominated plain villages. The last study (by Dalel Benbabaali) was by a paper factory in Telangana, in an area protected for tribals, but which employed Adivasis and Dalits as casual labour, and from which (or to which) there was no seasonal labour migration.