Bonded Labour, Agrarian Changes and Capitalism: Emerging Patterns in South India


  • Isabelle Guérin

    Senior Research Fellow
    1. University Paris I – Sorbonne, Paris, Cedex 05, France
    2. Institute of Research for Development (Research Unit Development and Societies), French Institute of Pondicherry, Pondicherry, India
    Search for more papers by this author

  • I sincerely thank my co-researchers, G. Venkatasubramanian, S. Ponnarasu and Marc Roesch, with whom the fieldwork on which this paper is based has been done. The fieldwork was carried out between 2004 and 2013 within the Labour, Finance and Social Dynamics research programme of the French Institute of Pondicherry. The latest fieldwork has been supported by the IOW project (Human Bondage in the Indian Ocean World: Roots, Structure and Transformations) funded by the French National Agency for Research (ANR). I also thank G. Muthusankar and G. Venkatasubramanian for their support in the elaboration of the map shown in Figure 1.
  • This paper was initially presented in the workshop ‘Agrarian Transformation in India: Its Significance for Revolutionary Politics’, held in Oxford in July 2011. Comments from participants in the workshop are gratefully acknowledged, and subsequently from Judith Heyer, David Picherit, Marc Roesch and G. Venkatasubramanian. The paper has also greatly benefited from comments of the editors of the Journal and the editors of the special issue (with, in particular, very careful readings from Jens Lerche and Alpa Shah) as well as two anonymous reviewers. I sincerely thank them for their helpful criticisms, comments and suggestions.


Drawing on a number of case studies from Tamil Nadu, this paper shows that bonded labour is not a relic of the past, but surprisingly contemporary. Refuting the tenets of the semi-feudal thesis, we argue that unfree labour can go hand in hand with capitalism, and that it can be initiated and sustained by capital itself in order to accumulate surplus value. Going against the tenets of the de-proletarianization thesis, we suggest that bonded labour is not always the preferred working arrangement for capitalism. Bonded labour should be examined in connection with specific historical contexts, the changing nature of the economy, the evolution of political forces and modes of socialization. I argue that bonded labour results from a specific regime of accumulation characterized by cheap labour, increased domestic demand sustained through household debt, as well as modes of conflict, contestation and worker identity formation that engage with both governmental programmes and consumerism.


Of the many debates on agrarian transformation and modes of production in India, the issue of labour, and especially the distinction between free and unfree labour, has been widely discussed over the past four decades. Is unfree labour a characteristic of feudalism that is destined to disappear with the transition to capitalism, or does capitalism in fact thrive on unfree labour? In recent decades, many studies have highlighted the gradual disappearance of agrestic unfree labour and the concomitant emergence of new forms of unfree labour. Jan Breman in particular has comprehensively discussed the continuities and differences between old and new forms of unfree labour. In feudal or semi-feudal societies, unfree labour was a source of rent through the monopolization of land and credit. Landowners were able to exploit peasants by their control over property rights and through usury, which guaranteed a perpetual cycle of indebtedness over generations, and further reduced the peasants' incomes. This in turn deterred lenders from investing in technological innovation, explaining the persistence of backward agriculture and semi-feudal production relations (Bhaduri 1973). Labour relations were embedded in mutual interdependence and a whole set of rights and obligations between the patrons and their clients. This was, of course, highly unbalanced and exploitative, but also gave clients some security in terms of daily survival, and informalized property rights (Breman 1974; Prakash 1990; Basu 2009; Pouchepadass 2009; Rammohan 2009). Economic factors, though important for labour-intensive tasks such as manual irrigation, were not the only driving force: the capacity of landlords to control a mass of servants was also a lever for prestige and power (Breman 2007).

New forms of bondage and unfree labour, which Breman defines as ‘neo-bondage’, differ in various ways. They do not guarantee a rent or help defend political status but are, instead, solely profit-driven. They are often short-term and non-hereditary, linked to seasonal migration and labour intermediaries. The ‘oppressor’ is not a landlord but a capitalist entrepreneur, defined here as an economic agent that exploits labour in the pursuit of profit and accumulation, while labour has to work for capital in order to subsist. Echoing Breman's work in Gujarat and Maharashtra on brick kilns and sugarcane harvesting, a number of empirical studies have shown the presence of neo-bondage in other parts of India and in a wide variety of seasonal work sites, such as saltpans, stone quarries and construction sites (irrigation canals, dams, road building). Apart from the agro-business industry (e.g. sugarcane harvesting), neo-bondage can also be found in rural or urban-based industries such as mines, rice mills, small-scale carpet and weaving workshops, gem-processing ateliers, bidi-making, fish processing, silver work and salt plains (Kapadia 1995; Byres et al. 1999; de Neve 2005; Lerche 2007; Breman et al. 2009; Srivastava 2009).

While most of these empirical investigations suggest that the feudal or semi-feudal thesis is obsolete, leftist parties such as the Communist Party of India (CPI), the Communist Party of India (Marxist) (CPI(M)) and the Communist Party of India (Maoist) (CPI(Maoist)) still view landlordism and agrarian power as the main enemies and that unfree labour is a characteristic of agrarian societies. Should we, instead, view unfree labour as an intrinsic feature of contemporary capitalism? The argument presented here, based on empirical work done in Tamil Nadu over the past 10 years by a French–Indian research team, is that the link between modes of productions and unfree labour is empirical and cannot be settled at the theoretical level. The empirical investigations presented here confirm that unfree labour is better understood as a capitalist labour relation rather than a pre-capitalist one, but also highlight exceptions and variations, and identify a number of important ‘drivers’ of such relations.

First, I argue that free and unfree labour categories are too rigid and limit the debate to theoretical disputes that run the risk of being divorced from empirical reality. The ground realities are much better understood in terms of a continuum of labour relations. First, forms and degrees of oppression and unfreedom vary greatly, from mild to the worst forms of bondage (Guérin et al. 2004). Second, and as argued by Lerche (2011), unfree labour relations should also be understood as a part of a continuum of labour relations, ranging from fully unfree relations to situations in which working conditions can be considered as ‘decent’, given that the definition of ‘decent’ work is subject to multiple definitions according to historical, social and political contexts (Lerche 2011).

Second, I argue that bonded labour should be examined in connection with specific historical contexts, the changing nature of the economy, the evolution of political forces and modes of socialization. More specifically, a detailed analysis of employers' and labourers' rationales, motivations and constraints shows that bonded labour results from a specific regime of accumulation characterized by cheap labour, increased domestic demand leading to the persistence of household debt, and modes of conflict, contestation and worker identity formation that engage with both governmental programmes and consumerism. Not only do specific capitalist accumulation strategies perpetuate bonded labour, but bonded labour is also directly or indirectly supported by public intervention and, in some cases, the desire for integration by the labourers.

This paper draws on fieldwork carried out with a research team1 between 2003 and 2010 in several residential and work areas in Tamil Nadu (see Figure 1). Research was conducted in three production sectors – brick kilns, sugar harvesting and rice processing. The sectors were chosen because in the decade of the fieldwork they were strategic sectors, either in terms of employing a significant number of labourers (brick), or as producing important staple food items (sugar and rice).2

Figure 1.

A map of the production sites, village sources and migration flows

Source: Author's elaboration, with the support of G. Muthusankar and G. Venkatasubramanian (French Institute of Pondicherry).

Residential area fieldwork was carried out in 15 Dalit (ex-untouchable) colonies in the Villupuram and Cuddalore districts (formerly south Arcot), which constituted ‘pockets’ of brick kiln moulders and sugarcane harvesters. For cane-cutting, we studied two production sites, the Cuddalore and Villupuram districts, which have specialized in sugarcane production for almost half a century, and a newer site near Madurai in the south of the state. For brick kilns, we also selected two sites: the largest industries concentrated in the Red Hills area on the outskirts of Chennai (Tiruvallur district) and smaller and more artisanal industries spreading over the Kancheepuram district. For rice mills, where labourers live and work in the same location, we compared distinct production conditions in the Red Hills area, Puducherry and Villupuram. Various data collection methods were used: statistical household surveys, ongoing qualitative interviews with various stakeholders (labourers, recruiters, managers and employers), and detailed analyses of production units and villages from which the workers came.

A comprehensive description of the three forms of bondage described here has been published elsewhere.3 Here, I focus my analysis on a comparison of bonded labour arrangements, their diversity and the underlying mechanisms producing bondage both in terms of labour and capital. The paper is organized as follows. The next section summarizes the main theoretical debates related to unfree labour, capitalism and feudalism. The subsequent section situates bonded labour in a wider context and briefly examines the current social, economic and political forces in Tamil Nadu today. A further section describes existing day-to-day bondage practices and sheds light on the varied degrees of restriction and intensity of bondage. An analysis then follows of the varying role of capital in shaping bondage in specific sectors and places. The penultimate section looks at the other side of the coin – labour – and examines the diversity of labourers' agency and struggles, ranging from extreme marginalization and domination to relative freedom and aspirations for integration. The paper concludes by summarizing our main arguments and examining their policy implications.

Unfree Labour, Feudalism and Capitalism

Over the past few decades, bonded labour has been subject to lively academic – and also ideological – debates, which have mostly focused on two major issues that are closely related. The first deals with definitions. What is unfree labour, and what is the distinction between free and unfree labour? Is indebtedness a sufficient condition to prove the lack of freedom or should unfree labour be defined by non-economic coercion? Is the line between freedom and unfreedom impossible to draw, since any labour contract necessarily involves coercion, as suggested by Banaji (2003)? When we use the terms free/unfree labour, are we talking about negative freedom – a worker is unfree if he or she does not have the ability to freely sell his or her labour? Or do we mean positive freedom – a worker is unfree if he or she has no bargaining power? Do we refer to freedom of contract or to freedom of exit from an employment relationship? As suggested by Mohan Rao, conflicting findings might also arise from the fact that people are simply not talking about the same thing (Rao 1999).

A second, closely related question is that of the links between capitalism and unfree labour. Are the arrangements that we label unfree labour mutually beneficial in an imperfect market context, as argued by some neoclassical economists (Bardhan 2001)? Is unfree labour inherent to capitalism, serving both to control and cheapen labour costs while preventing the formation of class consciousness, as suggested by the de-proletarianization thesis, whose most prolific representative is Tom Brass (see, e.g., Brass 1990, 1999, 2002)? Or is unfree labour a relic of tradition and a symptom of feudal or semi-feudal modes of production that are supposed to disappear with the transition to full capitalism, as argued by those proposing the semi-feudal thesis?

The latter, classical Marxist view is widely shared by radical leftist political parties (Brass 2002; Shah and Harriss-White 2011: but see also contributions by Basu and Das, Harriss, Lerche and Shah, this issue). Classical Marxist political economy postulates that the development of capitalism requires capital, but also ‘free’ labour. Labour is supposed to be free in a double sense: ‘free’ – or dispossessed – from the ownership of production, obliging workers to work for their subsistence; and ‘free’ to sell his or her labour power to an employer, and therefore ‘free’ from any coercive power. A ‘free’ worker is a worker who is available for production. The contrast is a feudal society, in which individuals are the landlords' property and are thus unavailable to industry.

Mohan Rao, for instance, views unfree labour and capitalism as essentially incompatible. He acknowledges that severe forms of exploitation occur, but argues that, historically, it was the freeing of labourers in its dual meaning that got capitalism started. To assert that unfree labour and capitalism can coexist or fuse would contradict the historical conditions of the emergence of capitalism. He rightly reminds us that in Marxist terms, the free/unfree dichotomy is used in the very restricted sense of a negative freedom to dispose of one's own labour freedom: ‘the free/unfree distinction refers to laws (or norms) regulating property rights in workers' labour power’ (Rao 1999, 101). Today, Rao argues, the term ‘unfree labour’ either gets misused and in fact refers to restrictions in workers' positive freedom – namely, their bargaining power – or refers to the continued existence of ‘semi-feudal’ modes of production. He considers India a typical case for this (Rao 1999, 123).

The semi-feudal thesis is also shared by leftist movements. Although their programmes and objectives differ in many respects, and while recognizing that landlordism superimposed capitalism in many regions, the CPI(M), the CPI and the CPI(Maoist) all emphasize that landlordism continues to be the most important form of exploitation of the rural labour force (see Shah, Harriss and Lerche, this issue). In their party programmes, very limited attention is paid to non-farm labour and migration, although both account for a very large share of rural households' income. There is no mention either of contemporary forms of bondage. Unfree labour remains linked to old forms of landlordism and tenancy.4

Besides these theoretical – and often very abstract – debates, there is a large body of literature inspired by political economy and economic anthropology, which is rooted in fine-grained empirical studies.5 This literature does not necessarily defend specific positions or try to establish a universal link between broad categories such as ‘feudalism’, ‘capitalism’ and ‘unfree labour’. Rather, it examines the concrete and historically given manifestations of unfree labour in its diverse forms. This paper draws on this literature to analyse how bonded labour interrelates with other labour regimes, how it emerges, evolves, disappears or, conversely, strengthens. This is examined in connection with specific historical contexts, the changing nature of the economy and in particular the rise of the non-farm economy, production techniques and productive forces, both national and international market configurations, state interventions, peasant mobilizations and power relationships.

Problematizing and contextualizing the relationship between unfree labour, capitalism and feudalism does not mean abandoning a theoretical perspective. A number of authors remain faithful to a Marxist perspective in terms of class conflict, while considering other forms of stratification (in particular, caste), thus recommending replacing the simplistic binary opposition of free/unfree with an approach emphasizing that, under capitalism, freedom/unfreedom is a matter of degree (Lerche 2011).

A detailed summary of the wealth of this body of knowledge is beyond the scope of this paper,6 but for our purposes it suffices to highlight its more salient findings.

With regard to definition, and the eternal debate on ‘voluntary’ versus ‘involuntary’ bondage, this literature suggests that defining free and voluntary choices makes little sense where there is intense vulnerability and oppression. Srivastava, for instance, argues that it makes more sense to examine bonded labour in terms of ‘costs of exit’: the appropriate question is not whether the workers are voluntarily in this type of contract, but what their alternatives are (Srivastava 2009).

As far as the explanatory factors of bondage are concerned, empirical analysis shows that neo-bondage emerges as the complex outcome of social hierarchies and discrimination against ex-untouchables and tribes, and specific modes of extraction of labour surplus. These correspond to the ‘low route’ of development, focusing primarily on cheap labour, that has been encouraged by the Indian government since Independence, and further reinforced in the context of neoliberalization and globalization (Byres et al. 1999; Breman 2007, 2010; Lerche 2007; Srivastava 2009).

Empirical evidence also highlights the ambivalent role of the state: it can either facilitate the disappearance of bondage by providing safety nets (Harriss 1992) or strengthen it by colluding with employers (Breman 2007; Guérin and Venkatasubramanian 2009). Workers' movements are noted for their essential role in the eradication of unfree labour and for improved conditions for labour (Byres et al. 1999; Bhowmik 2009; Prakash 2009; Rammohan 2009), but also for their absence. This absence is not due to lack of consciousness or false consciousness among bonded labourers, but because their struggles are mostly individual and isolated (Breman 2007).

Inspired by this scholarship and following Lerche (2007) in particular, rather than claiming that there are universal links between unfree labour, capitalism and feudalism, I examine the diversity and complexity of their interrelationships in a given context. As far as definitions are concerned, though coercion certainly exists in any wage relationship, I consider that the category ‘unfree’ is still valid as a category, distinct from ‘free labour’. However, what counts is not necessarily debt as an indicator, as argued by Brass (1990), but its wider consequences (Srivastava 2009, 133). My view is that bonded labourers are those whose freedom, wages and bargaining power are significantly restricted by debt. ‘Significantly’ is, of course, a rather vague term, but it is difficult to be more precise given the diversity of the degrees of restriction. In line with previous work, I recommend approaching bondage in terms of a continuum, from mild to harsh forms (Guérin et al. 2004: see also Rawal 2006; Lerche 2007; de Neve, forthcoming). I also recommend using the term ‘bondage’ rather than ‘unfree’, which helps to avoid misunderstandings with regard to the diversity of definitions surrounding free/unfree labour.

Going against the tenets of neoclassical economics, which consider that, ‘all things being equal’, debt bondage can be a ‘win–win’ arrangement for employers and labourers, I argue that such arrangements are shaped by and constitutive of highly unequal power relationships that result in extremely high levels of exploitation. Refuting the tenets of the semi-feudal thesis, I argue that unfree labour can go hand in hand with capitalism, and that it can be perpetuated and sustained by capital itself in order to accumulate surplus value. Nuancing the de-proletarianization thesis, I suggest that bonded labour is not always the preferred working arrangement for capitalism. Not all capitalists use bonded labour and those who do have a variety of motives. Maximizing or sustaining profit is, of course, the ultimate objective, and the main purpose of bonding labour through debt is to control, cheapen and discipline labour power. Bondage can be resorted to in response to various constraints related to production, market and balance of power, as illustrated in the three industries studied here. I also argue that the debates on bondage have neglected the role of consumption. I will show that consumption must be considered both as an additional factor for wage advances and also as a mode of resistance and struggle.

Bonded Labour in Context

Bonded labour, understood as one of a number of forms of non-decent work, should be set in the broader context of the globalization and neoliberalization of the Indian economy, which has accelerated sharply since the early 1990s. This has led to a significant decline of public investment in agriculture, withdrawal of quantitative restrictions on the import of agricultural products, the privatization of state-owned assets and government services, the dismantling of the structure of public banking, the emergence of global value chains controlled by multinational corporations and so on (Kennedy 2004; Bowles and Harriss 2010; Harriss-White 2010; Ramachandran and Rawal 2010). Recent measures in social protection, as generous as they might be, are not only barely accessible to the most marginalized people, but are entirely supported by the state, while capital remains exonerated from any cost sharing (Kannan and Breman 2013, 28).

The state of Tamil Nadu illustrates the growth of both liberalization and social safety nets. The state is the fifth-largest economy in India in terms of GDP, tenth in terms of ‘human development’, and is often seen as a model for combining economic growth and social progress (Vijayabaskar et al. 2004). The apparent success of the Green Revolution and the growth of the non-farm sector, the demographic transition, and significant progress in health and education could suggest that Tamil Nadu has accomplished its agrarian transition and the absorption of rural labour without great difficulty. However, these trends are accompanied by a number of factors that facilitate the existence of considerable forms of exploitation of the workforce.

The first factor is the emphasis on social policies at the expense of labour regulation. If Tamil Nadu is one of the most welfarist states in India, it is also known for its populist rhetoric that builds on the ‘people’ while leaving intact the privileges of the ruling classes (Subramanian 1999; Harriss 2001). Where social policy measures are implemented – highly unequally across the state, probably depending on local political conditions – they help significantly to raise living standards and wages (Djurfeldt et al. 2008; Harriss et al. 2010; Heyer 2012). Tamil Nadu is also one of the pioneer states in adopting specific measures for the unorganized sector – with, for instance, a Construction Workers Welfare Board in existence since 1994. However, the TN Social Security Act (Act 29 of 2006) is very thinly spread and poorly implemented (NCEUS 2007) and labour protection, as elsewhere in India, it still off the agenda (Lerche 2010; Vijayabaskar 2011). In the absence of formal legislation to protect labour, debt and personal relationships are the ‘best’ way for both employers and workers to secure their relations.

Moreover, the implementation of government programmes is in large part carried out through clientelism and gatekeeping relationships (Harriss-White 2003; Harriss-White and Janakarajan 2004). It is essential to bear in mind the specificity of state–society relations in order to understand the ambiguity of labour relationships. Rights exist on paper, but are implemented mainly through personal bonds of allegiance. Employers and/or recruiters still monopolize, or at least control, labour and access to resources, including development programmes. As we shall see later, and as observed by Picherit (2012) in Andhra Pradesh, labour relationships are not restricted to labour, but also entail real or anticipated ties of dependence and protection.

The decline of agricultural labour and a vast reserve of vulnerable and malleable labour is another essential component enabling bondage. One of the main effects of the Green Revolution has been to further deepen class, caste and gender differentiation (Harriss-White and Janakarajan 2004). Marginal farmers have benefited very little from the productivity gains of high-yield varieties. They have been excluded from capitalist farming such as sugar cane, and have suffered the most from the financial liberalization that began in the 1990s. For marginal farmers, informal lending is the most common source of funding (Harriss-White and Colatei 2004; Ramachandran and Rawal 2010; Guérin et al. 2012a), and this includes wage advance, as we shall see below.

Another factor has to do with consumption. This is still poorly studied (Cavalcante 2009), while I believe it is key for the future of the Indian countryside, both with regard to the agrarian transition and labour. In the villages studied here, whether for agricultural wages or sharecropping, payments in kind have become very rare. The costs of education and health keep on growing. A culture of consumerism is also emerging, stimulated by mass advertising campaigns (largely through television: Kapadia 2002) and facilitated by urbanization – not because of the rural exodus but, rather, through circular migration. The persistence of caste hierarchy as an organizing principle does not exclude evolutions and aspirations for change, including among the most marginalized. An increasing number of households aspire to acquire motorcycles, household appliances, manufactured clothing and so on. Social and religious rituals, especially weddings, may demand lump sums that may amount to several years of a household's earnings. In many villages in the studied area, Dalits want to finance their own temples and festivals to assert their independence from higher castes. Consumerist behaviours, of course, vary greatly in intensity and nature. They show a desire to integrate into global society – for instance, through durable goods – while at the same time helping to strengthen caste as an identity when they occur through social and religious rituals. ‘Consumerism’ is itself a problematic concept – should we see motorcycles or gold jewels (a fundamental component of marriage costs) as consumption or as productive assets? The consumer market is also highly segmented – the working poor and Dalits do not consume in the same way as landowners, employees and upper castes. The point I want to highlight here is that our understanding of labour arrangements, and bonded labour in particular, should take into account the significance of consumption. As I shall argue below, consumerist behaviour, by encouraging workers to demand ever-increasing numbers of wage advances, reinforces their own exploitation.

The lack of employment regulation, the continuing decline of agriculture, the existence of public programmes that adhere to patronage patterns and finally the emergence of consumption, are all factors enabling debt bondage. In order to understand the exact nature of bondage and its diversity, both in form and degree, one must now look at the more local level, and analyse the social, regional and sectoral specificities of labour regimes.

From Mild to Harsh Forms of Bondage: A Continuum

I have suggested above that the free/unfree dichotomy should be replaced by a continuum running from the mildest to the severest forms of bonded labour (Guérin et al. 2004; Lerche 2007). The three sectors studied by us in Tamil Nadu are good examples of this continuum.

Brick Moulders and Cane Cutters: ‘Mild’ Forms of Bondage

Brick-making and cane-cutting can be characterized as ‘mild’ forms of bondage. Debt restricts labourers' freedom and significantly cheapens wages, as we shall see below, and this is why I qualify such labour relations as bondage. However, in the case studies from both sectors, labourers enjoy a relative degree of freedom.

Both set of case studies are typical examples of neo-bondage arrangements described by Jan Breman (1974). The workforce is migrant and seasonal. Payment is by piece rate and issued at the end of the season. Production and processing units do not directly recruit workers and their families but, rather, local, native intermediaries who are often from the same community. Wage advances attract workers, and are mainly distributed during the slack season (from August to November–January), sometimes immediately at the end of the working season. Brick kiln labour recruiters follow the instructions of brick kiln owners, fixing the amount of labour required and paying part of the advances. For cane cutters, it is the sugar mill owners – and not farmers – who control the process.

During the work season, both brick producers and cane cutters are only given a weekly allowance, intended to cover their basic needs. Accounts are settled at the end of the season, with the season's production deciding the total remuneration, and a deduction of the sum of advance and weekly allowances. Some leave with a small amount, others have settled their accounts, while still others, more numerous, remain in debt at the end of the season and will have to come again for the next season.

In both sectors, most of the workers are Paraiyar, an ex-untouchable community viewed as one of the most marginal groups in Tamil Nadu. Cane cutters are mostly men. A few women and children participate, assisting with domestic chores and helping bind cane bundles, but their role remains marginal. In brick kilns, by contrast, men, women and children work. In 2004, a household survey of 300 workers from both sectors shed light on their vulnerability and extremely poor living and working conditions:7 miserable wages, very long working days, child labour, lack of hygiene, permanent harassment and restricted freedom. One-third stated that they had no freedom of movement, while others said that they were unable to negotiate more than a few days off.

In 2004, brick kiln workers were paid the equivalent of around INR 40–50 per day. This meagre income came at the cost of very long working days (12–16 hours, half of which were at night). Taking into account the length of the working day, the hourly wage was between 2 and 4 rupees. By comparison, in 2004 daily wages in agriculture were around INR 80 (for men), for 7–8 hours, which amounts to around INR 10–12 per hour. Of course, the comparison is difficult because agricultural workers' employment is much more irregular, especially in ‘dry’ (i.e. non-irrigated) villages.

Data collected in November 2012 show that cane-cutting wages are still significantly lower than for daily wage labourers. Cane cutters earn the equivalent of INR 80 per day for 12–14 hours of work (i.e. around INR 6 per hour), while in the villages studied daily wages in agriculture were around INR 250 (for men) for 6–8 hours of work (i.e. around INR 30–35 per hour). Daily wages in construction are similar.

In both sectors, workers are circular migrants. Production is seasonal and lasts 6–8 months per year. This allows the workers to leave the workplace for several months and to undertake other livelihood activities in the remaining part of the year, of which agriculture daily wage labour (‘coolie’) and marginal farming are the most common. They may also borrow from other sources. Bondage may be temporary – in response to illness, marriage or death – or chronic, which is most often the case for workers from dry villages. Exit options do exist – the most common probably being local agriculture and construction in nearby cities – but are limited. Workers not only have to have paid off their debt, but must also be in a position to access other opportunities. This requires mostly contacts and networks, in a context in which labour markets remain extremely fragmented (Harriss-White 2003). Although labour relations are outside all forms of legislation, they are highly structured, so that not just everyone can leave to take up work just anywhere. Whatever the destination, workers rarely take the risk of leaving alone. Intermediaries are needed to find a place to live and an employer who is willing to recruit, and to withstand harsh competition from other workers.

Rice Drying: A Severe Form of Bondage

Rice drying is a step up in terms of exploitation, typified by very harsh forms of bondage (Roesch et al. 2009). In contrast to the previous cases, workers are hired permanently and live permanently at the production site. Whole families are confined in production units made up of storage sheds and courtyards for rice processing. Men, women and children work day and night in shifts to boil rice, to spread it out in the sun and to monitor the drying. Some have been there for two or three generations. Their social exclusion is almost total. Not only does their confinement cut off all ties to the outside world, their family and their native village, but the exhausting pace of work and the living conditions prevent any form of family life worthy of the name. All of the bonded labourers we met were Irulars, a previously forest-dwelling tribal community.

Opportunities to leave the enclosures are extremely rare and strictly controlled. However, many workers do not even consider leaving: they have no contacts in the world beyond and often fear this external world, which many of them have never known. Working days last from 13 to 16 hours, partly at night, and sometimes longer for very big orders. Payment is piece rate, partly in cash, and in kind. In 2006, workers were paid the equivalent of INR 37 per day on average (INR 2.5 per hour). Under NGO pressure, some makeshift schools have been organized around production sites, but it seems that most of the children work. They have no specific tasks, but they help their parents to clean the drying area and then to spread the paddy. Housing and sanitary conditions are often extremely poor. There are very frequent cases of tuberculosis due to overcrowding and a lack of proper hygiene conditions, especially among young children and the elderly.

Here too, debt is a central element of the labour relationship. Recruitment is not only often based on an initial advance, but extremely low wages force the workers to regularly request new advances. Most of their total debts represent several years of wages and are thus impossible to pay back (Roesch et al. 2009). Exit options are therefore non-existent.

It is, however, interesting to note that across the three sectors, the above types of working arrangements are found only in certain units and for certain workers. The following section will consider the diversity of these working arrangements, and offers explanations for it.

Capitalism, Neoliberalism and Bondage: Various Paths and Multiple Purposes

Debt as a labour management method combines social, technical and economic factors. Even if the working relation is mainly of an economic nature, present-day bondage relationships arises only in communities where the verticalized ties of subordination, historically and socially rooted both in employers and workers' consciousness, are still sufficiently strong to make subordination acceptable (Srivastava 2005; Breman 2007; Servet 2007). In the three sectors studied here, the great majority of labourers are ex-untouchables or from tribal backgrounds.

Sector-based constraints should also be taken into account. Some production systems are more suited than others to bonded labour owing to technical constraints and/or market structures. By this I mean intensity of competition, technological factors leading to highly labour-intensive processes of production, which demand continuous production, seasonal variations demanding specific employment modalities to ensure the loyalty of the labour force, and also the role of capital. For employers, the main advantage of shackling labour with debt is to ensure the continuous presence of a cheap and docile labour force, either for the production season (cane-cutting, brick-moulding) or permanently (rice drying). I shall now discuss in more detail the technical aspects of each sector.

Manual rice drying requires intensive, continual labour that determines the final quality of the rice, the amount of broken rice, conservation longevity, texture and taste. The rice is dried in the sun immediately after soaking and boiling to prevent mould. The grains are spread under the sun for several days, but turned over very carefully to prevent chipping and to guarantee grain homogeneity. In units using bondage, labour is organized such that the same workers carry out the full cycle of activities (soaking, boiling and drying), all of which demand their continuous presence. Soaking and boiling take place at night, so that drying can begin at sunrise. Each cycle lasts for about 3 days and is repeated without a break to optimize operations.

In the case of brick production and cane-cutting, bonding labourers through debt is well suited to a labour-intensive, continuous and seasonal production process. Moulded bricks are destined for firing in kilns within the same sites, and supply disruption would be costly to the kiln owner. In the case of sugar, as soon as the cane is cut, it has to be pressed and processed into sugar in the sugar mills. Here too, supply disruptions are costly. The advances and end-of-season payments guarantee the presence of workers throughout the season, and the piece rate ensures a certain degree of productivity, while transferring most of the risks on to the workers. Whether cane-cutting or brick production, workers and recruiters are unanimous on one point: given that the work is very physically demanding, the advance is the only way to force labourers to work.

Comparing different forms of labour management is useful for understanding the mechanisms of bondage. Across the three sectors, not all units use debt and not all workers are bonded. First, debt is only used where there is mass production. Managing non-bonded labour on a larger scale is problematic. Bigger units are likely to seek higher profits, and usually have more competition because they target regional markets, while small units sell locally.

Bonding labourers with debt is the standard for ‘modern’ brick production, which uses a permanent concrete kiln. But this applies only for certain tasks, those requiring continuous presence: moulding, firing and loading/unloading bricks. Some tasks are irregular – for example, transporting moulded bricks to the cooking chimney is done only every 2–3 days – and in this event workers are hired daily and do not get an advance. Their daily wage is twice as high as that of the bonded labourers. The difference is one to four if we compare hourly wages. Since the 1970s, when the first ‘modern’ brick kilns emerged, employers have been using a wages advance to attract and discipline labour. Since then, the advance has kept on increasing (and has done so much more quickly than wages), probably in response to the continuous growth of the construction sector and the risks of labour shortage (Guérin et al. 2012b).

In the case of sugar production, there are two very distinct labour management systems: in some cases, cane cutters are migrants tied with debt, but we also found local labourers, paid daily and without advances. Differences between migrant and local labour systems are due to the varying degree of vertical integration of the value chain. The sugarcane sector is based on a triadic system comprised of farmers, sugar mill owners and public authorities (Landy 2006). Public authorities supervise production very closely and play an essential, incentivizing role for farmers, in particular through ensuring guaranteed minimum prices and providing subsidized credits. The state also supports sugar mills, for example, through tax exemption and subsidies. In turn, producers depend on the sugar mills to accept their produce and they begin the cutting only when they have a contract with the sugar mill. In the ‘local labour’ system, farmers depend from mills' orders for the sale but manage the harvest themselves and mostly use a local workforce. This is the case, for instance, in south Arcot (the Villupuram and Cuddalore districts), which converted to sugarcane production half a century ago. Most independent farmers cultivate small plots and have a local workforce on which they rely. In the ‘migrant labour’ system by contrast, farmers are fully dependent on the sugar mills, including for the organization of the harvest. The sugar mill owners employ a massive harvesting workforce. The workforce is managed by the mill owners: they are in charge of recruitment and payment through their own recruiters, who are officially registered by the sugar mills and who serve as the only point of contact between the mills and the harvest labourers. This labour system is found in the south of Tamil Nadu (Madurai district), where cane production is a more recent phenomenon. Our estimate is that 80 per cent of its sugarcane cutters are migrants, mainly from the south Arcot region. Apart from living conditions – cane cutter migrants live in highly crowded, unhygienic makeshift camps – the wage level is a fundamental difference between the two types of workers. Our calculations indicate that migrants earn 40–85 per cent less than locals, mainly owing to the commissions of the labour intermediaries (Guérin et al. 2009). Using migrant labour with previous cane-cutting experience and bonded through debt has been a means supporting the introduction of cane production on a large scale in an area where previously it hardly existed.

Diversified labour management systems were also found in the rice sector. While the previously discussed rice mills in north-west Tamil Nadu (Tiruvallur District) routinely resort to bonded labour, those in Villupuram district and the State of Puducherry hire daily labourers. Tiruvallur rice mills have opted for a continuous production process carried out by the same workers, who are therefore obliged to be present at all times. By contrast, the rice mills of Villupuram and Puducherry use either labour rotation for each production cycle (Villupuram), or division of labour tasks between men and women (Puducherry). In these cases, workers are neither bonded nor indebted, and are paid regularly. Most also engage in other income-generating activities, notably in agriculture, with rice drying as a secondary occupation. Wages are just as poor (between 2 and 3 rupees per hour) but the working conditions are much better, and workers are willing to be poorly paid for this secondary employment. In other words, cheapness of labour is secured without bondage.

Technical constraints account for the differences in labour-use systems. The two production systems differ in various aspects, the first of which is quality. Bonded labourers are used for higher production quality, which requires meticulous drying and permanent monitoring. Capital and market limits also come into play: rice mills in Tiruvallur have larger production capacities and depend on merchants, while in Villupuram and Puducherry, small units sell directly to the local market. Spatial constraints in relation to rice-producing areas and markets also come into play. Rice mills in Tiruvallur district are further away from rice fields, which results in added expense and all kinds of cost-cutting strategies. Meanwhile, due to their proximity to large urban centres and a more active job market than elsewhere, mills struggle to retain their workforce. Employers have got around this difficulty by recruiting from the Irular community, which is renowned for its rice-drying expertise, but also its social exclusion, as discussed below.

Over the past few years, the rice sector has changed in two ways. First, bondage has declined with mechanization. The first drying machines appeared in the early 2000s, and in 2006 around 20 per cent of production units were mechanized. Mechanization is even more prevalent today, but manual drying is still used, probably by more than half of the sector. At the same time, working conditions in manual drying units are worse than ever. Rice processing units are highly diversified according to their position in the treatment process. Some specialize only in one stage (drying or hulling), while others deal with all of its stages. Some plants use a completely manual process, while others are partially or entirely mechanized. Diversity of processing units also depends on their position in the supply chain, which in turn depends on their storage capacity and their degree of reliance on merchants. The most marginalized units specialize only in manual drying, and are small, traditional processing units (nerkalam). Debt bondage prevails here, and recent changes in the market have further increased pressure on workers, for two main reasons. First, consumers have increased quality expectations. The quality and uniformity of the drying process are critical for the taste, visual appearance and conservation quality of the rice. Mechanization is completely out of reach for small units, owing to excessive investment costs. Their only alternative is to demand even greater worker presence and vigilance than before. The second reason for increased pressure on workers is the growing role of traders, encouraged by the gradual liberalization of the rice market. Rice buyers are either sole traders or some of the modern processing units described above and who are also involved in trading. Faced with intense competition due to the liberalization of the sector, buyers seek all possible means to cut costs and do not hesitate to crack down on all production units and harass workers to make them work longer. As the tasks undertaken by the workers are very physically demanding, they can hardly produce more per unit of time. Of course, it can be argued that small-scale units are doomed to disappear and that their disappearance will de facto solve the bondage problem. This may be true, but how many generations will be sacrificed before reaching this point?

Labour and Bondage: From Exclusion to Hopes of Integration

For the employers, bonded labour relations are driven by a variety of strategies to keep labour costs down in labour-intensive industries, specific production conditions and market structures. For the labourers, bonded labour relations are also shaped by a variety of constraints. While poverty and discrimination remain primary ‘push’ factors, variable levels of economic and social exclusion and a diversity of labourers' constraints, rationale, motivations and forms of resistance (in intensity and nature), also matter. Here too, the three case studies point to a continuum, ranging from bonded labour as a form of protection against extreme marginalization, to bonded labour as a tool for aspirations of equality and integration.

For the Irulars, bonded labour is seen as a form of protection. In a context in which they have been driven from their land, and cut off from their previous livelihoods and their moorings, bonded labour in rice mills presents permanent employment, a roof and a safeguard against the risks of a social, economic and political life from which they are entirely excluded. Traditionally, Irulars working in rice mills were forest-dwellers (mostly from Javadhu Hill in Salem district in Tamil Nadu), who specialized in hunting rats and snakes, gathering honey, beeswax and medicinal plants, and tree root digging. Some were also agricultural daily wage labourers. Since the 1950s, as with many other communities specializing in forest activities, they have been gradually forced to move to the plains and to take up other occupations. Most have faced many difficulties in terms of finding employment and with social integration. For some of them, rice mills have served as a form of refuge. Irrespective of questions of debt, their vulnerability and social isolation is such that they feel that they have no alternative. In other words, social exclusion is the cause of their bondage.

By comparison, the agency of brick moulders and sugarcane harvesters is much stronger. I shall restrict the analysis to brick workers, due to a lack of space. In contrast to rice dryers, it is the desire for upward social mobility that reinforces the bondage of brick moulders, even if ultimately this desire locks them into debt traps and dependency upon recruiters and employers. The case of brick migrants illustrates two interesting elements. First, the nature of the work and the wage levels are certainly important considerations, but many other criteria are equally valued by the labourers. These include job security: their difficult and badly paid job includes a 6-month employment guarantee and this is extremely highly valued. Any other services that the employer and the recruiter are likely to provide also come into play. In the villages that have specialized in brick kiln migration for a long time, and echoing Picherit's observations on jobbers in the construction sector in Andhra Pradesh (Picherit 2009, 2012), labour intermediaries have over time become the key figures of village daily life, with respect to job opportunities but also regarding conflict settlements or politics, including access to government programmes or NGOs. These include employment schemes, subsidized housing, food subsidies, free durable consumer items such as televisions, fans, grinders, bicycles for pupils, microcredit and so on. Seasonal migrants generally have very little access to government programmes and NGOs, but for whatever little they do receive through these channels, the role of labour recruiters is often decisive. Most of the well-established labour intermediaries are involved in politics, either through alliances or directly as party representatives. Far beyond their role as labour recruiters, they are expected to help workers to take advantage of the few rights they have in the village. Debt bondage thus facilitates the access to a wide range of resources. In other words, and even if labour relations are much more a straight economic relationship compared to older forms of bondage, elements of patronage remain, not with employers but with labour recruiters, who are very often from the same communities as the labourers themselves. Labour thus remains embedded within wider relationships.

The second lesson from the brick kilns deals with labour struggles (Guérin et al. 2012b). Some workers do have some bargaining power, in particular those who have experience and who work in the kilns every year. They manage to avoid the worst production sites and limit the extraction of added value by labour intermediaries. However, they also use their positions to obtain higher advances, which feed their growing pursuit of consumerism. As shown by various ethnographies, most rural circular migrants remain engaged in political struggles ‘in the rural areas they identify as their first home’ (Gidwani and Sivaramakrishnan 2003, 349; see also Picherit 2012). Here, this translates into expenses related to agriculture and to social and religious rituals. These expenses are instrumental in maintaining and even sometimes strengthening labourers' status and their position in local hierarchies. At the same time, and since economic returns are extremely limited, these expenses lock them into debt traps. These costs also lead workers to demand larger advances, which in turn increase their dependence on the sector. The percentage of workers who return indebted at the end of the season, and who so are forced to return the following year, has continued to grow between 2000 and 2009 (Guérin et al. 2012b). My most recent field visits in January 2013 indicate that the gap between advances and wages is continuing to increase, which in turn reduces exit options. Some of the bonded labourers are not the poorest of villagers, in so far as wage advances are not only used for daily survival but also for consumption and agricultural and social investments. Debt is still, however, used to cheapen wages, to impose hard working conditions and to restrict exit options. As a result, as paradoxical as it might be, I suggest that bonded labour is not only compatible with rising aspirations for equality and integration, as exemplified by the desire for consumption, but it also relies on these aspirations.


The comparison within and between sectors sheds lights on the diversity of articulations and interactions between bonded labour and forms of production and capital. Bonded labour is found in highly labour-intensive industries, such as brick kilns and manual rice-drying units, but also in sectors with higher level of capital intensity (e.g. sugarcane processing). Bondage ensures cheap labour and willingness to work for very long days on very painful tasks, possibly on a seasonal basis (brick-moulding and cane-cutting), but not exclusively (rice drying). Bonded labour can be used to cope with labour shortages in the case of rapid market expansion (brick kilns), to facilitate the management of a massive workforce in the implementation of integrated vertical industries (sugar cane) or to increase the intensity of work in a highly competitive environment (manual rice-drying units).

The common point of the three case studies analysed here is that debt acts as a substitute for any legal regulatory framework, serving to some extent as a ‘contract’. In a setting such as India, where labour is overwhelmingly informal (Srivastava 2005; Kannan and Breman 2013), debt further encourages informality and the absence of formal labour rights while maintaining non-decent wage and working conditions.

On the labour side, as already documented by many studies, I observe that massive poverty and the casual character of employment relations are the main reasons why the land-poor and landless have no other choice but to seek out cash advances to cope with insufficient and irregular incomes, to have a job guarantee and possibly to invest in agriculture. I would, however, argue that bonded labour is also sustained through consumption and consumerism. Labourers, including the very poor, are also consumers, albeit to varying degrees. In the rice mills in particular, destitution and social isolation reach such a degree that consumption is restricted to basic physiological needs. Bondage results from their social exclusion. For cane cutters and brick moulders, however, and in particular for those who migrate regularly, significant increases in living standards are observed, related to the acquisition of durable consumer items and investments in prestigious social and ritual events. In those cases, bondage is strengthened by workers' desire for social mobility, however relative that mobility is in reality. In the villages studied here, investing in prestigious social and religious events is the primary source of social status for circular migrants, alongside access to development programmes and social benefits. These aspirations increase the self-exploitation of the labourers: although wages are essential, above all, their requirements are met through increases in the advances that recruitment agents and employers provide. Maintaining relationships with employers and/or labour intermediaries also allows hope for better access to development programme resources, since these are distributed through relations of allegiance, often with employers and/or recruiters as the central link (see also Picherit 2012).

Our field-related analyses suggest that bonded labourers are not always the poorest of villagers. Does that mean that as researchers we do not need to be concerned with the labour relations of which they are part? If brick moulders are able to spend hundreds of thousands of rupees on their daughters' marriages, should we still feel sorry for them? It is obvious that the main concern is with those who are in the most abject forms of servitude. But I would nevertheless call for debt as a tool of exploitation and freedom restriction to be condemned – the working conditions of both brick moulders and cane cutters are still extremely poor and their exit options limited.

The political and policy implications of my arguments are, of course, central. The semi-feudal thesis promotes the conclusion that bondage is caused by feudalism, and therefore that its elimination requires fighting against the class domination of feudal landlords. The neoclassical thesis views labour bondage as beneficial mutual arrangements within imperfect markets and anticipates its disappearance with the elimination of market distortions. It is true that ‘all things being equal’, debt remains the least bad solution, both for employers and workers. But it is precisely the legitimacy of this ‘all things being equal’ that must be questioned. For the de-proletarianization thesis, the enemy is neither feudalism nor market imperfections, but capitalism: the abolition of bonded labour should take place within a broader objective of a transition to socialism.

The analysis proposed here, which is in line with other studies combining political economy and field economics, shows that the understanding of bonded labour – and the issue of its eradication – cannot be dissociated from a broader analysis related to the whole landscape of labour, which includes both the lack or the inefficiency of labour regulations and the imbalance between capital and labour (Breman 1996; Lerche 2007). It should also be located in a more general perspective foregrounding dependency and exclusion from civil rights (Breman et al. 2009; Srivastava 2009; Breman 2010). The persistence of bondage is strongly related to the increased casualization and the informalization of labour relationships, and to the intensification of circulation and seasonal migration. It is also an indicator of the increased monetization of commodity exchanges and social relationships, and the development of mass consumption. Debt bondage is enabled by the decline in the availability of agricultural labour and lack of social protection, but also by the growing consumption needs of labourers. Paradoxically, increasing aspirations for equality and integration are helping to reproduce the very harsh conditions of capitalist exploitation and extraction of surplus value.

Abstract theories, as useful as they might be to build models and establish causality, should also take into account the complexity of societies and their constant evolution. This is what I have tried to show in this paper, highlighting how contemporary forms of bondage do not fit into frozen – and to some extent dogmatic – theoretical frameworks.


  1. 1

    See the acknowledgement note.

  2. 2

    For more details on the importance of these three sectors, see Guérin and Venkatasubramanian (2009) for brick kilns, Guérin et al. (2009) for sugar cane and Roesch et al. (2009) for rice.

  3. 3

    See note 2. See also Guérin et al. (2004) for a comparison of brick kiln and sugar cane, and Guérin et al. (2007, 2012b) for specific aspects of brick kilns. On cane cutters in Tamil Nadu, see also Marius-Gnanou (2008).

  4. 4

    See, for instance, CPI(Maoist) (2004, 13) and CPI(M) (undated, 14, 20).

  5. 5

    See, for example, the pioneering work of Jan Breman, starting from 1974 and continuing until 2010 so far (Breman 1974, 2007, 2010), collected essays (Byres et al. 1999; Breman et al. 2009), articles and monographs (Jodhka 1994; Kapadia 1995; Lerche 1995; de Neve 2005; Picherit 2009), reviews and essays (Harriss 1992; Jodhka 2004; Srivastava 2005, 2009; Lerche 2007).

  6. 6

    For an overview, see Lerche (2010).

  7. 7

    For more details, see Guérin et al. (2007, 2009) and Guérin and Venkatasubramanian (2009).