Debt-Bonded Kiln Workers and Their Intent to Return: Towards a Labour Geography of Smallholder Farming Persistence in Cambodia

: Despite the increasing preponderance of non-farm work in Cambodia, labour migrants across a range of working conditions remain linked to their rural homesteads through durable ﬁ nancial and social arrangements. This article explores this phenomenon through the case of debt-bonded brick kiln workers in Phnom Penh, formerly smallholder farmers in villages. Drawing on the ﬁ eld of labour geography, the article ﬁ rst examines the process by which labourers became debt-bonded, thus situating them within the country ’ s broader agrarian transition and recasting peasants as rural labour. It then explores workers ’ perceptions of rural life, suggesting that the unfreedom of kiln work, contrasted with the ﬁ xedness and potential for mobility in rural life, makes workers aspire to return to their land. The article ultimately highlights how the persistence of smallholder farmers can be understood as an issue of poor work under neoliberalism in Cambodia, and draws light on the agency of labour in understanding this.


Introduction
Cambodia is undergoing a period of economic transition, with industry taking over agriculture in terms of the largest share of GDP (World Bank 2015). Industrial employment has concurrently risen, from just 4.1% of total employment in 1991, to 19.6% in 2017 (ILOSTAT Database 2017b), whilst agricultural employment fell as a percentage of total employment from 79.9% to 42.4% in the same period (ILOSTAT Database 2017a). Despite this, 42% of the country's labour force remains in agriculture (NIS 2016) and 82% of those living in rural areas 1 continue to undertake some form of agricultural production (NIS 2015). This phenomenon of "peasant persistence" is documented across South-East Asia (Rigg et al. 2016), and is also connected to debates centred on agrarian change within the realm of critical agrarian studies literature (Bernstein 2006b;Edelman and Wolford 2017;McMichael 2007). The debate centres on the question of whether agrarian transition, that is the transition from feudal relations of agrarian production to a system of capitalist agricultural commodity production, necessitates the demise of peasant populations (Bernstein 2006b). Related to this is also the question of the developmental prospects for smallholder and small farmers, specifically in the face of the deepening marketisation of subsistence across the global South (McMichael 2007).
In Cambodia, agriculture is undergoing a period of distress, leading to an ongoing labour migration from rural households, both internally and overseas, with older family members remaining tied to rural homesteads and relying on remittances from younger family members that have migrated for work (Bylander 2014;Derks 2008;Parsons 2017;Lawreniuk & Parsons forthcoming). Yet the question of "peasant persistence" remains under-addressed (Rigg et al. 2016). Put simply: why are more small farmers not leaving their land behind altogether, given the opportunities in the non-farm sector and the decreasing gains from agriculture?
This article contributes to this debate through a focus on a distress-induced case of rural-urban transition: debt-bonded labourers in Phnom Penh's brick kilns, formerly small farmers in villages across the Mekong basin who were forced to leave due to mounting debts. Drawing on recent labour geography literature which seeks to highlight the lived experiences of labour and open up the notion of labour agency (see Coe 2013;Coe and Jordhus-Lier 2011), we explain why the majority of kiln workers are looking to return to rural life, despite the difficulties of agrarian production that they left behind, and a lack of evidence that bonded kiln labourers generally return to agriculture.
In doing so, we argue through an exploration of kiln workers' experiences that work in the kilns is experienced as unfree, and perceived as transitory, in contrast to the apparent fixedness of rural life. This is despite evidence from the kilns which highlights the inescapable nature of kiln debt. Furthermore, we suggest that despite experiences of rural distress, kiln workers see the potential for mobility in rural life, in contrast to a dearth of possibility in the kiln. In order to develop this thesis in the context of agrarian transition in Cambodia more broadly, we draw on Lerche (2007) and Gu erin (2013), to affirm that bonded brick workers in Phnom Penh constitute a capitalist labour relation, and within this, a severe form of unfree labour, due to the complete lack of alternatives that are open to them. The experiences of kiln workers are also tied to broader structures of capitalism in Cambodia, thus we seek to embed the transition of workers within the contexts of which they are a "constituent process" (Coe and Jordhus-Lier 2011:218), and to suggest that their aspiration to return to the land represents a form of "spatial agency" in the making of agrarian capitalism in Cambodia. Albeit, a form of agency that is born of distress rather than an active assertion for land-based livelihoods (Bernstein 2006b(Bernstein , 2013. As such, through an especially acute case of "emergent late capitalist precarity in East Asia" (Rigg et al. 2016:130), we ultimately aim to shed light on how in certain cases, trajectories out of distress-ridden agriculture can lead to further insecurity, and thus how rural life can come to represent a viable alternative. Through such analysis, the paper also offers a means of exploring the question of "peasant persistence" beyond political economy factors alone, and highlights instead the centrality of subjectivities in understanding trajectories of social and economic change.
"Peasant Persistence" in Cambodia: A Question of Labour The question of whether agrarian transition requires or necessarily compels the demise of the "peasant" has been central to Marxian debates on agrarian change over the past century. The concept of a so-called "classical" transition is centred on Marx's analysis of the English case, and highlights a shift from feudal social relations-comprising predatory landlords and peasants and characterised by low productivity and rent-seeking-to capitalist class relations, driven in some part by a productivity revolution in agriculture (Bernstein 2006a;Marx 1976). Through this process, the former feudal peasantry are understood to give way to class relations of large, landed farmers and wage workers, "free" from owning small parcels of land and able to sell their "labour power" as a means of reproducing themselves. Yet the complexity of ensuing trajectories of change in different regional contexts, and Marx's own later work on agrarian change outside of the English case (see White 2018), has certainly shown us that "infinitely diverse combinations of elements of this or that type of capitalist evolution are possible" (Lenin 1964:33). Even a brief glance at literature on global agriculture today highlights that contrary to the English case, small farmers or "peasants" remain dominant in the majority of countries.
Literature in the 1970s reinvigorated an exploration of this conundrum, with accounts focusing both on the politically salient nature of peasant politics to broader processes of change underway in post-colonial nations, thus contradicting accounts of the backward peasantry (Wolf 1969), as well as the particularities of so-called peasant logic vis-a-vis more structuralist understandings of social change (Scott 1985). Such literature emphasised the active role of peasants as "important historical agents" (Edelman and Wolford 2017:962) in reshaping broader trajectories of agrarian change, thus highlighting the need for an understanding of social transition that allows for the agency of the "peasantry". More recently, following the onset of neoliberal globalisation, the apparent continued persistence of "peasantries" has compelled a debate around whether peasant farming represents a viable alternative to neoliberalism (Desmarais 2002;McMichael 2007;Shiva 2016), or an exemplification of the poverty of work under late capitalism (Arrighi and Moore 2001). Put simply, are smallholder farmers staying rooted to rural life because of a desire to remain agrarian and autonomous from urban market capitalism? Or, are they doing so due to problems with the nonfarm work that they are moving towards?
The latter view is represented notably by Bernstein, who has argued that notwithstanding the complexity and variety of trajectories of agrarian change globally, rural capitalism has been dominant since the 1970s, characterised by the demise of "predatory landed property", a productivity revolution in agriculture, state-led development, and in many instances, the globalisation of agricultural value chains through global agribusiness (see Bernstein 1996). He has thus suggested that the term "peasant", denoting a subsistence-centred cultivator that is largely insulated from market pressures, is an anachronism, given that small farmers across the world for at least the past four decades have been reliant upon markets in some form or other in order to undertake agrarian livelihoods Debt-Bonded Brick Kiln Workers and Their Intent to Return (Bernstein 2006b). With this in mind, Bernstein argues that the central question which plagues agrarian transition today is one of labour, not the peasantry: how do the "classes of labour" that have emerged in both rural and urban areas across the global South through the onset of neoliberal globalisation reproduce themselves?
Bernstein's thesis is instructive in enabling us to reframe the question of "peasant persistence" in Cambodia to one instead of the reproductive strategies of rural labour. Evidence from across the country suggests that contrary to depictions of productive and organised small farmers emerging from elsewhere in the South (La Via Campesina 2017), agriculture predominantly comprises subsistence-based rice production (Saruth et al. 2014), thus the country is dominated by smallholder farmers. Agriculture across the country remains predominantly small scale, with landholdings of under a hectare comprising 58.9% of the total in 2015, and landholdings under three acres comprising 83.4% (NIS 2016:19). Levels of modernisation and crop diversification are also relatively low, 63.4% of all agricultural land remained rain-fed only in 2015, and only 7.6% of land was used for wet and dry season cultivation (NIS 2016:21). National agriculture growth increases over the past decade are thus shown to largely be the result of increased land use, and the majority of rural households are forced to reproduce themselves through labour markets, largely outside rural areas (Ovesen et al. 2012;World Bank 2015). 2 As such, smallholder farmers are both subsistence-oriented in production, but rely upon markets to reproduce themselves through waged work. In addition to this, Cambodia has been repeatedly named one of the most vulnerable countries in the world to the impacts of climate change, and Cambodian agriculture has been subjected to repeated environmental shocks, largely in the form of either flooding or droughts (Morton 2014;UNESCO Bangkok 2016).
The plight of smallholder farmers-turned-labourers becomes clearer if we situate them within Cambodia's broader economic trajectory. The country has undergone profound political shocks in the past century, emerging after a four-year period of state-led genocide under Pol Pot's Khmer Rouge regime in 1979 to over a decade of unrest and stagnation, before coming under the control of the United Nations for a brief period in the early 1990s, then a coalition government, and finally under the current regime led by the Cambodia People's Party from 1997 onwards (Strangio 2014). This regime is characteristically both authoritarian and neoliberal in character, with the latter having strengthened the former (Springer 2009). The country thus bypassed any form of state-sponsored agrarian development scheme in this period, unlike much of the rest of Asia (Beban and Gorman 2017). Since the 1990s, the government has repeatedly indicated that it is on a path to supporting agricultural modernisation among small-scale farmers and increasing rice exports in particular (RGC 2013). Yet to date, there remains very little by way of state support for agro-technology, marketing in commodity value chains, and for mitigating against the impacts of environmental shocks (ADB 2014). In addition, the state's move towards land privatisation since the early 1990s has led to a "scramble for land" (Gironde and Peeters 2015). The impacts of this upon rural livelihoods have been profound in many areas, as prior to titling, farmers relied on easy access to both collective and homestead lands to undertake a range of livelihood activities such as fishing, foraging and cultivation (Gironde and Peeters 2015;Hughes 2003).
The overall result is that whilst there is evidence of more recent industrial and services-led growth having lifted rural people out of poverty, such poverty reduction is shown to be decidedly insecure. Specifically, whilst the proportion of poor in the country fell from 50% in 2007 to 21% in 2011, and the majority of those lifted out of poverty were in rural areas (World Bank 2015), a fall off of just $0.30 cent/day would push the poverty rate back to 40% (ADB 2014). A key reason for this is the type of work that has flourished in the country, as Rigg et al. (2016:130) highlight across Southeast Asia, "while work has proliferated, whether it is secure is not so certain". Unemployment is incredibly low in Cambodia at 2.7%, however the nature of most work is insecure, with over 80% of the country's workforce employed in the informal sector (ILO 2016), where work is shown to comprise long hours, poor wages and high levels of labour discipline (Derks 2008). This is coupled with "inadequate social protection", with the International Labour Organisation noting that there is little sign of state investment in this regard (ILO 2016:8). What does is exist is largely funded by NGOs, applied unevenly, and focuses primarily on formal sector workers (Cook and Kwon 2007;Cook and Pincus 2014).
The lack of secure transition from agriculture to non-farm work in Cambodia thus needs to be understood in light of the particular character of work in the country, and smallholder farmers need to be understood as labourers rather than peasants. Such a reconceptualisation delinks smallholder farmers-turned-labourers from necessarily occupying an agrarian space, and asks instead how they understand and deploy space in order to reproduce themselves in a context of agrarian distress and primarily informal non-farm work. Yet given the difficulties of agrarian production the question remains: why are smallholder farmers that are also undertaking waged work, and are therefore among Bernstein's "classes of labour", choosing to remain tied to their rural land? In answering this question, we draw on literature from the labour geography tradition to begin to re-examine brick kiln workers' intentions to return to agriculture.

Recasting the Intention to Return to the Land as Labour's Agency
Labour geography has emerged as a field concerned with explaining how labour is an active agent in the making of broader structures of power, and in opening up what we understand by labour's agency within such processes (Coe and Jordhus-Lier 2011). Andrew Herod (1997:16) initially raised the notion of labour's active reshaping of "economic landscapes" to suggest that "the production of the geography of capitalism is not always the prerogative of capital". Herod thus highlights the spatial dynamics of labour's active shaping and reshaping of capitalism. Drawing on Harvey's thesis of the "spatial fix" of capital-capital's reconfiguration of space to overcome crises of accumulation-Herod highlights "labour's spatial fix"-labour's "desire to implement in the physical landscape their own spatial visions of a geography of capitalism which is enabling of their Debt-Bonded Brick Kiln Workers and Their Intent to Return own self-reproduction and social survival" (1997:17). Space, or rather the production of space, thus becomes a means through which labour's agency is enacted.
Herod's work has been widened in a number of directions in what is now a broader field of labour geography (see Castree 2007). For our purposes in this article, we focus on two areas development: types of agency and contextualising agency. Firstly, Katz (2004:242) has highlighted the myriad means through which workers can be understood to enact types of agency vis-a-vis broader structures of power and oppression, asking us to move beyond "finding 'resistance' in each discursive and cultural practice" undertaken by marginal communities. Instead, "resistance" is reworked into three overlapping conceptualisations of agency: "resilience, reworking and resistance" (Katz 2004:241). "Resilience" 3 comprises "restorative and strengthening acts", sometimes providing the groundwork for more structural forms of transformation but not necessarily; "reworking" refers to remaking "oppressive and unequal circumstances" without necessarily toppling them; and "resistance" refers more specifically to acts which "subvert or disrupt ... conditions of exploitation and oppression" (Katz 2004:242). Katz's thesis draws inspiration from the work of an earlier generation of anthropologists such as James Scott (1976) and Aihwa Ong (1988) that sought to valorise everyday and informal acts of resistance, thus contesting scholarship which centralised unionled workers' movements as the only legitimate means through which labour can enact resistance. Yet Katz's thesis offers us a means of moving beyond what she sees as an "industry" of finding "oppositional practice" in the everyday lives of the marginalised, highlighting instead how acts of resilience or reworking can also be made legible outside a more prescriptive notion of "resistance".
In developing Katz's thesis, a second, instructive approach emerged from Coe and Jordhus-Lier's review of agency within labour geography literature, where they highlight the need "to reconnect conceptions of labour agency into the webs of wider relations with other social actors and institutions in which they are inevitably embedded" (2011:221). Coe and Jordhus-Lier elevate four areas of concern with regards to this thesis: global production networks, the state, community politics, and labour intermediaries, with an emphasis on how these structures mediate and often limit the agency of labour in refashioning landscapes of work. Thus their focus is similar to that of Katz (and in fact they draw on her work at the start to anchor their discussion), in that they seek to nuance our understanding of labour's agency, more notably vis-a-vis external factors, as opposed to Katz who focuses on the type of agency itself.
Taken together, Katz (2004) and Coe and Jordhus-Lier (2011), as well as Herod's (1997) foundational work, offer a means of exploring the case at hand. Whilst Castree (2007:855) has argued that "in situations where workers are subject to coercion ... they have little or no agency", the more variegated understanding of agency put forward by Katz, along with the spatial dynamics of such agency highlighted by Herod, suggest that we can in fact theorise forms of agency among even the most oppressed, be they "the intentions or consequences of worker's agency" (Coe and Jordhus-Lier 2011:216, emphasis added). As such, the imaginaries, desires, and strategies of workers, even if unrealised, become part of the notion of agency in this rendering, and contribute to a fuller understanding of the spatial dynamics of agrarian-industrial transition. Coe and Jordhus-Lier highlight the need to situate such agency within the relational context of marginalisation from which it emerges. In articulating these concerns with those of critical agrarian research, the variegated concept of "agency" allows us to explore how smallholder farmers-turned-labourers imagine and deploy space as a means of resilience, of reworking, and of resisting the broader systems of exploitation that they find themselves in. As such, in seeking to explore the question of agrarian transition, or the social reproduction of labour across agrarian and nonagrarian work (Bernstein 2006b), among small farmer-turned-kiln workers in Cambodia, we proceed through a labour geography lens that acknowledges a more open understanding of labour's agency.
The next section highlights the research methods undertaken to inform analysis within this article, before the following section looks in more detail at the nature of transition in the case in question: debt-ridden small farmers who become bonded labourers on Phnom Penh's burgeoning brick industry.

Research Project and Methods
This paper is part of a broader research project entitled "Blood Bricks", which examines the nexus between climate change and modern-day slavery in Cambodia. Specifically, the project focuses on how climate change-induced rural indebtedness leads small farmers from villages across Cambodia to urban, insecure work as debt-bonded labour in Phnom Penh's Brick kilns. The project's emphasis is on the particular ways in which climate change deepens livelihood insecurity, leading to insecure labour relations on brick kilns, and the everyday experiences of such insecurity.
Research methods comprise two stages. Firstly, semi-structured interviews with 51 brick kiln workers across brick kilns in the environs of Phnom Penh. These were supplemented by 31 further qualitative interviews with other actors such as kiln owners, union leaders, former kiln workers, residents around the kilns and Buddhist monks to both triangulate information and provide additional perspectives. Analysis draws on qualitative data to take seriously the lived experiences of workers on brick kilns, highlighting perceptions, narrative accounts and commentary from other actors to enliven the perspectives of kiln workers. Secondly, a total of 308 quantitative surveys were conducted in three villages that comprise high levels of out-migration to brick kilns. Village A is located in Kampong Cham province and situated along the banks of the Mekong river, Village B in Prey Veng province around 50 km east of the Mekong river, and Village C in Prey Veng province just under 10 km east of the Mekong river. All brick-sending households were surveyed in each village, and then a randomised sample of other households was also sampled. These surveys were further supplemented by qualitative interviews with a sample of labour-sending households and other figures in the three villages.
Data collection comprises both qualitative and quantitative data, however the former are primarily drawn on in order to develop the arguments in this article.

Debt-Bonded Brick Kiln Workers and Their Intent to Return
All interviews were conducted in Khmer and later transcribed into English; all interviewees' names have been changed to protect their identities.

Farming, Debt, and Migration: The Context of Agrarian Distress
Beginning in the rural origins of labour transition, our research reveals the complex and articulating factors leading to indebtedness, and later debt bondagenamely poor social protection, poor state support for agrarian development, and a proliferating microfinance industry. We also highlight the catalysing impact of environmental change upon deepening rural insecurity. In doing so, we begin to re-embed kiln workers within the broader trajectories of change and relations of power from which they emerge (Coe and Jordhus-Lier 2011).
The majority of bonded labourers in Phnom Penh's brick kilns began their lives as smallholder farmers in villages around the Mekong basin region of Cambodia. Kiln interviews revealed that the earliest year bonded workers left their villages was in 1982, but primarily since 2005, and that they did so due to mounting debts in their home villages. Such debt is shown to have been incurred largely through household expenditure on productive forces in agriculture-predominantly chemical inputs or hiring-in labour-and on medical expenses. Leaving the former aside for a moment, the latter very much speaks to the impact of low levels of social protection upon insecure labour. Van Damme et al. have undertaken a study on the extent to which "out-of-pocket" health expenditure in Cambodia can lead to debt. They conclude that "[i]n Cambodia a disease episode, even a relatively short one such as dengue in a young child, frequently causes catastrophic health expenditure leading to debt in house-holds with precarious livelihoods" (2004:278). This argument is certainly evidenced across a number of brick kiln workers that we interviewed. Kravann 4 is a female bonded brick kiln worker in Phnom Penh and has worked here for two years, since medical expenses forced her to do so. She recounts: "I used to stay at home and grow rice ... But when my husband became sick, I sold all the farmlands. After that I had no choice besides working in this brick factory ... [my husband] had a stroke from high blood pressure. Half of his body has been paralyzed, and since then I have become the breadwinner for my family." The incidence of a single household health issue pushed Kravann into selling her land, and despite this, she still accrued debts from her husband's healthcare expense which led her to a life of bonded labour on the brick kiln. As such, the thin veil of social protection can leave smallholder farmers largely unprotected after even a single occurrence of poor health.
The other primary expense that drove rural labourers into debt is the attempt to develop agricultural productivity. Data from villages reveal that such debt is incurred both by households that are seeking to simply maintain existing levels of production, and by those looking to expand production, largely through increasing the number of harvests per year. As such, there is evidence of farmers attempting to develop their productive capacities and failing to do so. Sangha, 5 a male bonded kiln labourer working in a brick kiln in Phnom Penh for one and a half years, highlights the difficulties of trying to expand farming: "It was really difficult to grow rice in the dry season because we just didn't have enough water. When it did work, it was because we used a lot of capital, taking on debts in order to pay for fertiliser and oil. However it was really difficult to sell crops." Sangha ended up selling a proportion of his land to repay his debts to local lenders, but the proceeds didn't suffice. Therefore, he borrowed the remaining debt from the kiln owner and his entire family moved to Phnom Penh to live on the brick kiln and repay his debt through bonded labour. This case speaks to the broader lack of state-support for agricultural modernisation highlighted earlier in this article. Our research shows that the take-up of cash crops 6 in the three labour-sending villages that we surveyed was very low at only 14%, which fits with the national scenario, where 82% of all crops planted were rice varieties as of 2013 (NIS 2015:21). As such, attempts to both maintain and expand agricultural production in the face of poor state support risks leaving small farmers in high levels of debt.
Yet the issue of why agricultural returns are unable to keep up with cultivation costs goes deeper than inadequate state spending alone. Data from kiln workers and labour-sending villages reveal that a key reason for agricultural failure and concurrent increasing indebtedness was ecological shocks, generally droughts and floods, leading in many instances to side effects that more directly affected crop yields such as rodent infestations. "High vulnerability to climate threats" (ADB 2014:1) is noted by the Asian Development Bank as one of the three main obstacles to agricultural growth in the country, and our evidence certainly foregrounds this as a key factor in exacerbating rural indebtedness. Across the three laboursending villages surveyed, 99% of respondents indicated that there had been notable environmental changes in their village over the past five years, with 83% of respondents noting varying temperatures, 76% noting increased incidents of drought, and 73% noting more varied rainfall patterns. The majority of kiln workers also cited ecological change as a factor of agricultural distress. Arunny 7 is a female kiln worker from Phnom Penh and has been working in the kiln with her family-a husband and four young children-for a decade now. She recounts how she first incurred debt in her village: I was a farmer ... [but] it was not profitable. There were droughts, and the water pump didn't provide enough water, but there were also times when it rained so hard that crops were damaged. I borrowed money ... for farming activities like buying gasoline and fertiliser ... then I borrowed from the owner of the brick kiln to pay it back. Now I don't think I'll pay him back, I only earn enough to put the children through school.
Arunny highlights how the shifting climate is deepening insecurity among small farmers, as their already adverse incorporation into systems of commodity production is exacerbated by climate change (Hickey and du Toit 2007;Mosse 2007;Taylor 2013).
In addition to climactic shocks acting as a catalyst, a key driver of rural debt is undoubtedly the proliferation of a deregulated rural credit industry from the early 2000s onwards. Micro-finance began in Cambodia under a "humanitarian NGO-Debt-Bonded Brick Kiln Workers and Their Intent to Return driven model" (Bateman 2017:3) in the early 1990s, designed to provide opportunities for demobilised soldiers to start up small-scale enterprises as a means of social reintegration following decades of war. However by 2000, the state-run regional micro-lending projects had been centralised into a single profit-oriented entity, officially becoming a private commercial bank in 2009. The shift to a profit-driven model, micro-finance necessarily entailed interest, thus loans were structured to increase surplus extraction from the indebted. This period saw remarkable growth in the sector, with the number of clients served by Micro-Credit Institutions (MCIs) rising from 300,000 in 2005 to almost 1.6 million in 2013 (Bateman 2017). Furthermore by 2011, the average lending amount had risen to 139% of the country's gross national income (Bylander 2015a), indicating unstable levels of indebtedness. Data from the three labour-sending villages that we surveyed reveal high levels of indebtedness, with 56% of households in Village A, 46% in Village B, and 67% in Village C surveyed indicating that one or more persons in the household held a debt. Among bonded labourers in the kilns, the pattern was almost without exception the same-small farmers who incurred rural debt borrowed an increased amount from kiln owners to repay local lenders, moving their entire family to the kiln in exchange to work off the debt bond. Crucially, kiln owners offer interest-free loans, but extract surplus in other ways, as the next section will detail. Bylander (2015b) has highlighted the particular interrelatedness of ecology, migration and debt, arguing that such migration needs to be understood as a responsive coping strategy rather than a longer-term diversifying strategy in the face of ecological shocks. This argument is particularly pertinent in the case at hand, as the smallholder farmers in question are unable to gain the required level of training to undertake more skilled work such as garment labour, and are indebted to the extent that they can extend their credit no further. As such, they are compelled into borrowing, and their lack of choice is central to understanding their particular form of unfreedom (Gu erin 2013).
Ultimately, smallholder farmers are shown here to be facing agrarian distress, shaped and driven by articulating factors both structural-inadequate state support, a pernicious microfinance sector-and environmental, in the form of climate and environmental change. Yet despite the insecurities inherent in an agrarian life, bonded kiln workers in Phnom Penh express the aspiration to return. The next three sections unpack narratives from kiln workers, offering an analysis of such aspirations as a form of agency, before the final section returns to the broader issue of agrarian change.

Kiln Workers' Perceptions of Freedom
We have to work here because of our debt; if not we would have chosen to leave this place long ago. This is our unsolved problem, but the loan from the kiln owner comes with no interest. If the owner permitted it, all of us would go back home because of our hard work. People generally prefer to rent land to grow rice or vegetable. But in our case, we're in debt and so we aren't able to do such things. If we had enough money we would repay [the kiln owner] the loan. (Kosal, male brick worker, Phnom Penh) Kosal's longing to return home offers a strong starting point from which to anchor our discussion. Kosal was forced to sell his rice fields after both his wife and mother became sick at the same time, and he faced mounting medical expenses. He grew indebted in his village, and his initial non-farm job working at a salt factory did not allow him to earn enough to repay his debt. As a result, he ended up on the brick kiln, after the kiln owner advanced him the capital to pay off his existing debts. Kosal now owes US$4000 to the kiln owner. To put this in context, in 2016, GDP per capita in Cambodia was US$1269.9. Kosal's mobility is severely restricted as a result of his debts, he is unable to leave the kiln and work elsewhere, even in periods where kiln work stops due to rains, and he is therefore forced to increase his borrowing for his family's basic expenses in such periods of stagnation. The work he undertakes is also gruelling. Kosal and his wife Rangsei perform a range of tasks, from breaking up large mounds of fresh clay using hoes, manually transporting clean clay into brick-moulding machines, where clay is pressed into moulds, and then carrying and stacking moulded bricks into the sun to dry. The brick-moulding machine is also incredibly dangerous, Kosal reports a number of injuries incurred by workers, including seeing a child lose an arm to the machine when placing clay inside.
The overall conditions of Kosal's work are thus difficult, exhausting, and unsafe, and it is in this light that Kosal expresses a longing to return to his village. He does so through tropes which highlight his lack of choice in the matter, and which thus implicitly paint the ability to rent land and grow vegetables as a means of obtaining some freedom, despite the fact that Kosal's own previous experiences as a smallholder farmer speak to the insecurity of such work. Kosal's perception of farming can therefore be understood as a form of resilience-a means through which he is able to cope with the difficulties of kiln life and assert an alternative spatial imaginary.
In situating Kosal's work within a broader context (Coe and Jordhus-Lier 2011), the contrast between restriction and freedom in Kosal's account also speaks to the proliferation of unfree forms of labour in Cambodia's insecure labour market. Rather than representing a backward or semi-feudal form of labour, one that will be eradicated or altered once capitalist relations of production penetrate further; "unfree" bonded labour on Phnom Penh's brick kilns is best understood as a capitalist labour relation. The debate on unfree labour takes its inspiration from Marx's theorisation of the transition from feudalism to capitalism, when unfree labour relations would give way to free labour, defined as being "doubly free"-free from owning the "means of production" and free to sell its labour power through a market (Marx 1976; see Lerche 2007 for full overview of debate). These concepts of "freedom" are arguably based upon inherent assumptions of both a linear progression in social relations, and a progression that necessarily betters relations of exploitation for the poor. Yet Brass (1999) and Banaji (2003) have contradicted this interpretation of Marx, suggesting that unfree labour is both compatible with and in Brass' case central to the means through which late Debt-Bonded Brick Kiln Workers and Their Intent to Return capitalism accumulates. 9 Lerche (2007:447) has further argued that we need to "acknowledge the fluidity of the actually occurring levels of unfreedom', and it is in this vein we approach unfreedom on the brick kilns. Bonded kiln workers in Phnom Penh are "those whose freedom, wages and bargaining power are significantly restricted by debt" (Gu erin 2013:411), thus they represent a severe degree of unfree labour. This is a crucial point for two reasons. Firstly, in seeking to explore the rural persistence of classes of labour, we suggest that the case of bonded brick workers offers a clear example of one of the many "less secure and more vulnerable types of work" which characterise the majority of the Cambodian labour market (ILO 2012:vii), rather than an exception, or pre-capitalist labour relation. Bricks from the kilns in question go into the construction of office blocks and luxury housing across Phnom Penh, which are funded by various domestic and overseas investors (Brickell et al. 2018); thus they are part of the region's landscape of growth. Secondly, in drawing on Coe and Jordhus-Lier's (2011) argument around how labour's agency is constrained by broader structures of power, we would suggest that the labour control exerted by kiln owners is itself a form of class relational power (Pattenden 2016), 10 bolstered by state support in a number of ways. Forced labour is explicitly outlawed in Cambodia's national Labour Code (Kingdom of Cambodia 1997). Despite this, further evidence form the kilns highlights how the state is complicit in enabling kiln owners to restrict the mobility of indebted workers by helping them to track down and even arrest and charge fleeing debt-bonded kiln workers. This is in a context where union membership among kiln workers is incredibly low, which according to a representative 11 from the local Building and Wood Workers Trade Union of Cambodia is due to the difficulties of recruitment because of the spatial control exerted over labourers on the kiln site by owners. Furthermore, in shaping "what it means to be a worker" (Coe and Jordhus-Lier 2011:223), the state's poor social protection coverage means that bonded workers are forced to take on all the costs of their social reproduction. Evidence from South India, where levels of social protection coverage are notably higher, have highlighted the transformative impact that such policies can have in weakening the power of capital vis-a-vis labour control, and strengthening labour's bargaining power (Carswell and De Neve 2014;Heyer 2012). As such, the power of kiln owners over bonded workers represents a form of class relational power, and the state is complicit in bolstering this power and weakening labour's ability to bargain. Kosal's longing to return to the land, manifested in a perception of farm life, therefore denotes a form of resilience in a context of severe unfreedom.

Temporal Dimensions of Perceptions of the Rural
In developing our understanding of worker agency as resilience, we would further suggest that although reproduction in agriculture and on the kiln are both shown to be insecure in different ways, kiln work is understood as a more temporally limited form of work, whilst workers look towards agriculture as a constant. As such, the spatial aspect of resilience expressed by workers is also articulated with a temporal aspect. Sangha, 12 a male brick worker who we heard from earlier in this article, highlights this through two seemingly contradictory narratives. Firstly, upon being questioned as to the difficulties of rural life he responds: ... yes it was very difficult to grow rice in the dry season because we didn't have sufficient water. At times it turned out well, but we needed to use a lot of capital to make it happen and this meant borrowing the money to buy oils and fertiliser. It was subsequently difficult to sell our rice and repay these debts.
The picture painted here is of rural poverty and coping-where Sangha and his family lack the means to undertake successful cultivation and are thus forced into borrowing to do so, with the vagaries of environmental change rendering profits from this insufficient to repay such debts. Sangha further contrasts rural debt to kiln debt and highlights the relative advantages of the latter: At home I undertook farming but I always lost any profits and so I came here to work. At the brick kiln, the debts don't accrue interest and so we don't have to expend as much energy. But if we borrow from an ... [MFI], we have to pay the interest.
Sangha contrasts the relationship between debt and the nature of work across two spaces-the farm and the kiln-and concludes that interest-free kiln loans allow him to work less energetically, thus freeing him in some sense from the physically demanding nature of farm work (Marx 1976). Yet crucially, upon being questioned about future plans, Sangha is unwavering: "... I won't live here forever, I will return to my home village anytime soon." The kiln site is understood here as temporary, whereas agrarian life is more rooted as a place of home. Whilst environmental vagaries, poor social protection and state support for agriculture, and a predatory MFI industry undoubtedly render rural life precarious, there is a sense in which such vagaries are also understood as constant, whereas the perceived short-termism of debt-based work may offer temporary respite from the drudgery of rural life, but is understood as decidedly transient. Upon being asked what he did at "home", Sangha responded decidedly, "Farming, we do farming", the implication being that this is a vocation, whilst kiln work is a temporary means of reproducing oneself before returning to this.
Mealea, 13 a female brick worker, highlights a similar sentiment in her distinction between kiln and farm work. She is under no illusions as to the difficulties of farm life, suggesting that opportunities at home are incredibly limited: "There is nothing we can do at our homeland beside growing rice." Yet despite this near-dismissal of rural life, upon being asked if she will return, she responds: "I want to go but I'm not sure. Maybe in three or four years. We will have to work here until we have enough money to go home to buy land and farmland." The kiln site is thus recast as a temporary means of earning to enable agrarian production. Brown (2017) notes a converse phenomenon among workers in a Special Economic Zone factory in Laos. He highlights how they deploy space as a form of agency through straddling the rural/industrial divide, remaining rural smallholder rice farmers to lower their costs of social reproduction whilst undertaking low-paid factory work in the SEZ, and inadvertently offer a subsidy to manufacturing capital as a result. Mealea similarly looks to deploy space across the rural/industrial Debt-Bonded Brick Kiln Workers and Their Intent to Return divide, albeit as a means of enabling her to return to an agrarian existence. Both Sangha and Mealea thus highlight that depictions of rural life represent spatialtemporal forms of agency, where kiln work is understood to be a temporary means of overcoming indebtedness, and also a potential means of accumulation, to enable workers to return to rural livelihoods.

The Rural as a Place of Social Mobility
The notion of rural life as holding promise for social betterment is affirmed by a number of workers in the figure of the rich villager. Rithy, 14 a male kiln worker, highlights this by describing the life of the rich villagers in his own village: "they do not come to work, they open grocery shops instead ... They never leave, they stay in the village by opening grocery stores, being merchants, undertaking cultivation." Rithy is in conversation with a female worker, Pisey, who concurs with his analysis: Rithy and Pisey highlight the fact that despite their own difficulties in reproducing themselves as smallholder farmers, there does exist a class of rural capitalists that exemplify rural accumulation. Notably, none of the means of rural accumulation that they list are agricultural commodity production, speaking to the relative dearth of commercial farming in the country (Beban and Gorman 2017;Ovesen et al. 2012). Nonetheless, the notion of the rich villager offers a material manifestation of what Rithy and Pisey look towards in seeking to return to their villages. Crucially, the depiction reveals how for these workers, village life offers scope for social mobility, where kiln work offers no clear pathway out. As Mitchell (1996:199) has argued with regards to labour's imaginaries of landscapes, "[w]e need to show ... precisely how it is that the landscape is formed and made known [by labour], and, certainly, how it functions as a system of social control and opportunity within capitalist economies". In this instance, the restricted mobility and hopelessness of kiln life is contrasted by a sense of some opportunity in rural life, and such a conceptualisation highlights another form of resilience for those working on the kiln. As Rithy and Pisey's friend Seyha, another female brick worker, comments: "The fortune will come eventually."

Resilience and its Significance for Agrarian Change: Some Conclusions
Kiln workers' imaginaries of rural life, as offering freedom, permanence and the opportunity for social mobility, offer insight into why they choose to, or aspire to, remain tied to rural life. This is despite the difficulties they face in making ends meet in rural Cambodia. This article has ultimately foregrounded how the imaginary of rural life set out by kiln workers is a form of resilience against the difficulties of kiln work, and thus read at a broader level, against the poverty of nonfarm work across much of Cambodia. As such, this resilience, whether realised or not, offers a glimpse into how agrarian change is being experienced by farmers that are moving away from rural life, and how such experiences shape their complex and divergent trajectories both away from and back to the farm. The expanded conceptualisation of agency developed by Katz allows us to both emphasise experiential analysis of agrarian-industrial transition, and to understand how it materially shapes patterns of change, in this case the overwhelming persistence of smallholder farming households in rural areas where household members also work in the non-farm sector. As such, this article highlights the salience of a variegated and contextually situated rendering of agency in understanding the trajectories of smallholder farmer paths away from agriculture, where such farmers are understood to be among burgeoning "classes of labour".
In bringing analysis of labour's "resilience" back to bear upon the broader question of "peasant persistence", there are three conclusions we wish to draw out overall. The first is with regards to the nature and extent of argument that we make here. Whilst we contend that the depictions of rural life put forward by kiln labourers are best understood as a form of resilience against the harsh conditions of kiln labour, and that such resilience is in itself a form of agency, we stop short, as does Katz (2004), of suggesting that they constitute a means of resistance. As Coe and Jordhus-Lier (2011:216) highlight, there is some ambiguity in Katz's typology as to whether agency represents "intentions or consequences" of such agency. In our case, we would stress that the way kiln work is structured-comprising sporadic periods without any work that require workers to borrow further to undertake even daily expenses-meant that there was very little evidence among workers we interviewed of successful trajectories back to the village, thus speaking to the structural barriers at play which are highlighted throughout the paper. Rather, in returning to the issue of agrarian change and moving to our second conclusion, we argue that labour's perceptions of rural life offer a means of understanding why ties to the land remain so prevalent among an otherwise transitioning population in Cambodia, and across Southeast Asia (Rigg et al. 2016). The specific tropes that we highlight within kiln labourers' resilience-freedom, temporality, and the potential for progression-speak directly to the problems of insecure work under advancing capitalism in Cambodia, and highlight to some extent why the transition to non-farm work is not being accompanied by a loosening of ties to the land, either materially or in the imaginaries of workers. The argument thus centres on the experiences of bonded urban labourers as a means of understanding processes of rural change, highlighting how "livelihood security is ... co-produced in the factories and the fields of East Asia" (Rigg et al. 2016:130). This brings us to the final argument around the marriage of these two literatures: critical agrarian studies and labour geography. We argue that in the Cambodian context, given the particular configuration of relatively under-developed agrarian productivity-raising measures and the shocks and stresses of environmental change, as well as the geographical nature of social reproduction across rural/urban spaces, there is a particular salience in viewing smallholder farmers through a labour lens. This allows us to be cognisant of labour's social reproduction across space and thus understand the ways in which labourers deploy space, or imaginaries of space, in seeking to reproduce themselves.
Endnotes 1 This figure is taken from the Cambodian Agricultural Census, which in 2013 surveyed 63% of the country's population, excluding "highly urbanized areas of Phnom Penh, as well as some extremely remote areas and restricted areas in some provinces that were not reached by the enumerators" (NIS 2015:72). 2 The exception to this is with regards to the tribal communities that continue to reside in mountainous and forested areas of the country. However recent evidence of "land grabs" in such areas does evidence the fact that even these groups are being compelled out of their isolated existence and into reproducing themselves through markets (Milne 2013). 3 The usage of the term "resilience" here is distinct from a broader literature on social resilience, which largely refers to the material attributes and capabilities of communities to deal with crises; see Maclean et al. (2014) for overview. 4 Interview 45, 15 November 2017. 5 Interview 9, 2 November 2017. 6 "Cash crops" refers in the Cambodian context to non-rice crops, as rice has traditionally been understood as a subsistence crop. This is slowly changing as the government is beginning to support measures to increase rice as an export crop (Saruth et al. ). However for the purposes of this study, "cash crops" excludes marketable rice, and refers to all other crops. 7 Interview 23, 10 November 2017. 8 8 Interview 5, 3 November 2017. 9 9 Brass and Banaji offer different perspectives against the orthodox Marxian thesis. Brass (1999Brass ( , 2003 suggests that unfree labour is actively reproduced by capitalism as a means of preventing class consciousness, denoting it as a form of "de-proletarianization '. Banaji (2003:82) suggests that it should be understood as one of the "multiplicity of forms of exploitation based on wage labour". 10 10 A similar argument is made by Coe and Jordhus-Lier (2011:222) using the "Global Production Networks" frame (see Bair 2009), where they suggest that "there are massively different levels of potential agency within functionally integrated economic networks". 11 11 Interview 14, 23 November 2017. 12 12 Interview 9, 2 November 2017. 13 13 Interview 20, 6 November 2017. 14 14 Interview 28, 6 November 2017.