Crisis and Differentiation among Small-Scale Sugar Cane Growers in Nkomazi, South Africa

Against a context of declining sugar output in South Africa as a whole, the sugar industry in the Nkomazi Municipality of Mpumalanga Province has increased its share of the South African market. It has achieved this over a period of significant change in ownership, with the transfer of at least 25 per cent of land growing sugar cane into black community ownership through South Africa’s land reform programme. The industry now claims that the majority of land used for sugar cane in Nkomazi is owned by the beneficiaries of land reform. This paper examines the historical and contemporary trajectories of sugar cane production in Nkomazi, focusing particularly on the changing status of production on black-owned land. Among small-scale growers, a crisis in operation and maintenance of irrigation has prompted on the one hand a process of land concentration and ‘accumulation from below’, visible in the emergence of medium-scale growers, and, on the other hand, a move by the sugar milling company to take more direct control of sugar cane growing through rental agreements with small-scale landowners. The latter draws on recent experience of ‘joint-venture’ sugar cane production on land transferred to black ‘community trusts’ under the government’s land restitution programme. The paper argues that the moves to medium- and large-scale farming are consistent with the changing livelihoods and aspirations of black South Africans since the end of apartheid, but also suggests contradictions between the emergence of black capitalist medium-scale farmers on the one hand and extension of corporate control of production on the other. While corporate agriculture offers advantages to some, in particular farm employees and a small number of black-owned contractors, it appears to offer little benefit to the majority of African ‘landowners’ while potentially blocking the further expansion of medium-scale growers.


Introduction
Black ownership of agricultural land and 'economic empowerment' through agricultural production has presented major challenges to the post-apartheid governments of South Africa. In particular, it has become increasingly clear that transfer of land ownership is but the start of a process of creating productive black-owned farms. The absence of technical, managerial and marketing support to the new owners of commercial farmland has been identified as a key factor in the relative lack of success of agriculture following land reform in South Africa. 1 Despite wide acknowledgement of this, government agencies lack the technical capacity and experience to support development of competitive farming enterprise in South Africa's highly capitalised commercial agricultural industry. As a consequence, the private sector has been the primary source of commercial and technical expertise to support the development of black-owned agricultural enterprise and government has promoted the formation of 'strategic partnerships' and 'joint ventures' between commercial actors and land-owning black communities and individuals.
Experience with such partnerships has not been encouraging, however, and it has proved difficult to sustain productivity while achieving more equitable and inclusive distribution of benefits. 2 In this context of policy frustration, the South African sugar sector, and the Mpumalanga industry run by the sugar-milling company TSB in particular, stands as an exception in which productivity of farmland transferred under land reform has been maintained. There is some continuity here with the Mpumalanga industry's broader involvement with small-scale African sugar cane growers in Nkomazi since the 1980s. In this paper we draw on a study of sugar cane production on black-owned land in Nkomazi undertaken between 2012 and 2014. 3 The work generated primary data from interviews and surveys and supplemented this with information from a variety of secondary sources including production records and financial data provided by TSB. We argue that, following a rapid expansion of African sugar cane production since 1994, reorganisation of local government meant that services previously provided by Bantustan (KaNgwane) government were withdrawn, leaving growers more dependent on industry support to deal with the challenges of infrastructure renewal and growing production costs. This transition has taken place at a time when small-scale sugar cane growers confronted a crisis, primarily in the operation and maintenance of irrigation systems. We identify two responses to this crisis. First, falling sugar cane incomes have precipitated a move towards concentration of land-holdings and the emergence of a class of medium-scale African sugar cane growers. Second, recuperation of failing irrigation schemes via state funding has been linked with direct management of sugar cane production by the milling company (TSB), using a model initially developed to operate large-scale joint-venture sugar cane farms on land transferred to community trusts under land reform. In this paper, we explore the origins and outcomes of these trends, examining who wins and who loses. We argue that, while direct management of sugar cane production by TSB offers clear benefits to the industry in terms of stability of cane supply and opens up better opportunities for incomes and careers for some, it has generated little benefit to the majority of African 'landowners' while potentially blocking the further expansion of medium-scale growers.

Settlement, Dispossession, Restitution and the (Re)Structuring of the Sugar Industry in Nkomazi/Onderberg
Through much of the nineteenth century, the area now covered by Nkomazi Local Municipality was an arena of settlement and struggle linked to the establishment and expansion of the Swazi state. From the 1840s to the 1870s, a period of Swazi supremacy over much of the area, people moved in and out of the area as control was contested between the Swazi and neighbouring African powers (Zulu, Pedi and Gaza-Shangaan) and the Boer republic centred on Lydenburg. By the period of decline in Swazi power in the 1870s, the area was populated by groups identifying with leaders of diverse origins who had been established by the Swazi monarchy in administrative/military settlements along the northern periphery of Swazi territory. As Swazi power waned, their frontier was defined with increasing specificity, first by the Boers in the 1860s and then by the British in the 1880s. Following construction of a road to link Johannesburg to Lourenço Marques (Maputo) in 1876, colonial government began surveying the area and 'farms' were demarcated for European settlement, initially to serve as resting stations for travellers. Few Europeans settled in the Lowveld, however, and at the turn of the century in 'Barberton District' there were only around 3,500 Europeans, mostly in the mining settlements near Barberton, to the west of modern Nkomazi, while the area up to the Mozambique border was settled by about 30,000 Africans, 4 albeit on land that was notionally 'white' farms. 5 With the passage of the 1913 Land Act, prohibiting black purchase of freehold land in 'white' areas, a Native Land Commission from 1914 to 1917 identified land to be 'released' for African settlement. In Barberton District two areas identified were: first, the unsurveyed 'crown lands' between the Swazi border in the south, the Lomati river in the west, and with a northern boundary marked by the confluence between the Lomati and Komati rivers (see Figure 1); and second, the surveyed area known as 'Tenbosch' farm, between the main road and the Crocodile river. Lobbying by European farmers sought to exclude 'Tenbosch' from the area allocated for African settlement, but Africans continued to farm the area as tenants of a succession of corporate owners of the land. The Mpumalanga (Eastern Transvaal) Lowveld was an area of late white settlement owing to the prevalence of malaria and tsetse fly. 6 African communities were able to maintain access to land until the 1950s, living through a combination of migrant wage labour in the Barberton mines or further afield, and agriculture that produced occasional surpluses of sorghum and maize. 7 Numerous communities, particularly those to the north of the region, had been impoverished by the effects of foot-and-mouth disease and the resulting confiscation and destruction of their cattle, and were consequently more reliant on wage labour in mines and on the growing number of European-owned local farms. 8 In 1951, the government undertook a formal designation of 'chiefs' and their areas of residence, 9 incorporating them as salaried government officials, as a prelude to their resettlement in African reserves. 10 In 1954 the Ngomane were forced out of Tenbosch. These and others from the area known as 'Coalmine' (Mhlaba/Mkatshwa/Mbambiso) and from west of Lomati River (Shongwe/Matsamo) were relocated -thus displacing other people (Mawewe, Mahlalela) -in the 'crown lands' strip of territory north of the Swazi border.
European settlement began in the 1960s and the sugar industry was established in the lowveld 'Onderberg' area in 1967 with the construction of the first cane-crushing mill by Transvaal Suiker Beperk (TSB) at Malalane (formerly Malelane), some 60 km east of Nelspruit, and 1 km south of the Crocodile river that forms the southern boundary of the Kruger Park. The sugar industry followed an 'estate' and 'outgrower' model, with the milling company (TSB) directly operating an 'estate' of some 5,000 hectares (ha) producing citrus as well as sugar cane while the remainder of output was purchased from white settler 'outgrowers'. Expansion of commercial sugar cane growing by white settlers in the 1970s was associated with designation in 1982 of a KaNgwane 'homeland' to administer the areas reserved for black settlement. Nkomazi became one of the three districts of KaNgwane, and the Nkomazi population was to be administered under seven 'tribal' authorities.
Starting in the 1980s, sugar cane production was extended to Africans through irrigated projects financed by the South African government and run by Agriwane, a parastatal under the KaNgwane administration. Production by small-scale growers expanded in three distinct phases between the early 1980s and the mid-2000s: first, an Agriwane-led period of development during the 1980s, second, the Nkomazi Irrigation Expansion Program (NIEP) Phase One during the mid-1990s, and third, the NIEP Phase Two or 'Land Bank Projects' during the mid-2000s. A total of 37 small-scale projects were created covering roughly 10,000 ha of irrigated land and incorporating about 1,200 small-scale growers into sugar cane production. Project sizes ranged from 41.2 to 839.8 ha and individual plot sizes ranged from 2 to 20 ha. Within projects, plot holders typically shared the pumps and distribution infrastructure used to deliver water from local rivers to farmers' fields.
With the establishment of majority rule in South Africa in 1994, the KaNgwane administration was dismantled and many of its functions transferred to the new Provincial Government of Mpumalanga in Nelspruit. The areas of African settlement were joined with those of European commercial farming to the north, to form Nkomazi Local Municipality. Under the terms of South Africa's land reform, a series of claims for land restitution were lodged by people who had been removed during the 1950s. As was the case at the time of their removal, claimants were identified in terms of their 'tribal' (now termed 'traditional') leaders, and, in all, seven such 'traditional communities' made claims. Four communities were involved in a joint claim -the Tenbosch land claim -subsequently extended by two more claims by the Matsamo and Mahlalela to form the Greater Tenbosch claim. In 2008 the claims were settled, effectively transferring to black communities a total of 61,202 ha for which the South African government paid the existing white farmers R2.8 billion (US$426.7 million). 11 Much of the land was commercial irrigated sugar cane, citrus or banana plantations, covering much of the area from Malalane east to the Mozambique border and from the boundaries of the former KaNgwane north to the Kruger National Park. The owner of the sugar cane mills, TSB, had encouraged landowners not to contest the land claim, ensuring that the claim was settled quickly. It also moved swiftly to put in place arrangements to farm the land through a number of joint-venture companies in which community trusts, to whom land ownership had been transferred, were partnered with the company.
In 2013 the sugar industry in Nkomazi was constituted by a single milling company (TSB) operating two mills: Malalane (built 1967 and extended several times) with nominal milling capacity of 1.83 million tonnes of cane per year; Komati (built 1994Komati (built , expanded 1998Komati (built and 2006 with nominal milling capacity of 2.5 million tonnes of cane per year. In addition, TSB runs a refinery at Malalane which takes the output from the Malalane Mill and also part of the raw sugar output from the Komati Mill. The remainder of Komati Mill's raw sugar output is exported via Maputo. Also at Malalane is Molatek, a TSB-owned animal feed enterprise using the molasses by-product from sugar mills. In early 2014 TSB was sold by its owner, Remgro group, to Rainbow Chicken Ltd (RCL), also owned by Remgro, as part of a restructuring of the group, a move the group claims will enable better integration between different elements of its agricultural business.
The Nkomazi industry is distinctive in the South African context -although similar to production elsewhere in southern Africa 12 -because it is entirely based on irrigated production, whereas the longer-established industry in KwaZulu-Natal is predominantly reliant on rainfall. In the 2013/14 season the two Nkomazi mills crushed a total of 4.07 million tonnes of cane, thus running at 94 per cent of their nominal operating capacity. The contribution of different types of producer to the sugar cane harvest is shown in Table 1. Commercial growers on freehold land account for about half of the crop. This diverse group includes several thousand hectares of 'land reform' land either awaiting final settlement of restitution (following purchase by government) or transferred to black ownership under the 'willing buyer-willing seller' land redistribution programme facilitated by government grants (the Settlement Land Acquisition Grant and Land Redistribution for Agricultural Development programmes) to prospective black buyers. The other half of the crop is split between two different sets of black landowners and  between contrasting scales of production. These are first, small-scale growers on projects on ex-Bantustan land nominally held under customary rights under the jurisdiction of traditional authorities, and second, community trusts who hold land transferred in settlement of restitution claims and for the most part farmed as large-scale units of two to three thousand hectares by TSB, either through joint-venture companies or other lease agreements. As we set out in more detail below, the effect of land reform has been to shift sugar cane production away from largescale white commercial 'outgrowers' to black landowners but also to directly-operated 'estate' production by the milling company.
Here it is important to note that all 'growers', on both large and small-scale, hire labour for most of the field operations in sugar cane: permanent staff for irrigation, casual labour for weeding and contract labour (via a labour contractor) for cane-cutting. Among surveyed small-scale growers, 85 per cent employed at least one permanent labourer, on average one per 5.4 ha. Workers' pay from small-scale growers averages R857 (US$101) per month and no labourers received the national statutory minimum monthly wage of R2,274 (US$267), possibly because many are undocumented migrants from Mozambique and Swaziland. Casual labour, typically for tasks such as weeding and edge cleaning, is paid an average daily wage of R34 (US$4) -also significantly below the daily minimum wage of R105 (US$12.35), and, based on the statutory hourly rate, would amount to only three hours of work in a day, compared to the six hours claimed by growers. Pay rates are higher on joint-venture farms (see below), but this does not apply to cane-cutting, which is done on all farms by contract labour, with the same labour contractors supplying both large-and small-scale growers. Cane-cutting is based on piecework, and it is not uncommon for cane-cutters -the great majority of these migrants from Mozambique -to earn in excess of the minimum wage. This is perhaps a reflection of the intensely physical and inherently dangerous nature of the job.
In the remainder of the paper we trace the development of sugar cane production on black-owned land, identifying the divergent trajectories of production at small-, large-and medium-scale.

The Rise and Fall of Small-Scale Sugar Cane Growers
The first phase of small-scale farming development emerged after state farms in the KaNgwane Bantustan had failed to generate significant employment and were heavily indebted. The rapid expansion of commercial sugar cane production to the north prompted demand within KaNgwane for irrigated land for small-scale commercial agriculture. Existing state farms were subdivided to create projects with individual plots and shared infrastructure. Early projects allocated larger plots (e.g. approximately 20 ha at 'Figtree A') and, sometimes separate pumps, to individual growers, whereas on later projects individual plots were smaller -typically 7 ha (e.g. Ngogolo) -and pumps and pipe infrastructure were shared. 13 Agriwane provided finance and extension workers oversaw production and the provision of contractors. Despite initial scepticism surrounding the decision to grow sugar cane, 14 growers who began farming in this period were largely successful and repaid initial loans earlier than expected. 15 The second phase of small-scale production expansion, the NIEP, came at the close of the apartheid era, and two new dams, Maguga on the Komati in Swaziland and Driekoppies on the Lomati at the Swaziland-South Africa border, were justified by an irrigation allocation of 7,191 ha for small-scale growers in KaNgwane. Together with the 2,349 ha of existing irrigated sugar cane, this expansion also justified TSB's construction of their second sugar mill (Komati). During the NIEP the total of 21 small-scale projects covered 5,924 ha -thus less than planned, and an under-utilisation of available water resources. However, during this phase KaNgwane was reincorporated into the Republic of South Africa and implementation and financing of projects was transferred to the new Mpumalanga provincial government. Critically, the transfer involved dismantling Agriwane, and appears to have disrupted the Agriwane-administered loan repayments by which small-scale growers were to have financed the maintenance and ultimate replacement of the irrigation infrastructure. 16 This is of particular importance since inadequate funding for infrastructure maintenance is later identified as a key cause of problems affecting small-scale projects. The final phase of small-scale sugar cane farming expansion took place in the mid-2000s through the NIEP Phase Two, also known as the 'Land Bank projects', due to the prominent role of the Land Bank in financing this group of projects. In this phase, a further six projects on a total of 1,723 ha were intended to take up the full irrigation allocation to the area, under-used under NIEP Phase One.
Interviews with individual African sugar cane growers in 2013-14 suggests that over the past 50 years the industry has transformed livelihoods, first by people's displacement and resettlement and then through their recruitment into small-scale commercial agriculture. The evictions and resettlement of people into Trust Lands from 1954 created conditions typical of Bantustans in South Africa: overcrowded villages with limited grazing and unproductive land. For many households the remnants of subsistence agriculture were destroyed or reduced to little more than the maintenance of a kitchen garden. For early participants, small-scale sugar cane farming in Nkomazi had profound effects. High productivity and net incomes allowed households to improve the standard of their housing, buy cars and trucks and pay for their children's education, often to university level. Now, however, this phase of small-scale sugar cane production appears to be nearing its end.
The past decade has been marked by declining production from small-scale sugar cane growers, and by 2012 a total of only 888 farmers delivered cane from 6,238 ha of land, 17 leaving approximately 39 per cent of land on small-scale projects unproductive that season. Of the 355 registered growers that delivered no cane, some were replanting their fields or had been affected by the early closure of the sugar mills after a hauliers' strike earlier in the season. However, a significant number had either ceased production individually or could not produce because they had plots on small-scale projects that had collapsed completely. TSB noted that only about 450,000 of a potential 850,000 tonnes of cane were being produced from small-scale production. Average yield was 65 tonnes per ha for 2012, compared to 80 tonnes per ha needed to cover production costs. A survey of 108 growers showed that over three years (2010-2012) almost half had produced less than 60 tonnes per ha on average, while only 27 per cent averaged over 80 tonnes per ha. Moreover, comparison of these groups over time showed this yield difference between 'winners' and 'losers' tending to increase. 18 A key factor underlying the current problems is the degradation of shared infrastructure and a lack of financial capacity to rehabilitate it. The permanent irrigation infrastructure of NIEP projects was designed with a 15-20 year lifespan in 1995-98, so could be expected to be requiring replacement by 2010-13. In practice, equipment and infrastructure degradation was accelerated by a series of physical and climatic shocks. Major floods in February 2000 damaged many irrigation pumps and river weirs across the region. Investment in repairs and replacements were delayed by government financial constraints, further exacerbated by lack of maintenance following the dissolution of Agriwane. This resulted in many growers having no irrigation for the remainder of the growing season and production declining sharply. A gradual improvement over the next three years was checked when a severe drought affected the area in 2004-06, resulting in restrictions on irrigation. The overall effect of this prolonged period of low productivity reduced growers' savings available for reinvestment and entrenched a decline in output from small-scale projects. These problems were compounded on the Land Bank projects that many in the industry regard as crippled from the start by extremely high debt levels alongside a catalogue of poor project design and implementation they attribute to 'unscrupulous contractors', including selection of land unsuitable for irrigation, inappropriate irrigation designs and poor standards of construction. In recent years a spate of thefts of electrical cables have halted irrigation pumps and created new costs that projects already burdened with falling incomes and high debt levels have been unable to cover. The deterioration in irrigation capacity meant a failure to irrigate between 30 and 40 per cent of the area of small-scale projects using water from the Komati River during the 2012-13 growing season. 19 The origins of small-scale sugar cane production projects in the 'homeland' era left a number of legacies that left small-scale producers ill-prepared for the newly liberalised commercial environment of post-apartheid South Africa. One of the most important was that many individual growers had a poor understanding of their financial liabilities -which in some cases involved debt obligations that did not even comply with commercial conventions on rates of repayment. A second problematic legacy was the reliance on shared irrigation infrastructure which depended on external (publicly funded) technical support, soon to be withdrawn. Third, hierarchical traditional authorities played a central role in the allocation and governance of the 'communal' land upon which small-scale projects were developed. Individual growers were issued a 'right to occupy' (a document widely known as an RTO or PTO -permit to occupy) by the relevant chiefs for their irrigated plots. Unsurprisingly, perhaps, chiefs claimed disproportionate areas of irrigated sugar cane land for themselves and their families. Thus, for example, the chief of the Mahlalela has become one of the largest 'small-scale' growers in Mpumalanga. The system also provides opportunities for rent -the Matsamo traditional authority charges an annual fee of R150 per ha for land used for commercial agriculture, for example. Growers' ability to evade rent payments is constrained by their dependence on the traditional authority for services, such as confirmation of residence or authorisation of financial documents, often essential to an individual's ability to engage with South Africa's commercial and government agencies (e.g. to open a bank account or obtain a driving licence). While still negotiating this 'homeland' legacy, growers have been confronted abruptly with a very different set of challenges posed by purchasing services from commercial contractors as well as devising effective processes of collective management of shared assets.
Against this generally difficult backdrop, a number of projects have collapsed completely, with unpaid debts, accumulating interest charges, and diminishing prospects of refinancing and restarting production. The majority of these were Land Bank projects affected by serious problems from the outset, as discussed above, but two were from the NIEP Phase One (Boschfontein One and Two). As with the Land Bank projects, these had design flaws -they drew water from a local dam that had insufficient storage to supply the irrigated area -but were also affected by a series of cable thefts and were unable to pay for repairs. With no way to irrigate, both projects failed.
To avoid permanent loss of sugar cane areas, TSB has worked with government to restructure failed projects. In particular, TSB has negotiated the restructuring of project debt and the provision of government recapitalisation funding on the condition that projects will be reorganised into cooperatives with the appointment of professional management. Under these arrangements farming operations are to be consolidated for the whole project area and individual growers will become shareholders in the cooperative, to which they will lease their individual plots. This model draws on TSB's experience of managing sugar cane production on its joint-venture farms on land leased from community trusts that are now the custodians of land transferred to black ownership under land restitution. In the next section we will briefly review that model.

Production after Restitution: The Joint-Venture Model
Since the start of majority rule in South Africa, the structure of the Mpumalanga sugar sector has been dramatically transformed by one of the most valuable restitution claims in the country. As a result, the Mpumalanga sugar industry has been exceptional in the extent of land transfer to black ownership and claims that 62 per cent of the land supplying sugar cane to TSB's two mills in Nkomazi is now under black ownership. 20 Given this high level of changing land ownership, and the consequent risk of land falling out of production, TSB has played a prominent role in settling land claims, including through the restitution (sale to government) of 6,000 ha of TSB's own estate.
In return, government has favoured TSB's proposals to maintain sugar cane production by entering into joint-venture agreements with three of the communities with restitution awards. Joint-venture production involves the creation of farming services companies based on 50/50 partnerships between organisations representing the claimant communities (trusts or community property associations, hereafter called 'trusts') and Shubombo Agricultural Services, the farm operations branch of TSB. The boards of the joint ventures are formed equally through employees of TSB and board members of the respective trusts. The joint-venture companies pay the trusts to lease the land. The lease payments contain fixed and variable elements, the latter being determined by profitability. The joint-venture companies have access to centralised TSB services such as human resources and finance and accounting, as well as agricultural inputs purchased in bulk. In addition to joint ventures, TSB also operates farms it leases directly from trusts. In total, approximately 10,000 ha of sugar cane is farmed by joint ventures and other leasehold arrangements. In 2013 three joint ventures delivered 859,487 tonnes of sugar cane from 7,598 ha of land. An average yield of 113 tonnes per ha indicates that the joint-venture farms operate towards the top of the range of productivity in the industry and, from this perspective, land restitution in the Mpumalanga sugar industry has been a success.
However, some non-TSB joint-venture partnerships, for example between the Mjejane Trust and Umlimi Holdings, were short-lived owing to poor productivity and failure to generate profits. Mjejane's relationship with Umlimi gained international attention when action by members of the community to reclaim the farm was reported as the first land invasion in South Africa, prompting comparison with the Fast Track Land Reform of Zimbabwe. 21 The Mhlaba community also broke off their joint-venture agreement with Umlimi and directly contracted a former Umlimi employee as farm manager under the oversight of the trustees. Even where joint-venture companies are accepted by land-owning trusts, it is not clear that all members of the land-owning community benefit from the arrangement. Two aspects of current arrangements seem problematic. First, the high levels of output achieved by the joint-venture companies have been accompanied by high fixed costs, particularly for administration and management, which average 17 per cent of turnover, compared to lease payments equivalent to 8 per cent. Concern that joint-venture companies were investing too heavily in production costs was evident in interviews with both members of the claimant communities and officials from the provincial government. A potential conflict of interest arises between the position of the miller as a partner in production, a site where profits are split, and as the sole owner of the sugar factory, where they wholly own the profits. Thus, production of additional cane yields, while generating diminishing returns to the joint-venture company, may also provide an opportunity to maximise the capacity utilisation (and hence profit margins) of the sugar mills.
In contrast, joint-venture managers attribute higher costs to 'enterprise development': the need to build capacity by awarding contracts to service suppliers from the relevant claimant communities as part of the joint-venture partnership. Although all contracts are, in principle, paid at the market rate, and represent a potential source of income generation arising from restitution claims, the numbers of black-owned service contractors from a given community is likely to be small. Enterprise development may therefore limit price competition for service contracts with a negative effect on joint-venture profit margins that could provide benefits to the broader community of claimants.
A further benefit that joint ventures are claimed to offer their land-owning communities is employment. On the operations of joint-venture farms, there are fewer jobs -typically one permanent employee per 18 ha, compared to one per 5.4 ha on small-scale farms -but these are more remunerative in terms of pay and benefits. Moreover, a policy of giving preference to claimant community members in recruitment of staff for joint ventures means new career opportunities: the 190 permanent staff on a 3,600-ha farm includes, as well as irrigators and pump operators, jobs for 'supervisors' with career opportunities of promotion to farm manager and three or four positions for 'student interns' whose technical training is paid by the company. It is therefore possible to see in the joint-venture operations the development of a new black professional and managerial class embedded within the sugar industry. For another group of workers, however, joint-venture farms bring little change: cane-cutting on large-scale sugar cane farms is undertaken by the same migrant labour, supplied via labour contractors, as on small-scale production.
At a fundamental level, however, income generated by joint ventures is irrelevant to most 'beneficiaries' of land restitution, due to governance problems in the trusts. The primary income for claimant communities has, to date, been in the form of lease payments made by the joint ventures to the relevant trusts. Across three joint ventures, land rental represents 90 per cent of community income, with the remaining 10 per cent derived from profit dividends. 22 In 2012 lease income paid to the three trusts by the joint ventures totalled R15 million (US$1.76 million). However, despite two of the joint ventures being in operation since 2007, the payment of dividends by the trusts to their members has been either sporadic or non-existent. In general, trust management lacks transparency and communication between trustees and their communities is poor. A 'beneficiary' of land restitution who is also a small-scale grower observed: I think they take a lot of it for themselves rather than to give to the community. Then they fight amongst themselves and then the committee will be dissolved then another committee will be elected but the same thing happens again. I think they should have given each family a piece of land to decide what to do with it rather than putting us together. I think all this putting us together is a big problem, even on sugar. The government seem to like us to be together. 23 This view suggests a certain continuity of current policy with the apartheid-era insistence that all African occupation of land be mediated by tribal authorities. 24 That it is deeply problematic is illustrated by continuing struggles over governance of the trusts. Contestation of membership lists and litigation between competing factions to nominate trustees mean that dividend payments have in some cases been suspended. Failure to hold regular annual general meetings and elections has resulted in two of three trusts involved in joint ventures being placed under legal trusteeship. In the case of the Mjejane, trustees have (illegally) sold farmland to individuals from nearby communities for residential development. These buyers of 'housing stands', as such plots of land are known, have also been registered as new members in the community trust, thereby strengthening the position of the incumbent trustees in the face of the challenge from those who seek to replace them. The sugar industry has begun to respond to the threat to joint-venture sustainability posed by these governance issues by appointing an agricultural non-governmental organisation (NGO) to work with the trusts.
As in the case of small-scale sugar farming on 'communal' land, traditional leaders have played a prominent role within the trusts. While chiefs do not sit directly on the boards of trusts, they may promote nominations of trustees sympathetic to their positions. Thus, for example, indunas (officials within traditional hierarchies) placed as trustees of the Mahlalela trust declared their intention to pay part of the leasehold income directly to the royal house to meet 'administrative needs'. Similarly, the incoming chief of the Mhlaba tribal authority sought to claim as his private property a farm transferred to restitution claimants. The embattled trustees of the Mjejane trust have also sought the support of one of the tribal 'princes' as a means of legitimising their continued position in the face of conflict.
The problems of amount and distribution of benefits to land-owning communities have so far not distracted TSB from adapting its leasing model to recuperate production on failed small-scale grower projects, and such 'cooperative' arrangements are stipulated as a condition of government recapitalisation of such projects. The cooperative contract is a vehicle to recover debt and runs for eight years, corresponding to the debt repayment schedule. An early example of its use is Langeloop Phase Two, a Land Bank project that collapsed earlier than most due to an outbreak of smut fungal disease. It suggests that while the arrangement is successful in restoring production -the project delivered 95.5 tonnes per ha from 101.3 ha in 2013 -financial benefits to the plot holders are limited to rental income of R100-R150 (US$9.7-13.6) per ha per month, giving a maximum of R12,600 (US$1,223) 25 per year for the average seven-ha plot held by an individual project member. As discussed in the next section, this level of income is far below the aspirations of both existing sugar cane growers and potential young growers.

The 'Emerging' Medium-Scale Growers
While restructuring of production into cooperatives has focused on collapsed projects in order to secure government recapitalisation, the crisis in small-scale production has also accelerated processes of differentiation among small-scale growers, as a minority increase their sugar cane areas by purchasing plots from their neighbours. In a survey of 108 small-scale growers, 31.5 per cent had increased the size of their sugar cane fields since beginning sugar cane farming. For this minority, average sugar cane area had increased from 7.44 ha to 16 ha, compared to a slight decline from 6.78 to 6.74 in the average area of those not purchasing land. As a result, this group -a third of the initial small-scale growers -not only averaged twice the size of individual land-holdings of the rest, but also held half of the total area of small-scale sugar cane production. 26 As with the creation of cooperatives, institutions linked to TSB have played a role in this process. Akwandze Agricultural Finance, the industry's regional financial and credit agency, developed an 'RTO consolidation' loan product for which a cane delivery contract provides collateral. This facilitates purchase of 'RTO' land documents for plots of irrigated sugar cane land and has to date financed about 100 transfers of land. There is also survey evidence that some growers have financed purchases of sugar cane land with income from non-farm business or salaries from public-or private-sector jobs.
Implications for diverging asset values may be estimated from the average cost of irrigated sugar cane land, on an upward trend from R20,000 per ha in 2005 to R41,364 in 2013, giving an average value of land and water of almost R600,000 for farmers accumulating land, compared to approximately R250,000 for those who are not. Despite the cost of land acquisition, growers accumulating land tend to have higher net earnings for a given level of productivity (Figure 2), suggesting a critical role of farm scale in enabling small-scale growers to cover production costs.
In addition to this process of consolidation among small-scale growers on ex-Bantustan land, a small number of 'emerging' black growers are now operating medium-scale (40-400 ha) sugar cane farms on freehold land. While in some cases these medium-scale growers have 'graduated' from earlier experience on small-scale projects, and some growers on the 'smallscale projects' now have land-holdings that would be classed as 'medium-scale', the designation 'medium-scale' marks important thresholds. Whereas, on small-scale projects the organisation of production is still subject to collective management of shared infrastructure and land tenure is characterised (nominally at least) as a 'right to occupy', 'medium-scale' growers have usually purchased or leased 'freehold' land bought by the government from white commercial farmers under the 'willing buyer-willing seller' land redistribution programme.
This increase in scale is not without problems, however, as it crosses a number of boundaries for which growers are not always well prepared, particularly with regard to tax status and the complexity of loan funding which may characterise land transactions. This is exemplified by the Siyathuthuka scheme, for which TSB recruited six of the best small-scale sugar cane growers from Nkomazi together with a former TSB extension officer to operate seven farms of about 40 ha each. Productivity levels have been high, and TSB notes in the 2012 annual report that the group has been outperforming commercial growers in the Malalane area. However, while the land had been purchased by TSB in order to create a 'share block' ownership scheme, subsequent disagreements between TSB and the Mpumalanga Department of Agriculture blocked implementation of the original plan, leaving the growers with little income from the scheme for 15 years while they service commercial bank (ABSA) loans taken out to finance the land purchase. The case underlines that an ability to manage financial leverage may be as critical as farming expertise for survival in the contemporary farming environment in South Africa. The significance of the emergence of medium-scale sugar cane growers in Nkomazi is highlighted by the lack, otherwise, of succession of ownership of sugar cane plots. Almost 72 per cent of small-scale growers are aged 50 years or over while less than 4 per cent are aged younger than 30 years. A workshop undertaken in August 2014 with small-scale growers suggested this reflects the altered aspirations of the children of small-scale growers. Higher levels of education coupled with the ending of apartheid-era restrictions on employment has increased the opportunities available to young people. When workshop participants were asked how high incomes would have to be to encourage a younger generation to enter small-scale sugar cane production, they said R15,000 (US$1,456) per month, or R180,000 (US$17,476) per year. Our survey suggested only five per cent of small-scale growers earn over R150,000 per year, discounting unofficial withdrawals of money for household labour. This suggests that for a younger generation of sugar cane farmers either productivity or farm sizes will have to increase markedly. A yield of 70 tonnes per ha would require 50 ha of land to achieve a net income of R15,000 a month. Even for those producing more than 80 tonnes per ha (average net income R9,000 per ha per year), a minimum of 17 hectares would be needed, some 77 per cent more than the average area (9.6 ha) currently farmed by small-scale growers. 27 In practice, many small-scale growers have experienced a collapse in their income levels in recent years. The mean net annual sugar cane income estimated in 2012 was R49,161 (US$5,783), and a quarter of those surveyed earned nothing in 2012. For a majority of small-scale growers the need for further government investment to renew their irrigation infrastructure seems likely to mean they will become shareholders in 'cooperatives' run by industry-appointed managers, with only a low level of rental income assured for the next eight years.
However, for a minority of small-scale growers, land acquisition has enabled an increase in scale that allows them greater capacity to repay credit and invest in order to raise productivity. Some of these have diversified business interests: in retail shops, in transport and as labour contractors. Others have non-farm income from salaries in government or private-sector jobs. Yet others are growers' children, educated and with well-paid jobs elsewhere, taking over and expanding their parents' sugar cane holdings. Others still are new entrants who see more or less speculative opportunities in acquisition of high-value irrigated land. As discussed above, 'emergence' to medium-scale production through acquisition of 'freehold' commercial farms involves adherence to tax regulations and its corollary: the near-mandatory embrace of the financial services industry, with the pitfalls and risks this carries for the unwary. For many others, however, the historic basis of small-scale production in 'rights to occupy' land administered by traditional authorities is now overlain by land acquisition via transactions in a de facto land market, raising questions about the nature of land ownership under commercial agriculture in ex-Bantustans.

Conclusion
At a time when the other two big South African sugar producers, Illovo and Tongaat-Hulett, have been reducing production in South Africa in favour of expansion of production elsewhere in southern Africa, 28 TSB has maintained production in Nkomazi and acquired a third mill at Pongola, to the south, enlarging its share of the South African sugar market to 30 per cent. Given the company's associations with white settlement on land from which black people were evicted in the 1950s, its continued operation depended on negotiating the challenges of land reform. The proactive role played by TSB in leveraging the settlement of land restitution claims and in enlisting government support for maintaining land and water use for sugar cane not only provided the conditions for keeping the sugar mills running, but also forced the rapid evolution of new relations of production and land ownership for corporate agriculture in South Africa. In particular, the emergence of TSB as the major agricultural services provider in Nkomazi places it in a position where it controls significantly larger areas of land and water for sugar cane supply than it did in the pre-reform era. 29 Corporate leasing of land owned by black communities has, at least in the short term, resolved the perceived tensions between land reform and agricultural productivity. Moreover, the model has been adapted to manage production on the newly formed 'cooperatives' funded by government to rehabilitate production on ex-Bantustan land under traditional land tenure. Yet further expansion of corporate management of cane production seems possible through TSGro, a new farm management joint venture between TSB and Nkomazi small-scale growers' associations. Launched in 2013, TSGro offers services such as irrigation, input supply and farm management not only to the cooperatives but also to other small-scale growers either as individuals or as entire projects. Two factors may threaten the long-term stability of such a 'settlement'. One is the extent to which land-owning constituencies feel they benefit, and the processes through which they can demand accountability for the use of the land they nominally own. As observed above, instability and contestation arising from lack of transparency and accountability are already evident in community trusts that hold restitution land, but have also emerged in the cooperative model in Langeloop. This suggests that, to succeed in the longer term, corporate management of agriculture may need to instigate rules of 'shareholder' governance to overcome apartheid's legacies of patrimonial 'tribal' power hierarchies.
The emergence of a group of black medium-scale sugar cane producers may pose other challenges to the emerging pattern of corporate control of sugar cane production. Although currently dependent upon sugar cane contracts, this group has a degree of financial autonomy, and, in some cases, technical capacity to diversify production away from sugar cane in pursuit of higher returns. The comparatively rapid expansion of this group, both by consolidation of small-scale plots and acquisition of freehold farms, may come to confront constraints on further expansion due to the large areas taken out of the freehold land market by corporate leases on restitution land but also potentially due to terms imposed by traditional authorities for greater access to irrigated land in ex-Bantustan areas. The political processes arising from these economic factors may come to dominate the future development of sugar cane production in Mpumalanga.
The model by which a broad mass of rural South Africans were to have benefited from sugar cane as small-scale outgrowers generated significant benefits for a generation of people linked to about 1,500 households in Nkomazi, but has been checked over the past decade by a combination of factors. Poor technical support in terms of project design and irrigation operation and maintenance has undermined small-scale productivity, and opened a space which has been filled by corporate managerial control backed by government finance. In effect the model of 'outgrower' as 'farmer' has been superseded by one of outgrower as 'shareholder', drawing from the experience of corporate sugar cane production on land transferred to black communities under the government's land restitution programme. The lack of benefits so far obtained by a majority of the land-owning shareholders under these arrangements threatens the longer-term viability of this model and highlights the parallel process of consolidation of medium-scale farming businesses. While encouragement of such agricultural entrepreneurs has at times been the goal of South African policy on land and agriculture, their emergence in Nkomazi identifies a number of potential contradictions with other aspects of that policy, particularly with respect to the disproportionate control of land and water use by agricultural corporations and the ambiguous status of commercial agriculture on land administered by 'traditional' jurisdictions established under apartheid. In this landscape it seems likely that any rhetorical policy commitments to small-scale agriculture are liable to be subordinated in practice to the potentially competing interests of the corporate owners of the industry, a growing black professional and managerial class within corporate agriculture, the 'emergent' blackowned, medium-scale farms-cum-businesses, and the rent-generating demands of traditional authorities in rural areas.