Between 2000 and 2002 the ‘Fast Track’ land reform in Zimbabwe saw the transfer of some 9.3 million hectares of white-owned farmland to state ownership for the use of black farmers. The process has been criticized as deploying mob violence to occupy land, as enacting unconstitutional legislation, and as serving to reinforce undemocratic government and undermine human rights (Hellum and Derman Reference Hellum and Derman2004). Nearly a decade later, the political significance of Zimbabwe's land reform and its consequences for the country's people remains the subject of intense debate (Mamdani Reference Mamdani2008; Scarnecchia et al. 2009). It nonetheless appears that the Fast Track reforms marked the end of racialized inequality of control of land that, from 1930 to 1980, had seen half of Zimbabwe's agricultural land owned by European settlers comprising only 5 per cent of the population. However, despite this significant change, and the resettlement of some 134,000 families on former white-owned farmland (Moyo and Yeros Reference Claassens2005: 195), land tenure in Zimbabwe remains strongly ‘dualist’. Resettlement areas are clearly designated as either ‘A1’ or ‘A2’. The former consist of ‘family plots’ (with additional rights to common grazing land) that are inheritable but non-marketable, while on the latter ‘commercial farms’ are held as 99-year leases, to be used as collateral to secure loans and with a prospect of eventual transfer through a land market. In the aftermath of land reform, Zimbabwe's agricultural area is, in effect, divided between two forms of land tenure (data from Moyo and Yeros Reference Claassens2005: 197): a land market governing about 25 per cent of agricultural land, and ‘Communal Areas’ covering 66 per cent, in which land sales are illegal (9 per cent of land ‘in transition’ remains unallocated). At this juncture, the future terms of tenure on A1 and Communal Areas are unclear, but the continuing dualism between ‘commercial’ and ‘communal’ categories of tenure is emphasized in government statements and highlights tensions over land markets and their role in land tenure.

These tensions are widespread in Africa, and symptomatic perhaps of the discomfort of policy makers with growing evidence of commoditization of land through ‘sales’ and rental arrangements – and their political contestation in the name of ‘autochthony’ and other forms of ethnic discrimination. Contemporary scholarship on land tenure in West and Central Africa increasingly suggests that access to land has become mediated through informal ‘markets’ operating within ‘customary’ tenure regimes (Woodhouse et al. 2000; Amanor and Diderutuah Reference Amanor and Diderutuah2001; Gray and Kevane Reference Gray and Kevane2001; Kasanga and Kotey Reference Kasanga and Kotey2001; Benjaminsen and Lund Reference Benjaminsen and Lund2003; Mathieu et al. 2003; Colin and Ayouz Reference Colin and Ayouz2006; Ubink and Quan Reference Ubink and Quan2008), a combination we have described elsewhere as a ‘vernacular’ land market (Chimhowu and Woodhouse Reference Chimhowu and Woodhouse2006). Here, by land ‘markets’ we understand (usually) monetized transactions (‘sales’ or rent), so that purchasing power becomes a necessary (if not sufficient) condition for those wishing to use land. We understand ‘customary tenure’ to mean access to land (or other resources, such as water, forest, grazing, etc.) mediated by membership of a community, usually defined in terms of kinship within a lineage with historic claims to particular lands. Such rights of access are governed by hierarchy within both lineage and the broader ‘community’ identified usually with a customary authority, or chief, with powers to arbitrate over the interpretation and enforcement of customary rights within a defined area of jurisdiction. The notion of ‘vernacular’ land markets therefore incorporates potential tensions between the economic dynamics of land commoditization and the political dynamics (both local and national) through which specific land transactions are legitimated or contested – typically through competing interpretations of ‘custom’ or ‘tradition’.

The characterization of ‘customary’ rights as excluding a land market is rooted in colonial (especially British) administration that, in the name of ‘African tradition’, actively suppressed land markets among Africans in the early twentieth century (Hill Reference Hill1963; Chanock Reference Chanock, Mann and Roberts1991; Cowen and Shenton Reference Cowen and Shenton1991; Amanor Reference Amanor, Ubink and Amanor2008). It continues to exert strong influence on land policy debates in Africa that are often still couched in terms of a dichotomy between, on the one hand, registered land titles transferable through land markets, and, on the other hand, ‘customary rights’ that explicitly or implicitly are excluded from market transfer. The logic of this dichotomy is perhaps strongest in Southern Africa, where the ‘non-market’ nature of customary tenure was energetically promoted by colonial governments on the ‘communal’ or ‘tribal’ lands to which Africans were confined in colonies with European settlement. It is perhaps also a consequence of this that, in comparison with West Africa, few detailed studies of land transactions within customary tenure have been undertaken in Southern Africa. Despite discussion of land tenure reform in many Southern African countries, it often remains a key assumption of contemporary policy that markets do not and should not play any role in mediating access to rural land. This assumption is bolstered by an enduring regional discourse that characterizes ‘communal tenure’ as being ‘embedded’ within the socio-cultural fabric of ‘indigenous’ groups in the region and as having by and large resisted commoditization (see Berry Reference Berry1993; Quan et al. 2004), in contrast with markets for labour and for agricultural products and natural resources inherited from the colonial era. Yet evidence mounts of the sale or renting of ‘communal’ land in Lesotho (Selebalo Reference Selebalo2001), in South Africa (Benoit and Jacus Reference Benoit and Jacus1997; Magni et al. 2002; Claassens 2005), in Malawi (Kishinda Reference Kishinda2004; Peters Reference Peters2004; Peters and Kambewa Reference Peters and Kambewa2007), and in Mozambique (IIED 2006).

In this article we draw upon a field study of processes of access to land in late 2005 in Svosve Communal Area, Zimbabwe, to examine the economic and political dynamics underlying land ‘sales’ and rental. The article then considers the contemporary situation in Svosve within a broader historical context of colonial and post-independence government policies on African access to land markets in Zimbabwe. We argue that, while the current economic crisis in Zimbabwe has greatly increased demand for land, and hence the driving force for land commoditization, the ruling party's political control of rural constituencies is predicated on reinforcing customary leaders as agents of government. This is likely to perpetuate a juridical separation of ‘communal’ and ‘commercial’ land, while in practice reinforcing the ‘vernacular’ and insecure character of land markets in Communal Areas.

BACKGROUND TO SVOSVE COMMUNAL AREA

According to their oral history, Svosve people arrived in what is now Marondera District (Mashonaland East) from the north in the mid-eighteenth century. They then remained based in this area despite the turbulent period of raiding and conquest between rival Ndebele, Ngoni and Rozvi groups in the nineteenth century (Beach Reference Beach1994). Following the suppression of the first uprising (chimurenga) against British occupation, in 1897, colonial administrators expropriated a major part of the Svosve territory for European farming and the people were split between three ‘native reserves’ (later called tribal trust lands), at Svosve, and at Chihota to the west and Wedza to the south-east. At independence, in 1981, the ‘tribal trust lands’ were re-designated ‘Communal Areas’ (CAs), and Svosve, with a total area of 11,000 hectares, is one of the smallest. With annual rainfall of about 750 mm, it is regarded as having a relatively favourable climate for growing crops. However, nearly 50 per cent of the reserve presents constraints to cultivation due to waterlogging in lower-lying areas (dambos), or steep slopes and shallow soils overlying granitic outcrops (Elliott Reference Elliott1989, and Figure 1). As a consequence, colonial administrators were dismissive of the area's agricultural potential. Nonetheless, African vegetable producers cultivating small irrigated fields at the margins of the dambos were important suppliers for the then urban settlement at Marandellas and neighbouring European farm settlements, and it seems likely that this provided an important source of income, not only on the larger reserve of Chihota (Bell and Roberts 1991) but also in Svosve, mitigating pressures on Africans to seek wage income on neighbouring European-owned farms. This is consistent with records showing that, in the 1930s, farms found it difficult to recruit labour from Svosve and tended to bring workers from other areas of Zimbabwe, or from Malawi and Zambia (Elliott Reference Elliott1989: 79).

From the 1940s Svosve Reserve was subject to the full range of modernizing interventions by which the colonial government sought to improve and stabilize African farming. Chief among these were the introduction in the 1940s of mechanical works to control soil erosion (contours and drainage strips), the enforcement of destocking of cattle, and the reorganization of farming into designated residential, arable, and grazing zones under the Native Land Husbandry Act (NLHA) in 1952–6. Elliott's detailed archival study (1989) provides a graphic illustration of the impact of these colonial interventions, which included the removal of 50 per cent of the adult male population – equivalent to 256 households – to other reserves in 1947, and a reduction in cattle numbers by two thirds between 1942 and 1957 (Table 1).

Analysis of successive sets of aerial photographs of Svosve shows that, of the estimated total of 5,797 hectares usable for cultivation or grazing, the area under cultivation was reduced from 71 per cent (4,149 hectares) in 1947 to 36 per cent (2,110 hectares) in 1965, recovering to between 50 per cent (2,896 hectares) in 1975 and 47 per cent (2,712 hectares) in 1981 (Elliott Reference Elliott1989: 111). This underlines the extent to which existing patterns of land holding in the Communal Area reflect dislocations caused by colonial efforts at demographic and environmental management.

Since independence in 1981, census data show the population of Svosve rising rapidly (Table 1), reflecting changes in Zimbabwe's wage economy, particularly since 1990 (see below). Although a majority of people keep cattle, livelihoods have become increasingly centred on growing food (maize, groundnut, millet, sorghum), and, in a minority of cases, tobacco and paprika. In addition, some 70 per cent of households interviewed in 2005 were cultivating vegetables, sugar cane, bananas and other fruit on the moist dambo margins in the dry season. These wetland areas were also used for the production of tobacco seedlings during the dry season for transplanting at the start of the rains. It seems clear that living conditions for many in Svosve are very difficult. In the sample of 82 households surveyed in 2005, 89 per cent had incomes (including the value of their own food production) lower than the current ‘total consumption poverty line’, an estimate of minimum total consumption expenditure required to keep a rural family of five out of poverty (TCPL=Z$156 million per year, and Z$100 000=US$1 at the time of this survey). In the same sample, 68 per cent of households had incomes lower than the ‘food poverty line’ (FPL=Z$90 million per year), the sum needed to provide 2,100 kcal/person/month for a family of five. Household income was strongly associated with size of land holding, with the lowest income quartile averaging less than 1 hectare, compared to an average of over 2.5 hectares for the highest quartile (Chimhowu and Woodhouse 2008). We explore below the dynamics of land access that underlie this pattern of poverty linked to low land holdings.

LAND GOVERNANCE IN SVOSVE COMMUNAL AREA SINCE 1981

The control of access to land in the Communal Areas has been integral to the evolution of the system of local governance. After 1981 the newly elected majority-rule government embarked on local government reform, and through the Communal Lands Act (1981) and the District Councils Act (1982) transferred authority to allocate rural land from customary chiefs to 55 elected rural district councils (RDCs). The RDC Land Use and Natural Resources Committee, charged with land allocation responsibility, was constituted by elected councillors representing constituencies at ward level, and operating through local development committees at ward (WADCO) and village (VIDCO) levels.

The new arrangements diminished the role of ‘tribal’ chiefs whose powers over land had been formally reinstated in the last decade of colonial rule, following the colonial government's re-designation of the reserves as ‘Tribal Trust Lands’ in 1969. The 1982 legislation required RDCs merely to ‘have regard to customary law relating to the use and allocation of land’ (Section 8 (2)), effectively downgrading the role of customary authority in land allocation. In practice, legitimate authority was in many localities shared between committees of the ruling party, elected councils and customary leaders, often with considerable overlap between the membership of these different bodies. Alexander (2006: 109–10) argues that the non-democratic, centralizing, tendencies of the state, manifest in the retention of budgetary control at central level and in expectations that local councils would implement centrally designed land use programmes – many little different to the unpopular NLHA schemes of the colonial period (Kinsey Reference Kinsey2004) – undermined the legitimacy of the elected bodies. The principal beneficiaries were customary authorities, who were often better positioned to distance themselves from failed or unpopular government projects and partisan political appointments. As a consequence, in 1994, the government's Commission of Inquiry into Land Tenure commented that ‘there is evidence that the dissolution of traditional authority and their role in land and natural resources matters at independence was premature, and currently, there is widespread resistance to VIDCO/WADCO structures as credible authorities over land and natural resources’ (Government of Zimbabwe 1994: 26).

The passage of the Traditional Leaders Act in 1999 formally restored customary chiefs’ land allocation role in Communal Areas (although still notionally subject to approval by the rural district council). The elected VIDCOs were abolished, and, in their place chiefs appointed ‘village heads’ to lead ‘village assemblies’. The village heads, or masabukhu (singular sabukhu: the ‘keeper of the book’ or the local tax register), were perhaps the real ‘winners’ from the restoration of customary authority, since their post, created under the NLHA, was not even regarded as a legitimate ‘customary’ role in some parts of Zimbabwe at independence (Alexander Reference Alexander2006: 135). They became increasingly central to local governance, first through the key capacity of masabukhu to deliver tax revenue, and subsequently through their assertion of rights to allocate land (Alexander Reference Alexander2006: 167–8). In Svosve in late 2005, each ‘village’, or more precisely ‘kraal’ of 20–25 households, was governed in land matters by a ‘kraalhead’ or sabukhu, reporting to Chief Svosve (the rank of sadunha (headman), formerly an intermediary between sabukhu and chief, having been discontinued). The number of masabukhu had increased to accompany the rising number of households, so as to maintain each ‘kraal’ at roughly the same size. The 66 masabukhu in Svosve in 2005 were members of one of the two ward committees (Svosve is split into two ‘wards’), chaired by an elected ward councillor. The chief chaired a ward assembly, and, together with the two ward councillors, was a member of the rural district council for Marondera. In 2005, two of the Svosve masabukhu were women, one of whom was also a ward councillor for one of the two wards covering Svosve Communal Area.

LAND TRANSACTIONS IN SVOSVE COMMUNAL AREA

This analysis draws on a study undertaken in Svosve in December 2005 by researchers from the University of Zimbabwe.Footnote ¹ Fieldwork was timed to coincide with the start of the rains to allow identification of land users in Bamhara, Neshamba, Bonda, and Muchemwa (Figure 1). The research included semi-structured interviews with heads of households and masabukhu, to explore the contemporary dynamics of land access and use in Svosve, and a questionnaire survey of a sample of 82 land-using households from 15 ‘kraals’ to generate basic information on household assets and livelihoods. The four principal dynamics identified were: a high demand for land, notably from outsiders settling in the Communal Area; an intensification of pressure on Communal Area resources as new occupants have been settled on the surrounding formerly white-owned farms; supply of arable and residential land to settlers through sale of grazing land by masabukhu; and renting out of land left vacant by absent land holders.

LAND MARKETS IN ZIMBABWE

The exclusion of land markets from Communal Areas is enshrined in Zimbabwe's post-independence legislation, supported by discourses of inalienability of land in African tradition. Yet historical evidence suggests a different reading of the role of land markets within African custom in Zimbabwe over the past century. In particular, it suggests a need to recognize that racial segregation of land occupation was more a cause, rather than a consequence, of differences between European and African land tenure institutions. This is underlined by Ranger's (1993) observation that the ‘communal’ label for the African reserves was only used formally after Zimbabwe's independence in 1981. It reflected an adoption by the nationalist government of a particular colonial interpretation of how African land tenure worked, but one that in practice, Ranger argues, had existed only at the start of the twentieth century, immediately following the military defeat of the chimurenga. It was a moment in which the pre-colonial logic of concentrated settlements around chiefs’ compounds was replaced by that of dispersal of households seeking opportunities for independent ‘peasant’ farming:

It is this limited power of chiefship that Ranger argues was interpreted by colonial administrators as a distinctly African form of consensual collectivity in which land was held ‘in trust’ by the chiefs for allocation to members of their community on the basis of need. From the outset, however, Africans were involved in colonial land markets. Much of Matabeleland had become the property of land companies that relied on extracting rent from Africans who lived and farmed on it (Alexander Reference Alexander2006). Conversely, the very earliest legislation of the colonial period, the 1898 Order-in-Council, contained a ‘Cape Clause’ that stipulated that ‘natives’ in the colony were allowed to own or dispose of land on condition that transactions were undertaken before a judicial officer who was responsible for ensuring that the native party understood the nature of the contract. Palmer (1977: 281) found evidence that, by the time the British South Africa Company ceded control of Rhodesia to the British Government in 1925, Africans had purchased 49,966 acres by this route. At around the same time, a number of African chiefs indicated to the government's 1925 Morris Carter Commission that they wished to purchase their native reserves in order to secure them against future expropriation (Cheater Reference Cheater1990). Although the 1930 Land Apportionment Act prohibited Africans from purchasing land in areas reserved for European settlers, the 30 per cent of land allocated to Africans under this Act was divided between 22.4 per cent in the form of ‘native reserves’ and 7.4 per cent as ‘native purchase areas’ in which Africans could purchase farmland up to 103 hectares in area. In effect, therefore, 25 per cent (2.9 million hectares) of land reserved for Africans was to be allocated through a market administered by a Land Board. Some 6,500 African farmers used this mechanism to buy between them about 1 million hectares of farmland (Yudelman Reference Yudelman1964), although this represented less than 70 per cent of the 14,000 farms notionally available in the Purchase Areas (Bourdillon Reference Bourdillon1987: 82).

The operation of a land market among Africans was further extended under the NLHA introduced in 1951.This was intended to modernize land tenure as well as production, and the land rights allocated by colonial officials under the NLHA were tradeable, with transactions to be registered with the District Commissioner. It is worth noting that, while opposition to the NLHA, and its subsequent abandonment, is often linked with the creation of landlessness, Phimister's (1993) account highlights sabotage by an African elite of ‘ploughmen entrepreneurs’ who saw the NLHA as reducing their arable areas, and who mobilized ‘freedom ploughing’ campaigns – kurima madiro (Nyambara Reference Nyambara2001: 782) – to occupy land beyond that allocated by colonial officials. This needs to be seen, therefore, not as opposition to a land market, but, rather, as a protest against its restriction: the NLHA imposed a limit on land consolidation of up to a maximum of three times the standard individual allocation. In contrast, it seems possible that this pressure for extension of formal individual tenure made apparent by the NLHA was seen as a threat by the colonial authorities, particularly as the 1960 Quinton Report saw freehold tenure for all races as a route towards ending racial segregation of land tenure (Alexander Reference Alexander2006: 68). The abandonment of the NLHA in 1961 should not be interpreted, therefore, as reflecting a simple opposition of African ‘tradition’ to imposed ‘modernity’. However, throughout three decades of government intervention, the consensual communitarian model of chiefship identified in the first decades of colonial administration had been nurtured as the alternative ‘African’ tradition, if only one to be overcome by the colonial modernization project. When modernization faltered and the NLHA was abandoned, it was to this African ‘tradition’ that the colonial administration turned, establishing chiefs as land authorities in the reserves, renamed as Tribal Trust Lands in 1969.

The transfer of control of land allocation to customary authorities in the Tribal Trust Lands was quickly followed by land sales to ‘squatters’ and ‘aliens’, as Ranger (1993) and Nyambara (2001: 785) have documented. Similarly, the existence of informal land markets under customary authorities in post-independence Communal Areas has been widely acknowledged (Bourdillon Reference Bourdillon1987; Elliott Reference Elliott, Eden and Parry1996; Saruchera Reference Saruchera2000; Chaumba et al. 2003; Moyo and Yeros Reference Claassens2005; Alexander Reference Alexander2006). However, this has been of only peripheral concern in recent writing on land in Zimbabwe, most of which has as its principal concern the redistribution of white-owned commercial farmland for black resettlement. The land transactions we have described in Svosve imply a commoditization of land, but opinion is divided on whether such transactions constitute a ‘land market’. Moyo and Yeros (Reference Bernstein, Moyo and Yeros2005: 185), for example, state: ‘In the 1990s, communal land was increasingly commoditized by payment of fees or political obligation to local MPs, ruling party members, chiefs, headmen and spirit mediums. In turn, land conflicts among “villagers” and “squatters” also intensified.’ In contrast, Andersson (Reference Andersson1999: 557) argues that although sabukhu have always charged for land allocations, land is ‘not a commercial commodity’ since communal and individual claims on land co-exist and thus payment does not provide legal security to individuals. This view reflects the particular historical context of Andersson's study, in the Save Communal Area where land allocations by masabukhu, although involving payments by new settlers, are understood as assertions of frontier demarcation in grazing areas between neighbouring masabukhu under conditions where land is not considered scarce. In such circumstances, where competition is for territorial control rather than for use of the land, not only is commoditization of land likely to be less intense, but also the security of the land allocation is more likely to be contested. In these respects, the case of Svosve in 2005 offers a sharp contrast, but the illegality of land sales, coupled with the increasingly tense political situation (see below) suggest many of these ‘vernacular land markets’ remain insecure.

The Fast Track land reforms have reproduced Zimbabwe's colonial dualism in land tenure. Access to a land market for a minority, with some 6,000 recipients of ‘A2’ commercial farms, averaging 312 hectares in size (Moyo and Yeros Reference Claassens2005: 197), contrasts with the 93 per cent of redistributed land allocated as A1 tenure, providing ‘use rights to a family plot plus common grazing land’. These ‘family plots’ are ‘inheritable but non-marketable’, thus reproducing the terms of ‘traditional land tenure’ officially operating in Communal Areas. The dualism was underlined by the then Minister of State for Special Affairs with responsibility for Land and Resettlement: ‘Ever since we embarked on the land redistribution exercise in 2000, 14,000 A2 farmersFootnote ² have benefited while more than 150,000 were catered for under A1. However, let me emphasize that offer permits are given to A1 farmers while leases are for A2 farmers’ (The Herald, 28 April 2007).

This legal dualism seems as much at odds with current realities of land commoditization as with those under colonial administration. Although the Traditional Leadership Act extends to the ‘A1’ resettlement areas a model of customary authority, in some cases imposing headmen and chiefs where elected officials had represented villages for the previous twenty years (Kinsey Reference Kinsey, Evers, Spierenburg and Wels2005), the dynamics of land commoditization appear similar to those we identified in Svosve Communal Area. In particular, while ‘A1’ resettlement has made land available to some who previously had little or no land, Kinsey's (2004) data on resettlement before Fast Track showed that 48 per cent of recipients of ‘A1’ plots had existing holdings larger than the permitted 1.96 hectares, and also that 48 per cent of A1 settlers continued cultivating the land they had occupied before resettlement. In both respects, resettlement practice contradicted the government's stated legal requirements, and contributed to a rapid socio-economic differentiation among ‘A1’ settlers (Kinsey Reference Kinsey2004: 1686). As in Svosve, there is evidence that land redistribution has done little to produce equitable land holdings. In the context of widespread collapse in state services to small-scale producers it seems likely that many of the poorest land holders will prove ‘too poor to farm’ (Bernstein Reference Bernstein, Moyo and Yeros2005: 85). Conversely, the enlarged land holdings that some have acquired through resettlement may fuel a land rental market. There is already some emerging evidence of this in some places (see Marongwe 2008). Significantly, Kinsey (2005: 148) found that over 60 per cent of ‘A1’ settlers regarded renting out of land as a matter ‘for the plotholder alone to decide’, suggesting they felt that ‘rights to farmland in resettlement areas are both individual and non-transitory in nature’. Further evidence for such a socially embedded attitude to land as individual property and marketable commodity is provided by Zimbabwe press reports of land allocated under land reform in Matabeleland South being rented out (at Z$15–20 million per month) for grazing herds of cattle owned by Zimbabweans living in South Africa. Government officials are quoted as saying ‘what is happening in Matabeleland South is just a tip of the iceberg, as the practice was rampant throughout the country’ (The Sunday Mail, 1 July 2007).

LAND AND STATE POLITICS

Government disapproval of renting out of land allocated under land reform, and the repeated ‘audits’ to ensure ‘productive’ use of both ‘A2’ and ‘A1’ land is frequently reported by the Zimbabwean press, and highlights the tensions within the government's policy of using land redistribution to stimulate investment in commercial agriculture, on the one hand, and to retain political and economic control on the other. On the occasion of the issue of the first ninety-nine-year leases to ‘A2’ farmers, in November 2006, President Mugabe emphasized that the leases ‘were registrable with the Deeds Office as in the case of title deeds’. This, he said, ‘would enable financial institutions or any lenders who would have registered his or her interest on the lease to recoup his or her monetary obligation from the lessee or any other person to whom the lease might be transferred’ (Daily Mirror, 10 November 2006). This apparent statement of private and transferable title to land contrasts sharply with the institution of customary tenure through which ‘permits’ to use land are to be administered for the vast majority of Zimbabwe's rural residents in Communal and ‘A1’ resettlement areas.

Here, despite the widespread operation of land rental markets and the sale of grazing land by customary authorities, discussed above, there is evidence that access to land may become increasingly a matter not only of purchasing power or historical occupation of land but also of political allegiance. Customary leaders, headmen and chiefs, although political actors in their own right, and capable of using land allocation for their own strategic purposes (for example, Andersson 1999), are increasingly subject to pressures not only from central government but also from the ruling party. The role of traditional leaders as agents of the state, implicit in the Traditional Leaders Act (TLA) of 1998 (Kinsey Reference Kinsey, Evers, Spierenburg and Wels2005), is reinforced by their state salaries and allowances and government-purchased vehicles (104 chiefs were allocated cars and ‘another 100 due to benefit’ according to The Herald, 21 July 2006). More significant, perhaps, are reports that political rallies by opposition parties in rural Manicaland were blocked by police on the grounds that the organizers had failed to get permission from the local customary authority: ‘when we go to the police, they tell us to get a letter from a traditional leader of that particular area before we can hold the rally. But the chiefs cannot give us the letters because they are afraid that ZANU-PF Youth militia will terrorize them afterwards’ (ZimOnline 2007a). The TLA (Article 45) expressly prohibits any role of traditional leaders in elections, but such reports confirm Kinsey's (2005: 137) observation that, in practice, they are active in managing the rural vote on which the ruling party depends.

In the longer term, it is possible this will further diminish the legitimacy of customary leadership, already at risk as a result of demographic change. Particularly in the many instances of Communal Areas, such as Svosve, but even more so in resettlement areas (Kinsey Reference Kinsey, Evers, Spierenburg and Wels2005; Spierenburg 2005) with large numbers of incomers, the cultural and political power of customary authority rooted in notions of historical links to a particular piece of land may already be weakened. Alexander's (2006) account of the failure of state building in rural areas in the two decades following independence suggests that the revival of customary authority in the 1990s was a reaction to the perceived top-down political control of the elected village structure introduced at independence. Thus it was a lack of democracy, rather than an affinity for tradition, that drove disaffection with elected village government in the 1980s. However, in the short term, the coercive power of customary authorities is likely to be increased not only by their role in providing access to land but also by the increasing dependency of many rural people on government for food supplies. The good maize harvest of 2005–6 was followed by crop failures in the following two years, which the FAO/World Food Programme Crop Supply Assessment Mission attributed to a drop of 40 per cent in maize yields in 2006–7 (relative to 2005–6), and a further drop in yield of 32 per cent in 2007–8 (relative to 2006–7). Adverse weather (for example excessive rain in December 2007 and January 2008) was only partly responsible for falling productivity, which was attributed principally to a lack of inputs, especially fertilizer. As a consequence, some four to five million people faced food shortages each year ( ZimOnline 2007b, Financial Gazette, 6 June Reference Peters and Kambewa2007; FAO/WFP 2008). The continuing crisis has resulted in increasing electoral opposition to the ruling party, despite its willingness to use its capacity for coercion in rural areas, and a ‘government of national unity’ was agreed with opposition parties in early 2009, shortly after record hyperinflation (231 million per cent per year) had forced the government to allow Zimbabwean businesses to trade in foreign currency.

ACKNOWLEDGEMENT

The authors wish to acknowledge funding of this study by the Economic and Social Research Council (UK) through the Global Poverty Research Group, a collaboration between the Institute for Development Policy and Management (University of Manchester) and the Centre for the Study of African Economies (University of Oxford). We also wish to thank Jean-Philippe Colin and two anonymous reviewers for their comments on earlier versions of the article.