Malawi + 5 more

HIV/AIDS and agrarian change in Southern Africa

Originally published

Michael Drinkwater

Presentation for the United Nations Regional Inter-Agency Coordination and Support Office Technical Consultation on Vulnerability in the light of an HIV/AIDS Pandemic

9-11 September 2003
Johannesburg, South Africa

Introduction: Food Security, HIV/AIDS and Social Change

In order to understand the impact of HIV/AIDS on farming systems and food security in Southern Africa, there are several parameters that are critical to attaining a realistic and relevant notion of the trends that are taking place. First, since I do not believe it is useful to separate conceptually a food security from a livelihoods perspective, I shall not attempt to, although I will focus on the aspects of food security and economic choices. Second, I shall posit that, substantially, in the rural areas of countries like Malawi and Zimbabwe, processes of depeasantisation and deagrarianisation (Bryceson 2003) are accelerating as a result of the impact of HIV/AIDS. These terms shall be explained in more detail subsequently, but they mean essentially, that farming systems and social structures, as conventionally understood, are breaking down and mutating into forms where poverty and vulnerability is considerably heightened. And third, the nature of these processes, their form and consequences, are different in each context.

Food security has conventionally focused on how households access the food and income they require for survival on an inter- and intra-seasonal basis. If nutritional security is entertained additionally as a concept, then food security relates to the pathway of food into the household, and nutritional security to the nutritional outcomes, once the internal factors related to storage, preparation, distribution, health and mother care, have been taken into account.

An holistic livelihoods model is vital to the analysis of food security, I would contend, for the principal reason that an understanding of what is happening to a household's and individual's assets is critical to knowing what is happening to the status of food security. An appropriate approach also needs to seek to understand social and economic trends in a dynamic way, and in this regard adding a rights dimension to a livelihoods framework is helpful, for instance, to understand the new forms of discrimination and vulnerability that are arising as a result of the HIV/AIDS pandemic.

There are two models that I shall refer to in this paper. The first is a reformulation of CARE's earlier household livelihood security model as a livelihood rights model, and the second is a dynamic variant of this that helps provide a vehicle for analysis. As illustration of the uses of these frameworks, I shall make reference to direct and indirect instances of their actual utilisation.

Cluster Analysis as an Insight to Understanding Livelihoods and Social Change

Within the context of the livelihoods model, a form of analysis that I have found particularly helpful to understanding the dynamics of social change has been defined as cluster analysis, for want of a better term. This analytical method was developed initially during participatory appraisals of farming systems in Zambia during the early 1990s, and then subsequently used in a study to analyse the impact of HIV/AIDS on farming systems during the same period.

The concept of cluster was intended to overcome the deficiencies of social analyses which focused either on the household or the individual as the unit of analysis. Typically the former, especially in its questionnaire variants, was seen - and remains - often weak in its more complete analysis of inter- and intra-household relations, whilst a focus solely on the individual is over-reductionist. A good livelihoods analysis should generate an understanding of context, social differentiation, and social disaggregation (gender, generational and other diversity differences), and the technique of cluster analysis assists this. Since single households are usually connected to others in overlapping relationships, the term cluster was coined to make sense of these. Its initial definition was: 'A group of producers between which there are multiple resource exchanges, usually based on the factors of kinship, labour and good exchange, and or common access to draught power' (Drinkwater 1992). Critically, these resource exchanges do not entail direct, reciprocal exchanges, such as characterise exchanges between households and individuals not of the same cluster. Adding a rights model, however, adds the poignant point that some of the worst and most systematic instances of rights abuses occur within these familial relationships. There is a great distinction between supportive relationships, such as of grandparents to grandchildren, for instance when the latter are orphaned, and of abusive relationships, such as when a husband's siblings take all property from a wife on his death and send her and her (female) children impoverished back to her village of origin.

Two studies which used this methodology to analyse livelihoods and the impact of HIV/AIDS were respectively the Zambian study of 1992 (Drinkwater et al 1992), and a subsequent exercise in Zimbabwe in 1999 (Westley ed, 1999). This latter study was less specifically focused on understanding the impact of HIV/AIDS, although this became a key finding of the exercise. Another study which also used a livelihoods framework, though less explicitly the technique of cluster analysis, to analyse the impact of HIV.AIDS on rural livelihoods and farming systems in Malawi, was conducted more recently in 2001 (Shah et al, 2002). All three studies shall be referred to in a comparative manner in this paper.

CARE's livelihoods model has always been built around the standard Chambers and Conway (1992) definition built around the capabilities, assets and activities required for a means of living. The most recent variation of this model (Fig 1) shifts the way livelihood outcomes are defined from the relative achievement of basic needs to the relative fulfilment of the human rights, or freedoms, fundamental to a life with dignity. Since the analyses drawn upon in the paper did not utilise this variant of the livelihoods model, some implications of this shift will be discussed subsequently.

Understanding the effects of HIV/AIDS on farming systems and social change

There are many causal factors at work in the shift that is occurring in rural livelihoods in Southern Africa. The nature, extent and rate of this shift varies too between contexts. One primary cause is the impact on livelihoods of the removal of state funded input and output marketing institutions, as a result of structural adjustment policies. In Zimbabwe, the erosion of democracy and the destruction of the country's economy have caused untold suffering. In Malawi, a high density, yet largely agrarian society, with few generators of economic growth, is imploding on itself as the capability of agricultural to support this population deteriorates. Zambia, like Malawi, suffers from the poor connectivity of its rural areas, but with larger land horizons and urban centres, the agrarian nature of Zambia's rural economy is less threatened than Malawi's, except through the largely ignored depredations of periodic livestock epidemics on rural household asset bases. And across all countries, though to differing degrees, and to a far greater extent in Zimbabwe and Malawi, HIV/AIDS, has accelerated processes of rural impoverishment and the breakdown of extended family relations, that have also been the foundation of traditional safety net mechanisms.

Fig 1: Livelihood Rights Model

In the livelihood analyses conducted in Zambia in 1992 and Zimbabwe in 1999, perhaps the most significant difference shown up by the cluster analysis technique with respect to the impact of HIV/AIDS on rural livelihoods, is the impact of illness and death amongst urban residents on rural households.

The Zambian study drawn upon was conducted in the Mpongwe area of Copperbelt Province, about 60 km south of the town of Luanshya. At the time, the area was defined as at an early impact stage, in that the majority of deaths that had occurred in the area from HIV/AIDS, were of people who had contracted the disease in urban areas, and with the failure of formal medicine, had, in effect, returned to rural areas to die. There were 34 cluster case studies in the study, comprising about 150 'households' in all, and in these there were 9 deaths recorded from HIV/AIDS, as well as undoubted others of young children. But the study showed that HIV/AIDS transmission was now occurring within the local area, and as such was beginning to affect household heads for the first time. Amongst the 34 case studies, five had primary producers (the household most responsible for food security) whose production had decreased in 1992/93, despite it being a good season following drought. In all five cases, the reason for decreased production was ill health, and all HIV/AIDS related (Drinkwater 1993). It was predicted that the continuation of this trend would be even more devastating on livelihoods, because the loss of household heads, following the long depletion of their productive assets, would result in the impoverishment of the remaining family members, and the likely breakdown of the family structure (Box 1). This is a trend, though inadequately researched, which now appears to be having a growing impact in Malawi.

Box 1: Mpongwe, Zambia, Case Study.

Maybin Malaya worked for 27 years at Mpongwe Mission Hospital as a driver. He retired in 1990 and became a successful farmer. He had oxen, planted 4 ha and produced 170 bags. In 1991/92, during the drought, he planted only 1.6 ha and produced 40 bags. The year of the study, 1992/93, with good rains, he planted 2 ha and still only produced 40 bags. His cattle, except for three cows, had now also died through disease. Maybin's first wife died of AIDS in 1983, following which he remarried. He himself now complained of coughing, chest pains, headache, fever and diarrhoea. From his second marriage he had had three children. Two had died, the third was very sick. His wife suffered from abdominal pains and was frequently attacked by absecces; she also appeared to have AIDS. Maybin had 6 children still living from his first marriage. Since March 1992 they had been living with their maternal grandmother and it seemed that she was now responsible for them. (Drinkwater 1993)

The Zimbabwe study of 1999, in contrast to the earlier Zambian study, showed the far greater importance of urban remittances to rural livelihoods. In this study, undertaken in a rural area a short distance away from the south-eastern Midlands town of Zvishavane, by definition better off households were supported by urban incomes. Typically, these remittances funded items such as agricultural inputs and school fees, and were critical in the maintenance of production and consumption levels across an extended family. The impact, therefore, of the remitter's death, since these were the people most likely to be first affected by HIV/AIDS, was devastating. (Westley 1999). Rural families lost their income support and gained orphans instead, as is illustrated in Figure 2.

Fig 2: Cluster Case Study, Zvishavane, Zimbabwe

Source: Drinkwater, 1999, Field Notes

When the Malawi study was conducted in late 2001, it drew from the methodology used nearly a decade earlier in Zambia. Both studies used combinations of qualitative and quantitative information, although there were distinctions in the methods. The Zambia study used a participatory appraisal methodology, with an introductory meeting, which developed a health calendar and social map of households, followed by the purposively selected cluster case studies. For these interviews, the livelihood security version of the framework shown in Fig.1 was used to develop a checklist, used more or less flexibly, but which allowed consistency in the quantitative information collected. In a series of separate focus group discussions with men and women issues arising from the case studies were explored further. In the Malawi study, focus groups at the village level were followed by semi-structured interviews with a random sample of 310 households.

The central finding of the Malawi study was the high number of households affected by chronic sickness and death, with the proportion varying between 22% and 64% per village. It can be assumed that if a cluster analysis had been utilised it would have found that virtually every cluster would have experienced some impact. The study looked at the impact of distance from trading centres on the rates of chronic illness and found no correlation; what it did find in terms of a positive correlation was the proportion of households affected and their mobility patterns and livelihood strategies. Where more people spent time outside their village for economic activities like petty trading and vending, there were higher rates of illness (Shah et al, 2002).

Ten years on from the Zambian study of an early impact community, the Malawi study of late 2001 is showing the effects of the full impact of HIV/AIDS. These effects are both social and economic, a challenge for assessments since it is far simpler to produce data on economic production than on the more complex processes of social fragmentation and human rights failure. The Malawi study has information on both the economic and social aspects. With respect to the household economy, the study concluded that the most immediate impact of chronic illness is loss of labour, something that over 70% of households so affected had experienced. This loss led to other problems such as delayed agricultural operations, affecting nearly half of the households that had experience chronic illness, changes in crop mix (26%), leaving land fallow (23%) and changes in source of livelihood (36%). The net result, as shown also in the Zambian study, is decreased agricultural productivity, even if chronic illness is not the only cause of this (Shah et al, 2002).

If these economic impacts of HIV/AIDS can be reasonably well predicted, it is the harder to analyse social consequences of the AIDS pandemic that depict its full impact on human lives. One of the most interesting findings of the Malawi study was of the frequency that changes in marriage partners was occurring, and thus the fluid and fragile identity of the household itself. The composition and location of households was subject to frequent changes, with in particular the status of women and children in the patrilocal system becoming increasingly insecure (Shah et al, 2002). This pheonomenon is not restricted to rural Malawi. A decade earlier, it is remarked in the Zambia study that, 'the institution of marriage is clearly in disarray'. In that study, one woman who had children from three men, and who received no assistance from her former husband, remarked that 'a husband is just a short-term relative' (Drinkwater 1993).

There are two major implications of this. First, is the growing abuse of the rights of women and children, and the second is the heightened extent of their vulnerability and poverty. On a recent field visit in Malawi (August 2003), a woman project manager took one look at the demeanour of a woman farmer showing us her dimba garden and asked her what had happened to her. She had been chased off a purchased farm in a resettlement area a year earlier by her husband's relatives, when he had died, being accused of being responsible for his death. Higher rates of death lead to increased accusations and preoccupation with witchcraft. Against custom, she had not been invited to a recent ceremony to unveil his tombstone. Instances of this kind of story abound. Again, the Malawi study provided substantial illustration, through the use of timelines, of orphaned children being repeatedly displaced and growing up with little love and attention, predominantly being exploited for their labour. A recent newspaper article in South Africa ran a story of adolescent orphan girls in northern Malawi being sold as 'sex slaves'.

The issue of labour exploitation is a particular facet of the Malawi, and no doubt other country experience, of the result of growing rural poverty and social fragmentation and dislocation. In a discussion with the team running a CARE public works program in Malawi, the team remarking on the challenges of targeting, stated that the wage labour rate on the safety net program was explicitly set above the local ganyu, or piece work labour rate, since this had become so exploitative (Devereaux 2002). Exacerbated again by the growing supply of orphans, widows, and other near destitute households, rural labour supply continues to grow, whilst demand, with asset decline, is diminishing. Of the poor households interviewed in the Malawi study, 55% were depending on ganyu labour for more than four months of the year. 'Once the household starts to depend on ganyu, it becomes very difficult for it to come out of it.' Those looking for ganyu labour take what opportunities they can, and since these are most commonly at the peak points of the agricultural cycle, their own crops are neglected and the yields shrink further. This 'downward spiral' of food insecurity means they end up consuming directly whatever small harvest they have, and have no food at all for storage (Shah et al, 2002).

This kind of trend is indicative of the processes of deagrarianisation and depeasantisation, to which Bryceson (2003) refers. These trends are highly pertinent to analytical work on HIV/AIDS in the Southern Africa region, and issues of food and livelihood security. 'Deagrarianisation' is 'a process of occupational adjustment, income-earning reorientation, social identification and spatial relocation of rural dwellers away from strictly agricultural-based modes of livelihood' (Bryceson 1996). More specifically, and critically, in the Southern African context, 'depeasantisation' is a variant of deagrarianisation, 'whereby peasantries' economic capacity and social coherence are being progressively undermined. They literally unravel as communities' (Bryceson 2003).

This social unravelling, it can be conjectured, is proceeding apace in rural Malawi, and I would imagine, in rural Zimbabwe. Whilst it is likely to be happening less so in rural Zambia, with its greater land horizons and still more agrarian nature of rural livelihoods, and similarly in rural Mozambique, the decade old Zambian study undertaken in the vicinity of the Copperbelt towns, nevertheless show that there are similar trends.

Implications of Depeasantisation for Rural Food and Livelihood Rights

The effects of HIV/AIDS on rural livelihoods, once the epidemic has reached full impact stage, is immense. Impact is experienced both in terms of the deterioration of household economies, and then in the unravelling of the social fabric of the lives of those affected or afflicted by AIDS. As illustrated in the Zimbabwean participatory livelihood analysis of 1999, it may be an urban death and the subsequent double blow of the loss of a remittance coupled with the gain of a new set of orphans, that sets off the chain of decline. In Lesotho too, the illness and death of a remitting male head is often responsible for initiating this process, whereas in Zambia and Malawi, there is not the same reliance on remittances and hence connection between urban illness and rural impact.

There is more complete documentation, thus far, of the economic impact of HIV/AIDS, especially with respect to agricultural production. 'The most immediate impact (of chronic illness) being the loss of labour, which, depending on the timing and duration of sickness, results in delayed agricultural production, land being left fallow, changing crop mix, a change in livelihood sources, an increased dependence on ganyu, and ultimately, decreased agricultural production and livelihood status' (Shah et al, 2002). This same kind of impact has been documented for countries like Zambia and Uganda. Farming systems change might well be a result too, with more intensive cropping systems being replaced by ones that are less intensive, productive and nutritionally diverse. An example of this is the replacement of diverse and complex systems of banana, legume, maize and vegetable cultivation in areas north of Lake Victoria in Uganda, by cassava and sweet potatoes, because of the lower labour requirements (Tumushabe, Bantebya and Ssebuliba, 1994). The extent of impact depends on how HIV infection is taking place and the economic roles of those affected. What the method of cluster analysis shows clearly is that the illness or death of one key producer can lead to the deterioration of the livelihoods of four or five households - thus if only 20% of households in an area are affected, it can be expected that the locality will be experiencing the full impact of the epidemic.

Fig 3: Livelihood Security Components from a Human Rights Perspective

If a clearer picture is emerging, and can be shown by the kinds of studies discussed here, of the economic impacts of HIV/AIDS, the process of social unravelling that takes place alongside, is often more opaque. In particular, what is least explored, is the impact on human rights, and the increasing levels of rights abuses and failures. This is one reason why a livelihoods analytical model should pay a greater attention to human rights issues and status, as is done in the models shown in Figs 1 and 3. In Fig 3, a livelihoods model developed by CARE UK is shown in a dynamic way, with a particular emphasis on the 'barriers to access', many of which may take place in the form of various types of discrimination (Sanderson 2003, personal communication).

Most critically of all, and as yet still far too ignored, the issues which are exacerbated by HIV/AIDS are first and foremost about gender, even in terms of the situations of orphans. If there are two factors which have had an overwhelming contribution to the spread of the HIV/AIDS pandemic in Africa, they are poverty and male sexual power. The latter is grounded in the fact that males are rarely held accountable for the systematic abuse of female human rights which occurs across the length and breadth of the African continent. In an urban livelihoods study I facilitated in the mixed Muslim and traditional cultures of Zinder, Eastern Niger, in February 2003, gender inequality was identified as the root cause of food and livelihood insecurity, often manifested in the casualness and callousness with which the verbal divorce practice of repudiation is carried out. Here in Southern Africa, CARE Lesotho/ South Africa are seeing gender inequity as the core underlying rights theme, which denies women access to all sorts of fundamental entitlements - land, pensions, their children, village citizenship, their own property - if their husband dies, or even divorces her (Abbot 2003, personal communication).

Box 2: Ugliness

When travelling to Mpongwe by landrover (in September 1993), we stopped to pick up a woman about 20 km before Mpongwe. She had a pile of pots wrapped in a chitenje cloth, some clothes in a basket, half a bag of unshelled groundnuts, 20kg of maize and three young children. Why was she leaving, we asked. Because she had become ugly, she replied. Her husband had just taken in a younger woman from a neighbouring village and kicked her out. With only the possessions she was carrying now, she was returning to her mother's village the other side of Mpongwe (Drinkwater 1993).

The story in Box 2 of the Zambian woman who became ugly is repeated again and again across the continent; women who with their children are ejected from their former homes, stripped of all their assets, including most especially their human dignity, and returned as paupers and outcasts to their villages of origin. They are afforded no protection, save that their parents can muster, and have no recourse. Their rights and those of their children are not protected. Those that abuse their rights in the most devastating of ways are not accountable.

This, most starkly, is the implication of depeasantisation, especially in the context of its acceleration through the AIDS pandemic. Thus far though, the economic consequences of gender inequality are clearer. As Topouzis notes in a paper that summarises work across much of sub-Saharan Africa: 'It appears that when a household has been affected by male adult mortality, surviving widows and their families often have few, if any, assets to dispose of in their time of need. Thus, food security coping strategies may disintegrate quite soon after male adult death and food consumption may decline sharply' (1998). But that the fundamental human rights - and food security issue - certainly in the context of Southern Africa and the impact of HIV/AIDS, is gender inequality, remains much less well understood and is still paid too little analytical attention - even if it has been signalled for attention by such high profile visitors to the region in 2003 as Kofi Annan, Steven Lewis, and James Morris.

If the Challenge is, 'Can Agrarian Livelihood Collapse be Overcome?', do we have the Right Analytical Tools?

A recent literature review analysis of livelihood trends in Malawi over the last 20 years showed that the cost of maintaining rural households post structural adjustment could well rival the cost of earlier marketing subsidies, which at least ensured higher levels of rural food security (Tango International 2003). Moreover, the present form of subsidies - food handouts, input handouts, and largely non-asset building public works safety net schemes - are all hugely dependency creating. Thus, the viability of rural livelihoods in Malawi has declined to the level where they are being sustained only by donor handouts. The rural maize production peaks of the 1999 and 2000 harvest, coincided with the peak of the starter pack scheme which delivered, ostensibly, seed and fertilizer starter packs to virtually every household in the countryside. In the following two seasons the scheme was cut back, and these were the food security crisis years from which the country is still emerging. In the circumstances of growing chronic food insecurity in rural Malawi, it is certainly arguable that cutbacks in the starter pack scheme played as significant, if not more so, role in the crisis than the vagaries of rainfall.

That this is now the situation adds some urgency to the search for pathways out of the current crisis of chronic food insecurity. This search requires to be based on analyses of livelihoods that are more profound than most of the assessments that have accompanied the recent crisis, and requires strategies that are somewhat less blunt than the present response approaches by all agencies, whether UN, NGOs, or donors, alike, and whether described as emergency or developmental relief interventions. Saving lives, as the emergency activities undoubtedly have, is one thing. Rebuilding livelihoods, especially in a context where it is not clear what this entails, is something very different.

At the onset of the emergency response to the crisis, I was struck by one comment made at a regional VAC meeting in June 2002, that defined the role of assessments as determining the deviation of food security from the 'normal'. This has been a constant refrain during the course of the 'emergency' interventions of the last year. Yet, having worked in Southern Africa for the past quarter century, and researched records of climate and peasant livelihoods over the century before that, one fact that is clear is that there is no such thing as a normal season, and no such phenomenon as a normal food security status.

In a literature review of livelihood trends conducted for Zambia, parallel to the Malawi study, McEwan (2003), cautions about over-exaggerating the impact of HIV/AIDS on rural livelihoods, since what figures there are suggest only about 11% of rural households, on average, are affected. Whilst undoubtedly an area such as that of Mpongwe, studied in 1993, and adjacent to the mining towns of the Copperbelt, would be more highly affected than other areas, what is shown by the cluster analytical method is that the death, chronic illness, or lost labour of one person, particularly if key to the economy of the cluster as a whole, will impact not one but several households.

This presents a significant analytical challenge for the future, since, in the meantime, we still retain the household as the unit of all quantitative and most qualitative surveys, even as the social phenomenon that is the household is mutating and at its most fragmented and fragile state in African history. So the questions are, how do we propose more adequately to understand and deal with levels of vulnerability and human rights failures that are at their highest historical levels?

These questions matter, since it is clear that in this situation, blind interventions, whilst in the short term saving life, can beyond the immediate food shortages of the past two years, be hugely inhibiting of the search for the kind of pathways that will facilitate the rebuilding of lives and human dignity.

In this vein, for example, gender empowerment activities, such as the capacity building and savings and loan, public works models that CARE Malawi is promoting. Women in the contract association groups as part of this program, were shown to be able to cope more effectively with the loss of labour of their husband, if he was chronically ill or had died, and to be recorded more respect by men, because of their improved income earning capacities. This changed status resulted in changes in roles and behaviour of men and women, with the result that both were contributing more fully to productive and reproductive activities. Female headed households were rendered less vulnerable through their ability to continue to earn and income after graduating from the program. 'They said it was better to do this rather than look for a man to remarry or engage in sexual prostitution for a livelihood' (Chalimba and Pinder 2002). This pithy statement succinctly sums up the argument for the need for new pathways, and for processes that will lead to real debates in communities about gender roles and social relations, and ways of shifting these to record more respect, motivation and benefit for all men, women and children.


Bryceson, D.F. 1996, 'De-agrarianisation and rural employment in Sub-Saharan Africa: A sectoral perspective', World Development, vol.24, no.1, 97-111.

Bryceson, D.F. 2003, 'Sustainable rural livelihoods in Sub-Saharan Africa: Sustaining what and for how long', Afika-studiecentrum, Leiden.

Chalimba, M. and Pinder, C. 2002, 'Gender power relations study and the impact of safety nets in Malawi', Vol 1, CARE International in Malawi.

Devereaux, S. 2002, 'State of disaster: Causes, consequences and policy lessons from Malawi', Action Aid, London.

Drinkwater, M. 1992, 'Knowledge, consciousness and prejudice: Adaptive agricultural research in Zambia', in Scoones, I. and Thompson, J. 1994, Beyond Farmer First, London, Intermediate Technology Publications.

Drinkwater, M. 1993, 'The effects of HIV/AIDS on agricultural production systems in Zambia', FAO, Rome.

McEwan, M. 2003, 'Changing landscapes and the outliers: Micro and macro factors influencing livelihood trends in Zambia over the last 30 years', CARE International, Johannesburg.

Shah, M K, Osborne, N., Mbilizi, T. and Vilili, G. 2002, 'Impact of HIV/AIDS on agricultural productivity and rural livelihoods in the Central Region of Malawi', CARE International in Malawi.

Tango International, 2003, 'Livelihood erosion through time: Micro and macro factors that influenced livelihood trends in Malawi over the last 30 years', CARE International, Johannesburg.

Topouzis, D. 1998, 'The implications of HIV/AID for rural development policy and programming: Focus on Sub-Saharan Africa', FAO and UNDP.

Tumushabe, J., Bantebya, G. and Ssebuliba, R. 1993, 'The impact of HIV/AIDS in agricultural production systems in Uganda', in Barnett, T. 1994, The Effects of HIV/AIDS on Farming Systems and Rural Livelihoods in Uganda, Tanzania and Zambia, FAO, Rome.

Westley, K. ed. 1999, 'Household livelihood assessment in Midlands and Masvingo Provinces', CARE International in Zimbabwe.