The HIV virus is particularly comfortable where wealth and poverty co-exist – it thrives on inequality (Msimang, 2003:109).
It is one of the paradoxes of the seemingly inexorable development of the AIDS epidemic in Africa, that whereas patterns of prevalence and impact reflect a range of social inequalities (as health and disease do anywhere), HIV/AIDS is also a great leveller. One of the reasons for this is the universal fact that no cure has yet been found, so all are vulnerable; but additionally in this case, Africa's abysmal level of development has for long left the majority of its people, whether rich or poor, without access to the life-prolonging drugs which have turned AIDS in developed countries into a manageable problem. Although this is now beginning to change – most particularly in South Africa where a struggle against the state and big pharmaceuti-cal companies has led to the importation of cheaper generics and a promise of treatment for all; and in some other African countries where foreign aid is supporting limited programmes of anti-retroviral distribution – the long period of incubation of HIV means that those presently dying from the disease reflect the situation as it prevailed up to a decade ago, with wealth and social position offering no prophylactic. The rich and famous are infected along with the poor and nameless.1
The levelling effect of AIDS has led many commentators to ignore the way in which ‘class’ inequalities are reflected in this crisis. The focus has been on other social divisions to which AIDS also speaks – most notably gender inequities, but also those of generation or location (rural compared with urban). If we want to understand or intervene more effectively in the AIDS crisis, then class needs to enter the analytical frame and it matters how ‘class’ is defined. I argue here that as much consideration needs to be given to a relational as an indexical definition (elaborated below). I consider what can be drawn out of both indexical and relational accounts of class in Africa, drawing particularly on the Tanzanian literature and on research carried out in Tanzania between 1995 and 2000.2
The seduction of statistics
The statistical record on AIDS in Africa is huge, albeit often based on estimated figures drawing on limited data accumulated in resource-starved conditions. Such data is collected by African governments and/or by international agencies, including private foundations, and was initially framed within a predominantly biomedical discourse. Amongst the reams of biomedical statistics are to be found occasional correlations of HIV prevalence or AIDS deaths with socio-economic indices.
When epidemiologists speak of ‘socio-economic differences’ it is in terms of quantifiable indicators – these encompass gender, generation and race but also ‘class’ indicators such as income, occupation, education, literacy or quality of life.3 Basing their accounts on survey data, they rest their analyses on the positivistic assumption that ‘the numbers speak for themselves’. There is little or no effort expended on the ‘why’ questions and they rarely follow up their surveys with qualitative work. On the whole, they cross-tabulate single indices with HIV or AIDS rather than exploring the links between them.
It is worth asking why such data regarding HIV/AIDS is included in studies which are anchored in medical rather than sociological discourse. Ainsworth and Semali, in a paper on Tanzania (1998), suggest that consideration of ‘socio-economic correlates’ has a threefold purpose: better targeting and effectiveness of preventive efforts; assessing the economic impact (on production and on the costs of treatment); and devising better coping strategies. Whilst using the same language of ‘measures’, the National Institute of Mental Health (2004) in the US justifies the linkage of socio-economic indices with health (and particularly HIV/AIDS) in slightly more sociological language. They note that,
choice of SES [socio-economic status] measure should depend on how researcher believes SES is linked to health damaging exposures, health protective resources and ultimately to health – e.g. is it exploitation, few tangible resources, lack of prestige or a combination that causes poor health? (2004:1).
Whilst indexical accounts can be shown to have limited explanatory value, their pervasiveness has given them a taken-for-granted aura in both the literature and policy formulation process around HIV/AIDS, and it is worth looking more closely at their inadequacies
Indexical Accounts of ‘Class’ & AIDS
Poverty as Explanation: a Vulgar Version of Class Analysis
For many of those who cite socio-economic variables as a factor in the transmission of HIV, ‘poverty’ seems to substitute for class. Not only is ‘poverty’ wheeled out by many international organisations as the context for the AIDS epidemic in Africa, but it may also be cited as an explanation:
The HIV epidemic has its origins in African poverty and until poverty is reduced there will be little progress in reducing transmission of the virus or an enhanced capacity to cope with its socio-economic consequences (Cohen, 2004:5).
… it is difficult to tell whether poverty causes AIDS or vice versa – or whether another variable, such as war, inadequate health and poor education, explains the relationship (Bloom et al. 2001 cited in WHO, 2002).
We need to ask what it is about impoverishment (i.e poverty as a process) that leads people into risky sexual behaviour, if that is the conclusion we wish to draw. And if we wish to understand how to change this situation, it may be necessary to ask what causes the impoverishment in the first place. The conceptual language of ‘poverty’ is inadequate to this purpose, as are static indicators of its significance.
Education as a Proxy for Class
In addition to poverty, the other variable which is often cited in epidemiological studies of HIV/AIDS is education. Ainsworth and Semali (1998) in a much cited work refer to several Tanzanian studies which use level of education as proxy for ‘socio-economic status’. This is a not an unreasonable equation, given that in Africa the well-educated still enjoy some prestige and it is mostly the better-off who can sustain children through to higher levels of qualification which are then the passport to well-paid jobs.
However, the reason why epidemiologists pursue such correlations is not to link education to class level but usually to imply that lack of formal education can be equated with ignorance, in particular of the causes and means of protection against the HIV virus and thus ‘the poor and uneducated have higher infection rates’ (Ainsworth & Semali, 1998:2). It is then often concluded that education is the solution to the problem and that the uneducated should be targeted. I have suggested elsewhere that this reasoning is doubtful, given the ways in which knowledge of AIDS has been spread through word of mouth and the media rather than through formal teaching, and also given the limited impact of formal educational programmes on the capacity of the most vulnerable to protect themselves (Bujra, 2000b).
Ainsworth and Semali's reason for raising this point is to challenge it in another way: ‘at least in the first decade of the HIV/AIDS epidemic in Africa, HIV did not follow this pattern’ (Ainsworth & Semali, 1998:2) – the better-off and more educated were actually more susceptible than the poor and uneducated.
The Changing ‘Class’ Character of the AIDS Phenomenon?
From the literature, it would appear that there has been a global shift in terms of the socio-economic patterning of HIV/AIDS but with Africa trapped for longer in an earlier mode. Bloom puts the case more generally: ‘it seems plausible that the rich are more at risk in the early stages of an epidemic … the poor at an increasing risk as the epidemic progresses’ (Bloom et al. 2001, cited in Garbus, 2003:33). A decade earlier Berer was citing statistics which seemed to support the same argument:
In 1985 some 70 per cent of Brazilians with AIDS had university degrees. By 1989 that figure had dropped proportionately to thirty per cent. In Zaire, physicians and manual workers seen at a Kinshasa hospital were equally likely to be HIV-positive in 1984. In 1986 the prevalence among the manual workers had doubled, while the level among the physicians had remained constant (Berer, 1993:48).
‘Class’ here is being understood as socio-economic status, with the evidence based on indicators such as education, income and ‘wealth’. The reference is to a set of status categories which are being held to coincide to produce ‘class’ groupings. However, the assertion that the patterns are changing suggests that these are more than analytical categories. Some awareness, behavioural change or organisational effort must have occurred to explain such dramatic patterns of change. Since there is a tendency for medical models of change to focus on individualised enlightenment and transformation (whose aggregation is then measured), this more collective shift is rarely considered.
To approach this question in a relational way we need to address the logic of linking high socio-economic status with risky sexual behaviour. In the literature it is sometimes related to the greater mobility and social freedom of the better-off and to their higher levels of ‘disposable income’ (Bloom et al. 2001). But having more ‘non-regular sexual partners’ is also statistically linked to higher educational levels in Africa (WHO study cited by Ainsworth and Semali) and affluence and education are generally linked in such studies. This sits uneasily with the assumption that over time the more educated learn better avoidance techniques, cutting down the number of partners and using condoms more consistently (Ainsworth and Semali, 1998:2; Garbus, 2004:42); that mobility is maintained, but its costs in terms of sexual behaviour are diminished through the power which knowledge brings. It is notable that despite considerable evidence of the higher prevalence rates amongst the more affluent and educated in Africa, there is never a suggestion that they be targeted to change their ways (compare sex workers or ‘youth’ who are prime targets for such action). And the evidence that those of high socio-economic status are modifying their behaviour is less than convincing, as I suggest below.
Relational Views of Class & the Question of AIDS
Sociological accounts of class are often grounded in indexical data, but they generally interrogate such data to address questions about the relationships amongst class-defined groupings. It is worth returning to classical definitions of class to emphasise this point. For Weber class is defined by market positionality, by wealth and income and by social prestige and power – but these indices produce more than abstract social categories; they create groups which are aware of their position in society and may act to transform it – as for example in the emergence of capitalist society. For Marx (to take the other distinctive approach to class analysis), the concept is held to relate directly to the relations of production in society but more particularly to processes of exploitation through economic power. The Marxist distinction between ‘class in itself’ and the more consciously aware and organised ‘class for itself’ is parallel to Weber's view of classes as more than simply statistical categories.
If Marx and Weber shared this sociological starting point, the context for their analyses was also the same – i.e. capitalist societies (though both had a general model that was applied to earlier or other societies as well). They diverge in many other respects. Weberian accounts of class (filtered through Durkheim and structural functionalism) fed more heavily into contemporary sociology and by extension into the epidemiological understandings already noted. A vocabulary of class emerged which was simplified into upper, middle and working classes and was focussed on status distinctions more than on the structure of capitalist society. Marxist analysis, with its emphasis on economic structure and political action and its designations of proletariat, petty bourgeoisie and bourgeoisie, was eventually marginalised. In either case, there is considerable debate around the relevance of capitalist class categories or ‘Western’ occupational labels to the African setting with its low levels of industrialisation and history of recent colonialism (see e.g. Copans, 1985; Lubeck, 1987; Bujra, 2000a).
Despite these reservations, I want to argue that a sociological and relational definition of ‘class’ is vital to understanding both the transmission of and response to HIV/AIDS in Africa and that this has little to do with class as socio-economic status and more to do with class as economic and social power. The limited extension of capitalist relations of production and property in Africa and its servitor role in global capitalist accumulation have not produced societies with a large wage-labouring class directly exploited by a bourgeoisie whose power has been consolidated. We are witnessing processes of class formation in a context more appropriately described as primitive capital accumulation. Peasant/subsistence cultivators, though still the majority, are subject to increasing and brutal pressure from encroaching capitalist farmers and multinational agrocapital (see Bryceson, 2004). A small industrial/commercial/financial sector, dominated by multina-tional capital, has called into being a minority wage-labouring class who are usually migrants from rural areas. There is the consolidation of a ‘political class’ which controls the state through often parasitic forms of accumulation such as land grabbing and corruption, and which is served by an artificially inflated category of state functionaries. Other elements such as salaried or independent professionals, the growing army of domestic servants or the burgeoning petty commodity producers of the urban informal sector, are dependent upon other class fractions for their survival. Conceptualising class difference in this way derives from a Marxist perspective, though it is worth emphasising that Marxist class analysis is not straightforwardly read off from occupational indicators. Class groupings are discerned by reference to their position in relation to production and their alignment in social and political struggles.
Indexical accounts often collapse the differences between petty commodity producers and wage labourers into terms like ‘the urban poor’, or conflate peasant with capitalist farmers, or fail to distinguish between professional wageworkers and ‘businessmen’, collapsing both into a category of ‘the middle class’. When it comes to analysis of the AIDS epidemic, it is these latter indexical categories that are more commonly used, thus disguising exploitative relations between classes and conflicts between class fractions. Whilst there is recognition of the contrast between low rural and high urban levels of HIV prevalence in Africa, the question of why and how this has happened is not considered. Primitive capital accumulation in land and partial processes of proletarianisation, with consequent separation of families through labour migration and the emergence of commercialised sex – all processes relating to class formation as well as the transformation of gender relations – now help to reproduce current patterns of HIV prevalence. The recording of high prevalence in mining areas and along major transport routes indicates the way HIV transmission reflects the penetration of market and capitalist relations. Even the flowering of petty entrepreneurialism links the emergence of class relations with HIV transmission as is noted in one trader's account of fishing camps around Lake Mweru in Zambia:
People come here from everywhere [to buy salt fish for trading] – Zimbabwe, South Africa, Nigeria, Malawi, Congo. When they meet, well you know what nature does. Wherever there are people, there are temporary marriages. Men cannot stay without their wives for very long and women may be desperate, so they sell themselves. HIV starts here and moves on (The Guardian, 7 January 2006).
Class formation and the progression of AIDS are linked in Obbo's ground-breaking work on Uganda, in which she traced the changing patterns of sexual networking in one of the most affected areas (Rakai) linking them to AIDS deaths (Obbo, 1993). In the early days, sexual networking was expanded in tandem with the explosive growth of trade and smuggling between Uganda, Tanzania and Kenya, with traders and their wives (especially in polygamous marriages) and sexual partners the first to be struck down. A second wave also affected the more affluent and educated: businessmen, bankers, civil servants – ‘people with money and power to put them in risky situations’ (Ibid. p. 951). It is clear that the initiators were mainly men, and that the women with whom they shared such situations were generally of a different class – sex workers, often from impoverished rural backgrounds, on whom men expended their ‘disposable income’. These relationships can reasonably be described as class relations, with considerable (purchasing) power on the part of male customers, the power to demand sex without condoms and to keep the price low (this is not to say that sex workers do not themselves have some leverage). As the death toll began to rise, behavioural changes were noted: ‘the tattered appearance of “sex workers“ at bar lodges in Masaka and Rakai towns suggested that many men were avoiding risky behaviours’ (Ibid. p. 952).
However another kind of sexual networking persisted, in what Obbo describes as ‘the [small] world of elites’. Men and women, who had usually been at school or University together, or who shared work locations or membership of the same religious or social associations had intermarried (the social reproduction of emergent classes). At the same time some were also engaged in intra-class liaisons, affairs with ‘people who were known and presumed safe’ (Ibid. p. 952). Deaths within this known group were now also inducing fear and ‘men increasingly stayed at home’ (Ibid. p. 953).
From what was then Zaire (now DR Congo), Schoepf offers case studies of women in both the sex-worker and the elite liaison situation, each experiencing the threat of HIV infection (1993). Looking at sexual networking from a different angle she also notes that:
AIDS is not only a disease of poor women engaged in sex work or long-distance lorry drivers. Men and women in these occupations are at very high risk … However, they do not form bonded groups. Their social networks extend into virtually every social milieu and along the trade routes to all but the most remote villages (Schoepf, 1993:1).
Besides transmission we also need to look at class-defined response to the epidemic. Following the contraction of state welfare systems in the wake of structural adjustment (see Poku, 2004), it appeared for a time as if the role of the state in strategies for prevention and treatment would be eclipsed by an influx of non-governmental organisations working in the AIDS field, generally funded by international agencies. In the short run such organisations became, amongst other things, a site for the reproduction and expansion of a class fraction of professional wage labourers (Bujra and Mokake, 2000). The levelling effect of AIDS (with heterosexual intercourse as the key mode of transmission and neither rich nor poor initially able to prevent the inexorable progress of the disease) has not led to a classless solidarity amongst all those affected. Despite the denial of life-prolonging drugs to nearly all, the better-off marked their exclusivity through better care and nutrition and had a greater facility to protect themselves. Some of the better-off afflicted did reach out to those in rags, but philanthropy is double edged – it both obliterates class distinction whilst it marks class boundaries.
The Tanzanian Case
The Epidemiological Account of Socio-economic Indicators for HIV/AIDS
In Tanzania, as elsewhere, research on AIDS has been framed by a predominantly medical discourse, with concern for the social determinants of the disease playing a minor role and such determinants presented in indexical mode. However, early local case studies, based on survey data, offer occasional citations of socio-economic indicators which confirm patterns already noted. Soderberg et al. for example (1994), writing of a very large (4,840) sample of blood donors in south-west Tanzania (Ilembula), note the usual higher levels of seroprevalence in more urbanised areas, especially those ‘on-road’ (major traffic routes). There is also a specific reference to occupation – farmers (mainly peasant cultivators) are reported as least likely to be infected. We are left to speculate as to the reasons for this, though there is the implication that they are amongst the least mobile and least wealthy in the population and hence least likely to have the opportunity for risky behaviour.
Another study (Kwesigabo et al. 1999) reports on data from north-western Tanzania (Kagera: the region of highest prevalence near to the Ugandan border). This sampled all hospital patients over a three month period in 1992 (1,422 cases). A breakdown of the sample in terms of occupation (listed without comment!) delivers the following striking contrasts:
| Occupational Category | % in Sample | % HIV Positive |
|---|---|---|
| Civil servants | 14 | 45.5 |
| Businessmen/women | 7.6 | 41.3 |
| ‘Peasants’ | 50 | 36.9 |
| Other, inc not stated | 27 | 44.8 |
| Adapted from Kwesigabo et al. 1999 |
Obviously the categorisation of occupations here is sociologically clumsy. ‘Peasants’ probably included at one extreme some who were employers of labour producing for the market; at the other extreme there was a majority who were basically small peasant or subsistence cultivators – in class terms quite different. As is suggested in Obbo's account, petty capitalist farmers will have more disposable income to marry second wives and to cultivate liaisons or to visit sex workers. The category ‘other’ is very large and distorts the whole table. Nevertheless, it is notable here that, as in the previous example, those with the highest status and resources also have the highest level of infection. Tibaijuka, also reporting from Kagera, confirms that ‘the majority of the victims were from relatively well-off families’, businessmen or traders with multiple partners (Tibaijuka, 1997:968).
Ainsworth and Semali also focused their work on Kagera. Their survey drew on a different kind of sample, taking all those who had died in the area during the period 1991-4, of whom nearly 40 per cent had a diagnosis of AIDS, and adding in all the adult members of the same households who remained alive. This allowed for a threefold comparative basis from which they drew the following conclusions about those who had died from AIDS: they were more likely to be female and to have died younger than those dying from other diseases; they were slightly more educated and much more likely to be non-‘farmers’ (note again the use of this conflated term5), especially white-collar workers or traders than either those who had died from other causes or were still alive. Presumably because fishermen were here selling their catches at good prices, ‘male fishermen and female merchants/traders have a significantly higher risk of dying of AIDS compared to farmers’ (Ainsworth and Semali, 1998:10). So from this study too, the greater prevalence amongst higher socio-economic categories is confirmed and a link is also made between occupation, gender and the transmission of HIV. What is lacking is any exploration of why these statistical patterns emerge.
A more recent overview of studies of HIV/AIDS in Tanzania by Lisa Garbus (2004) cited a wide range of similar work, though the statistical record on which she draws is notably sparser than her parallel account of South Africa (Garbus, 2003). That which exists is based on sociologically limited categories – studies which report that ‘the rich’ are more likely than ‘the poor’ to access health care by bringing their children to clinics, having them vaccinated or paying attention to their nutritional status (based on a 2003 study by Schellenberg et al. in southern Tanzania). The more highly educated were also more likely to mention condoms as a protective measure and to insist that they would use them.
The implication that this is due simply to their greater levels of health knowledge needs to be questioned. The imposition of user charges and the cut-backs in the health service following structural adjustment means that the vast majority of people cannot afford to access a considerably diminished public health care system – income as much as knowledge is implicated here.
Conversely the usual categories amongst the poor are cited as vectors for the transmission of HIV: sex workers and food retailers, especially those working in areas of high male incomes such as mining who were suffering high rates of prevalence (from a WHO study reported in 2003 of the gold-mining area round Lake Victoria). A case where class exploitation more clearly enters the picture as a factor in the spread of HIV/AIDS is Garbus's reference to a 2003 TAMWA study of domestic servants from fourteen regions of Tanzania: sixty per cent of women reported having sex with male members of families employing them, of whom half had been forced.
Garbus notes that Tanzania's GDP per capita is one of the lowest in the world and that ‘Tanzania's industrial sector is one of the smallest in Africa’ (Garbus, 2003:20). Though she does not link these two factors, she does report that Tanzania's industrial sector is feeling the pinch of AIDS sickness and deaths: ‘business closures related to the HIV/AIDS impact have been noted’ by the Tanzanian Chamber of Commerce and Industry and by the government. These have been occasioned by falling revenue following loss of skilled labour and higher recruitment and sick leave costs (Ibid. p. 58). Big companies like Brooke Bond and Standard Chartered (foreign multinationals) and Tanzanian Breweries (originally a parastatal, now privatised, one of the largest industries in Tanzania) had HIV/AIDS policies (Ibid. p. 85) though the practical import of these is not described. Reading between the lines here, it is possible to see that the wage-labouring class is being badly hit by the epidemic but that this is also impacting on the owners of industry and threatening the accumulation of capital (for a recent review see Bujra, 2004). Since wage-workers are usually migrant workers, the transmission of infection to rural areas is a common consequence of sexual liaisons in towns.
In 1992 the first World Bank study of Tanzania had noted a paucity of national data relating HIV/AIDS to socio-economic variables. A very recent study (THIS, 2005) remedies this by reporting on an extensive national sample survey of the population at large. Interviewing over twelve thousand (with women slightly more than half the sample) they broke down their sample into the usual variables of gender, rural/ urban residence and education, but also added ‘wealth’. This indicator involved the listing and quantitative measure of household assets – consumer durables, residential condition of house and access to facilities like water or electricity. Then ‘individuals were ranked according to the total score of the household in which they resided. The sample was then divided into quintiles from one (lowest) to five (highest)’ (Ibid. p. 9). This is a measure of personal wealth (not capital), though it actually disclosed the relative poverty of the vast majority. The measure tells us nothing about sources of income or about social relations between people of different levels. Occupation is not one of the indices in this study – though the claim is made that 71 per cent of men and 63 per cent of men were in ‘agricultural occupations’ (Ibid. p. 21). It is clear that the vast majority of those interviewed were relatively poor peasants whose subsistence would not have come through wage employment, and whose ‘assets’ were minimal. Some respondents would have been entrepreneurs or self-employed craftspeople, whilst others were wage labourers, and amongst these would have been relatively highly paid professional workers. The division of respondents into ‘wealth quintiles’ effectively disguises even the resource inequali-ties between these different social categories as, by definition, each quintile has an even number of members.
Leaving this aside, what we can draw from this extensive study does not contradict earlier and more particular research findings. Unsurprisingly, wealth and educa-tional levels coincided again here, and the higher the wealth quintile the more likely the respondents were to claim fewer sexual partners, to claim to have used condoms in risky sexual encounters (extra- or non-maritally) and to have been for HIV testing – all suggestive of future-oriented and responsible behaviour which would be anticipated in ‘educated people’. The authors take the responses at face value, though it is likely that such respondents would have been more likely to know what answers were expected and more articulate in asserting their moral standing. However, this study also incorporated an HIV testing element, and from this it can be seen that it is those in the highest quintile who had the highest rates of prevalence. The authors do not even comment on this finding, but it is clear that their presumption that education and wealth would be protective is confounded. What is more, youth of this highest wealth quintile conceded to more risky sex than any other category, whilst the young women amongst them were more likely to have had sexual relations with men ten or more years older than themselves and more of them also reported that they had been coerced into sex. Two speculative conclusions might be drawn here. First, is that the pattern of higher prevalence being associated with higher wealth is not just an early historical aberration – it continues to be reproduced. And second, there is a chilling intersection of socio-economic position with gender relations here, suggestive not just of greater disposable income and power amongst better-off men, but of young women who are struggling to find acceptable outlets for their sexuality in situations of relative freedom. Given the geographic mobility of socially mobile populations there is also perhaps a pattern of loosening customary controls.
Relational accounts of social class & AIDS
Epidemiological categories of socio-economic status are generally based on the aggregation of individualised data, but one study cited by Garbus describes ‘community risk’ rather than that of individuals. Following work in Northern Tanzania, high community risk was linked to ‘a relatively robust and mobile economy [in which] the local population comes into contact with a larger sexual network, some of whom may already be infected’ (Garbus, 2004:17, based on work by Bloom et al. 2002). Viewing communities as collective actors often leads to calls for ‘community empowerment’ in the struggle against AIDS (e.g. UNAIDS, 2004), whereas what is most striking about ‘communities’ is that they are divided on the lines of gender, class and other social divisions (Bujra and Baylies, 1999) – they cannot be held collectively to be at risk. Lwihula et al. in an early study of Kagera (Lwihula, 1993) showed that local people saw not ‘communities’ but ‘young businessmen’ trading in illegal goods across the border, and their philandering ways, as an explanation for the spread of the disease. It is the link between class and gender that is striking again here – masculine power gaining an added impetus through an emergent entrepreneurial class category.
Setel's more sociological study of Kilimanjaro showed how ‘the term “businessman” had taken on pejorative connotations in many contexts – particularly in light of AIDS’ (Setel, 1996:1173) – such men were usually young, often in conflict with their elders and their footloose lifestyles were ‘replete with opportunities for sexual excess’. Setel explores the emergence of this class category in terms of a transformation of rural modes of production, an increasingly class-differentiated peasantry, and with it a collapse of customary modes of socialisation and social control. AIDS in this area was linked to discourses of ‘cultural change and loss’, especially ‘a decline in the moral character of manhood’ (Ibid. p.1176).
That a similar process can lead to women being regarded as especially blameworthy is explored (sociologically) by Weiss (1993). Haya women from Kagera have migrated for generations to large towns in East Africa where they operate successfully as sex workers. These women are migrants in the sense that they return home with savings to purchase land. In the customary system, their access to land would have been via fathers or husbands within a patrilocal pattern of marriage residence (see also Tibaijuka, 1997). Land ownership offered women a rightful place to build their graves – and increasingly their deaths were from AIDS. Men viewed their independent capital accumulation with disfavour and linked it with their blame of women for spreading disease. At the same time men, are also independently mobile (as traders and entrepreneurs) and mobility has come to have a class significance:
Automobiles … are not only highly charged representations of class differences in Kagera, they are also central icons of the very process of class formation (1993:21).
It would be surprising if emergent class groupings did not also play some part in the response to the AIDS epidemic, focused as this has been in the latter part of the 1990s on NGO activity. I have already mentioned the ways in which the increased funding of NGO AIDS action allowed for the consolidation of petty-bourgeois elements, whilst arguing that philanthropy is also a mode of class control. In Dar es Salaam we noted that some of the first sufferers came from affluent backgrounds:
Despite the seemingly obvious association between AIDS and poverty in Dar es Salaam, AIDS activism was not generated in such contexts. It came out of the leafy suburbs and the campus, the social milieu of the educated and the comfortably-off, those whose world was shattered by the realisation that their income, their modern knowledge and familiarity with scientific facts could not protect them from this invisible killer (Bujra and Mokake, 2000:154).
… the emergence of welfare NGOs represents the privatisation of the provision of public goods. Those who run NGOs would formerly have been employed by the state (funded through taxes and foreign aid); now they must freelance to earn their keep. The object of their class concern (the poor, the powerless, the sick and uneducated) cannot pay for services previously provided in a limited way through the public purse, or for their innovative expansion. This petty bourgeoisie must thus transform itself. They become private entrepreneurs touting for funds in the world aid market (Bujra and Mokake, 2000:156).
privileged members of the urban establishment who have simply and cynically cashed in on the epidemic … another local saying with a double meaning states that ‘AIDS is business’ (Setel, 1996:1176).
We observed similar exchanges in Dar es Salaam, with a similar process whereby ‘victims’ of one socio-economic category are transformed into clients of another, represented by NGO staff. ‘The AIDS organisation becomes a stage on which class relations are symbolically marked out at the same time that they are occasionally transcended’ (Bujra and Mokake, 2000:167). I also argue that one of the dynamics going on here is ‘a class project of uplift and enlightenment’ (p.166) though clients may resist the implications of this.6
Lushoto & the Limits to Philanthropy
Material collected in a rural area of Tanzania (Lushoto, in the North-east, 1995-2001) as part of a larger Gender and AIDS project, showed that there were limits to philanthropy, though not to class attitudes in the context of AIDS. We carried out a baseline survey in each of six local areas in Tanzania and Zambia, of which Lushoto was one (Baylies and Bujra, 2000). The survey of 100 local people was designed as a structured sample, divided evenly between men and women, with inclusion of age categories and finally a division in terms of ‘socio-economic status’. The methodol-ogy used here suffered from many of the faults already described above; it used socio-economic indicators rather than documenting class difference per se because the dimensions of the latter could not easily be agreed for such a wide-ranging study and in advance. However this data was then backed up by in-depth ethnographic work. The measure used in the survey to determine the position of particular interviewees had to be based on observable data and was therefore prone to subjective evaluation. We set up divisions of ‘upper’, middle and ‘low’ based on the appearance of house, surroundings and apparent lifestyle. Fourteen per cent of our sample was to be sought out from the ‘upper’ category.
In Lushoto I was working in a mountainous rural area near to the district capital (Bujra, 2000b). One village, composed of a series of hamlets strung across the hills was the focus. The people were predominantly peasant/subsistence farmers, mainly Muslims of a single ethnic group (Sambaa). A few men were growing coffee or other cash crops and there were a sprinkling of shops, a couple of bars, a sugar pressing workshop (for making beer), and a small maize grinding mill. The village was on the steep upper slopes with nearly every square inch under cultivation but with considerable soil erosion. On the lower slopes outside the ‘village’ there was better land, especially in a few valley bottoms, and here larger farms had been carved out, originally by German settlers, but now occupied by African small capitalist farmers. It was not easy to find our quota of ‘high’ level interviewees, amongst whom half had to be women (nearly all wives of rich men) and two young people (still dependent on rich fathers). We found a few in the higher hamlets and descended to the lower slopes to fulfil our quota.
Reanalysing the data now in terms of this variable, I note that it did have some value in so far as high socio-economic level was linked to other variables. Compared to middle and low categories, these people had a higher educational level, with one in five having had some secondary education and none were without education at all (compared to nearly 30 per cent in the lowest category). In terms of their mode of livelihood, two owned factories (timber, saw-mills); one was a professional wage-worker, the others were engaged in commercial farming and cattle husbandry, combined in some cases with trade or shop ownership. They were also three times more likely to be Christians than the lower categories and though predominantly still Sambaa, nearly thirty per cent had been born outside and were of other tribal origins. Both Sambaa and others had often come to this area to take up professional or administrative jobs and then bought land.
In terms of AIDS it was notable that the higher the socio-economic status, the more likely it was that the respondent would acknowledge knowing someone with (or who had died of) AIDS (71 per cent of the high group, only 28 per cent of the low group), thus confirming a picture of the higher socio-economic groups suffering greater prevalence. This person was also far more likely to have been a close relative (rather than an acquaintance). Although they tended to present themselves as of respectable moral standing and suggested that faithfulness and the avoidance of ‘immoral’ behaviour was the main form of protection from this disease, they were also more likely to agree that condoms could protect – twice as likely as the lowest group. They were marginally more likely to acknowledge that they had used a condom themselves (36 per cent compared with 25 per cent).
None of this tells us anything about the relations between classes – and it is fair to point out that direct class relations were limited. Even though some of the large farmers employed labourers, these were more likely to come from the local town than from up the hill. Sambaa village people avoided wage labour so long as they were able – in our sample only 11 per cent were wage-labourers of whom just over half were casual farm labourers. Comparing the educational qualifications of men and women in the high category, it is clear that they were marrying people of similar standing. Unless they were from the village, they had not chosen spouses from there. A process of class cultural reproduction was clearly in evidence here, expressed through marriage choices, educational attainment and style of life (furniture and ornamentation of houses, leisure activities of younger people and so on).
Our research project entailed a dimension of action research and towards the end of the research I became involved in supporting the establishment of a girls’ group intended to support young women in avoidance of risk. The group was set up together with a prominent village woman who happened to be a Christian, moderately well-educated. To find funding for the group's activities (the establish-ment of a small plot for growing vegetables, an opening ceremony at which they were to perform a play about AIDS and to which village leaders and dignitaries from the town were invited, etc.) we decided to look for ‘benefactors’ (wafadhili) amongst the better-off on the lower slopes and in the nearby town as well as those in the village who were trading crops or owning shops. Despite the fact that this category was more affected by AIDS than those who were much poorer, our efforts were almost totally unsuccessful. A few excerpts from field notes give an indication of the stratagems which were devised to avoid giving to this worthy cause. We visited an elderly man, the owner of two timber factories, operating lorries and cars and living in a substantial house built of bricks and tiled, on the lower slopes. The room in which we had interviewed him had a telephone, easy chairs and a large coffee table, photographs, carvings, a bookshelf full of books, embroidered curtains etc.
We avoided making an outright demand for money but made it clear that we needed help and that even ‘two tins of maize seed’ would allow the girls to start their garden. The old man's face took on a carefully blank expression as we went along. [When we invited him to the opening ceremony and said that we were also inviting the local MP]… he suddenly became animated and said how helpful [the MP] had been to the (Lutheran) church choir, buying them instruments and also arranging for a recording session in Dar 7 … When I heard this I was more hopeful … for the old man [had] evidently got the point we were making obliquely, that he might give us something. There was a pause – a long pause. And then he said: ‘Welcome again’… Another long pause and he went out – ostensibly to check whether they could come to the ceremony or not. He came back, saying that unfortunately that day they were invited to Mtae with the choir, etc etc, but as they wouldn’t be able to come he thought a little something might help – and he dug his hand in his pocket and brought out – Sh1000! (c £1.15). I then asked him if we might put his name down as a benefactor, but he said not to do this now, maybe later (author's field notes, 1996).
He said that he would try his best to come [to the opening ceremony], but we left the second house empty-handed too. Indeed when I was saying that the girls needed only two tins of maize seed to plant but it was difficult because their parents had none to spare because of last year's disastrous harvests, the old lady said that they had got ‘nothing’ too! Actually they have some very good valley land and they are growing a great variety of crops on it, using hired labour (author's fieldnotes, 1996).
F's house … is very large and set in lovely gardens, with a dog. We knocked at the front door and it was opened by F himself. [He invited us in but, despite our explanations he was unresponsive to our pleas for help] … we gave him the invitation, but he did not drop his cold manner. After a while his wife joined us, a very fair woman, almost white, and three small grandchildren. Their house is beautiful inside, with panelled wooden floors and panelling part way up the walls, nice furnishings and carpets, polished cupboards and tables, soft lighting. The wife seemed warmer and greeted E with something like pleasure, and E began to explain [our purpose] again. Then the wife said that we were doing it all wrong – we should have written a letter straightforwardly to ask for money … When we said about the girls going off to towns and being in danger she said, with the sweetest of smiles, that she ‘didn’t think – well it was just that some girls were like that, they would misbehave anyway – it's their way’… And when we told her about the vegetable plot she sneered and said that there was no profit to be had from a plot, better they do something else. Eventually she said that she had some thread somewhere [for a sewing project], she would look for it and let us have it [she never did]. But as for a donation – that depended on ‘bwana’ [her husband]. He meanwhile had gone out, looking bored. We thanked her and … left (author's fieldnotes, 1996).
It seems clear that, in this area, those with ‘disposable income’ preferred to spend it in ways other than the support of potential fellow sufferers from a fatal and devastating disease. It is also clear that they were deliberately distancing themselves both socially and emotionally from their impoverished neighbours, their behaviour in this respect contrasting with the early AIDS activists in Dar es Salaam. There might be many explanations for this, but surely one of them is class interest. By and large these are people in the accumulative stage of class formation rather than the consolidated phase – when expression of class position through philanthropy or as a bid to rescue or co-opt the lower orders might become more important.
Conclusion
The AIDS epidemic in Africa is properly perceived as a tragedy and those affected by it as victims of tragedy. I have tried to show here that class analysis is very significant to comprehending and addressing the AIDS crisis and in particular the dynamics of its levelling as well as its divisive character. But in order for class analysis to be usefully employed we need to move beyond positivistic methods culminating in indexical accounts of class and towards a more relational and sociological understanding. Class defines itself in dynamic interactions; it is more than a set of static statistical categories – and this is reflected in relation to AIDS as indeed much else. My argument here has been that class formation and consolida-tion in the context of the extension of capitalist relations of production frames both the transmission and response to HIV/AIDS in Africa. On the one hand, it is evident that massive geographical movements of population in response both to coercive pressure (land alienation, taxation) and to the need to secure livelihoods (initially predominantly affecting men but later also women) have led to novel processes of class formation as well as transforming gender relations. Taken together, these processes created situations of risk once the HIV virus entered the picture. On the other hand, there is a particular issue about social mobility. As higher class groupings are formed through entrepreneurial activity, commercial farming or the creation of a professional cadre of wage-labourers, such categories begin to consolidate by intermarrying (often later marriages), planning education for their children, and developing class cultural attitudes to justify their rewards. In the process their social isolation from other categories becomes more marked. This has its consequences for the transmission of HIV as social controls over sexual behaviour are attenuated and women make bids for more sexual autonomy, whilst men may use their greater disposable income to buy sex or additional wives. But it is also at this class level that philanthropy emerges as a way of confronting a problem which affects all to different degrees.