Abstract
Because redistribution concerns ‘who gets what and from whom’, redistributive conflicts revolve around ‘who should get what and from whom’. Individuals as well as states distinguish between deserving and undeserving claimants. People may favour people they know over strangers, kin over non-kin, or some kin over other kin. This paper uses data from survey experiments to show that young South Africans distinguish between deserving and undeserving claimants on both the state and kin. The hierarchy of desert with respect to public welfare is clear and intuitive, with people who cannot look after themselves being considered more deserving than those who can. Deservingness with respect to different categories of kin – i.e. the ‘radius’ of responsibility for kin – varies less markedly, but with some variation between racial or cultural groups. Deservingness with respect to both public and private support is affected dramatically by the attitude and reciprocity of the claimant, with the important exceptions of mothers who should be supported unconditionally. Public and private support appear to be complements not substitutes for each other, in that people who believe that the state should support people in need are also more likely to believe that kin should do so also.
Introduction
Conflict over the distribution (and redistribution) of resources is shaped by the perceived justice of the distribution. ‘Who gets what’ generates conflict insofar as it differs from perceptions of who should get what. In many societies there is a widespread perception that some people get less than they should, i.e. less than they deserve, and they can thus make legitimate claims on others for redistribution. The idea that ‘deserving’ poor people can be distinguished from ‘undeserving’ poor people recurs in public discourse across both time and societies. The distinction is clearest with respect to public provision, governed by formal rules of eligibility. ‘Social welfare policy cannot be fully understood without recognizing that it is fundamentally a set of symbols that try to differentiate between the deserving and undeserving poor’ (Handler and Hasenfeld, 1991: 11, quoted in Van Oorschot, 2006: 23). The distinction applies also to private support, especially between kin. Indeed, it is likely that the legitimacy of claims on the state is related in some way to the legitimacy of claims on kin. This paper uses survey data from the South African city of Cape Town to examine patterns in the perceived legitimacy of claims made on kin and the state, and the relationship between these. The paper thus contributes to our understanding of when, how and why the uneven expansion of public provision is likely to lead to conflict.
Across most of north-west Europe and the European diaspora, the construction of welfare states in the twentieth century was driven in part by the declining capacity or willingness of both kin and local community to provide for deserving categories of poor people (De Swaan, 1988). As people declined to recognize the legitimacy of claims made on them by strangers and even kin, so these responsibilities were institutionalized through the welfare state. Welfare-state-building entailed not only decommodification, as Polanyi (1944) recognized (see also Esping-Andersen, 1990), but also – to varying extents – defamilialisation (Orloff, 1993; Esping-Andersen, 1999). Fewer and fewer elderly people lived with their children and more and more elderly people received public ‘home help’. Esping-Andersen cited data showing that this process was most advanced in the social democratic welfare regimes: In Denmark, by the mid-1980s, only 4 percent of the elderly lived with their children, and 22 percent of the elderly were covered by public home-help (Esping-Andersen, 1999: 63, 71).
The process of defamilialisation was far less advanced outside of north-west Europe. In Japan, two-thirds of elderly people lived with their children in the mid-1980s, and barely one percent received public home-help (1999: 63). Kin remained even more important across much of the global South. Many people continued to live in extended families or to recognize obligations to more distant as well as closer kin. Observing the power of kinship claims and responsibilities in Africa in the mid-twentieth century – at the same as welfare states were being built in the advanced capitalist societies of the global North – Fortes famously described kinship as entailing ‘inescapable moral claims and obligations’ (Fortes, 1969: 242). In many societies, the power of kinship was tied to a conception of personhood that was far removed from the individualism of western liberalism. Sahlins refers to the ‘mutuality of being’ that is specific to kinship (Sahlins, 2011). In its idealized form, this understanding of kinship precluded even accounting between kin. Echoing Fortes (1969: 246), Pitt-Rivers writes that ‘yours and mine no longer exist; there is only ours’ (2016: 444, quoted in Bjarnesen and Utas, 2018: S8).
In practice, even in Africa, kinship was never as binding as Fortes’ idealized account suggested. Many urbanizing migrants did remain tied into strong kinship networks, but others saw urbanization as an opportunity to embrace the individualist consumerism associated with modernity and to shrug off responsibilities to kin (Ferguson, 1999; Jeske, 2016). In the 1990s and 2000s, claims on extended kin increased as a result of the AIDS pandemic, with sick adults and orphaned children becoming the dependents of grandmothers and other kin (Seeley et al., 1993; Foster, 2000). As a result, according to both qualitative and quantitative research, many kinship claims have become fluid and negotiable. Whilst it is rare for kin to leave children without support and care, many adults can no longer count on kin for support or care or even shelter, and must turn instead to neighbours, friends and fellow church congregants (Ross, 2003; Seekings, 2008a; Harper and Seekings, 2010). Whereas older survey questions probed the support provided by some kin, new survey questions reveal the numbers of kin who do not provide support (Madhaven et al., 2017).
The changing import of kinship in Africa has fueled the expansion of public provision. This is most evident in the case of South Africa, where social assistance was originally introduced for selected categories of its white (and coloured) citizens when it became clear that kin support no longer sufficed. The South African state justified the exclusion of its black subjects primarily on the grounds that black South Africans did and should provide adequate and appropriate support for their kin. It was only when it began to become clear (in the 1940s) that kinship was failing growing numbers of elderly black women and men that old-age pensions were extended to them (albeit with greatly reduced benefits). By the end of apartheid (in 1994), racial discrimination in the value of the old-age pension had been ended, and most elderly women and men received a pension that supported entire households, including unemployed adults as well as grandchildren (Møller and Sotshongae, 1996; Sagner and Mtati, 1999). In the 2000s, social assistance expanded in South Africa through the provision of means-tested grants to more and more poor mothers with children, who could not support themselves and whose kin could or would not support them adequately.
Elsewhere in Africa, political elites have tended to resist proposals to expand tax- or aid-financed social assistance or social grants, on the grounds that (first) kin should take primary responsibility and (secondly) that poor families should work harder (see for example, Kalebe-Nyamongo and Marquette, 2014). When the African Union adapts global statements on social and economic rights, it invariably adds discussion of the importance of family and of responsibilities alongside rights: Poor people are not seen as deserving of state support in part because they are deemed to be deserving of familial support. Public expenditure on social protection remains lower in Africa than in other parts of the world (ILO, 2017; World Bank, 2015). Nonetheless, social assistance programmes have expanded, especially for the elderly (through modest pensions) or for multi-generational households lacking working adults. The most thorough recent estimate of the scope of social assistance across sub-Saharan Africa suggests that 60 million people benefit directly from cash transfer programmes in 37 countries (Hickey et al., 2019; see also Garcia and Moore, 2012).
The normative bases of deservingness
Popular assessments of whether claimants are deserving of support from state or kin generally entail norms besides the need of the claimant. Van Oorschot, in his pioneering analyses of quantitative data on public opinion across Europe, identifies four additional criteria shaping perceived deservingness with respect to public provision: ‘Control’ (over need, i.e. whether or not the person is considered responsible for his or her situation), ‘attitude’ (i.e. whether the person is likeable), reciprocity and identity (i.e. is the person one of ‘us’?) (van Oorschot, 2000; 2006). The elderly usually qualify as deserving of public support under all five criteria: They are rarely seen to be responsible for their infirmity or ungrateful for support, and are believed to have earned support in their old age; most people identify with them (because they are our parents or grandparents, or even us in future years), and they have clear needs, exacerbated by infirmity. Poor single mothers may not be seen as qualifying under any but the last criterion. In the USA, black people have often been considered less deserving (i.e. by white Americans) because they are seen to be responsible for their poverty (control), through their laziness (attitude), and because ‘they’ are not part of ‘us’ (identity). Across much of Europe, immigrants are regarded in much the same light. Van Oorschot (2000) found that the actual level of need was the least important of the five dimensions of deservingness.
Turning to kin support in the advanced welfare-capitalist societies, Finch’s pioneering research in Britain in the 1980s suggested that there was no clear consensus around the obligations people had to their kin. Kin support was negotiated between individuals, without clear rules – but also with careful attention to the legitimacy of declining to support someone. The closeness of the kin relationship was one principle informing support, but other principles reflected more the characteristics of the individuals involved. Did they get on well? Was the relationship reciprocal? Was it possible to maintain a healthy balance between dependence and independence? And was it the right time in the lives of both kin for one to assist the other? ‘Obligation, duty and responsibility. . . are commitments developed between real people, not abstract principles associated with particular kin relationships’ (Finch, 1989: 181; see also Finch and Mason, 1993).
Previous research in post-apartheid South Africa using quantitative data suggests that the dimensions of deservingness with respect to public provision are very similar to those discussed by Van Oorschot (2000, 2006) for European societies. South Africans express a clear normative hierarchy of desert, ranking the elderly and disabled as most deserving, and young able-bodied adults as least deserving. Individual-level characteristics (such as whether someone drinks heavily) also shape perceptions of deservingness. Perhaps surprisingly, there is little or no evidence of racial discrimination, i.e. people do not seem to discriminate in favour of other members of the same racial group and against members of other racial groups (Seekings, 2007, 2008b, 2010). Survey data consistently show very strong support for increasing the value of the old-age pension, even if this required tax increases (Seekings, 2008b), but deep ambivalence over support for young mothers, some of whom are viewed as abusing grants intended for their children by spending them on themselves. 1
With respect to private support, between kin, the most careful qualitative studies have focused on decisions by pensioners (i.e. the recipients of relatively generous old-age pensions from the government) as to whether or not to support younger kin. Sagner and Mtati (1999) concluded that there was a clear ‘normative hierarchy of who should be cared for’: grandchildren first, adult children later, stepchildren later still, and other kin last. More recently, using survey data, Harper and Seekings (2010) found that the range of kin on whom young black men and women could make claims was not much wider than it was for young white or coloured men and women. Moreover, the kin on whom young black people could make a claim were almost all maternal kin (but see Mkhwanazi and Block, 2016). Sagner and Mtati (1999) also found that decisions to support kin were shaped also by the behaviour of the claimant. Pensioners justified ‘their decision not to help a particular needy (grand)child or kinsperson’ by reference to the ‘unreasonable behaviour of the person needing support, be it that s/he had often eschewed her obligations in the past, or that s/he had severely defied gender and age-related roles’. They quote pensioners complaining about children who work but nonetheless fail to contribute to others, whether financially or in other, symbolic ways: ‘Children of today, they are more occupied with themselves’, said one; ‘they are just children by name, they are snakes’ (1999: 405–7). Bray (2009) similarly found that sibling support was conditional on behaviour.
Perceived ‘obligation’ – whether by state or kin – depends on the perceived deservingness of a claimant. The existing qualitative research in South Africa suggests that – in South Africa as in Europe – a claimant’s deservingness may be conditional not so much on need as on his or her character and behaviour (‘attitude’, for Van Oorschot), including past or prospective reciprocity (which might be thought of as one dimension of ‘attitude’).
This paper examines the legitimacy of claims on both the state and kin in contemporary South Africa, using new experimental data from a large sample of young people in the city of Cape Town. I confirm the findings of previous studies that showed a clear hierarchy of perceived deservingness with respect to state programmes, as well as the importance of the claimant’s own behavior or attitudes. With respect to claims on kin, I show that there are minor differences in the perceived deservingness of different categories of kin (with mothers consistently viewed as most deserving of support). The behaviour of kin, especially with regard to reciprocity, is of considerable importance (although less so with respect to the claims of mothers than with regard to other categories of kin). Finally, the paper suggests that deservingness in relation to the state and kin are complements to each other, not substitutes. The factors that render some people deserving of public support are similar to those that render them deserving of kin support. The importance of an individual’s behaviour – rather than need – in assessments of deservingness suggests that there is a strong possibility of conflict when individuals are seen to behave inappropriately.
Methodology
This paper analyses data collected through vignettes (or ‘survey experiments’) included in a survey conducted in Cape Town in 2009. The use of vignettes took off in attitudinal research in the USA in the 1980s and 1990s (Sniderman and Grob, 1996; Gaines et al., 2007). Vignettes entail presenting respondents with a description of a hypothetical situation, or story, and then asking one or more questions related to the situation. The most important characteristic of a good vignette is its verisimilitude, i.e. it must appear to be true in the sense that the story, while hypothetical, is also credible and familiar. The objective is to get people to imagine a situation which might easily happen, or does indeed happen frequently, allowing people to comment on it as if it was a true situation. The value of vignettes is that the details of the situation or story can be manipulated, in order to see how the details of the manipulation affect respondents’ answers. Different respondents get different versions of the vignette. If the assignment of versions of the vignette to respondents is random, then we can use the results to analyse the effects of variations in the details of the story on the respondents’ answers to following questions, in relation to the varying characteristics of the respondents themselves. The method is experimental in that we observe the consequences of selective interventions in the specification of the vignette. Vignettes thus allow for the combination of ‘the distinctive external validity advantages of the representative public opinion survey with the decisive internal validity strengths of the fully randomized, multifaceted experiment’ (Sniderman and Grob, 1996: 378).
Vignettes are especially useful for the analysis of norms. Morality is often situational. Faced with a general question in the form ‘what is the right thing to do if . . .’, people will often prefer to respond ‘it depends’, i.e. it depends on factors that are not specified sufficiently in the question. Vignettes take much of the uncertainty out of the situation by specifying the situation, i.e. they allow for norms and beliefs to be contextualised. Vignettes try to approximate reality in another sense also. People’s views on the morality of a course of action are typically shaped by their interaction with other people in a process of formal or informal deliberation. Vignettes can incorporate this through attempts to persuade respondents to change their mind in response to new information.
This paper analyses data from two vignettes. The first, probing perceptions of who is deserving of financial support from the state, had been used previously in South Africa to probe the effects of race on perceived deservingness (Seekings, 2008b; 2010). The second, probing perceptions of who is deserving of support from kin, was new.
Both vignettes were put to a sample of almost three thousand young men and women, aged 20–29, across Cape Town in 2009. The survey was the fifth wave of a panel study of young people. The panel originally, in 2002, comprised a representative sample of just over 4,500 young people, then aged 14–22. The original sample was drawn using a three-stage sampling design: First, a random sample of neighbourhoods was selected; secondly, a random sample of residential units was selected within each neighbourhood (including through the use of aerial photographs and on-the-ground enquiries in shack settlements); thirdly, in the selected households, all young adults were interviewed, unless there were more than three in the household, in which case only three were selected, also randomly. By the fifth wave (i.e. the fifth round of interviews) attrition had reduced the size of the panel by one-third: Approximately 3,000 of the original participants were re-interviewed in 2009. Full details of the original sample and attrition up to 2009 are available elsewhere (Lam et al., 2009), as are the data. 2 In summary, the individuals remaining in the panel in 2009 comprised reasonably good samples of young black and coloured men and women. High attrition rates among young white people mean that the data from white respondents need to be treated with caution. 3 In this paper we report unweighted data. Data were analysed using Stata. The modelling results are from logistic regression models, reporting odds ratios. The author is an ‘outsider’ to South African society in that he grew up in Britain, but has lived and worked in South Africa for about thirty years. This research was prompted by very many discussions with people from diverse social backgrounds about the challenges of navigating through the social landscape of claim-making.
There are two important limitations to the data analysed in this paper. First, the data concern reported attitudes with regard to the legitimacy of claims not actual behaviour. However lifelike the situations described in the vignettes, respondents were asked to respond to stories, not report their actual experiences, nor was there any direct observation of what happens in practice. It is likely that, for many individuals, there is some divergence between what they say should happen and what they or others actually do. Secondly, the data are from young people in a metropolitan area. It is likely that the young, urban respondents are less likely to acknowledge the claims of – and obligations to – more distant kin than older people and people in rural areas. It is also possible that the young, urban respondents have different ideas about the kinds of behaviour that are considered inappropriate and render a claimant undeserving. We should be cautious in generalizing from our sample to the general population.
Claims on the state
The perceived deservingness of citizens with respect to public provision was examined through a vignette based originally on research by Sniderman in the US (Sniderman and Piazza, 1993). Respondents were presented with a brief description of an individual, and were then asked whether the respondent considered that this individual (the ‘subject’) was deserving of a government grant (like South Africa’s old-age pension). If the respondent said that the subject was deserving, then he or she was asked whether the government should give the subject a small sum (‘about R240 per month’, which at the time was the value of the government’s Child Support Grant, i.e. about US$ 24 per month, in 2009), a larger sum (‘about R1010 per month’, which at the time was the value of the old-age pension, or about US$ 100 per month), ‘more than R1010 per month’, or some other amount.
The initial description of the subject took the following form: ‘[Tracy] [is sick]. She is [28]-years-old, [is single with two small children], and lives in a [rich white] neighbourhood.’ Each of the italicized parts of this description was varied. The subject might be male or female, aged 18 or 28 or 38 or 48 or 58, with three possible familial statuses (single without children, single with small children, married with small children). The vignette did not specify directly the race or class of the subject, but rather specified the kind of neighbourhood where the subject lived in terms of race and class. Neighbourhoods were described as either (a) rich and coloured, (b) poor and coloured, (c) rich and white, or (d) poor and African (black) – i.e. referring to the various combinations of race and class that continue to define most neighbourhoods in post-apartheid Cape Town. The name given to the subject corresponded to the kind of neighbourhood. The vignette also specified why the subject needed support. He or she ‘does not want work’, ‘cannot find work’, ‘is sick with AIDS’, ‘is sick’, ‘was abandoned by husband and cannot find work’ (only used for subjects described as married women), or ‘needs to look after sick and elderly parents’ (also asked only in relation to female subjects). In most versions of the vignette the nationality of the subject was not specified, but in some versions the subject was described as either British or Nigerian or Zimbabwean, to assess the effect of immigration on perceived deservingess. We did not ask about elderly people, because survey data shows a general consensus that they deserved public support. 4
The responses to the vignette on public desert showed that most young men and women in Cape Town share a common vision of a hierarchy of desert even among working-age adults. Respondents’ assessments of the deservingness of the claimant in the vignette were analysed through a series of multivariate logistic regression models. The full results are not reported here, because they broadly conform with the results of previous experiments using the same or similar types of vignette (Seekings, 2008b; 2010). In summary, subjects who were sick were seen as most deserving, whilst subjects who did not want work were least deserving. Older subjects were more deserving than younger ones, subjects with children were more deserving than subjects without any. Foreigners – especially Nigerians – were seen as less deserving. Subjects from rich neighbourhoods were seen as less deserving than subjects from poor neighbourhoods. The gender of the subject was not relevant. The experiment confirmed that the situation of the claimant (or what Van Oorschot terms ‘control’, i.e. responsibility for need) had a big effect. The proportion of respondents who considered the claimant as deserving varied between 24 percent (for claimants who did not want work) to 72 percent (for claimants who were sick with AIDS). A multivariate regression model, which controlled for all of the other characteristics of the claimant, suggested that a claimant who was sick with AIDS was more than nine times as likely to be deemed deserving than someone who did not want work.
The radius of responsibility towards kin
The second, new, vignette set out a situation in which a person (henceforth the ‘subject’) was asked for accommodation by a kinsman or woman (or in some variants, non-kin) (henceforth the ‘claimant’). Respondents were not asked whether they would accommodate the claimant, but rather whether they thought that providing accommodation was the right thing for the subject to do. The vignette opened with a standard introduction: ‘In South Africa today, many people help family (kin) or neighbours or other people when they have problems. I am going to describe a situation to you, and then ask you whether you think that people should help someone in this situation. When I say should, I mean: is it the right thing for someone to do.’ The basic form of the vignette was as follows: ‘[Joyce]’s [house is not big, but there is some extra space; she/he has a job and can pay her/his expenses]. One day, [Joyce]’s [brother’s friend] phones [Joyce] and says that he wants to move to Cape Town [to get better health care]. Should [Joyce] say ‘yes’, her [brother’s friend] can stay with her?’ The vignette entailed four factors that varied: the gender of the prospective subject (Joseph or Joyce); their capacity to accommodate someone (i.e. the size of the house and financial resources, with three values); the relationship between them and the claimant; and the reason why the claimant wanted to move to Cape Town (two values: to look for work or to access better health care). The primary focus of the vignette was the significance of the relationship between the subject and the claimant: Do people recognize stronger obligations to close kin than to distant kin? The relationship variable had eight values: brother, sister, father, mother, grandmother, male cousin, female cousin, and non-kin (specifically, a brother’s friend). All 96 possible versions of the vignette were used.
Table 1 reports descriptive statistics for the different variables in the kinship vignette. Overall, two-thirds of our respondents said that it was right that the subject accommodated the claimant. Only one in six said that the subject should refuse to help the claimant. In general, respondents recognized the legitimacy of claims on kin.
Kinship claims, descriptive statistics.
N=2915.
The gender of the subject made no difference to respondents’ assessment of what was the right thing for the subject to do. The circumstances (i.e. wealth) of the subject were consequential, however: When the subject was described as having resources, the respondent was more likely to say that the subject had an obligation to the claimant. Rich subjects are more likely to be seen as having responsibilities to kin than poor subjects. The reason why the claimant was coming to Cape Town was not relevant. 5
Crucially, the reported relationship between the subject and the claimant was significant. Obligations were weakest to non-kin (although half of our respondents said that providing accommodation was the right thing to do, even when the claimant was described as a friend of the subject’s brother). Obligations were strongest to close kin (parents and siblings) or grandmothers. Obligations to cousins were weaker than to close kin, but stronger than to non-kin. The pattern in the descriptive statistics was reflected also in the results of a multivariate regression model, reported in Table 2 (showing odds ratios). Respondents were three and a half times more likely to say that the subject should accommodate his or her mother than that the subject should accommodate his or her brother’s friend. There were some differences according to the race of the respondent. Fathers were placed lower down the hierarchy of desert by young black respondents, whilst white respondents placed grandmothers far lower down the hierarchy. Neither of these findings is surprising. First, many young coloured and black men and women have little or no contact with their fathers. The declining role played by fathers (and paternal kin) is associated with declining obligations to them. Secondly, it is ‘normal’ among white South Africans that grandparents be independent of their children and grandchildren, providing for their old-age through savings (including contributory pension programmes) and living in old-age homes when they can no longer live on their own.
Kinship desert, regression model.
Logistic regressions; odds ratios reported; standard errors in brackets.
p<0.05 **p<0.01 ***p<0.001 X not significant N=2915.
The experimental data provide further evidence that kinship remains important, in that most people recognize a normative responsibility to support others, but the radius of kinship is uneven. Obligations to mothers and (for many people) grandmothers are much stronger than to other kin.
The effects of character and behaviour on the legitimacy of claims
After respondents had answered the question about whether the subject should accommodate the claimant, we tried to persuade the respondent to change his or her mind by providing additional information about the claimant. Our persuasion experiment differed depending on whether the respondent’s initial answer. Respondents who had initially said that the host should accommodate the claimant were told either that the claimant was ‘an irresponsible person and cannot be trusted’ (i.e. an example of what Van Oorschot calls the ‘attitude’ of the claimant) or that ‘last year, [the claimant] did not help [Joyce/Joseph] when [Joyce/Joseph] asked [him/her] for a small loan’ (i.e. an example of what Van Oorschot called ‘reciprocity’). After being given one or other piece of additional information, the respondent was asked whether he or she still believed that the subject should accommodate the claimant. Alternatively, if the respondent had initially said that the subject should not allow the claimant to stay, then we tried to persuade the respondent to change his or her mind and find that the subject should allow the claimant to stay. Some respondents were told that the claimant was a responsible church-goer (‘attitude’), others that the claimant had lent the subject money last year when the subject asked for a loan (‘reciprocity’).
The data from the persuasion experiments provides more insights into the norms underpinning negotiations over support for kin. First, consider the respondents who had initially assessed that the subject should accommodate the claimant. When told that the claimant was irresponsible and untrustworthy, 45 percent of the respondents changed their mind and said that the subject should not accommodate the claimant; only 32 percent held to their initial position. When told that the claimant had declined to help the subject last year, 25 percent of the respondents changed their mind; 54 percent held to their original position. Overall, we persuaded more than one-third of the respondents who initially favoured assistance to change their minds. One in five respondents was unsure, and 43 percent stuck to their guns and continued to favour assistance. We have no evidence on why additional information about reciprocity was less consequential than additional information about the claimant’s character, but it is possible that respondents imagine non-damning reasons why the claimant did not help the subject out previously, such as the claimant having no money him or herself.
The picture was very similar among respondents who initially said that the subject should not accommodate the claimant. When told that the claimant was a responsible church-goer, one in three respondents changed their mind and said that the subject should help the claimant. The same proportion changed their minds when told that the claimant had previously provided a loan to the subject. In total, one in three changed their minds, 44 percent stuck to their guns and 22 percent were unsure.
The data on persuasion and dissuasion suggest strongly that information on the ‘attitude’ or character of the claimant has strong effects on assessments of desert. Evidence relevant to reciprocity also matters. Claimants’ past generosity to the subject affects the subject’s obligations to the claimant – although perhaps not symmetrically, in that not helping in the past is less consequential than helping.
Whilst one version of the experiment focused on reciprocity directly and the other on the claimant’s behaviour in other respects, the two versions might both be understood in terms of reciprocity. Across much of Africa, reciprocity entails not only being prepared to support people who would support you, but also behaving in ways that are deemed to be appropriate. If a subject assists a claimant, then the claimant has a reciprocal obligation to behave in certain ways. Our experimental data suggests that the implicit form of reciprocity (i.e. behaviour in return for support) is more important than direct reciprocity (i.e. support in return for support).
Respondents’ susceptibility to persuasion or dissuasion varied according to some aspects of the initial description of the subject and claimant in the vignette. When the claimant was the subject’s mother or grandmother, it was much harder to persuade respondents to switch from an initial endorsement of assistance. When the respondent was a sibling or cousin, it was much easier. When the claimant was a mother or grandmother, we persuaded fewer than one in four respondents to change their minds. When the respondent was a sibling or cousin, we persuaded double this proportion to change their minds. In other words, the pattern in initial assessments of desert was evident also in how resolute respondents were when faced with negative information. Respondents not only considered mothers especially deserving but were also much more likely to stick to this assessment regardless of the reported character or past behavior of these mothers. More respondents recognize obligations to mothers (and, except for white South Africans, their grandmothers also) and view these obligations as relatively unconditional. Obligations to other kin are somewhat weaker and are much more conditional. 6 Indeed, insofar as they are conditional, ‘obligations’ to other kin should perhaps not be considered as obligations at all.
The conditionality of ‘obligations’ to kin (other than mothers and perhaps grandmothers) was underscored by evidence from the final stage of our vignette. Our persuasion/dissuasion experiment was followed by an extension to the vignette. Respondents were told that the host had accommodated the claimant, and that the claimant had then behaved in certain ways. We then asked our respondents whether the host should continue to allow the claimant to stay, or should the host evict the claimant? Respondents who answered ‘yes’ (or that they were unsure) in either of the first two stages of the vignette (i.e. in the initial assessment of desert or when faced with positive additional information about the claimant) were told that, when the claimant came to live with the subject, the claimant either (a) sat at home and did not look for work or (b) did not help in the house. Faced with this further information, more respondents said that the claimant should be asked to leave than said that the claimant should be allowed to stay. Conversely, if told that the claimant had (c) looked for work or (d) helped with cleaning and cooking, respondents were almost unanimous in saying that the subject should continue to accommodate the claimant. As in the earlier stages of the vignette, few respondents thought that mothers and grandmothers should be asked to leave, even if they behaved badly. Much higher proportions of respondents thought that the continued accommodation of siblings or cousins should be conditional on their good behavior.
These quantitative data suggest a clear pattern in the norms around kin support among young people in Cape Town. Most young people acknowledge that they have strong (i.e. largely unconditional) obligations to their mothers and perhaps grandmothers. Their ‘obligations’ to other kin are weaker, but more importantly are far more conditional on the behavior and attitudes of the claimant kin. Claims made by close kin can be dismissed as illegitimate if they behave in ways that are considered inappropriate, including if they fail to fulfil reciprocal responsibilities.
The relationship between assessments of desert with respect to the state and kin
Including vignettes on both public and private support meant that we can probe the relationship between norms around each of these. We ran a series of regression models to assess whether assessments of public desert (i.e. of deservingness with respect to financial assistance from the government) informed assessments of kin obligations. The results are summarized in Table 3. For each dependent variable (i.e. assessed public desert and assessed private desert) we ran separate models with controls for the specification of one or both vignettes, and with controls for the characteristics of the respondent. In every case, we found that black respondents were twice as likely to assess a claimant as deserving in one vignette if they had assessed the claimant as deserving in the other vignette. There was a slightly less powerful relationship among our coloured respondents. Among our smaller number of white respondents we could not discern any statistically significant relationship.
Relationship between public and kinship desert.
Each cell reports the coefficient from a different regression model. Standard errors in brackets. Results for white respondents are for small n; not only are the results not significant, but we cannot be confident even whether odds are >1 or <1.
This suggests that deservingness for public support and deservingness for private support are seen as complements not substitutes. Respondents who endorse public provision in the vignette are much more likely to endorse kin support also, and vice versa.
The general literature on the roles of state and kin in providing support often implies that the state has a responsibility to support categories of people who are deserving of public support but not of private support. For example, the responsibility of the state to provide care as well as income for elderly people may in part be due to the normative assessment that working-age adults are no longer responsible for their elderly parents. Our data from Cape Town suggests that this is not the case within this population. Overall, the kinds of people who are considered deserving of public support are the same kinds of people who are considered deserving of support from kin.
This has important possible implications for public policy. First, it renders more likely conflict over private support. Kin are likely to quarrel with each other over the behaviour of dependents and the expectations of breadwinners (as the articles in this issue reveal). Secondly, there appears to be little support for universal provision, whether public or private. There might be broad support for public pensions and private support for the elderly and children, but there is considerable ambivalence about both public and private support for working-age adults. Some working-age adults might be deserving of either public or private support (or both of these), but others are widely considered to be undeserving, typically on account of their behaviour. Given these attitudes, we would not expect wider support for a basic income grant – payable to all citizens – that was unconditional in the sense that it was not dependent on good behaviour. The experimental data are consistent with the widespread ambivalence towards young mothers who receive child support grants. In an unpublished countrywide survey conducted in 2015, more than half of all respondents agreed (or agreed strongly) that ‘young women spend too much of their child support grants on beer or other alcoholic drinks’; only one in five disagreed (or disagreed strongly) (see also Blake, 2018). A third possible implication is that any expansion of public provision is likely to ‘crowd out’ private support, i.e. that private support shrinks when public support expands. Some research suggested that this might be the case in South Africa (Jensen, 2004), and the expansion of public provision in the 2000s was coincident with declining remittances (i.e. financial transfers) between households (Leibbrandt et al., 2013). Trade unions have long implied that there are crowding-out effects in that they have demanded the expansion of public support so that their members do not need to support kin privately.
It needs to be noted that the vignette for private support focused on accommodation not explicit income support, although the provision of accommodation generally entails some additional expense also. It is possible that respondents might see their financial obligations to kin as being reduced if the state provides for them. We did not ask about – or specify a vignette about – the direct implications of public provision (most obviously through an old-age pension) for kin support. It is even possible that ‘obligations’ to mothers or grandmothers are acknowledged more widely than to other (mostly younger) kin in part because older people are more likely to have pension income, which might be shared around, whereas young kin are more likely to be financial dependents.
Conclusion
Across much of Africa, including South Africa, both the welfare state and kinship are in flux. Welfare states are being expanded as the legitimacy of public provision has grown (although much more slowly than advocates of social protection, including foreign aid donors and international organisations, would like). This trend is in part due to the perception that kin support is contracting, i.e. the state should assume responsibilities that kin had previously shouldered but are no longer doing so. This does not mean that kin support has become insignificant. South Africa is distinctive in its combination of both an extensive welfare state – which pays grants or pensions every month for about 30 percent of the total population – and a high private dependency rate, as many people support a variety of close and (less often) more distant kin. The coexistence of selective public provision and widespread private support means that the legitimacy of claims made on either becomes more consequential: Most individuals who receive either public or private support are perceived to have responsibilities of their own, especially with respect to the kinds of behaviour that are considered appropriate (as other papers in this issue demonstrate).
This paper examined evidence from survey experiments on norms around the claims that citizens make on state and kin. The paper does not report or assess evidence on actual behavior. It found, first, further evidence that there is a clear perceived hierarchy of desert with respect to public provision. Poor people of working age are much more likely to be considered deserving of public support if they are not seen to be responsible for their poverty, due to disability, sickness or the demands of care-work. This pattern accords broadly with the actual programmes that exist in contemporary South Africa: Disability grants and child support grants are the programmes that pay grants to working-age adults. The data cannot uncover the nature of the relationship between programmes and attitudes. It is possible that attitudes have aligned with the existing programmes, just as it is possible that the set of programmes was legislated because of popular expectations.
The paper also presented the first experimental data on norms around support for kin. We found that the relationship between the claimant and the subject mattered. Mothers have the strongest claims. Grandmothers have strong claims, except among white South Africans. Fathers have weak claims among black South Africans, but stronger ones among coloured and especially white South Africans – corresponding to the changing realities of kinship. Richer kin have more extensive obligations than poorer kin: ‘Affordability’ matters.
Most importantly, the persuasion and dissuasion experiments showed that ‘obligations’ to kin are shaped strongly by the character and behaviour of the claimant, with the exception of mothers (and to a lesser extent grandmothers). Support for most kin is also somewhat reciprocal, in that claimants who had previously helped the subject had stronger claims and claimants who had declined to help the subject had weaker claims. Obligations to mothers (and grandmothers) tend to be stronger and less conditional than obligations to siblings or cousins. This is important because it underscores the potential for conflict. Eligibility for public programmes (such as the Child Support Grant) is not conditional on the character or behaviour of the recipient. Even if a recipient spends the grant on drink, this is the right of the recipient. The public are more discriminating: Many individuals may be considered as deserving, but some are not. This helps us to understand why there is a high level of public indignation over the supposed ‘abuse’ of some grants (especially the Child Support Grant), as grant recipients’ behaviour is judged negatively. The conditionality of deservingness might prompt indignation with respect to the ‘abuse’ of government grants, but is likely to introduce conflict into kinship relationships, as people decline to support kin because of their judgements over the claimants’ character or behaviour.
The negotiability of kin obligations does not mean that kinship has lost its normative importance. As other scholars have noted, idioms of kinship remain culturally potent across much of Africa, continuing to frame solidarities and mutual assistance between even non-kin (Sahlins, 2011; Bjarnesen and Utas, 2018; Spiegel, 2018). But the character of solidarity has changed.
People – possibly excepting white people – who recognize the legitimacy of a citizen’s claim on the state are also more likely to recognize the legitimacy of a person’s claim on kin (and vice versa). It does not seem to be the case that people with a parsimonious approach to kinship responsibility are more likely to endorse public provision as a substitute, nor that many people view state responsibilities as limited because the family should provide. Rather, young people in Cape Town tend to favour both public and private support for deserving people and to oppose both for undeserving people.
The data analysed in this paper shed specific light on intergenerational relations. This paper adds to the existing literature on kinship norms in South Africa with further corroboration that kinship and family are not simply shrinking, or shrinking uniformly, but are rather changing unevenly. Faced with many demands by kin (and others) for usually scarce resources, South Africans of all ages are forced to ration their support. Earlier research suggested that grandmothers (and others) prioritise grandchildren, and that obligations to and of maternal kin are often much more important than those to or with paternal kin. This paper shows that, among young people, perceived obligations to mothers (and grandmothers) remain pressing and generally unconditional, whilst support for other kin has become highly conditional and negotiable.
The context of contemporary South Africa contrasts with the societies across the global North where most research on deservingness has been conducted in that South Africans must negotiate with many claimants. Very high unemployment rates (especially among young adults) as well as AIDS-related mortality have resulted in high rates of ‘dependency’. Most South Africans with resources – whether money or (in urban areas) housing – are under considerable pressure to distinguish between who they will and who they will not support. How and why they do so are, however, not so different to the situation in European societies as analysed by scholars such as Van Oorschot and Finch. ‘Need’ itself is not enough to distinguish between claimants. With respect to access to public resources, (racial) identity is not significant (although we did not probe discrimination against cross-national immigrants). But ‘control’ and ‘attitude’ (including reciprocity) are important. Only mothers – and, less clearly, grandmothers – appear to be largely exempt from the negotiability and conditionality of support.
Finally, it is important to bear in mind that the data analysed in this paper was collected from young people. These young South Africans do not think that people like them have automatic claims on either state or kin, despite the unfavourable economic context. As some claimants are deemed more deserving than others, so claims on kin must be negotiated and claims on the state are questioned. Under South African conditions, redistribution intensifies as well as reduces the likelihood of conflict over who gets what, including conflicts between kin.
Footnotes
Acknowledgements
An earlier version of this article was published as ‘The Legitimacy of Claims Made on Kin and State in South Africa’, CSSR Working Paper 416 (Cape Town: Centre for Social Science Research, University of Cape Town). This research was funded in part by UKAid through the UK’s Economic and Social Research Council, grant ES/J018058/1 for the Legislating and Implementing Welfare Policy Reforms research project.
Funding
The author(s) received no financial support for the research, authorship, and/or publication of this article other than from UKAid (see above).