Inequalities are stark and obvious in post-apartheid South Africa. How to analyse inequalities, however, is far less clear. In Class, Race and Inequality in South Africa (Seekings and Nattrass Reference Seekings and Nattrass2005), we combined original analysis of quantitative data with critical use of a wide range of secondary historical, anthropological and sociological studies to examine both continuities and changes in the South African social structure over the second half of the twentieth century.

Of the various arguments made in our book, the one that provokes the most criticism is our identification of an ‘underclass’ in post-apartheid South Africa. Focusing on this, Callebert (Reference Callebert2014)Footnote ¹ argues that our analysis ‘assume[s] a fundamental divide in South Africa's economy based on socio-economic exclusion’, ‘fail[s] to capture the many ways in which people cross these divides in making a living’ and has ‘problematic policy implications’. In his account, we offer a ‘bifurcated’ or ‘dualist’ analysis of the South African economy. ‘Access to formal sector jobs’ is the ‘new and fundamental divide that runs through South African society … . In this [that is, our] argument, being a labourer no longer puts one among the lower rungs of society, but is a privilege … .’ For Seekings and Nattrass, he writes, ‘the true socio-economic divide … is between those with and those without access to jobs and other income-earning opportunities’. We are guilty, it seems, of proposing not only that ‘the poor’ should become ‘low-paid labourers’, but that ‘non-unionized low-paid jobs’ should be created ‘at the expense of better-paid unionized employees’.Footnote ²

Callebert's article raises some important points, but it is based on a gross misrepresentation of our argument in Class, Race and Inequality. We do not present a ‘dualist’ or ‘bifurcated’ analysis of South African society. Our analysis of the class structure identified nine or ten classes, on the basis of occupation and earnings from wealth or entrepreneurial activity, not two as implied in Callebert's charge of dualism. We combined these into three (not two) composite categories or strata (see Seekings and Nattrass Reference Seekings and Nattrass2005: 254, Figure 7.1; 337, Figure 9.4). Our ‘underclass’ formed one part of the poorest or most disadvantaged of these three strata. Moreover, this disadvantaged stratum comprised also what we called the ‘marginal working class’ (as well as a residual ‘other’ category) (see also Seekings and Nattrass, Reference Seekings and Nattrass2002: 2–3; Seekings, Reference Seekings2003a, Reference Seekings2003b, Reference Seekings, Bridge and Watson2011). Our analysis thus recognized that many working people (and their dependants) experience disadvantages and poverty in a similar way to the households that we classified as part of the ‘underclass’.

Callebert employs the discursive device of labelling working people as ‘labourers’: ‘When one considers the conditions and pay of many of South Africa's labourers, it does seem difficult to argue that they are privileged.’ Yes, it is hard to argue that ‘labourers’ in marginal or precarious employment – for example, on farms or construction sites – are privileged in any substantial sense relative to the households we included in the ‘underclass’. That is why we did not argue this. As Callebert himself acknowledges elsewhere in his article, our discussion of the ‘underclass’ entailed drawing a contrast primarily with the formally employed, mostly skilled workers who comprise the core membership of most trade unions, not the working poor. It is the unionized school teacher, car worker and even bus driver who are privileged relative to the working poor and underclass (although they are not nearly as privileged as the much richer, upper classes).

Having misrepresented our argument, Callebert proceeds to criticize it using a series of arguments. We allegedly underestimate redistribution from working people to the ‘underclass’, either within households or through remittances between households. We allegedly fail to notice that households employ complex livelihood strategies. We allegedly ignore the ‘structural connections’ between different kinds of households in terms of the political economy of capitalism. And, finally, we assert (allegedly not only without evidence, but also mistakenly) that our supposedly distinct classes have divergent or contradictory interests.

South Africa's poor comprise both working poor and unemployed, and many poor households include both working poor and unemployed people. The poor thus stand to benefit from both job creation and higher wages (unless higher wages result in job destruction). Many unemployed people live in households with working people, and would benefit indirectly if such working people received higher wages. Unfortunately, Callebert seems to have misread our discussion of precisely who comprised our ‘underclass’. The households that we classified into the ‘underclass’ were precisely those households shut out entirely from waged employment. We did not suggest that all unemployed people fall into our ‘underclass’. Our ‘underclass’ comprised ‘people who are not only unemployed in a society where unemployment means poverty but also lack the capital to give them a significant chance of securing employment in the future’ (2005: 298). In the 1993 survey that we used for our empirical class analysis, one half of all unemployed individuals fell into other classes, especially the classes that we labelled ‘intermediate’, ‘core working’ and ‘marginal working classes’. Our analysis recognized that many unemployed people live in the same households as working people.

Callebert's assertion that ‘Seekings and Nattrass do not ask how income is redistributed’ within or between households is incorrect. We emphasized that one half of the unemployed are co-resident with working people. We also discussed remittances transferred between households. We calculated that, in 1993, ‘about 70 percent of the unemployed were dependent primarily on inter- or intrahousehold transfers, almost all from wage-earners’ (ibid.: 291). Even when we limit our analysis to underclass households, one-third of their aggregate income came, in 1993, from remittances (ibid.: 291: Table 8.6). We even provide an analysis of net transfers through remittances between classes (292: Table 8.7). We might have provided more ethnographic detail on these transfers (although Callebert seems to have missed Chapter 6 of our book entirely), but it is incorrect to claim that we ignored them. We also pointed out that remittances accounted for a much smaller proportion of the income of the underclass than public transfers. This gap has become more pronounced over time. Leibbrandt et al. (Reference Leibbrandt, Finn and Woolard2012: 23–4) show that, by 2008, the poor had come to rely overwhelmingly on public transfers and only minimally on private remittances. The direct economic ties between workers in formal employment and ‘underclass’ households have, for the most part, collapsed.

Callebert, citing recent ethnographic research, correctly points out that households – especially poor households – employ complex livelihood strategies. We agree. Indeed, not only did we include profiles of such households in Chapter 6, but our class analysis explicitly recognized that many households – including households in the ‘underclass’ – engage in various forms of informal income-earning activity. In our class schema, households could be classified in the ‘underclass’ as long as their total income from self-employment was less than the value of an old-age pension (an admittedly arbitrary threshold). ‘Underclass’ households were thus defined by their exclusion from wage labour and their meagre earnings from informal business; they were not defined by non-participation in informal business.

Poor households do not employ the same complex livelihood strategies as in the middle of the century, which is the period on which Callebert has himself done important research (focusing on dockworkers in Durban). Callebert seems to underestimate the processes of economic and social change that have transformed Southern Africa over the second half of the twentieth century (changes which occupied most of the first half of Class, Race and Inequality): de-agrarianization (with a steady decline in the proportion of the population enjoying any meaningful access to livestock or crop farming, and related shifts in agrarian norms and social relations), urbanization, the transformation of the structure of employment (with the steady decline in unskilled formal employment, the growth and later decline of semi-skilled work, and the steady expansion of semi-professional, white-collar and skilled work) as well as the raising of the ‘colour bar’, and, finally but crucially, the rapid growth (from the 1970s) in open unemployment. These processes resulted in a massive increase in the economic heterogeneity of South Africa's African population, reflected most obviously in growing inequality from about 1970 in the intra-racial distribution of income.

Does it matter that there are some – albeit diminishing – economic links between working people and underclass households, or that underclass households pursue complex livelihood strategies? Unfortunately, Callebert never presents a clear alternative analysis of the class or social structure of contemporary South Africa, but he seems to imply that formally employed, skilled workers, the working poor and the unemployed are all part of the same class, presumably defined in contradistinction to the capital-owning bourgeoisie. This argument is more plausible with respect to metropolitan townships such as Soweto (outside Johannesburg) or Khayelitsha (outside Cape Town). It is much less plausible when we broaden our gaze to encompass the whole of South Africa. Callebert seems not to have considered the geographical location of the underclass and other marginal classes, shown in Figure 7.5 of our book (2005: 259). In metropolitan areas, in 1993, only one in six households fell into the most disadvantaged of the three strata we identified in our class analysis – that is, either the marginal working class or the underclass (or a residual other class). In rural areas, however, 60 per cent of households fell into this disadvantaged group of classes. The underclass, as we analysed it, was a largely rural class in location. These were not, for the most part, households in Soweto or Khayelitsha that included even part-time working people or successful participants in the informal sector. They were households in rural areas where there was little or no prospect of either wage employment or substantively remunerative informal self-employment (including in agriculture).

Callebert argues that all non-rich households share some structurally similar position, for example because the unemployed subsidize the reproduction of cheap wage labour and function to depress wages. Callebert here makes the argument about ‘cheap labour’ made famously by Harold Wolpe (Reference Wolpe1972) but ignores entirely Martin Legassick's prescient acknowledgement at the time (Reference Legassick1974) that the rural poor were largely surplus to the needs of South African capital. Any suggestion that capital needs, in the 2000s, as many as six or eight million unemployed people is unfounded. Moreover, the presence of massive unemployment has not prevented many formally employed workers securing real wage increases.

Finally, Callebert contests our argument that there is any conflict (or even difference) between the interests of formally employed, mostly urban trade union members and those of the mostly rural, unemployed households that comprise the underclass. At the same time, he dismisses our advocacy of an expansion of low-paid employment. He seems to envisage a future in which the benefits of higher wages paid to formally employed workers will trickle down to the poorest through remittances and through a magical and massive growth in employment. Both parts of his vision are clearly fantasy. The value of remittances has collapsed, as formally employed workers look after their immediate families rather than more distant kin. A high-wage strategy has resulted in job destruction, most obviously in labour-intensive, tradable sectors such as clothing manufacture (Nattrass and Seekings, Reference Nattrass and Seekings2013).

South Africa's poverty and unemployment crises are clearly rooted in a long history of dispossession and disadvantage, as we show in Class, Race and Inequality, but they also have more immediate roots in the collapse of formal un- and even semi-skilled employment. Perhaps the most telling indication of this is the skills composition of the members of unions affiliated to the Congress of South African Trade Unions (COSATU). According to COSATU's own surveys of their membership, the proportion of members in unskilled occupations declined from 30 per cent in 1994 to just 6 per cent by 2008, whilst the proportion in either unskilled or semi-skilled employment fell from 60 per cent to 22 per cent (Bischoff and Tshoaedi Reference Bischoff, Tshoaedi, Buhlungu and Tshoaedi2012: 52). In so far as unskilled employment still exists, it is largely informal. Yet Callebert would have us believe that the trade unions are the organizations that are going to protect the interests of the rural (and urban) poor!

Because Callebert does not provide any clear alternative analysis of the class structure of South Africa, it is difficult to know precisely how he believes that inequality and stratification should be analysed. I suspect that he simply reads the class structure off the capitalist character of the economy, in a crude (and truly dualist) Marxian manner: by which in a capitalist society, there is a capital-owning bourgeoisie, and everyone else – from teachers to the rural unemployed – is in the exploited working class. In Class, Race and Inequality we set out various reasons why the analysis of class should not be limited to the crude Marxian distinction around the employment relationship. Different working people (and their dependants) are in very different social and economic positions. They have (in Weberian terms) very different life chances. We identified an underclass on the basis of ‘systematic disadvantage’ relative to ‘the bulk of the working population’ (2005: 298): our underclass comprised households whose members had little prospect of finding employment or, we might add, establishing other livelihood strategies that yielded more than meagre earnings.

Does Callebert believe that poor rural households who lack social capital, educational credentials, access to credit or land, and experience of waged work, are in a similar class position to urban households whose members have passed matric, have well-paid jobs and ready access to work opportunities, and have savings and can easily raise credit? It seems that he does. He argues that the former would benefit as much or more from paying higher wages to the latter as they would from our proposed strategy of expanding low-wage employment and reducing unemployment. In the kind of fantastical world imagined by Callebert, poor, rural households would somehow benefit more if well-paid urban workers in formal employment were paid an extra R2,000 per month than they would through the provision of new low-wage jobs paying R2,000 per month. Because it is not credible to argue this on the basis of benefits trickling down through remittances, Callebert invokes the argument that any expansion of low-wage employment necessarily undermines the position of currently better-paid workers, causing welfare losses to the ‘working class’ as a whole. In this neat and tidy world, there are no fundamental conflicts of interest between the formally employed and either the unemployed generally or the underclass in particular. It does not matter that trade union membership rates are highest not among the working poor, nor among even semi-skilled and skilled workers, but among workers whose monthly earnings in 2005 were reported to be between six and eleven thousand rand. It does not matter that unions' members are overwhelmingly the more highly skilled, better-paid workers in formal, secure employment, and that the unions have almost no members in the poorest half of the population. The logical extension of his argument is that, because all non-capitalists have the same interests, then the unemployed and especially the predominantly rural underclass can rest assured that all of their interests are being looked after by trade unions.

Sadly, Callebert's fantasy is worse than a fantasy. In the real South Africa, higher wages paid to many employed workers have not trickled down to the underclass. A high-wage strategy has not reduced inequalities, poverty or social exclusion. Employers have inevitably responded by continuing to shift to more skill- and capital-intensive technologies, when they can, or (in many cases) to shut down, when they cannot. The result has been stagnant formal employment, and declining formal employment in labour-intensive, tradable sectors. A central theme in our book was the growth of unemployment, which we believed had not received adequate acknowledgement from either scholars or (in practice) policy makers. Arguments such as Callebert's serve to perpetuate the invisibility of the underemployed, and especially of those unemployed who are chronically excluded from the labour market. Callebert's analysis is dangerous because – whilst failing to present significant original or primary evidence, ethnographic or other – it implies that policy makers do not need to worry about unemployment or exclusion. In this scenario paying higher wages will solve all problems. This is not just nonsense, it is dangerous nonsense.