First Published:
17 Mar 2026, 3:56 pm
Details
Reviewed by
Astrid Wood, Alan Mabin, Matthew McKeever, Nomonde Gwebu and Owen Crankshaw
Book Review Forum: Urban inequality: Theory, evidence, and method in Johannesburg
Crankshaw Owen, Urban Inequality: Theory, Evidence, and Method in Johannesburg, Zed Books, 2022. Paperback £26.99, Hardback £85.50, Ebook £29.59, ISBN 9781786998958.
Introduction
Astrid Wood – Newcastle University, UK
Owen Crankshaw’s Urban Inequality: Theory, Evidence and Method in Johannesburg is a major contribution to debates on labour market restructuring, urban inequality and post-Apartheid social change. Building on more than two decades of research, the book substantially extends arguments first developed in Crankshaw’s influential 2008 article in Urban Studies, offering a comprehensive and empirically grounded account of how inequality has evolved in South Africa’s largest city-region. In so doing, it challenges some of the most widely accepted assumptions in urban theory while providing a thoroughly researched longitudinal analysis of Johannesburg.1
The book develops conceptual, methodological and empirical contributions that together advance both urban studies in general as well as understandings of inequality in Global South cities. Conceptually, Urban Inequality offers a sustained critique of social polarisation theory and the post-Fordist city. While Johannesburg appears to be a paradigmatic case for these theories of labour market inequality – it has experienced deindustrialisation, maintains an abundant supply of low-skilled labour and provides little protection for the unemployed – Crankshaw demonstrates that the expected hollowing out of middle-income employment has not occurred. Instead, the labour market has undergone pronounced professionalisation with strong growth in middle- and high-income non-manual occupations. Inequality has concurrently deepened through persistently high unemployment among low-skilled workers. For Crankshaw, low levels of educational attainment for many remains a key factor to explain unemployment. The book thus reframes urban inequality as a divide between those who are employed and those who are excluded from formal employment altogether rather than as increasing wage dispersion among the employed. This conceptual move unsettles dominant theoretical frameworks and suggests that debates over polarisation and professionalisation have often obscured the central role of unemployment in shaping inequality in post-Apartheid cities.
Methodologically, the book makes two important interventions. First, it argues for analysing inequality at the scale of the city-region rather than within municipal boundaries alone. By extending the empirical frame beyond the City of Johannesburg to encompass Gauteng province, Crankshaw demonstrates that employment change and occupational restructuring are shaped by functional linkages across the region such as job relocation and commuting patterns. Second, the book offers a rigorous critique of conventional measures of segregation, particularly the dissimilarity index, showing how they can misrepresent change over time. By developing alternative measures, spatial mapping and longitudinal analysis, Urban Inequality provides a more robust methodological framework for examining segregation in contexts of rapid demographic and occupational change.
Empirically, the book provides one of the most comprehensive accounts to date of labour market change, racial inequality and spatial restructuring in Johannesburg from 1970 to 2011. Drawing on population census data and statistically representative labour force surveys, Crankshaw shows that the growth of professional and managerial employment outpaced population growth among White South Africans thereby enabling substantial occupational mobility among well-educated Black South Africans. This in turn drove significant residential desegregation in formerly White neighbourhoods where Black residents formed a majority by 2011. The book also documents the persistent inequality rooted in unemployment and spatial exclusion especially in the southern parts of the city. The result is a nuanced account of the tensions between social change and structural constraints in the post-Apartheid city.
The commentaries that follow engage with Urban Inequality from complementary perspectives. Alan Mabin situates the book within South African urban scholarship, emphasising its meticulous empirical foundations while raising questions about spatial scale, data limitations, and the political implications of the analysis. Matthew McKeever highlights the book’s broader theoretical significance for urban sociology, particularly its challenge to Global North theories of inequality, while offering a critical discussion of its engagement with positivism, causation and theory-building. Nomonde Gwebu reflects on the book’s implications for understanding contemporary urban inequality in South Africa and its relevance for future research. In his response, Owen Crankshaw clarifies his theoretical and methodological positions and reiterates how his findings challenge established accounts of urban inequality and racial change. Together, the commentaries and response underscore the importance of Urban Inequality as both a definitive study of Johannesburg and a wider intervention in debates on how inequality is conceptualised, measured and explained in cities of the Global South.
Commentary I
Alan Mabin – University of the Witwatersrand, South Africa
Eighteen years ago, Crankshaw (2008) published an innovative article in Urban Studies that argued both that professionalisation rather than social polarisation, and a significant degree of desegregation, characterised Johannesburg’s post-Apartheid employment and consequent spatial trajectories. Urban Inequality, published 14 years after the article, extends the research, data, methods, argument and conclusions in four major ways. In the first part, the book engages more substantially with theories of inequality, providing an important critique of shortcomings in the international literature. The second part moves to the case of the largest city-region in South Africa: it goes beyond the borders of the City of Johannesburg and adds the whole of Gauteng province (to which Johannesburg is central but it also includes Pretoria and much more) to the analysis of employment change and of inequality. The earlier work took the story up to 2001, while the book adds new data from the 2011 census. It also develops new forms of compiling a spatial view using mapping as well as other data techniques. And the book also adds much more detail in support of the key arguments made in 2008, rather than amending those arguments and conclusions.
In writing about Johannesburg’s northern suburbs over a decade ago I made central use of Crankshaw’s earlier work – describing it as ‘meticulous’ (Mabin, 2015: 409). Although I did not add the point at that time, I do think that feature arises from the author’s original studies being in pure sciences before he turned to sociology. The book is probably even more meticulously researched and written than his earlier work.
Exploring changing relationships among employment categories, and incomes, Crankshaw argues that it makes no sense to examine only the City of Johannesburg, which includes around a third of the population of Gauteng province. There is a data challenge in doing this, but it is in my view reasonably resolved for the period up to 2011 by the use of data for 23 ‘magisterial districts’ which together make up an area rather close to that of Gauteng province – at least for the province in 2011. I like Crankshaw’s argument for the larger scale though I might phrase it differently – ‘Johannesburg is functionally part of all the cities in Gauteng’ (p.28) which I take to mean that the diverse parts of the province are all so closely functionally linked that employment and division of labour shifts will best be seen only at larger scale, since ‘employment trends [in one part] could … be caused by changes within the province, such as the movement of certain kinds of jobs from Johannesburg to Pretoria’. Also, Gauteng is far removed from the other large cities in the country (Cape Town and Durban in particular) – though there is a limit to that side of the argument for places such as Rustenburg, Emalahleni (Witbank) and Sasolburg which are perhaps as intimately connected to the rest of the larger city-region. Perhaps it is a surprise that Crankshaw does not engage with ways in which other actors define the ‘Gauteng City-Region’, in particular the Gauteng City-Region Observatory (GCRO – for example see https://www.gcro.ac.za/about/the-gauteng-city-region/).
Crankshaw shows that among those employed there has not been social polarisation in the sense of a shrinkage of middle-income earners but rather an expanding in less well-off and much better-off categories. Deindustrialisation in Gauteng has not had that result, a point probably of global interest, with the rapid growth of service-related middle income and high-income jobs. But huge growth in the proportion of people unemployed means a new form of social polarisation – that between all those employed and all those unemployed with the second huge and growing part of the population facing immiseration. How non-formal work affects that situation is a matter of considerable debate in South Africa (e.g. Financial Mail, 2025; Lehohla, 2025). But evidently the biggest social issue facing the country is that of polarisation between the unemployed and the employed and its connection to education. Crankshaw’s later views and conclusions in this field would also be of significant interest.
How does all this affect the social geography of the city? That is the question posed in the second part of Urban Inequality. Yet there is very significant shift in the geographical scope of the work at this point: from the use of data for the whole of Gauteng and not only the City of Johannesburg in the first part, the second reverts to a focus on the latter alone. I do think there is a potential problem with that, for as recent GCRO work shows it is in other extensions of Gauteng, to the north of the province or city-region, and perhaps further afield, that the full concern of polarisation between the employed and those without regular employment reaches the extremes (Mosiane et al., 2025).
An important argument in Urban Inequality concerning how segregation may be measured, suggests that the familiar and often applied dissimilarity index is deeply flawed and can misrepresent results in relation to change over time: it appears under many circumstances to exaggerate the extent to which segregation persists (pp.163–184). While Hamann (2022) is critical of Crankshaw’s view here: I find the book’s argument convincing. That relates to the sometimes controversial but empirically well supported conclusion that Crankshaw reaches, that ‘the extent of racial desegregation in Johannesburg’s formerly whites-only neighbourhoods is substantially more than is generally accepted’: the African, coloured and Indian population reaching 56% in 2011 (p.16). At the same time the book emphasises that the deepening area of inequality and indeed of degrees of social if not purely racial segregation in the city, is that between the zones of exclusion in the south of the city which are overwhelmingly black and where unemployment is the key social issue, and the zones of relative privilege in the northern areas, where the growth of middle income non-manual employment has driven racial desegregation.
In several already published reviews, each of which provides a synopsis of Urban Inequality, some suggestions were made for extension of the work. Most of those reviews provided little critical evaluation (e.g. Ballard, 2023; Hamann, 2022; Mulibana, 2025). Hamann’s (2022) review suggested that bringing the analysis up to the more recent present and extending the spatial analysis to include the rest of Gauteng would potentially explore new directions of change both in aspects of inequality and in spatial segregation. A key difficulty with this first suggestion is the quality of the 2022 census. Local small scale data from that census has not been published except for police station precinct data – which in at least some cases suggests that the census may have been wildly inaccurate (e.g. in the police precinct where I live in Cape Town which has considerable visible population growth, the published result suggests a decline of several thousand between 2011 and 2022; in the Midrand precinct in the northernmost areas of the City of Johannesburg the police precinct data suggests a decline of over 100,000, which is clearly nonsense). As the leading scholar of inequality and segregation in Johannesburg since the demise of formal Apartheid, I will be intrigued to know how Crankshaw might suggest the analysis be brought closer to the present.
A more critical review of Urban Inequality by Cooper (2023) – which includes a detailed and readable synopsis of the arguments of the book – makes a number of comments to which a response by the author would again be valuable. Cooper writes that ‘there are important political implications of these occupational class structure transformations since the 1980s, which require further debate in South Africa’ (Cooper, 2023: 110). These implications include, according to Cooper, weakening of the trade union movement that was so important to the end of formal Apartheid and achievement of electoral democracy. Cooper also calls for greater integration of the analysis into emergence of a ‘global knowledge based economy’ in recent decades – that might of course demand attention for the decade and a half since the effective 2011 end of the data analysed in the book. Cooper also wants more attention to ‘capitalist agency’ in the changes that have occurred (Cooper, 2023: 114).
Finally, and this is beyond what one expects Owen Crankshaw to do himself, Cooper suggests that ‘it would clearly be most valuable over the next few decades for “southern” scholars to produce case studies of urban inequality in many more cities of countries of the South’ (Cooper, 2023: 122). I hope this small contribution to discussion of Urban Inequality adds to the momentum for exactly that. Certainly the book itself should do so, as ‘an excellent and rigorous book that is of great value to South African urban scholars and urban inequality researchers more generally’ (Ballard, 2023: 269).
Footnote
1. Although urban and spatial inequality are presented as interchangeable urban refers to a more narrowly defined area, city, town or neighbourhood (excluding rural or countryside settings). Spatial refers to a more loosely defined or an undefined unit of space.
References
Ballard R (2023) Post-Fordism and urban inequality: The case of Johannesburg. City 27(1–2): 267–269.
Cooper D (2023) Transformations of race and class in South Africa: Discussion article of book on urban inequality: Theory, evidence and method in Johannesburg by Owen Crankshaw. South African Review of Sociology 53(1): 104–123.
Crossref Web of Science Google Scholar
Crankshaw O (2008) Race, space and the post-Fordist spatial order of Johannesburg. Urban Studies 45(8): 1692–1711.
Crossref Web of Science Google Scholar
Financial Mail (2025) ‘South Africa’s informal economy is becoming a giant.’ 28 August. Available at: https://www.financialmail.businessday.co.za/fm/features/cover-story/2025-08-28-south-africas-informal-economy-is-becoming-a-giant/ (accessed 24 February 2025)
Hamann C (2022) Book review: Urban inequality: Theory, evidence, and method in Johannesburg. Journal of Asian and African Studies 57(8): 1690–1691.
Crossref Web of Science Google Scholar
Lehohla P (2025) ‘Debating the Labour Force Survey – A Response to Fourie’s Critique.’ Available at: https://iol.co.za/business-report/opinion/2025-06-17-dr-pali-lehohla-debating-the-labour-force-survey–a-response-to-fouries-critique (accessed 24 February 2025)
Mabin A (2015) In the forest of transformation: The ‘northern suburbs’ of Johannesburg. In: Harrison PGotz G, Todes A et al. (eds) Changing Space, Changing City: Johannesburg after Apartheid. Wits University Press, pp. 395–417.
Mulibana L (2025) Book review: Urban inequality: theory, evidence, and method in Johannesburg. South African Geographical Journal 107(1): 132–133.
Mosiane N, Murray J, Peberdy S, et al. (2025) Landscapes of Peripheral and Displaced Urbanism. Gauteng City-Region Observatory.
Commentary II
Matthew McKeever – Haverford College, USA
This is a book that lives up to its title: a theoretical objective, pertinent qualitative and quantitative evidence, and an innovative methodology that allows the author to advance debates over the nature of inequality in the largest urban area in South Africa. It delivers key insights on all fronts, and is essential reading for any sociologist or geographer examining spatial inequality.
This book matters for two reasons. The first is that it elucidates an empirically based explanation of contemporary inequality in Johannesburg. South Africa is the most urbanised country in Africa, the most economically developed, and the most unequal. These ‘stylised facts’ make clear that cities are key to understanding inequality, and no city in the region is more important than Johannesburg. The book examines the changing contours of inequality there through a focus on residents occupations, thier prime source of income for residents. Using data from the census and other representative sources starting from 1970, Crankshaw documents the long-term trends of a dramatic rise in high- and middle-income non-manual work, and the comparatively lower gains in both middle-income manufacturing jobs and low-income jobs that require little education. What he shows is, first, that in the post-Apartheid era those with good education stayed employed by moving into the newly created non-manual jobs. Second, that these trends persisted across race. Both the well-educated White workforce as well as those in other population groups who had the necessary education were able to move into these expanded non-manual occupations. And third, that the lower growth in middle- and low-income jobs raised unemployment for those workers with less education, who are disproportionately Black South Africans (following the book, the term ‘Black’ here refers to all population groups except Whites).
These findings have important implications for understanding shifts in racial inequality. Advantageous non-manual jobs expanded more rapidly than the White population, which meant that qualified Black workers could move into them. Though many did not have the required education, it meant that racial inequality could decline, albeit less than might have been predicted following independence. There are some interesting parallels here to theories of educational inequality, which examine the declining relationship between class or race and educational attainment as universal schooling becomes more widespread. Using the theory of Effectively Maintained Inequality (Lucas, 2001), for example, one might expand on this analysis to look for changing patterns of inequality within occupation. If Whites are no longer are the majority group in certain higher status positions, might they instead shift to controlling the market for better jobs within an occupation?
There are some data limitations in this part of the analysis. The data only represent a portion of the South African labour market. The ability of workers to migrate, not just South Africans but southern Africans in general, means that any urban analysis will only capture a portion of the story of national inequality. One other limit is that earlier data, particularly those collected by the Apartheid-era state, are less trustworthy. Having worked with some of the census data from the 1970s and 1980s, it is clear that they failed to capture all population groups equally. The analysis is designed in a way that minimises these concerns, about which there is little one could do, and the causal story of occupational inequality relayed in the book clearly explains what others studying the country have described.
Later chapters tie these processes explicitly to the spatial order of the city, showing how economic development and housing policies led to historically specific relationships between jobs and housing. Crankshaw compares the northern suburb of Sandton, the central city area, and the southern suburbs of the city (Soweto and others), demonstrating how contours of economic possibility do not follow the same pattern found in other post-industrial cities. There has been less concentrated poverty despite the rise of unemployment because those holding high- and middle-income non-manual occupations have not uniformly fled urban neighbourhoods. Reinforcing this, housing policy did not lead to a concentration of inexpensive housing in less economically developed suburbs. Combined, housing policies and economic restructuring have led to patterns of inequality in Johannesburg that are distinct from those found in other global cities.
This ties into the second reason this book is essential reading, which is that case studies should also advance a theoretical understanding of urban inequality. Most theory on this topic is built on studies of a small set of cities in the Global North. Yet the presumption that New York or London might tell us all we need to know is absurd. Looking at how a post-Fordist economic shift has changed urban inequality in a society not central to the literature lays bare its theoretical deficiencies. As Crankshaw argues, much of this research has argued that cities have experienced concentrated inequality because of increasing social polarisation that resulted from post-industrial labour markets experiencing job growth at both poles of the earnings distribution. An alternative perspective in the literature discussed in the book is that increased professionalisation in the labour market has led to workers with higher education doing well, while others are left unemployed or competing for ever fewer low-income positions. These perspectives are set at odds to one another, as they both attempt to explain an outcome of increased urban inequality through different foci. In my reading, what Crankshaw’s research shows is that these distinct perspectives are talking past each other rather than in conflict in how they explain changes in inequality. While the studies taking the latter perspective make an argument based in economic shifts and policy changes that are historically specific, those arguing for the former are often not doing so, and thus cannot account for the causes of changing inequality. This book shows that one must sufficiently attend to place, documenting inequality and constructing an empirical understanding of what political and social forces are shaping it, in an explicitly theoretical way. The book succeeds because theory is not a distinct objective in the book, set aside in an introductory or concluding chapter, but rather intertwined with methods in every chapter. In arguing for a sociology that is tied to the particular historical setting and politics, the book expands our understanding of urban inequality more generally.
There are some other theoretical claims that I found less convincing. The book at times makes the argument that problems in the literature are partially due to a positivist orientation, but I would disagree that the story is actually that complicated. The book itself uses differing critiques of positivism. On pages 11–12, Crankshaw offers a definition of positivist research as that which focuses exclusively on the empirically measurable features of society, with the explicit rejection of theory and conceptual categories until after the data are in. If positivism is like grounded theory, but for more quantitative folks, few urban sociologists would agree that is what they are doing, as it is an approach that has been subject to critique within sociology for decades (see, e.g. Parsons’ (1937) critique of utilitarian theory). It is alternatively hinted that positivism is behind the improper use of statistics. On page 10, Crankshaw writes that ‘statistical associations … are used as evidence for the existence of causal mechanisms’, a critique made again on page 159. Such an argument is related, but fundamentally different. Research that involves ‘the casual use of statistics…’ (p.196) in claiming causality is inherently problematic, and again this is a critique that has been made for decades regarding the use of statistics in sociology (see, e.g. Freedman, 1991). Insofar as previous work in urban inequality has argued that statistical correlation alone shows a causal relationship, it should be rejected as bad science. But it is important to separate out a critique of scientists from a critique of science. The solution to lazy work is better research from more clear-eyed researchers.
In the end, I would argue that this book is important because it is an example of such work. In using different types of empirical evidence to verify when a causal explanation makes sense, the book shows what the hard work of doing good social science looks like. It does not extrapolate from one city to the world, nor does it simply use national statistics to create causal explanations. The real theoretical claim here is that social change, in particular changes to the occupational distribution and housing policy, will shape inequality in an urban area. And, if you want to know how any particular area is changing, you have to go and look at how those forces are creating change in that particular area. New York cannot explain Johannesburg any more than it can explain Lagos or Beijing. The book does an excellent job in exposing this weakness in much of urban sociology, and thus advances both the empirical and theoretical understanding of inequality.
References
Freedman D (1991) Statistical models and shoe leather. Sociological Methodology 21: 291–313.
Crossref Web of Science Google Scholar
Lucas S (2001) Effectively maintained inequality: Education transitions, track mobility, and social background effects. American Journal of Sociology 106(6): 1642–1690.
Crossref Web of Science Google Scholar
Parsons T (1937) The Structure of Social Action. McGraw-Hill.
Commentary III
Nomonde Gwebu – University of the Witwatersrand, South Africa
Urban Inequality aims to ‘provide new insights and causes of trends in racial inequality that go beyond the analysis of state policy alone’ (p.13). The work swiftly satisfies this aim, and proceeds to apply its data, methodology and comprehensive reflections to go well beyond this initial intention. The research contributes to a robust understanding of how urban inequality ought to be conceptualised and measured, which, as the notoriously most unequal nation in the world is of particular interest in the context of South Africa. The evidence presented offers a credible definition of urban/spatial inequality, a concept that is often applied in vague and inconsistent ways.
The term racial inequality or racial segregation is widely accepted to describe racial segregation coupled with the privileging of one race over another (or others; p.169). The concept of urban/spatial inequality, although similar, includes a geographical aspect, while still referring in some sense to people, and therefore needs further clarification.
While Cole et al. (2018) and Turok (2018) apply spatial inequality to describe the unequal distribution of public resources and amenities, Beier (2023) uses the spatial/urban inequality framing to describe a more generalised poverty, of which poverty of housing is a part.
When the term racial inequality is used, the unit of measurement is clearly the individual, and it is generally understood that what is sought is a measure of access to public spaces, resources and amenities (p.169). While the concept of urban/spatial inequality is evoked in a similar way, it is less clear what precisely is being measured and to what end (p.26). While scholars may achieve internal consistency in their own research, there is a need to define an overarching essence of what research really intends to grapple with when measuring urban inequality. An initial understanding of the subjects which are either equal or unequal is a good place to begin.
In South Africa, high- and middle-income earners are responsible for the purchase or rental of their own housing, while low-income earners may become eligible for State assistance. Strictly speaking, equality would be achieved by the State if it delivered housing that was equal to the market offering, in the same way that equality of water means equal access to water of the same quality. The housing market being divided in this fundamental way makes absolute spatial equality an untenable situation, and therefore renders its opposite, spatial inequality, somewhat empty without further clarification.
Urban Inequality presents the missing units of measurement that the term has been in search of, namely, a measure of the relative race, occupational class and geographical movement of individuals over a specified timeframe. As one of its profound achievements, the study explicates an impactful definition of the term urban/spatial inequality. Crankshaw presents his findings in a relatively ‘dispassionate’ tone, while also rejecting simplistic positivist attributions of causation (p.11). Instead Crankshaw views ‘statistical correlations as the object to be explained rather than as evidence for a hypothesised causal mechanism’ (p.196).
Existing literature frames the challenge of overcoming Apartheid’s spatial legacy as persistent in various ways. Huchzermeyer and Karam (2006) ask Informal Settlements: A Perpetual Challenge? while Charlton (2018) laments the State housing response mechanisms in Confounded but Complacent: Accounting for How the State Sees Responses to Its Housing Intervention in Johannesburg. It is a small step from the notion of a stubborn challenge, to the rhetoric or outrage driven conclusion that there has been no notable change at all, as seen in the title The New Apartheid (Mpofu-Walsh, 2021); or, by Thomas Picketty, who claims there has been a regression in desegregation between 1994 and 2015 (Allison, 2015); and others who draw parallels between post-Apartheid South Africa and the colonial era (Farkash, 2015). Crankshaw argues that certain accepted urban inequality measurement methodologies obscure results and should not be used to compare segregation in cities with different proportions of races in the population.
Crankshaw’s submits that:
stronger growth of high-income, middle-class jobs and middle-income, non-manual jobs benefitted well-educated workers of all races. The result was the dramatic growth of the black middle class, on the one hand, and extremely high unemployment rates among poorly educated African and coloured workers on the other. (p.15)
The development of the housing market for the high-income, middle-class has continued well beyond the researched 1996–2011 timeframe. While the national average property price was R700,000 in 2015, properties within estates in previously white neighbourhoods, like Sandton, had an average sale price of R2 million (De Kock, 2016). These contexts have created a ‘labour market spatial mismatch between the excluded ghettos and the new job opportunities increasingly found in Sandton’ (p.15).
The geographical concentration of private housing for high-income, middle-class residents has also driven the morphology of housing for the unemployed who ‘have been concentrated to a greater extent in the erstwhile racial ghettos of the southern suburbs … under conditions of extremely high unemployment, these racial ghettos have therefore been transformed into ghettos of exclusion’ (p.149).
The data presented in Urban Inequality confirms the view that ‘professionalisation’ best describes how the restructuring of the labour market has affected the patterns and distribution of income in Johannesburg. Professionalisation proposes that ‘poverty among the poorly educated inner city black residents was caused by growing unemployment, which was caused, in turn, by the decline in all manual jobs in the inner city’(p.3). The theory of social polarisation on the other hand submits that ‘poverty among inner city black residents is due to their displacement from middle income manual jobs and their concentration in a growing number of low-skilled and low-paid service jobs’ (p.166). The data does not show an increase in the volume of low-skilled and low-paid service jobs; the theory of social polarisation therefore cannot apply (p.5).
The research presents a nuanced, data-led rendering of the changing nature of social inequality in Johannesburg over the last 20–40 years. While the book does engage with existing debates and their key arguments; its distinguishing feature is its ‘focus the on the logical connections between method, evidence and theoretical descriptions’ (p.11). The dispassionate presentation of the data makes a significant contribution to the socio-politically charged domain of inequality research in South Africa. The book shows that while weaving moral judgements into the telling of the story of inequality may intend to give a face to the data, or to humanise the study; these judgements risk shaping the way the data is collected and processed; or else obscuring the nature, interpretation or value of the findings.
References
Allison S (2015) Thomas Piketty’s South African Conundrum, Business Maverick. Available at: https://www.dailymaverick.co.za/article/2015-10-04-thomas-pikettys-south-african-conundrum/ (accessed 28 January 2026).
Beier R (2023) Why low-income people leave state housing in South Africa: Progress, failure or temporary setback? Environment and Urbanization 35(1): 111–130.
Crossref Web of Science Google Scholar
Charlton S (2018) Confounded but complacent: Accounting for how the state sees responses to its housing intervention in Johannesburg. Journal of Development Studies 54(12): 2168–2185.
Crossref Web of Science Google Scholar
Cole M, Bailey R, Cullis J, et al. (2018) Spatial inequality in water access and water use in South Africa. Water Policy 20(1): 37–52.
Crossref Web of Science Google Scholar
De Kock P (2016) Estate living on the rise in South Africa, April 2016.
Lightstone Property. Available at: https://www.lightstoneproperty.co.za/news/Estate_Living_on_the_rise_April_2016.pdf (accessed 24 February 2026)
Farkash A (2015) The ghosts of colonialism: Economic inequity in post-Apartheid South Africa, Global Societies Journal 3: 12–19.
Huchzermeyer M, Karam A (2006) Informal Settlements: A Perpetual Challenge? UCT Press.
Mpofu-Walsh S (2021) The New Apartheid. Tafelberg.
Turok I (2018) Worlds apart: Spatial inequalities in South Africa, In: Smith M (ed.) Confronting Inequality. Jacana Press.b, pp. 159–182.
Author response
Owen Crankshaw – Bristol University, UK
Before responding to the reviewers, I would like to summarise the important findings of my study and how they challenge our existing knowledge of urban inequality. The case of greater Johannesburg provided an excellent test case for the social polarisation theory because it met the three conditions that were argued to be necessary for social polarisation, namely (i) deindustrialisation, (ii) an available supply of low-skilled workers and (iii) limited unemployment benefits for workers. However, the findings show that the workforce has undergone professionalisation alongside very high levels of unemployment among low-skilled workers with limited unemployment benefits. These findings are therefore an important challenge to theories of social polarisation and professionalisation. They suggest that the absence of these three conditions does not cause professionalisation. Instead, the professionalisation of the workforce is probably caused by the mechanisation and automation of manual work, which reduces the demand for low-income and middle-income manual workers. The result is much more employment growth among high-income and middle-income non-manual jobs than among low-income manual jobs, hence the professionalising pattern of growth.
This study also has implications for the widely held belief among many scholars that racial inequality in South African cities has not changed substantially. The implication then, is that the features of the Apartheid city have not changed because the causes of Apartheid remain intact. In specific terms, these scholars argue that the formerly whites-only neighbourhoods have not been substantially desegregated, and that there has been no improvement in the living standards of black (coloured, Indian and African) residents with the result that racial inequality has not changed substantially.
This study of Johannesburg demonstrates that these arguments are fundamentally false. Although it is true that there are high levels of poverty among black residents, this poverty is caused largely by unemployment, which in turn was caused by employment trends that are not the direct result of racial discrimination during Apartheid. It was the slow rate of employment growth of manual middle-income and low-income jobs that was the main cause of unemployment among black workers who had not completed high school. Granted, Apartheid education policy was an important reason why more black workers did not complete high school. But this is the historical effect of an Apartheid policy that interacted with the subsequent slow employment growth of middle- and low-income manual jobs. If circumstances were different and employment in manual jobs had grown more than in other jobs, the result would have been very low levels of unemployment among black residents and therefore much less poverty in the post-Apartheid period.
Although it is also true that white residents have largely retained their high standard of living, this has not prevented black residents from being employed in better paid jobs and living in the formerly whites-only neighbourhoods. The reason for this was that there was a dramatic growth in employment for high-income professionals, managers and technicians. But because the white population did not increase in size, 86% of all these new high-income jobs went to well-educated black residents. By 2011, about 65% of all high-income middle-class jobs were held by black workers, a substantial change in the form of racial inequality since the Apartheid period.
An important result of the growing black high-income middle class is that many of them have moved to live in formerly whites-only neighbourhoods in Johannesburg’s inner city and suburbs. The result was that, by 2011, white residents were in a minority of only 44% in the formerly whites-only neighbourhoods. Again, this high level of racial desegregation represents a substantial erosion of the Apartheid geography of Johannesburg.
Alan Mabin makes the important point that I could have presented employment trends for the City of Johannesburg as well as for the Gauteng Province as a whole. My initial statistical analyses were conducted solely on the City of Johannesburg, but I was concerned that these results would be criticised because they could be caused by residential movement in and out of Johannesburg. I therefore chose to include statistics for the whole of the Gauteng Province, a decision that was bolstered by my finding that the employment trends by occupation were much the same in both the City of Johannesburg and Gauteng Province. In retrospect, I could have presented employment trends for both areas.
Nomonde Gwebu reminds us that many scholars still argue that Johannesburg’s racial geography has not changed substantially since the Apartheid era. One reason for this belief is that much research on racial residential desegregation has relied on the dissimilarity index to measure the extent of change in racial residential segregation: these studies all show that the dissimilarity index has changed only slightly, and scholars have therefore concluded that the extent of racial residential desegregation is not substantial. As I stated above, these findings contradict my results, which show that by 2011, just over half (56%) of residents in the formerly whites only neighbourhoods were black (African, coloured and Indian).
I argue that the dissimilarity index cannot be used to study changes in racial segregation if the racial composition of the city changes over time. Following the work of statisticians, I argue that the dissimilarity index has not declined because the population of black residents has grown so much faster than the population of white residents. Furthermore, growing numbers of black residents also live in the formerly blacks-only neighbourhoods, which have very few white residents. The dissimilarity index measures segregation by calculating the numbers of black residents who would have to move in order to achieve perfect desegregation. So, this growing number of black residents who live outside the formerly whites-only neighbourhoods has largely counteracted the arithmetic contribution of the number of black residents who do live in the formerly whites-only neighbourhoods. The result is that the dissimilarity index has changed only slightly, even though the formerly whites only neighbourhoods have been substantially desegregated.
Matthew McKeever has interpreted my criticism of positivism to mean that previous research has relied too much on correlations because they are embedded in a positivist framework. He argues that my criticism is not relevant because these researchers are not being positivists, just myopic and lazy.
My response is that very few quantitative researchers will admit to being positivists. Instead, their commitment to positivism is implicit and revealed by their treatment of evidence in the study of causation. So, I agree with Matthew McKeever that positivists do not accept that correlations alone are proof of a causal relationship: Positivists argue that for statistical correlations to constitute evidence of causation, a number of conditions must be fulfilled. This position is summarised in popular Sociology textbooks as follows (Babbie, 2013: 93). First, the values of the variables must be statistically correlated. Second, the event measured as the causal variable must precede the event measured as the effect variable. Third, the statistical correlation between the variables must not be a ‘spurious’ correlation in the sense that the correlation is actually caused by one or more other variables. A fourth condition, which is usually implicit, is that the causal relationship measured by the statistical correlation must make sense in terms of a causal mechanism or social theory (Neuman, 2000: 52).
These conditions are perfectly reasonable within the terms of a positivist philosophy of science. But criticisms of this positivist model of causation are not concerned with these kinds of conditions. Rather, it is concerned with the faulty logic of this model of causation. Porpora (2008: 198) argues that this positivist model of causation makes it seem as if the statistics are the important part of the argument, when in truth it is evidence of the qualitative nature of the causal mechanism that makes the causal model logical and persuasive, even though it is tucked away as the final condition of the causal model. He asks us to consider the familiar example that we are taught as an illustration of how the positivist model of causation eliminates spurious causes.
For example, there is a correlation between ice-cream sales and deaths due to drowning: the more ice cream sold, the more drownings, and vice versa. There is, however, no direct [causal] link between ice cream and drowning. The third variable at work here is season or temperature. Most drowning deaths occur during summer—the peak period for ice-cream sales. (Babbie, 2013: 94)
Porpora argues that this example only makes sense because of our knowledge concerning the qualitative nature of the causal mechanisms at work. We are presented with a statistical correlation and the ludicrous logic that this correlation means that there is a causal connection. But the example is only ludicrous because of our knowledge of the qualitative nature of ice cream. We know that ice cream is harmless and therefore cannot cause a person to drown. Without this qualitative knowledge, we could not make a judgement about what causal mechanism could cause the statistical correlation. The same can be said of our choice of the third variable, namely the seasons (or temperature). We chose the ‘seasons’ variable to find out if the correlation between ice cream eating and drownings was spurious because of our knowledge about the qualitative nature of the seasons and how they influence our swimming practices. So, contrary to the argument made by Babbie (2013) and Neuman (2000), the statistical correlations themselves tell us nothing about causation. Instead, it is our knowledge of the qualitative nature of causal mechanisms that informs our understanding of causation. For this reason, critical realists, such as Porpora, argue that statistical correlations are the starting point of a causal enquiry, not the end point. Consistent with the logic of critical realism, this enquiry therefore uses qualitative research methods to discover and describe the mechanisms that might explain statistical correlations. This contrasts with the positivist approach of using statistical correlations as evidence for the action of qualitatively defined causal mechanisms.
Finally, it is worth adding that the positivist model of causation is really a method that measures the number of times that a causal mechanism was activated. It does not provide qualitative evidence that describes the causal mechanism itself. The description of the causal mechanism used in positivism models is usually hypothetical and based on existing knowledge that may or may not be relevant. If the knowledge is relevant, the statistical correlation is not evidence of how the causal mechanism operates, but simply a measure of how often it was activated (Crankshaw, 2014: 512–513).
ORCID iDs
Astrid Wood https://orcid.org/0000-0002-3155-1396
Alan Mabin https://orcid.org/0000-0003-3191-2056
Matthew McKeever https://orcid.org/0009-0000-7869-8232
Owen Crankshaw https://orcid.org/0000-0002-9006-5005
References
Babbie E (2013) The Practice of Social Research, 13th edn. Wadsworth, Cengage Learning.
Crankshaw O (2014) Causal mechanisms, job search and the labour market spatial mismatch: A realist criticism of the neo-positivist method. The Journal of Critical Realism 13(5): 498–519.
Crossref Web of Science Google Scholar
Neuman W (2000) Social Research Methods: Qualitative and Quantitative Approaches, 4th edn. Allyn and Bacon.
Porpora D (2008) Sociology’s causal confusion’. In: Groff R (ed.) Revitalizing Causality: Realism About Causality in Philosophy and Social Science. Routledge, pp. 195–204.
