Changes in the Racial Division of Labour during the Apartheid Era¶
Owen Crankshaw (1996). Changes in the Racial Division of Labour during the Apartheid Era. Journal of Southern African Studies, 22(4). Taylor & Francis.
| Type | journal article |
| Scope | Economics and economic history of apartheid |
| Method | descriptive quantitative |
| Period studied | 1965–1990 |
| Regions | South Africa, urban formal sector, Pretoria-Witwatersrand-Vereeniging |
| Themes | labour markets wages, manufacturing industrial policy, unions strikes, education human capital, inequality poverty unemployment, methods data sources, business and state |
| Citation key | crankshaw1996 |
Argument¶
Crankshaw revisits the debate over how far economic growth eroded the colour bar. He argues that earlier estimates of the African middle class diverged because scholars used inappropriate neo-Marxist class schemes and broad official occupational categories. He builds an eclectic occupational classification, combining neo-Weberian and labour-process insights, and applies it to the Manpower Survey data tapes. Advancement into professional work was largely confined to state-employed semi-professionals such as teachers and nurses. Advancement into private-sector routine white-collar jobs was much greater than into the skilled trades, where white unions and the state, together with employer mechanisation and job fragmentation, restricted Africans. The overall pattern is a 'floating' colour bar, in which Africans advance only as whites move up.
Evidence and method¶
Manpower Survey data tapes (1965-1990, c.600 occupations, biennial) reclassified into 11 categories; Manpower Survey 1990 published tables; SAIRR race relations surveys on teacher qualifications; 1979 Fine Spamer Associates report on desegregated facilities; 1985 HSRC case studies of firms (Human, Rainey and Rajah); earlier published estimates of the African middle class (Wolpe, Davies, Charney, Nolutshungu, Simkins and Hindson).
Place in the debates¶
- liberal vs revisionist on economic growth and racial inequality / the colour bar (see The job colour bar, white labour and its erosion). Finds substantial but uneven African advancement, which partly vindicates the liberal view that the colour bar eroded. Still endorses the revisionist 'floating colour bar' idea: whites moved up ahead of Africans, and the racial hierarchy was fractured rather than dismantled.
- size and measurement of the African middle class (see The job colour bar, white labour and its erosion). Attributes the wide divergence in estimates (e.g. 121,948 vs 1,315,800 for 1970) to flawed class theory and coarse official occupational categories. Offers a finer reclassification that yields far lower figures than Wolpe's.
- role of the state, white unions and employers in shaping the racial division of labour (see The job colour bar, white labour and its erosion). The apartheid state expanded African semi-professional jobs by staffing segregated services. White unions, with state and employer compromise, restricted African entry to the skilled trades until the 1980s.
Findings¶
-
African advancement into managerial and professional jobs was negligible by 1990, but substantial in semi-professional jobs. (p. 12)
The African proportion of semi-professional employment grew from 24 per cent in 1965 to 41 per cent, a level that was only 4 per cent less than the white proportion
-
The semi-professional share is driven by teachers and nurses in state institutions, so the state expanded African middle-class employment. (p. 13)
Since most teachers and nurses are employed in state schools and hospitals, these results therefore show that the state played an important role in expanding semi-professional employment for Africans.
-
Teachers, technicians and nurses made up two thirds of semi-professional jobs in 1990. (p. 14)
three occupations therefore account for just over two thirds (67 per cent) of all semi-professional employment.
-
African teachers were far less qualified than white teachers. (p. 14)
By 1982 only 3 per cent of all white school-teachers did not hold at least a post-matric teaching diploma. In contrast 85 per cent of all African school-teachers did not hold such a qualification.
-
The African share of routine white-collar employment doubled between 1965 and 1990. (p. 14)
the African proportion of routine white-collar employment doubled between 1965 and 1990, rising from 15 per cent to 31 per cent
-
Wolpe's 1970 estimate of the African new middle class is badly inflated; recalculated with his own criterion it is one sixth as large. (p. 15)
I arrived at an estimate of 211,282 for the African 'new middle class' in 1971, an estimate that is only one sixth of Wolpe's estimate of 1,315,800 for 1970.
-
African artisans were almost absent until the Wiehahn reforms, after which the share rose from 2 to 19 per cent. (p. 16)
After the Wiehahn reforms in 1979, however, the proportion of African artisans rose from a mere 2 per cent to 19 per cent (Figure 3).
-
Employers mechanised and fragmented the skilled trades in a compromise with white unions and the state, expanding African semi-skilled operative jobs instead of apprenticing Africans. (p. 17)
Instead of apprenticing African workers, employers struck a compromise with white unions and the state which allowed them to mechanise and fragment the skilled trades.
-
Africanisation of front-line supervision followed the fall in white semi-skilled numbers. (p. 18)
the proportion of African employment in the front-line management jobs of charge-hand, supervisor or foreman increased from 13 per cent in 1965 to about 30 per cent in 1989
-
White employment in artisan and routine white-collar jobs declined as Africans entered, which Crankshaw reads as whites moving up (the 'floating' colour bar). (p. 19)
white employment grew from about 437,000 in 1965 to a maximum of 602,000 in 1979 (Figure 5). During this period, the proportion of African routine white-collar workers grew from 15 per cent to 22 per cent.
-
Factory workplaces show almost no racial overlap, unlike offices. (p. 22)
In the case of a motor vehicle manufacturer, 96 per cent of semi-skilled machine operatives were African and only 4 per cent were white. Conversely, all artisans were white.
Evidence ledger¶
Sources used¶
| Source | Kind | Period | How used |
|---|---|---|---|
| Manpower Survey data tapes (biennial Manpower Surveys 1965-1990, ~600 occupations by race and sex) | household or labour survey microdata | 1965-1990 | Author re-analysed the raw data tapes to reclassify ~600 occupations into 11 categories and derive race-specific employment series and shares (Figures 1-10). |
| Manpower Survey 1990: Occupational Information, Report No. 02-01-01 | official statistical series | 1990 | Source for 1990 racial and sexual breakdowns across semi-professional, routine white-collar and skilled trade occupations (Tables 2-4). |
| Summary of the Results of Manpower Surveys No. 13 (1979), No. 14 (1981), No. 15 (1983) | official statistical series | 1979-1983 | Establishes that the Manpower Surveys excluded 'independent' homeland firms and bureaucracies (Bophuthatswana, Transkei, Venda, Ciskei) from specific years. |
| Population Census occupational categories (as used by Wolpe, Davies, Nolutshungu, Charney) | population census | 1960-1970s | Critiqued as too broad (eight occupational groups) to measure African advancement and the basis of divergent middle-class estimates. |
| Interview with Mrs Brelage (Department of Manpower) and Mrs de Jager (Central Statistical Service) | interviews oral history | not stated | Provides information on Manpower Survey sampling frame, sample sizes and the ~90 per cent realisation rate. |
| Survey of Race Relations in South Africa: 1982 | secondary literature compilation | 1982 | Supplies teacher qualification data (3% of white vs 85% of African teachers lacking post-matric diploma in 1982). |
| Race Relations Survey 1988/89 and Institute for Education Planning (OFS) figures for 1991 | secondary literature compilation | 1988-1991 | Used to estimate that 42 per cent of African teachers in the late 1980s lacked a post-matric teaching qualification. |
| Report on Asiatic, Black and Coloured Advancement 1979 | company records or reports | 1979 | Company survey data on desegregated facilities by sector (Table 5) and black supervisors of whites in offices vs factories. |
| Human, Rainey and Rajah, Occupational Mobility and Wage Differentiation: A Qualitative Study (HSRC, 1986) | secondary literature compilation | 1985 | Establishment case studies (bank, supermarket head office, motor manufacturer, factory) showing racial integration in clerical but not manual work. |
| Prior estimates of the African middle class (Charney, Davies, Nolutshungu, Simkins and Hindson, Wolpe) | secondary literature compilation | 1951-1990 | Table 1 estimates are compared with the author's reclassified Manpower Survey figures (e.g. 211,282 vs Wolpe's 1,315,800). |
| Roukens de Lange publications (Manpower Supply and Demand 1992; Employment in South Africa 1984) | secondary literature compilation | 1984-1992 | Cited on survey error in Manpower Survey fluctuations and on the conflation of unskilled manual and service categories. |
| Report of the National Manpower Commission on High-level Manpower in South Africa | commission or official report | not stated | Cited as an example of grouping managers, professionals and semi-professionals into one category. |
| My Career/My Loopbaan 1991 | official statistical series | 1991 | Cited for post-matric certification requirements by professional associations to justify classification of sales occupations. |
New data¶
Author reclassified ~600 Manpower Survey occupations (1965-1990 biennial tapes) into eleven occupational categories by race (and sex), producing new employment series and shares by race for each category, plus the white workforce occupational profile for 1971, 1979 and 1990.
Inference¶
Quantification: Descriptive tables.
Descriptive time-series and cross-sectional comparison of race shares across a reclassified occupational scheme built from Manpower Survey microdata; conceptual critique of earlier estimates; inference about causes (state employment, white unions, labour process) is interpretive, supported by supplementary company survey and case-study evidence.
Key claims¶
| Claim | Basis | Strength |
|---|---|---|
| African advancement into professional jobs was largely restricted to semi-professional occupations (teachers, nurses, priests), pointing to the state's role. (p. 644) | Manpower Survey tapes (Figure 1) and Manpower Survey 1990 report (Table 2); state employment inferred from the fact that most teachers and nurses work in state schools and hospitals. | Partial data |
| The African share of routine white-collar employment doubled from 15 to 31 per cent between 1965 and 1990, much more than into skilled trades. (p. 645) | Manpower Survey data tapes reclassified (Figure 2, Table 3). | Systematic data |
| African artisans were almost absent until after the Wiehahn reforms (2 per cent to 19 per cent), and remain concentrated in the least skilled trades. (p. 647) | Manpower Survey tapes (Figure 3) and 1990 Manpower Survey report (Table 4). | Systematic data |
| Earlier estimates of the African middle class diverged mainly because of inappropriate class theories and broad official occupational categories; the author's reclassification yields 211,282 for 1971, one sixth of Wolpe's estimate. (p. 646) | Comparison of Table 1 estimates with author's re-analysis of Manpower Survey data using Wolpe's criterion. | Partial data |
| African advancement followed a 'floating' colour bar: white routine white-collar and artisan employment fell in absolute terms while whites moved up into more skilled jobs. (p. 652) | Manpower Survey absolute employment series (Figures 5, 6, 10); upward mobility inferred from absence of white growth in lower occupations, with no individual mobility data. | Partial data |
| The lack of African artisan advancement was due to a compromise among white unions, employers and the state allowing mechanisation and fragmentation of trades. (p. 648) | Asserted, referring to the author's separate book (Crankshaw 1996) and not demonstrated in this article's data. | Asserted |
| Routine white-collar workplaces were less racially segregated than manual workplaces. (p. 652) | Fine Spamer 1979 company survey (Table 5) and HSRC 1985 establishment case studies. | Illustrative or anecdotal |
The author's caveats¶
Manpower Surveys exclude agriculture, domestic servants and (from 1979-83) firms in independent homelands; some fluctuations reflect survey error, especially after 1987 when the CSS took over and drew a new sample; supervisor and security series are inconclusive; the data give no insight into inter- or intra-generational mobility; no contemporary data on workplace desegregation; unemployment excluded.
Gaps in the evidence¶
The argument on 'floating' colour bar and white upward mobility is inferred from aggregate occupational stocks, not from individual mobility histories, wages, or cohort data. The causal claims about white unions, employers and the state rest on a separate book and are not tested here with archival or union records. The case-study and Fine Spamer evidence is small and dated; linked firm-level or household panel data (e.g. later October Household Surveys or census microdata) could test integration more systematically.
Replicability: Data described but not public.