Cheap labour thesis and the function of the reserves¶
Was racial labour control, including migrancy and the reserves' subsidy to wages, designed to supply and sustain cheap African labour for capital?
Works engaging the debate: 78
Positions¶
- Functionalist cheap-labour and articulation thesis (Wolpe, Legassick, Burawoy, Magubane, Johnstone)
- Qualified or periodised versions: reserves later became dumping grounds for surplus labour
- Critics: the thesis is reductive, ideology or white security mattered more, and 'cheap' labour was not cheap (Dubow, Lipton, Feinstein, Posel)
Evidence base¶
Most of this debate has been argued from a small set of reused quantitative series, official commission reports and secondary syntheses, not from new primary data. The most heavily used series is Francis Wilson's 1972 reconstruction of African gold-mine wages, 1911-1969. It underpins the static-real-wage claim in Massey (1983) (combined with WNLA annual reports), the racial earnings ratios in Wilson (2011) (12:1 in 1936 rising to 21:1), the wage tables in Horner and Kooy (1980a) and Lemon (1984a), and the monopsony claim in Labor Markets in South Africa... (n.d.). Mine labour supply by territory comes from Chamber of Mines, WNLA and TEBA recruitment records. Harington et al. (2004) gives the only systematic annual series, 1896-1996, across about twenty source territories. Massey (1983) adds WNLA foreign-share, desertion and turnover series for the 1970s and 116 Mafeking blacklist records. Jeeves and Crush (1992) uses Chamber labour-supply data for 1933-39 and married-quarters counts (under 1,500 in 1931; 0.7 per cent in 1974). The size of the migrant system and the role of the reserves were estimated indirectly from published population censuses. Nattrass (1976c) and Nattrass (1976b) apply a standard-population residual method to the 1936, 1946, 1960 and 1970 censuses (1970 Report 02-02-02 by magisterial district), arriving at about 1.49 million male oscillating migrants in 1970. Nattrass (1976a) combines census employment counts with Agricultural Censuses Nos. 34, 44 and 46 to estimate net African outflow from white farms, 1951-1970. Stavrou (1988) tabulates agricultural census series on farm numbers, company ownership and tractors from 1951 to the 1980s. For the apartheid period, the annual SAIRR Surveys (Horrell et al. (1972); Horrell (1968)) supply pass-prosecution totals, foreign mine-labour shares and poverty-datum-line comparisons; these figures are themselves compiled from Hansard replies and departmental reports.
Archival work is concentrated in a few case studies. Central Archives NTS and GNLB files, the Chamber of Mines Archive and the Barlow Rand archives underpin (Jeeves (1975a)), (Jeeves and Crush (1992)), (Moodie (2005)), (Breckenridge (1995)), (Duncan (1990a)), (Bonner and Shapiro (1987)) and Edgecombe (1987). Municipal and employer-association records (FCI memoranda, Johannesburg NEAD files) underpin (Posel (1990a)), and DNA files underpin (Evans (1997)). Cape magistrate, trader and WNLA records, together with censuses, underpin (Beinart (1982)). Crush (1982a) draws on Swaziland National Archives file DO9j2. Commission evidence carries much of the interpretive weight: Low Grade Mines and Economic and Wage Commissions in Johnstone (1971), Fagan and Van Eck in Legassick (1974), Tomlinson evidence in Jeeves and Crush (1992) and Roth (1995a). A large share of the literature presents no data at all. This includes the theoretical modelling of Lundahl (1982a) and Curtis (1984), the logical critiques of Kantor and Kenny (1976), (Posel (1983)) and Nattrass (1991), and the reviews (Freund (1987a)) and Nattrass (2006). Only (Heintz (2002)) and Mariotti (2012b) are econometric, and neither tests the reserve-subsidy mechanism directly.
Contested evidence¶
The central functionalist claim is that reserve production subsidised wages below family reproduction cost. It has almost never been measured. Massey (1983) concedes that secondary exploitation cannot be quantified and relies on comparing 1977 wages with the SAIRR household effective level. Nattrass (1976c) estimates that the modern sector saved R393m in housing costs (4 per cent of 1970 GDP), but this is an illustrative calculation from assumed marginal costs. Critics answer with equally thin evidence. Kantor and Kenny (1976) reads van der Horst and the Keiskammahoek Rural Survey to argue that pre-1930s wages were set by the difficulty of attracting labour. Crush (1982a) shows from DO9j2 that Swazi reserves were deliberately sized for subsistence plus 50 per cent population growth, so they were not designed to force labour out.
The Chamber of Mines' costing of family housing for the Tomlinson Commission is read in opposite ways. In Roth (1995a) and Roth (1995b) it shows that stabilisation was not viable and that the mines feared reserve collapse. In the cheap-labour literature it is evidence of the saving that migrancy delivered. Jeeves and Crush (1992) shifts the question from mining economics to politics, using archival evidence of the state's 3 per cent cap on stabilisation. Edgecombe (1987) shows that Natal collieries built a settled family workforce through debt peonage, which undercuts any single-industry logic. The flat Wilson wage series is read as monopsonistic wage suppression by Johnstone (1971) and Labor Markets in South Africa... (n.d.), and as an outcome of class struggle by Massey (1983). Nattrass (2006), reporting Feinstein, argues that cheap labour was not cheap once productivity costs are counted, but the review shows no series. Duncan (1990a) asserts that dividends disprove the mines' claim of a binding cost constraint without presenting dividend figures; Johnstone (1971) relies on the Chamber's own working-cost calculations.
On employer interests, (Posel (1990a)) uses FCI memoranda and the 1956 survey (38 per cent of firms used labour bureaux) to argue that influx control did not serve differentiated capital. Freund (1987a) cites Webster's foundry poll and Bernstein's survey (45 per cent favouring some control) to argue that manufacturers still wanted it. Nattrass (1976a) uses census residuals to show that white farms, not only the reserves, supplied labour, which complicates the dual-sector model. Beinart (1982) uses magistrates' and WNLA records to argue that homestead heads, not only recruiters, controlled migrant wages, and that 1920s rural production recovered alongside mass migrancy.
Underused and missing sources¶
Several sources that could settle points are rarely or never used. These include reserve household production and income series: Simkins' 1918-1969 reserve output estimates are cited only in passing by Nattrass (1991), and Tomlinson's district data are used mainly descriptively. Also missing are mine-level wage books, working costs and dividends, which would test the cost-constraint argument made in Duncan (1990a) and Johnstone (1971). Labour-bureau placement registers would test (Posel (1990a)) and Evans (1997), which reports falsified placement statistics. Pass-law prosecution series by district would also help. No evidence ledger links remittances or reserve output to wage levels over time.
First, published population census tabulations, including population distribution 1904-1960 and 1960/70 tables, would allow the standard-population migrant estimates of Nattrass (1976c) and the farm outflow residuals of Nattrass (1976a) to be replicated and extended by district. Second, the agricultural censuses, 1949-83, could test the shortage-to-surplus and mechanisation claims of Jeeves and Crush (1990) and Stavrou (1988), and the reserve-decline premise. Third, SAIRR Survey extracts for 1961-76 hold the pass-prosecution and foreign-mine-share figures used by Horrell et al. (1972) and Hackland (1980).
Strength of key claims¶
| Basis | Share of key claims |
|---|---|
| Systematic data | 5% |
| Partial data | 52% |
| Illustrative or anecdotal | 24% |
| Asserted | 19% |
Comparative figures for all debates are on the Literature overview.
Best-evidenced works¶
Harington et al. (2004), Massey (1983), Jeeves and Crush (1992), Nattrass (1976c), Beinart (1982), Jeeves (1975a)
Most-used primary sources¶
Holdings are matched by machine against the archive's file manifest; a dash means no match was found, not that the project certainly lacks the source.
Works and their positions¶
| Work | Position |
|---|---|
| Baldwin (1975) | Holds that the ideological goal is real but that the policy serves to maintain white domination and a controlled cheap labour supply. Removals and migrancy are mechanisms of labour control; economic considerations are increasingly important. |
| Beinart (1982) | Accepts the broad framework but says it gives capital and the state too determinant a role. Migrancy and chiefly survival also arose from homestead, chiefly and popular pressures to protect communal tenure and control wages. |
| Bond (2004) | Accepts it and extends it with gender: migrant labour was subsidised by rural African women, and capitalism 'looted' the bantustans. |
| Bonner and Shapiro (1987) | Claims the company sustained pre-capitalist relations more directly than perhaps anywhere else in South Africa, via the estate and chiefly collaboration. |
| Breckenridge (1995) | Rejects the view that the minelords' defence of gold was merely 'cynical hypocrisy' to get cheap servants. He says they acted against their immediate material interest, though by 1931 they embraced devaluation partly to cut the real cost of labour. |
| Cobbett and Nakedi (1987) | Argues migrants and bantustan commuters are the cheapest tier of a segmented labour market, with decentralisation incentives and the avoidance of unions built on that cheapness. |
| Crush (1982a) | Takes a qualified, critical position. Rejects the high-level abstraction of 'mode of production' and the passive treatment of African societies, but accepts a 'de-mystified' articulation framework as a starting point for historical-geographic analysis. Also rejects the universalism of world-systems analysis. |
| Crush (1982a) | Argues against conflating the origins of reserves with their later labour-reserve role. Reserve creation was shaped by local struggles between African, settler, Boer and company interests. Swaziland shows land was not reduced below subsistence in order to force labour out. |
| Crush (1982b) | Rejects world-systems universalism and the high-level 'mode of production' abstraction. Keeps a modified, de-mystified articulation approach that treats African societies' internal dynamics as active. |
| Crush (1982b) | Argues reserves had varied origins and that the functionalist reading fits only some cases. Swaziland shows reserves sized for subsistence and preserving chiefly power, not designed solely to force labour out. |
| Crush and Namasasu (1985) | Follows the structuralist view (Clarke, Legassick) that reserves subsidised wages and reproduced and controlled labour, though it says they were not created by or at the behest of mining capital. |
| Curtis (1984) | Rejects both. Against the neoclassical view, he holds that apartheid is not simply irrational or profit-reducing. Against traditional Marxian analysis, he holds that it is not uniformly functional. Its effects are uneven and contradictory. |
| Curtis (1984) | Accepts that labour-power was cheapened but extends the thesis. The mechanisms carry surplus-value costs, can reduce productivity, and affect migrants and non-migrants, miners and farmers, and men and women differently. |
| De Clercq (1979) | Argues that the dual system guaranteed employers a cheap, racially divided labour force, with concessions to a shrinking white, coloured and Asian stratum. This is a radical, class-based reading of segregation. |
| De Kiewiet (1941) | Treats cheap native labour as the central resource of South African mining and industry, comparable to wool or wheat elsewhere, and links high white wages directly to low black wages. |
| Duncan (1990a) | Rejects the mines' claim that costs prevented better treatment of workers. Dividends show 'viable alternatives to the cheap labour policy' existed, and the mines chose bonded, poorly paid migrant labour because they believed it paid more. |
| Duncan (1990b) | Rejects the mines' claim that their cost structure could not bear better treatment. Dividends show viable alternatives existed, and the mines chose bonded, cheap migrant labour because it paid more. Their propaganda convinced many that there was no alternative. |
| Edgecombe (1987) | Argues that the ideological rigidity of the post-1948 government mattered, but that the changing economic circumstances of the coal trade (domestic orientation, ISCOR/ESCOM) explain why the government 'felt free' to intervene in 1955-56. |
| Evans (1997) | Sides with Dubow: reserves arose from fears of urbanisation and social control rather than a calculated cheap-labour strategy, though they later became labour reservoirs. |
| Freund (1987a) | Rejects Lipton's claim that manufacturers were indifferent or opposed to apartheid labour policies. He argues that wage differentials, racial hierarchy and migrant or 'homeland' labour remained attractive to employers, including in the 1980s. |
| Freund (1987b) | Rejects Lipton's claim. Manufacturers opposed state rigidity over hiring but valued cheap, disorganised black labour, differentials, separate townships and, in some cases, migrant labour. |
| Freund (2009) | Criticises the benevolent view of manufacturing held by Nattrass, Lipton and, in part, Feinstein. He credits Legassick with showing that mining set the pace for manufacturing. |
| Hackland (1980) | Treats foreign mine workers as a cheap labour supply that also keeps up the reserve army of labour in South Africa. Their decline in the 1970s is a political threat, though cushioned by high unemployment. |
| Harington et al. (2004) | Endorses De Kiewiet's view and says the industry needed sufficient, stable labour numbers. It treats resettlement of families on the Reef as economically suspect to owners and politically unacceptable to governments. |
| Harries (1986) | Jeeves (as reported) denies that organised capital designed a cheap migrant system and stresses competition and white protectionism. Harries credits the 'struggle' point but contests Jeeves's liberal assumptions, giving more weight to the magnates' pass laws and wage-depressing measures. |
| Heintz (2002) | Notes that labour-discipline models say influx control lowered the cost of disciplining workers, but that firm evasion, rising White wages and turnover costs undermine the profitability gains. Treats the question as unresolved and stresses stability over profit. |
| Higginson (1992) | Argues that states and employers proletarianized Africans without creating a stable working class, and that the rural areas increasingly could not reproduce industrial labour to employers' specifications. Emphasises the tension between the migrant system and the need for a settled workforce. |
| Hindson (1987) | Defends the materialist thesis as period-specific (late 19th century to 1930s; 1930s to late 1960s). Later industrial opposition to influx control does not show that it was not functional to capital earlier. |
| Horner and Kooy (1980a) | Argues that the reserves began as a subsidy to wages below reproduction cost. They are now better seen as pools of surplus labour, though their precise function is described as unclear. |
| Horner and Kooy (1980b) | Reserves allowed below-reproduction-cost wages in early development. They now function as pools of surplus labour whose inhabitants depend almost wholly on wage labour. |
| Horner and Kooy (1980c) | Holds that reserves allowed wages below reproduction cost and super-profit in early development. They now function more as pools of surplus labour. |
| Horrell (1968) | Documents low African wages, migrant mine labour and wages in border areas below urban rates, but does not theorise; the Black Sash and others criticise the migrant system's effects on family life. |
| Horrell et al. (1972) | Documents low mine and farm wages, a widening white/black wage gap and a migrant system dependent on foreign Africans (75.86% of WNLA labour), but offers no explanatory thesis. |
| Jeeves (1975a) | Accepts that coercive labour legislation sustained cheap migrant labour, but stresses dependence of capital on the state and the low-grade-mine cost logic. |
| Jeeves and Crush (1990) | Holds that white agriculture was a bastion of low-wage labour, sustained by state coercion and by subsidies funded from the gold mines. |
| Jeeves and Crush (1992) | Rejects it as one-dimensional and as confusing cause and effect. The authors say it ignores how workers and rural households shaped the system and that mining capital did not design migrancy as a perfectly profit-maximizing arrangement. |
| Johnstone (1971) | Strongly supports it. Cheap, rightless African labour was the foundation of Rand profitability, and the mining capitalists actively created and operated the wage colour bar. |
| Kantor and Kenny (1976) | Sceptical. Migrant labour is 'not always functional to capital'. Influx control restricts the labour supply and raises its cost. The mines gained but manufacturers and many employers did not, and farmers benefited most from influx control. |
| Kantor and Kenny (1976) | Notes the argument is older than Wolpe and Legassick (Houghton, Macmillan). Van der Horst argued the Reserves raised bargaining power. The authors call the Wolpe thesis internally inconsistent, since a declining subsidy should push wages up. |
| Kanyenze (2004) | Supports it: migrant labour is a 'structured cost advantage for South African industry'. Employers kept hiring and firing flexible at little cost, and host states did not bear the reproduction costs of labour, which fell on rural households in the supplier states. |
| Kraak (1995) | Accepts its insight on cheap African labour and the reserves but criticises it as treating capitalist transition monolithically. Follows Hindson, Morris, Kaplan and O'Meara in stressing differentiated labour power, with an urban stabilised proletariat alongside migrants. |
| Legassick (1974) | Supports a refined version: apartheid sustains the low value of black labour-power by keeping 'necessary wants' low and shifting reproduction costs to the reserves. |
| Lemon (1984a) | Sets out both views and accepts that they often coincide. He sees ideology as paramount but modifiable where it conflicts with economic aims. His conclusion that the new system is an efficient adaptation of apartheid to capitalism leans towards a functional reading. |
| Lemon (1984b) | Takes a middle course: the two coincide in many respects, with ideology modified where it conflicts with economic aims. His conclusion nonetheless stresses the system's functionality for capital. |
| Lemon (1984c) | Notes both views and sides mainly with ideology as paramount, modified where it clashes with economic aims. He accepts that the aims often coincide, and concludes that the new system is 'a highly efficient adaptation' to capitalism. |
| Lundahl (1982a) | Holds that unskilled African wages in the white economy are tied to the marginal product of labour in the reserves, or to pass-law-limited migration. Land alienation and reserve over-crowding therefore reduce the supply price of labour. |
| Lundahl (1982b) | Holds that land alienation and the creation of reserves produced the 'unlimited' labour supply. Rejects the assumption of constant returns to labour in the reserves, since population pressure produced diminishing returns and falling per capita incomes. |
| Mare (1981) | Leans revisionist and structural. Removals follow changes in labour use by industry and agriculture and by state functions serving capitalism. Business silence is read as evidence that the people removed are surplus to production. |
| Mariotti (2012b) | Describes the long-standing preference for low-wage foreign labour, which kept domestic Africans out of mining. Draws on Wilson, Leys and Crush et al. rather than adjudicating the debate. |
| Massey (1983) | Supports the neo-Marxist view that reserve production subsidises capitalist wages (secondary exploitation), but argues that wage levels are set by class struggle rather than the reproduction cost of labour. |
| Massey (1983) | Rejects it: employers "paid as little as possible", and a 'traditional wage' was only a rationalisation; wages were determined by the state of class struggle. |
| Massey (1983) | Revisionist: the economic basis of separate development is to make all black workers 'foreign' migrants in order to sustain low wages and control. |
| Minkley (1990) | Qualifies it. Manufacturers sought cheap African labour, and migrancy cut wage and reproduction costs for many firms. But migrancy's instability and uncontrollability, and workers' rural attachments, undermined productivity. Many 'free labour capitalists' wanted a stable workforce and a break with migrancy. |
| Moodie (2005) | Accepts Johnstone's account of a monopsonistic wage system as 'capitalist collectivism' but adds that it also prevented competition between high- and low-grade mines. Says Johnstone's account of social relations at the point of production needs refinement. |
| Murray (1980) | Sceptical; finds the formulation paradoxical and possibly merely rhetorical, and suspects it carries an 'economism'. Prefers to say that obligations have broken down and the conjugal tie has been burdened. |
| Murray (1980) | Endorses the periphery-core framework as the proper basis, but wants anthropological fieldwork to give it empirical content. |
| Nattrass (1976a) | Agrees with the general thrust but argues the dichotomy is too simplistic. White farms were themselves major labour suppliers to the non-agricultural sector, and the reserves were filled partly by people evicted from farms. |
| Nattrass (1976b) | Holds that the system has benefited the modern sector (no social overhead investment, lower wages, weakened worker organisation) while imposing inefficiencies. Frames the wage effect cautiously as 'it can be argued'. |
| Nattrass (1976b) | Finds that outside mining there appear to be no marked employer preferences for or against migrants. Migrants have broadly similar access to well-paid jobs, except Transkeians and foreigners. |
| Nattrass (1976c) | Presents the case that temporary migration held Black wages lower and weakened organisation as an argument ('it can also be argued'), and lists housing cost savings to the modern sector. She is cautious and does not test the claim. |
| Nattrass (1976c) | Finds no marked employer preference for or prejudice against migrants outside mining, since migrants from most groups obtain high-wage jobs at similar rates. |
| Nattrass (1991) | Notes the hypothesis was reformulated in relative terms, which makes it untestable. She contrasts the radical subsistence-wage view with the liberal supply, demand and bargaining view, and argues that recent work has undermined the early radical presumptions. |
| Nattrass (2006) | Reports Feinstein's view that radicals failed to grasp that 'cheap' black labour was not really cheap, because it undermined the productivity growth needed for employment and output growth. Nattrass presents this approvingly. |
| Nattrass and Seekings (2010b) | Qualified: African labour was 'cheap' relative to white labour and British wages but not by global comparison; industrialisation rested on a combination of cheap and expensive labour. |
| Olivier (1929) | An early statement of the view that land dispossession, taxation and labour controls were designed to supply cheap African labour to mines and farms. He sees the labour 'shortage' as created by foreign capital's demand, not by African idleness. |
| Petterson (1951) | Accepts that the mines, operating on 'small margin of profit', must have a constant supply of cheap unskilled labour, secured by the poll tax and legally limited wages. This is stated as fact rather than argued against alternatives. |
| Phakathi (2012) | Accepts that racial coercion produced cheap labour for the mines, but stresses its contradictions (the 1922 white strike, wage-bill threats, worker resistance) and argues it became dysfunctional and unsustainable. |
| Pogue (2006) | Treats the state, pass laws, hut taxes and the recruitment monopoly as supporting low-wage African labour, with the state aligned to mining capital, following Yudelman's co-optation of white labour. |
| Posel (1983) | Argues the changing conditions of reserve reproduction were at most a necessary condition; Wolpe's own account shows that the United Party alternative was equally viable, so the transition requires political explanation. |
| Posel (1990a) | Rejects both Lipton's claim that manufacturing and commerce opposed migrant labour policy and the Wolpe/Legassick view of full functional fit. Sides partly with Hindson that employer interests were differentiated, but denies that influx control matched them. |
| Posel (1990a) | Employers' preferences were skill-dependent: urbanised workers for semi-skilled work, and cheap, docile 'raw' migrants for the numerically predominant unskilled work. Many employers refused to raise wages and so tolerated shortages. |
| Roth (1995a) | The reviewed book challenges the thesis that mining companies welcomed reserve impoverishment, arguing they feared losing labour permanently to urban areas. Roth endorses this as a demolition of a misconception. |
| Roth (1995b) | As reported by Roth, the book challenges it: the mining industry did not simply welcome impoverishment of the reserves and feared losing its labour permanently if the reserves declined. Roth seems to accept this. |
| Sharp and Spiegel (1990) | Accepts the general link but calls it 'a very blunt instrument'. Following Bozzoli, it argues that male domination also has pre-capitalist and local roots and varies with resources and histories. |
| Sharp and West (1984) | Shows the copper company benefiting from replacing Whites with lower-paid Coloured workers, an example of employers using race-based labour rules to cut costs and tie labour. |
| Sitas et al. (1984) | Cheap labour, sustained by migrancy and the reserve subsistence subsidy, underpinned Natal's industrialisation. After the war it was eroded by reserve debilitation, relocation and mechanisation, though it persists in decentralised areas and in clothing and textiles. |
| Smith (1972) | Supports a cautious version. Industry's quick concession of wage increases in the 1973 strikes and high returns on investment suggest wages were below free-market levels. He notes that labour economics is contested (Horner) and that capital/labour substitution rates are not known. |
| Stavrou (1988) | Largely endorses it. Quotes Legassick and Wolpe that the reserves allowed surplus farm labour to be dispersed and removed cheaply, and treats separate development as entrenching capitalist agriculture. |
| Tatham (1990) | Describes Wolpe's argument that the reserves subsidised labour costs and that apartheid responded to the decline of the reserves' reproductive function. She notes that later Africanist work (Harries, Delius, Beinart) qualified the view that migrancy and segregation were simply capital's creation. |
| Taylor (1978) | Supports the view that the mines sought cheap contracted migrant labour on the assumption of rural subsidy and a recruiting monopoly, and that the Chamber expanded recruitment areas rather than raise wages. |
| Trapido (1970) | Strongly in favour. Depressed African wages produced the surplus that, via taxation, tariffs and railways, financed mining profits, the state and manufacturing, and they attract foreign investors. |
| Truth and Reconciliation Commission, Reparation and Rehab... (1998) | Sides with the revisionist view that business broadly benefited from migrant labour, influx control, landlessness and the repression of unions. It treats business opposition as late, weak and indecisive, and says the mines provided the blueprint for grand apartheid. |
| Van Schoor (1986) | Firmly radical: segregation is a product of the mining revolution and capitalism's demand for cheap landless labour. It is not a Boer atavism or an accidental departure from Cape liberalism. |
| Wilson (2011) | Sides with the revisionist view: the migrant labour system, developed on the diamond and gold mines and later the model for apartheid urban labour policy, kept black wages static in real terms and generated poverty in the reserves as it generated wealth in the cities. |