Farm labour: mechanisation, labour shedding, wages and labour control¶
What caused farm mechanisation and the shift from labour shortage to surplus and shedding, and how were farm labour, wages and paternalism regulated?
Works engaging the debate: 44
Positions¶
- Habakkuk-type labour scarcity and wage push
- Policy-induced factor-price distortion via subsidised capital (Van Zyl, Binswanger)
- Labour control motives (Marglin-type; De Klerk)
- Post-1994 minimum wages and ESTA as job-destroying vs protective
Evidence base¶
The debate has been argued mainly from official serial statistics that cover only white-owned (and Asian and Coloured) farms. The backbone is the Agricultural Census, also issued as the Report on Agricultural and Pastoral Production. Hendrie's compendium of the 1950-74 census tables, (Schirmer (2004))'s provincial machinery counts for 1945/46, 1960, 1976 and 1979, and Africa (1976)'s implement and capital-formation series for 1936-74 all rest on it. So do (Stavrou (1988))'s farm-number and company-ownership tables for 1960-80 and the regional employment series in Vink and Kirsten (2001) and Roberts (2009). Population censuses for 1921-1970 supply the African farm population. Simkins (1984b) interpolates them, against Agricultural Census employment and the 1937-39 Native Farm Labour Committee wage benchmark, into a 1923-69 series of labour supply, employment and per capita cash-and-kind income. Nattrass (1976a) uses the same censuses to estimate residual out-migration of more than 1.5 million Africans from farms between 1951 and 1970. Knight (1976b) dissects census coding changes and undercounts for 1946-70. Price-side evidence comes from the Abstract of Agricultural Statistics and related sources. These include capital formation growing at 4.0% a year against employment at 0.71% a year over 1950-80 in Van Zyl et al. (1987a). Kirsten et al. (2007) uses domestic and border price series to build nominal rates of assistance, peaking at a five-year average of 31% in 1980-84. Liebenberg (2012b) rebuilds Fisher output and input indexes for 1910-2010 from ledger files and the 1968-94 Survey of Agricultural Mechanization Sales.
On the micro side the debate leans on a small number of farm surveys. The most important is de Klerk's 1982 survey of 61 Western Transvaal maize farms, with recalled data for 1968-81 (De Klerk (1984c)) (De Klerk (1984a)) (De Klerk (1984b)). Others are the Department of Agriculture maize production cost surveys of about 80 farms on a rotation basis, used by Van Zyl et al. (1987a), and the Albany series: Roberts in 1957, Antrobus with 81 farms in 1977, and Roberts with 40 farmers in 2008 (Roberts (2009)). There are also the Natal postal wage survey of 1974/75 set against 1962/65 (Standish (1976)), the 2003 HSRC sample of 64 farmers and 64 workers (Atkinson (2007)), and Nelson's survey of 108 sugar farmers relayed in Budlender (1984). Labour-control arguments draw on qualitative archives. These include Native Affairs Department and Native Labour Bureau files (Bradford (1990)), Natal Agricultural Union minutes and magistrates' files (Mazower (1990)), Native Recruiting Corporation records (Crush (1993)), Farmers Weekly letters (Schirmer (2004)), and Bantu Affairs files in Weenen (Sato (2012)). Post-1994 claims rest on StatsSA/NDA employment series, the October Household Survey and the Nkuzi evictions survey (Vink and Kirsten (2001)) (Atkinson (2007)) (Roberts (2009)).
Contested evidence¶
The same Western Transvaal survey supports opposite readings. De Klerk (1984c) and De Klerk (1984a) compute that combines were cheaper on nearly all farms above about 250 tonnes. They conclude that a rural labour push, scale and foreign technology drove adoption, and that tax write-offs and state policy were at most minor causes. Schirmer (2004) reads farmers' stated reasons in the same survey as evidence of a control motive. Kirsten et al. (2007) cites de Klerk in support of tax-induced combine adoption, against de Klerk's own conclusion. The factor-price distortion case in Van Zyl et al. (1987a) and Van Zyl et al. (1987b) rests on a declining labour cost share in the cost surveys and on aggregate capital-to-employment growth. The link to tax concessions and subsidised credit is asserted, not tested, and Vink and Kirsten (2001) relays it second-hand from Townsend. Africa (1976) uses national census series to argue that regular employment did not fall despite heavy post-war tractorisation, and that labour shortage was not the prime cause. Schirmer (2004) reads the 83% rise in tractors in the same period as labour substitution. Stavrou (1988) likewise finds no aggregate displacement before the 1970s. The timing of the shift from shortage to surplus also divides authors. Simkins (1984b) shows employment falling from over 90% to below 80% of labour supply after about 1955 and argues the Agricultural Census undercounts farm residents thereafter. Jeeves and Crush (1990) asserts, without data, that mechanisation cannot explain the shift. Standish's Natal survey treats shortage as a product of low wages (Standish (1976)). Knight instead regresses the SAAU 1969 shortage survey on regional census earnings for 34 regions and finds no clear wage link (Knight (1976b)). The employment series themselves are disputed: Hendrie (1976a) shows casual counts swinging between 592,000, 770,000, 328,000 and 779,000, and Knight (1976b) finds the census, the SAAU survey and the population census mutually inconsistent. Removal totals range from SPP's 1.13 million from white farms (Hallett (1984)) to up to 1.4 million (Mare (1981)) and at least two million (Mabin (1991b)), and none of these states its method. Productivity evidence cited for distortion or inefficiency is also contested. Lipton and Lipton (1994) relies on the Thirtle index. Williams (1996) recomputes it and finds a drought-year base effect. Liebenberg (2012b) cuts multi-factor productivity growth for 1945-2010 from 1.84% to 0.86% a year. On post-1994 regulation, (Roberts (2009)) and Atkinson (2007) attribute casualisation and evictions to the BCEA and ESTA on the basis of farmer self-reports and the Nkuzi survey. Yet Roberts's own Albany census figures show labour costs falling from 43% to 17.6% of current expenditure between 1957 and 2002, and Vink and Kirsten (2001) documents a secular decline dating from long before the legislation.
Underused and missing sources¶
The decisive gap is farm- or district-level panel data. Liebenberg (2012b) notes that census microdata are no longer accessible. Rarely or never used anywhere are Land Bank and cooperative loan ledgers, machinery price series, and Margo-era records of tax write-off claims; these would measure the user cost of capital. Also unused are the 1968 labour tenant registration lists (Sato (2012)), labour bureau placement registers and NRC district recruit counts linked to farm origin (Crush (1993)). Payroll records and worker-side household data would test the wage and paternalism claims in De Klerk (1984b) and Atkinson (2007). Post-1994 Labour Force Survey and QLFS panels, Department of Labour inspection records and court eviction records, used with a control-group design, would test the effects of minimum wages and ESTA instead of relying on farmer opinion.
Strength of key claims¶
| Basis | Share of key claims |
|---|---|
| Systematic data | 10% |
| Partial data | 58% |
| Illustrative or anecdotal | 21% |
| Asserted | 11% |
Comparative figures for all debates are on the Literature overview.
Best-evidenced works¶
Simkins (1984b), De Klerk (1984c), Liebenberg (2012b), Kirsten et al. (2007), Roberts (2009), Knight (1976b)
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 |
|---|---|
| Africa (1976) | Rejects labour shortage as the prime driver. Stresses the availability of implements, postwar prices and terms of trade, and industrial demand for raw materials. |
| Africa (1976) | Holds that regular farm employment stayed roughly constant, so mechanisation had no large displacement effect. It notes this implies underemployment, and notes that Coloured and Asian workers did leave the farms. |
| Africa (1976) | Treats the 1969 shortage as regional and its causes as obscure, and suggests surplus labour exists on white farms. |
| Atkinson (2007) | Draws on all four. Accepts the revisionist account of how a cheap, unfree labour force was created (1860-1980). Criticises it for neglecting paternalism and the psychological dimension, and holds that farms were never fully 'total institutions'. |
| Atkinson (2007) | Attributes the decline to several causes: subsidised capital, consolidation, abolition of labour tenancy, and tenure and minimum-wage laws after 1994. Warns against attributing it to mechanisation alone and treats the relative weight of push and pull factors as unresolved. |
| Atkinson (2007) | Argues that minimum wages and ESTA had perverse effects, raising transaction costs, cutting in-kind benefits and prompting pre-emptive evictions. She prefers incentives to rights-based compulsion. |
| Atkinson (2007) | Sees it as a deeply unequal but real source of social capital and a 'micro-welfare system'. Argues that ESTA misread its voluntary, normative basis, while recognising its coercive and disempowering side. |
| Beinart (1994) | Large farmers shed resident workers to gain economies of scale and to pre-empt labour legislation, minimum wages and unionisation. Protective legislation has itself contributed to a further phase of extrusion. |
| Binswanger and Deininger (1993) | Argues that mechanisation of large farms was induced by subsidised capital and cheap credit, tenancy restrictions and the threat of reform, not by labour scarcity. |
| Bradford (1990) | Rejects the view that 1948 was a turning point. Argues that pre-1948 Pact and UP regimes already intervened on behalf of large farmers with coercive labour schemes that apartheid inherited and expanded. |
| Bradford (1990) | Argues farmers' main demand in the 1930s was not influx control but state help in procuring foreign immigrants. Large farmers were not uniformly short of labour; the shortage fell on smaller farmers and labour tenants. |
| Budlender (1984) | Labour availability and control are the main determinants. Sugar farmers said labour availability mattered more than relative factor prices. Mine wage increases in the 1970s are linked to the Western Transvaal and Natal shifts, with farm size and capital access also important. |
| Budlender (1984) | Mechanisation displaces workers indirectly, through farm consolidation, attrition, loss of casual reserve labour and substitution by women. The cost is therefore hidden and felt in the reserves. |
| Budlender (1984) | Notes Young's questioning of the economic advisability of Anglo's mechanisation. Argues the gap between invention and innovation reflects capital access and scale rather than technical unavailability. |
| Budlender (1984) | Rejects a Luddite position. Technology can ease arduous work, but under current economic and political conditions it is chosen for farmer benefit, with harmful effects on health, skills, ecology and employment. |
| Clynick (1990) | Treats the post-war white 'maize revolution', with mechanisation and capitalisation, as the main spur to dispossessing labour tenants. It places the emergence of a landless black proletariat in the 1950s rather than in the 1930s. |
| Cobbett and Nakedi (1987) | Attributes it to mechanisation, drought and indebtedness, but also to workers leaving intolerable wages and conditions, rejecting a purely 'voluntary' reading. |
| Crush (1993) | Yes. Against Lemon, Greenberg and Lipton, Crush argues that farmers' unions, progressive and unprogressive alike, fiercely defended the protected labour supply, partly out of fear of losing skilled workers and of mine wage competition. |
| Crush (1993) | Qualifies Lipton. The progressive-dominated SAAU was unwilling to override the unprogressives or to accept mine demands, so the progressive/unprogressive split does not explain the state's response. |
| Crush (1993) | Treats the timing as secondary to the general shift from labour shortage to oversupply across the 1960s-80s. Notes Marcus's two-phase view and does not settle the dispute. |
| De Klerk (1984a) | Takes farmers' labour-shortage complaints seriously but concludes mechanization was chiefly driven by foreign technology and economies of scale, stimulated by farm enlargement and men's reluctance to take farm jobs; so a rural labour push rather than an urban labour pull. |
| De Klerk (1984a) | Finds state policy (tax write-offs introduced 1977, removals, labour control boards) at most a minor cause; mechanization and consolidation were rather an informal, perhaps unintended, element of relocation. |
| De Klerk (1984a) | Argues farmers sought greater control over the labour process, and that the loss of strength-based 'skills' weakened the bargaining position of reserve communities, though remaining machine operators gained marketable skills. |
| De Klerk (1984b) | Finds that causation between mechanization and loss of permanent workers per family ran in both directions (an urban labour pull and a farm push). It argues that mechanization raised wages but lowered family incomes and intensified seasonal poverty. |
| De Klerk (1984b) | Rejects it for this sample: there was effectively no relationship between workers per family and the permanent wage rate. |
| De Klerk (1984c) | Accepts farmers' labour-shortage claims only partly. Concludes the 'rural labour push', with economies of scale and new foreign technology, outweighed the 'urban labour pull'. Gives state policy and relocation little weight, though real wages rising and real interest rates falling in the 1970s are treated as contributing factors. |
| De Klerk (1984c) | Argues mechanization was partly a way for farmers to gain control, by using smaller teams, women, children and resident families, and by undercutting the bargaining position of reserve communities. |
| De Klerk (1984c) | Treats farm mechanization as an informal, unintended extension of the relocation programme. Finds influx control did not always immobilize farm workers. |
| De Klerk (1990) | Holds that cheap credit, negative real interest rates and the 1977 tax write-off promoted labour-substituting machinery after 1970. Judges that some over-mechanization has probably occurred, and sees little further capital-for-labour substitution ahead. |
| Hallett (1984) | He ties the move from 1950s labour shortage to 1970s surplus to 1960s mechanization, together with influx control, which blocked the usual rural-urban shift. |
| Hendrie (1976a) | Documents the post-1969 decline in regular, casual and domestic farm employment and the fall in farming units, but offers no causal explanation. A follow-up analytical paper is promised. |
| Jeeves and Crush (1990) | Sceptical of the conventional mechanization account. It says mechanization was unevenly introduced and 'cannot explain all of it', and points to other causes such as rural radicalism and state policy. |
| Kirsten et al. (2007) | Links combine-harvester adoption in maize to preferential tax treatment and says it coincided with the peak of forced removals of permanent farm labour. Wage-driven or labour-supply explanations are not tested. |
| Knight (1976a) | Farm wages were depressed by occupational and geographic restrictions on mobility. Reported farm labour shortages reflect excess demand at low wages, and the pass laws kept labour on white farms despite urban wage differentials. |
| Knight (1976a) | Finds a large employment response to wage increases in sugar and timber and considerable scope elsewhere. Notes that tractors in the post-war period replaced oxen rather than labour, and that wage rises can both cause and be caused by mechanisation. |
| Knight (1976b) | Holds that post-war mechanization (tractors) raised output but did not reduce absolute farm employment. Sugar and timber show large labour-shedding responses to wage rises, so wage-induced mechanization is likely if farm wages rise. |
| Liebenberg (2012a) | Attributes the post-1961 fall in farm employment mainly to mechanization of harvesting and capital-favouring policy, noting labour legislation after 1992-97 accelerated labour shedding; cites Van Zyl et al. on depreciation tax incentives and says the continuing decline is not conclusively explained. |
| Liebenberg (2012b) | Attributes the post-1961 fall in employment to mechanisation of harvesting, supported by tax incentives on capital (citing Van Zyl et al.), with minimum wages and security of tenure legislation accelerating shedding in the 1990s. |
| Lipton and Lipton (1994) | Rejects this view. Argues apartheid probably reduced agricultural employment by encouraging mechanization on White farms and suppressing smallholders, so South Africa 'deagriculturized' prematurely. |
| Mabin (1991b) | Argues that the state wanted to eliminate labour tenancy but struggled to do so because the agents of removal were the landowners, not officials. White farmers were not unanimous in supporting the change. |
| Manona (1984) | Holds that farm migrants, being landless, migrate permanently and treat town as home, unlike homeland oscillators who build rural homesteads and retire there. |
| Manona (1984) | Treats mechanization, land rationalization, restrictions on stock and share-cropping, and evictions as push factors. Also stresses voluntary pull factors (independence, wages, schooling) and the role of kin networks. |
| Mare (1981) | Attributes the move to a smaller settled workforce plus seasonal contract labour to mechanisation and agri-business consolidation of farms. State policy runs alongside this. |
| Mazower (1990) | Builds on Morris's reading of urban migration as a form of class struggle rather than an impersonal urban pull. It differentiates what is being resisted (landlord exploitation, intra-family exploitation) while also stressing real urban wage and job opportunities. |
| Mazower (1990) | Criticises Morris's state-centred approach for glossing over farmworker actions and local variation. It argues that analysing local enforcement shows relatively little state action in the 1940s. |
| Mazower (1990) | Adopts the 'hidden struggles' literature (following White) but insists on differentiating targets of resistance, and treats migration as an ambiguous form of resistance. |
| Mazower (1990) | Argues that the Smuts government did not intervene because agriculture's weight waned relative to manufacturing and mining and because the NAD favoured better wages, whereas the Nationalists, who had strong agrarian interests, only gradually moved towards organised agriculture's demands. |
| McCann (2001) | Argues that labour loss to the mines pushed farms toward low-labour maize and that wheat farms mechanised under labour costs. He also stresses subsidies from the mine-based economy for white farms, so both labour and state factors matter. |
| Nattrass (1976a) | Argues for a combination. Pull from labour scarcity in non-agricultural jobs drew White, Coloured and Asian workers out. Push from farmer eviction of squatters and labour tenants, and substitution of casual or migrant labour, worked mainly on Africans. |
| Nattrass (1976a) | Rejects both simple readings. The controls let the state allocate African labour between competing fractions of capital, within the long-run constraint that farm employment must fall. |
| Nattrass (1976a) | Supports Wilson's claim of increasing reliance on migrants once regional patterns are accounted for, using 1970 gross flow estimates. |
| Roberts (2009) | Sides with the economic school (Antrobus 1984). Labour shedding in Albany is explained by relative profitability and the price-cost squeeze, though the author calls Marcus's purely political reading 'not an objective view'. Legislation is seen as accelerating inevitable adjustments. |
| Roberts (2009) | Albany evidence is read as consistent with the competitive-model prediction of disemployment, especially of unskilled labour, though confounded by ESTA and the price-cost squeeze. Conradie's small-effect findings for Western Cape grapes are noted as contrasting. |
| Roberts (2009) | Finds casualisation in Albany (regular share 61% in 1957 to 38% in 2002), attributed mainly to lower transaction costs of casuals under the BCEA/minimum wage. This contrasts with Conradie's Hex River evidence of no legislative effect. |
| Roberts (2009) | Farmers moved to cash largely for administrative convenience before and after the 10% deduction cap. The author is sceptical of Shabodien's 'hallmarks of feudalism' claim, treating it as exaggerated. |
| Sato (2012) | Holds that tenant evictions were less orchestrated by the state than black-spot removals and were often carried out by individual farmers, especially after the mid-1970s. |
| Schirmer (2004) | Sides with Marglin: labour control was a significant, sometimes independent motive, made possible by state protection from competition. He does not claim that cost was irrelevant. |
| Schirmer (2004) | Stresses continuity: paternalism and the conflict it produced persisted throughout 1940-1980, rather than giving way to wage relations. |
| Schirmer (2004) | Combines both: cheap state credit and marketing-board prices made machines affordable, while farmers' dissatisfaction with labour supplied the motive. He argues that the 1970s shortage was minor and that farmers over-reacted. |
| Schirmer (2004) | Argues that improved productivity was often an unintended consequence and that control-driven mechanisation reduced productivity gains, citing World Bank and Thirtle et al. on modest growth. He does not test this himself. |
| Sharp and Spiegel (1990) | Links mechanisation, population growth and changed payment forms on white farms to the removal of dependants and relocation to Qwaqwa. |
| Simkins (1984a) | Accepts the argument that white farmer pressure helped bring in the labour-bureau/influx control system, but notes that labour demand levelled off within a decade, making it unnecessary to farmers. |
| Simkins (1984b) | Cites Morris's argument that White farmers pressed for it. Adds that within a decade farm labour demand had levelled off, making the system unnecessary to farmers. |
| Simkins (1984b) | Attributes it both to wage differentials and, very importantly, to state resettlement under the 1936 Land Act anti-squatting provisions, aimed at areas of 'surplus labour' and rising black/white ratios. |
| Standish (1976) | Argues the shortage is created by low farm wages. Industrial attraction is a problem only because of those wages, and shortage can be remedied only by raising pay past a motivating threshold. Wages are not the only factor, since hours, family residence and plots of land matter too. |
| Standish (1976) | Rejects farmers' claim that higher wages raise turnover and cut labour supply. The share of farms finding labour adequate rose consistently with wages. Turnover rises with pay only up to a threshold, which is explained by workers working until they have saved enough for a stay in the Homelands. |
| Standish (1976) | Contrasts negligible real farm wage growth with 8.44 per cent per annum real growth in mining since 1965, and argues that the sector's claimed shortage cannot be remedied by such small increases. |
| Stavrou (1988) | Presents both. Leans towards the Marxist view that machinery raises capital intensity and management control over labour. Finds that mechanization has so far not caused large aggregate labour displacement, but De Klerk's maize study shows sharp declines in harvest labour. |
| Stavrou (1988) | Cautions against overstating labour bureaux; the continued application of the pass laws by the white community is what mainly kept farm workers from leaving. |
| Van Zyl et al. (1987a) | Attributes the post-1970 fall in labour per cultivated hectare mainly to mechanization of harvesting. The authors say tax concessions, interest-rate subsidies and general agricultural subsidies accelerated it, with urbanization a contributing factor. They stress factor-price distortions over labour-supply or political explanations. |
| Van Zyl et al. (1987a) | Argues capital is scarce and unskilled labour abundant in South Africa, so subsidized capital produced over-mechanization that departs from the economic optimum. It recommends policy reform towards labour-using technology. |
| Van Zyl et al. (1987a) | Finds real cash and in-kind pay both rising, with cash faster. Real total farm pay grew faster than average real income per worker nationally over 1970-1985, which cuts against the claim of stagnant farm wages, though levels are not analysed. |
| Van Zyl et al. (1987b) | Stresses capital subsidies (tax write-offs, subsidised interest) and mechanization of harvesting, with urbanization as a contributing factor. Argues some farms are over-mechanized relative to South Africa's factor endowments. |
| Van Zyl et al. (1987b) | Shows an early complementarity phase (1945-70) followed by substitution after 1970, using an Ishikawa-curve framework and elasticities of substitution. |
| Van Zyl et al. (1987b) | Finds real cash and in-kind pay rising, with cash growing faster. Says farm pay growth compares favourably with national and black-worker income growth, and cautions about comparing farm and other wages. |
| Vink (2012) | Argues the post-1970 employment decline after combine harvesters was more probably due to fiscal policies lowering the cost of capital than to the marketing boards. Boards' price support may have raised earlier employment. He calls for more research. |
| Vink and Kirsten (2001) | Relies on Townsend (1997): a close relationship between the machinery/labour ratio and relative factor prices after 1950. Policy distortions (credit, tax, R&D allocation) drove labour-saving technical change, and the removal of subsidies in the 1980s partly reversed it. |
| Vink and Kirsten (2001) | Suggests, mostly on anecdotal evidence, that labour and tenure laws raise farm labour costs; admits the employment effect is unclear because the cost of capital has also risen. |
| Williams (1996) | Criticises the Binswanger-Deininger-Feder schema as evolutionist. Argues that state interest-rate, tax and pricing policies in the 1970s encouraged combine harvesters and labour displacement, and that mechanization initially raised labour demand. |