Labour Supply in the South African Economy and its Implications for Agriculture¶
J.B. Knight (1976). Labour Supply in the South African Economy and its Implications for Agriculture. SALDRU Farm Labour Conference, September 1976, Paper No. 18. Southern Africa Labour and Development Research Unit (SALDRU), School of Economics, University of Cape Town.
| Type | conference paper |
| Scope | Economics and economic history of apartheid |
| Method | descriptive quantitative |
| Period studied | 1946–1979 |
| Regions | South Africa, Homelands/reserves, White agricultural areas, Witwatersrand gold mines, Natal sugar belt, Western Cape |
| Themes | agriculture land, labour markets wages, mining migrant labour, inequality poverty unemployment, bantustans removals influx, demography urbanisation, methods data sources, manufacturing industrial policy |
| Citation key | knight1976 |
Argument¶
Knight asks whether South Africa is running out of relatively unskilled African labour, and what that means for white agriculture. He reconstructs African population, labour force, non-farm employment, farm employment and Homeland underemployment from the censuses and employment statistics, adjusting for census under-reporting and changing definitions. He finds that the data are too weak for firm conclusions, but that 'residual labour' and underemployment in the Homelands appear to have risen in absolute terms since 1951. Mine wage rises after 1971 are explained mainly as an attempt to cut dependence on foreign labour and recruit South Africans, not as the result of a general scarcity. Case studies of mining, sugar and timber show employers responding to higher wages by mechanising. Knight concludes that specific shortages will coexist with general surplus, that pass laws and a fragmented labour market have kept farm wages depressed, and that competition between mining and farming for labour is likely to grow only gradually.
Evidence and method¶
Population censuses 1946-1970 (including unpublished Department of Statistics tabulations), employment statistics, agricultural censuses and Reports on Agricultural and Pastoral Production, the S.A.A.U. 1969 farm labour survey, Sadie's demographic adjustments, Vermaak's HSRC activity rates, Nattrass's migrant estimates, the Economic Development Programme 1974-79, the Tomlinson and Marais commissions, Wilson on gold mine labour, the Chamber of Mines ex-miner attitude survey, the Maree and de Vos Ciskei/Transkei survey, the Froneman committee summary, and firm-level case studies from one mining group, a sugar estate and a timber company.
Place in the debates¶
- Labour surplus vs labour scarcity (is South Africa running out of unskilled labour?) (see Unemployment: causes, measurement and labour market institutions). Sceptical of a general scarcity. Argues that residual labour and underemployment have grown in absolute numbers, that shortages are specific (physically fit young men, particular regions), and that the data cannot settle the question.
- Causes of mine wage rises after 1971 and the fall in South African recruits (see The mine migrant labour system: origins, recruitment, foreign labour and stabilisation). Attributes the rises mainly to concern about dependence on foreign labour (Malawi, Mozambique) and an attempt to recruit nationals. The gold price made them possible but was not their cause. Rejects the mining industry's view of a drying-up labour pool.
- Farm labour market: shortage, wages and pass laws (see Farm labour: mechanisation, labour shedding, wages and labour control). 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.
- Causes of farm mechanization (wage-induced vs technical) (see Farm labour: mechanisation, labour shedding, wages and labour control). 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.
- Backward-sloping supply curve and migrant behaviour (see The mine migrant labour system: origins, recruitment, foreign labour and stabilisation). Treats the backward-sloping supply curve as a widely held view and offers a Todaro-type probabilistic migration explanation of why Homeland underemployment can coexist with fewer South African miners.
Findings¶
-
Census under-reporting of African employment cannot be corrected satisfactorily, so level estimates are very uncertain. (p. 16)
There is no satisfactory way of taking account of census under-reporting of African employees.
-
Homeland population growth far outstripped growth in agricultural production, pointing to rising underemployment in the reserves. (p. 44)
that the absorption of population by the rural 'Homelands far outstripped the growth of agricultural production
-
Manufacturing and mining African earnings diverged sharply after the war, making industry the preferred destination. (p. 54)
those in mining, having been roughly equal before the war
-
Real mine wages were flat for decades, supported by migrant labour, efficient recruiting and monopsonistic behaviour. (p. 56)
Real earnings in mining remained roughly constant from 1951 to 1971, as indeed they had since 1911
-
The mines increasingly relied on foreign recruits; the South African share of gold mine labour fell from over half to under a quarter. (p. 57)
The proportion fell from over half before the war to 22.4% at its lowest, in 1972.
-
With a fixed gold price and collusion among mines, the response to labour needs was foreign recruitment rather than higher real wages. (p. 57)
Given the fixed gold price between 1949 and 1969, monopsonistic collusion among mines in not competing for labour, and a belief in the backward-sloping supply curve, real wage increases were resisted
-
The post-1971 rise in mine wages was aimed at attracting South African nationals because of the political risk of foreign labour dependence. (p. 59)
The increase in mine wages since 1971 must be seen primarily as an attempt to recruit more South African nationals.
-
The mining industry's conventional wisdom, that no pool of involuntarily unemployed men remains, conflicts with Knight's evidence of large and rising residual labour. (p. 60)
It is a commonly held view in the mining industry that there is no significance pool of suitable, involuntarily unemployed male labour remaining in the Homelands.
-
Regular farm workers earned less than both manufacturing workers and miners, though real wages probably rose slowly to 1972/3. (p. 66)
The average earnings (including benefits-in-kind) of regular farm workers were consistently lower not only than those of manufacturing workers but lower also than those of miners (which excluded benefits-in-kind).
-
Pass laws held farm labour in place and did not eliminate farm labour shortages at prevailing wages. (p. 67)
The pass laws have not prevented persistent shortages of farm labour
-
Coal mining moved steadily away from hand-loading, mainly for technical reasons, with wage rises of the 1970s adding to the trend. (p. 75)
have declined from accounting for about 95% of total production in 1955, to 60% in 1965, to 40% in 1975
-
Post-war farm mechanisation (tractors) did not reduce employment; its effects were substitution for draught animals and expanded output. (p. 79)
For instance, the tractor replaced oxen rather than labour.
Evidence ledger¶
Sources used¶
| Source | Kind | Period | How used |
|---|---|---|---|
| Population Census 1970 (Report 02-02-02 and related reports; Single Ages 1941 to 1970, Report 02-05-05, Table D1) | population census | 1970 | Basis for African population, age structure, 'economically active' counts, non-farm and farm employment, unemployment and Homeland population estimates. |
| Unpublished Department of Statistics tabulations of African population of the Bantu Homelands (1970 boundaries) | population census | 1946-1970 | Used for Homeland population, density and masculinity ratios and the adjusted-for-undercount series. |
| J.A. Vermaak, Die Vraag na en Aanbod van Mannekrag in die Republiek van Suid-Afrika, Deel 1 (HSRC, 1974), from unpublished 1970 census data | secondary literature compilation | 1970 | Supplies African age-specific activity rates applied to other census years' populations. |
| J.L. Sadie demographic evaluations and projections (SAJE June 1970; Projections of the South African Population 1970-2020, Industrial Development Corporation) | secondary literature compilation | 1946-1980 | Source of census-undercount-adjusted African population estimates and projections used to adjust the labour force. |
| Economic Development Programme for the Republic of South Africa 1974-1979 (Office of the Prime Minister) | commission or official report | 1973-1979 | Source of official labour force, non-farm and farm employment projections critiqued and then used for residual labour forecasts. |
| Department of Statistics employment statistics (South African Statistics 1970/1972/1974; Bulletin of Statistics) | official statistical series | 1946-1974 | Employer-based employment series compared with census employment by sector and used for growth rates and productivity. |
| Statistics of Houses and Domestic Servants 1970, Report No. 11-03-06 | official statistical series | 1970 | Sample survey of households used to estimate domestic servant numbers (754,000) for comparison with census. |
| Agricultural Census / Report on Agricultural and Pastoral Production (1953/4 to 1972/3, incl. 1958-59, 1968-69, 1970-71) | agricultural census | 1953-1973 | Series on regular, casual and domestic farm employees and earnings, used to assess farm employment trends and reliability. |
| South African Agricultural Union survey, 'The Farm Population and Labourers of South Africa' (1970, mimeo) | employer association records | 1969 | Alternative estimate of regular and casual farm employees, labour shortage percentages by region, used to cross-check agricultural census. |
| South African Agricultural Union, 'Bantu Labour Policy' (mimeo, October 1974) | employer association records | 1974 | Illustrates farmers' stated wage and labour policy and expectations about agreement labour. |
| Natal Agricultural Union, 'Service Conditions of Bantu on Farms in Natal' (mimeo, 1973) | employer association records | 1971-1973 | Cursory poverty datum line comparison and survey of Natal farm wages; recommendation to raise wages. |
| Second Report of the Commission of Enquiry into Agriculture (Marais Commission), RP 84/1970 | commission or official report | 1970 | Gives benefits-in-kind ratio for farm labour and evidence on scope for agricultural mechanisation and labour economising. |
| Tomlinson Commission, Summary of the Report of the Commission for the Socio-economic Development of the Bantu Areas (1955) | commission or official report | 1950-1955 | Source for Homeland incomes, livestock numbers, carrying capacity and the 357,000 farming families used to size productive Homeland agricultural employment. |
| Report of the Native Economic Commission 1930-32 (UG 22-1932) | commission or official report | 1930-1932 | Qualitative evidence of growing impoverishment, land shortage and landlessness in reserves. |
| Social and Economic Planning Council report (1946) | commission or official report | 1946 | Qualitative corroboration that reserves were overpopulated. |
| Froneman Committee on employment of foreign Africans (1963, unpublished; summary by Ken Owen, SAIRR 1963) | commission or official report | 1963 | Estimates of foreign African numbers and employment in 1960 and recommendation to replace foreign labour. |
| Aerial photograph and sample checks of Homeland census counts (Smit 1965 thesis; Schulze 1969; Hattingh 1973) | secondary literature compilation | 1951-1970 | Establish serious undercounting in inaccessible rural Homeland areas in earlier censuses. |
| Jill Nattrass, 'The migrant labour system and South African economic development', SAJE March 1976 (Tables 4, 7, 10) | secondary literature compilation | 1946-1970 | Estimates of temporary migrants from rural areas (635,000 to 1,357,000) and migrant life-cycle used for rest-period and labour-pool calculations. |
| Maree and de Vos survey of four Ciskei and Transkei districts (1965 and 1968) | household or labour survey microdata | 1965, 1968 | Provides direct estimates of underemployment rates (about 22% of labour force) in the Homelands. |
| Francis Wilson, Labour in the South African Gold Mines 1911-1969 (1972) | secondary literature compilation | 1911-1969 | Source of real mining earnings series and gold mine labour numbers to 1956, plus the shortage/abundance phases. |
| Mine Labour Organisations (N.R.C.) Ltd., Reports and Financial Statements (1966-) | chamber of mines or recruiting records | 1966-1972 | Supplies South African versus foreign African gold-mine employment shown in Table 16. |
| Chamber of Mines Human Sciences Laboratory attitude survey (M.M. Masipa, 'Attitudes towards gold mining of ex-miners in other occupations', mimeo) | author survey or fieldwork | c.1973-74 | Shows pay was the main reason for dissatisfaction and what would lure ex-miners back, supporting wage-increase-as-recruitment argument. |
| Redelinghuys, HSRC 1974 study of Tswana work-seekers at labour bureaux (Rustenburg area), 1971/2 | household or labour survey microdata | 1971/2 | Identifies customary rest periods and factors influencing duration of unemployment among migrants. |
| Department of Statistics Bulletin of Statistics, South African Statistics and Agricultural Reports (real earnings series in mining, manufacturing, farming) deflated with urban White CPI | official statistical series | 1950s-1975 | Figure 2 average real earnings of African employees in mining, manufacturing and regular farm work. |
| Official Yearbook of the Republic of South Africa; Department of Agriculture, 1975 Abstract of Agricultural Statistics (Tables 80, 89) | official statistical series | 1960-1975 | Homeland agricultural GDP growth, producer prices and agricultural production volume, used to compare production and population growth. |
| Donald Pursell, 'Bantu real wages and employment opportunities in South Africa', SAJE June 1968 | secondary literature compilation | 1958-1966 | Attribution of 1958-61 and 1964-66 manufacturing wage rises to Wage Board initiatives. |
| P.J. van der Merwe, 'The Bantu in the South African economic system' (mimeo) and 'The economic influence of the Bantu Labour Bureau System', SAJE 1969 | secondary literature compilation | 1951-1969 | Source for Homeland male age-group data (Table 13) and for the view that farm workers evade pass laws. |
| Interviews and case studies of a gold/coal mining group, a large sugar estate and a timber company; author's impressions of S.A.A.U. | interviews oral history | 1972-1975 | Evidence on managerial responses (mechanisation, task sizes, employment cuts) to wage increases and shortages. |
| R.P. Plewman, 'Minerals and Manpower', Journal of the South African Institute of Mining and Metallurgy, Oct. 1974 | secondary literature compilation | 1974 | Data on age distribution of Black gold miners and the share of unskilled workers by degree of mechanisation. |
New data¶
Author constructs adjusted series of African labour force (standardised to 1970 activity rates and adjusted for undercount), adjusted non-farm and farm employment, converted casual farm employment, 'residual labour' and underemployment estimates for 1946-1970 with projections to 1980, plus Homeland labour pool estimates; derived from official data rather than new collection.
Inference¶
Quantification: Descriptive tables.
Descriptive reconstruction of labour force, employment, residual labour and underemployment by permuting assumptions about census undercounts and definitions; cross-checking of competing official sources; indirect Homeland evidence; wage series comparisons; a simple regional regression of shortage on wages; and case studies of firm responses, combined with theoretical reasoning (Todaro-type model).
Key claims¶
| Claim | Basis | Strength |
|---|---|---|
| Census and official data are too inconsistent (changing definitions, undercounting) to measure African labour force, employment and unemployment precisely; changes in female economic activity definition distorted 1960-70. (p. 8) | Comparison of census definitions across 1946-1970 (coding instructions, interviews with Dept of Statistics) and undercount checks. | Systematic data |
| Census undercounted the African population by roughly 10% before 1970 and 3.9% in 1970, and the labour force by even more. (p. 12) | Sadie's demographic adjustments and local aerial-photograph checks. | Partial data |
| African non-farm employment grew faster than the African labour force in the 1960s (about 3.4-3.8% per year) and is projected to do so, but this rests on shaky data. (p. 20) | Census and employer employment statistics, EDP projections (Tables 5-6). | Partial data |
| Statistics on farm employment are unreliable and show no growth, indeed a decline, in African employment on White farms. (p. 26) | Agricultural census 1953/4-1972/3, S.A.A.U. survey 1969 and 1970 census compared in Tables 8-9. | Partial data |
| Residual labour and underemployment (largely in the Homelands) rose in absolute numbers since 1951, though trends in rates are ambiguous. (p. 38) | Derived series from adjusted labour force minus employment, with Tomlinson-based assumption of constant productive Homeland farm employment (Tables 11, 14). | Partial data |
| Homeland agricultural production grew far more slowly than rural Homeland population, implying absorption of labour into unproductive reserve activity. (p. 44) | Official Yearbook GDP and price data, Homeland population series, Tomlinson livestock/income data and Native Economic Commission. | Partial data |
| Mining's declining South African recruitment resulted from low mine wages, monopsony and reliance on foreign labour, not exhaustion of the labour pool; wage rises after 1971 aimed to attract nationals. (p. 59) | Wilson's wage series, Table 16 on gold mine employment, attitude survey of ex-miners, Homeland labour pool calculations. | Partial data |
| Wage increases in the 1970s induced sizeable labour-saving mechanisation in sugar and timber, less in gold mining; scope exists in White agriculture should wages rise. (p. 80) | Case studies from one mining group, sugar estate and timber company plus the Marais report. | Illustrative or anecdotal |
The author's caveats¶
Author stresses repeatedly that data do not permit clear-cut answers: census definitions changed, undercounting is large and uncertain, census and employer statistics diverge, agricultural census employment series are erratic, Homeland data are unpublished and sparse, the price deflator is inappropriate, and adjustments rest on extreme or untested assumptions; the Maree-de Vos survey cannot be extrapolated nationally.
Gaps in the evidence¶
The argument needs reliable, consistently defined labour force surveys and age-specific Homeland data with independent undercount estimates, which did not exist then. Direct sample surveys of Homeland employment, time use and incomes, and farm-level wage and employment panels, would test underemployment and the farm-mining competition hypothesis better than residual calculations. Mining company wage and recruitment records by origin and reasons for worker choices were partly available but are used mostly second-hand; some archival and survey data may be accessible today.
Replicability: Data described but not public.