Skip to content

Mines and Migrants in South Africa

Literature › Works

Robert E. B. Lucas (1985). Mines and Migrants in South Africa. The American Economic Review, Vol. 75, No. 5. American Economic Association.
Type journal article
Scope Economics and economic history of apartheid
Method econometric
Period studied 1946–1980
Regions South Africa, Malawi, Mozambique, Lesotho, Botswana, Swaziland, homelands
Themes mining migrant labour, labour markets wages, business and state, bantustans removals influx, methods data sources, unions strikes
Citation key lucas1985

Argument

Lucas estimates an econometric model of South African mine labour demand, 1946-80, for the gold, diamond, coal and other-minerals sectors separately. He argues that the Chamber of Mines exercised monopsony power over an upward-sloping, inelastic labour supply before 1974. The "labour reserve" view of southern Africa is therefore not supported. The 1970s wage rises were made possible by higher mineral prices and driven by a wish to stabilise the workforce and answer "starvation wage" criticism. After 1974 the mines showed a significant preference against foreign labour, which cost productivity in gold. The job colour bar made black and white labour complements. The diamond sector, with its independent wage policy since 1964, is the exception.

Evidence and method

Annual data 1946-80 on output, employment, capital stock, wages and prices by mining subsector, plus nonwhite labour force proxies. The estimates are 3SLS and 2SLS with principal components of supply-side shifters for the sending countries. The paper also draws on interviews with industry people, Chamber of Mines reports, and secondary works (Wilson 1972, Lipton 1980, First et al. 1977).

Place in the debates

Findings

  • Mine wages for nonwhites tripled in real terms in 1971-78. (p. 2)

    Between 1971 and 1978, wages of more than one-half million nonwhite laborers in the South African mines tripled in real terms.

  • The workforce switched from majority foreign to majority domestic in the same period. (p. 2)

    In the same period, the nonwhites employed in the mines switched from being 62 percent foreign to 62 percent domestic.

  • The Malawi recruiting ban in 1974 sharply reduced Malawian employment on the mines. (p. 3)

    the number of Malawians employed in the mines fell from 140,000 in 1974 to 2,000 in 1977.

  • Section 10 workers shunned mining for better-paid manufacturing work, so mine labour came mainly from those without urban rights. (p. 3)

    this meant mining attracted virtually no section 10 workers who naturally preferred manufacturing.

  • Forced removals of farm squatters to the homelands expanded the labour pool and eased mining's domestic supply by 1979. (p. 4)

    forced relocation of massive numbers of squatters from white farms to homelands in the late 1970's

  • A large rise in the real gold price made wage increases possible. (p. 4)

    By 1974, the real price was 2.9 times the 1971 level and by 1980 the increase was 6.7 fold

  • The wage initiative stemmed from Anglo American. (p. 4)

    The initiative to raise wages apparently stemmed from the Anglo American Corporation and particularly from new directors appointed about 1971

  • The racial earnings gap on the gold mines was very large. (p. 4)

    average earnings of whites in gold mining in 1970 were over 20 times greater than those of nonwhites.

  • The wage tests are consistent with monopsony exploitation in gold before 1974. (p. 10)

    All of the evidence is thus consistent with exploitation of monopsony power, at least by the dominant gold sector, before 1974.

  • The switch to domestic labour was costly: the mines would have wanted considerably more workers absent the policy. (p. 12)

    the gold sector would have wanted to employ some 70,000 more nonwhites and the coal sector nearly 20,000 extra as mineral prices rose

  • In diamonds, eroding the colour bar hurt white labour, as the white union had feared. (p. 12)

    It seems the erosion of the job color bar indeed had precisely the effects on white labor the Mine Workers' Union has always feared.

  • The labour reserve supply curve was upward-sloping and inelastic. (p. 14)

    the supply curve on which the industry operated has been shown not only to be upward sloping with respect to mine wages of nonwhites

Evidence ledger

Sources used

Source Kind Period How used
Annual time-series dataset (1946-1980) on mining output, employment, capital stocks, wages and prices by subsector (gold, diamonds, coal, other minerals), compiled by the author from unnamed published sources official statistical series 1946-1980 Provides all variables (output, employment, k, e, wages, prices) for the estimated labor demand and production function systems.
Published Divisia quantity indices for other minerals (several spliced together) official statistical series 1946-1980 Measures output for the 'other minerals' subsector with 1957 base of 1000.
South African cost-of-living index and cost of food index official statistical series 1946-1980 Deflates capital expenditure into constant 1970 rand stocks, real wages and prices, and the manufacturing expected wage.
Nonwhite male population ages 18-49 and nonwhite employment/earnings in manufacturing, construction and South African Railways and Harbours official statistical series 1946-1980 Proxy for nonwhite male labor force and expected manufacturing wage in the South African labor supply equation.
Supply-side exogenous variables for labor-sending countries (population, rainfall, wages, crop prices, employment) used as 23 underlying variables reduced to principal components international trade or finance data 1946-1978 Principal components shift the overall labor supply function in equation (7) estimated in Table 2.
Lucas (1983), 'Emigration, Employment and Accumulation: The Miners of Southern Africa', Discussion Paper No. 4, Harvard secondary literature compilation 1946-1980 Larger study containing the supply-side data and treatment of the 23 variables underlying this paper's demand model.
Interviews with industrialists and industry/recruiting participants interviews oral history c. early 1980s Support claims about Chamber split over wages, recruitment policy and motivations; not systematically reported.
TEBA recruitment data, first 11 months of 1981 (share of novices by country) chamber of mines or recruiting records 1981 Shows low novice shares among foreign recruits, taken as evidence of bias against foreign dependence.
Transvaal and Orange Free State Chamber of Mines, Annual Report 1957 chamber of mines or recruiting records 1956-57 Documents easing of European labor shortage in coal by 1956, justifying omission of pre-1956 coal white-labor observations.
Merle Lipton (1980), 'Men of Two Worlds: Migrant Labour in South Africa', Optima secondary literature compilation to 1980 Source for the 42 percent share of black workers with section 10 rights and for qualitative account of the industry's wage rationalizations.
Francis Wilson (1972), Labour in the South African Gold Mines 1911-1969 secondary literature compilation 1911-1969 Cited for the claim that real black mine wages did not rise 1911-1971 and the history of foreign labor reliance.
Legislation: Bantu Urban Areas Act of 1945 (section 10) and 'master and servants' acts legislation gazette 1945-1970s Defines the section 10 labor market segmentation and the binding contracts that make labor supply sluggish.
Ruth First et al. (1977), The Mozambican Miner secondary literature compilation 1970s Cited for debate on whether the post-1974 decline in Mozambican recruits reflected South African or Mozambican decisions.

New data

Author assembled annual 1946-1980 time series on output, employment (nonwhite and white), capital stock and equipment, earnings, and prices for the gold, diamond, coal and other minerals subsectors, plus a 23-variable set of supply-shifters for labor-sending countries; not published in the article (about 33 observations per equation).

Inference

Quantification: Econometric (causal design).

Structural econometric model of derived labor demand (quadratic and translog production functions, flexible accelerator) estimated by three-stage least squares on 1946-1980 sector time series, with tests of monopsony via pre/post-1974 wage coefficient differences and Gallant-Jorgenson cross-equation restrictions, supplemented by a two-stage least squares labor supply equation and qualitative institutional narrative.

Key claims

Claim Basis Strength
Mine labor supply was upward-sloping and inelastic (elasticity below one) to 1974, contradicting the 'labor reserve' view. (p. 1101) 2SLS estimate of overall supply equation (Table 2, 1948-73, 26 observations) with principal-component shifters. Partial data
The Chamber, especially in gold, exploited monopsony power in 1946-74. (p. 1102) Larger absolute wage coefficients pre-1974 in labor input equations (Table 1), consistency with production function estimates (Table 3) and Table 2 elasticity. Partial data
After 1974 the mines showed a significant preference not to rely on foreign labor, which would otherwise have meant roughly 70,000 more nonwhites in gold and nearly 20,000 in coal. (p. 1104) Coefficient on post-1974 dummy in nonwhite labor input equations (Table 1), plus TEBA novice shares in 1981. Partial data
Localization of the labor force reduced productivity in gold as inexperienced South African novices replaced skilled foreign workers. (p. 1104) Gold production function estimates (Table 3), post-1974 domestic labor term. Partial data
In diamonds, stabilization and erosion of the color bar after 1964 raised nonwhite productivity and lowered white productivity and demand. (p. 1104) Diamond production functions (Table 3) and labor input equations (Tables 1, 4) with post-1964 interaction terms. Partial data
White and nonwhite labor are complements across sectors, with the color bar requiring extra whites for added nonwhites; white labor is quasi-fixed. (p. 1105) Tables 1 and 4 coefficients and positive lagged dependent variables. Partial data
No systematic evidence of wage discrimination against nonwhites by the mines or of monopoly-induced labor restriction by DeBeers. (p. 1106) Gallant-Jorgenson restriction tests and re-estimation with a psychic-cost term; correlation of diamond sales and production. Partial data
Mine wages tripled in real terms 1971-78 following the Chamber's decision, after rising gold prices, to raise wages, with the initiative reportedly from Anglo American. (p. 1096) Narrative from interviews and industry knowledge; no cited documents. Asserted

The author's caveats

States that space does not permit a full discussion of data sources and compilation; notes collinearity prevented estimating the translog other-minerals equation, no serial-correlation correction in production functions, and calls the wage discrimination test 'simple'. Small samples (25-33 observations) are evident from the tables.

Gaps in the evidence

The article does not name the underlying statistical series (e.g., Chamber of Mines annual reports, Department of Mines, Statistics reports), so the data cannot be checked; archives of the Chamber and Anglo American would test the claimed motives for the wage increases and foreign-labor policy. Mine-level or firm-level panel data on wages, recruitment and productivity (exists in Chamber and company records) would offer better identification than 33 annual aggregate observations with many parameters. Direct data on actual supply-side conditions (TEBA recruitment by origin) would test the excess supply claims.

Replicability: Data poorly documented.