Skip to content

Emigration to South Africa's Mines

Literature › Works

Robert E. B. Lucas (1987). Emigration to South Africa's Mines. The American Economic Review, Vol. 77, No. 3. American Economic Association.
Type journal article
Scope Economics and economic history of apartheid
Method econometric
Period studied 1946–1978
Regions South Africa, Botswana, Lesotho, Malawi, Mozambique, South African homelands
Themes mining migrant labour, labour markets wages, agriculture land, bantustans removals influx, comparative theory, methods data sources
Citation key lucas1987a

Argument

Lucas builds a simultaneous econometric model of temporary migration to South Africa's mines from five labour-supplying areas and its effects at home. Migration responds positively to the expected mine-versus-home wage differential, except where quotas or recruiting bans bind. Labour withdrawal lowers subsistence crop output in the short run, but migrants' invested earnings raise crop productivity and cattle holdings in the long run (Lesotho is the exception). Emigration also raises plantation and estate wages in Malawi and Mozambique, which sets sending-country employers against the mining houses. This conflict explains state interventions such as quotas, forced labour and Malawi's 1974 recruiting ban. The model is estimated on annual time series for 1946-78.

Evidence and method

Annual time-series data for 1946-1978 compiled from national archives, statistical offices, agriculture ministries and university libraries in Botswana, Lesotho, Malawi and South Africa. They cover mine employment, wages, crop output, cattle numbers, rainfall, estate and plantation employment and wages, and fertilizer sales. Secondary sources on Mozambican forced labour (Vail and White, ILO reports) and on TEBA recruiting supplement them. Source details are in an appendix available from the author on request.

Place in the debates

Findings

  • Mine recruitment was very large relative to the male population of the main suppliers around 1970. (p. 2)

    was close to 50 percent for Botswana, and about 15 percent even for Mozambique where the proportion was lowest.

  • About 60 percent of mine earnings were remitted or paid as deferred pay at home, which underpins the real-wage measure used. (p. 4)

    a substantial fraction (approximately 60 percent on average) is remitted or sent as deferred pay for spending in the home country.

  • From 1974 the mining houses, acting through their recruiting agency, preferred domestic workers, so foreign supply was no longer demand-clearing. (p. 3)

    the South African mining houses, acting in concert through their centralized recruiting agency, exercised a preference for domestic workers

  • Labour supply to the mines responded to wage differentials; it was not an infinitely elastic reserve, which motivated centralised monopsonistic recruiting. (p. 10)

    Clearly, southern Africa has not been a labor reserve for the mining houses, in the sense of infinite elasticity of labor supply

  • Crop-production estimates contradict a simple surplus-labour model: wage employment reduces subsistence output in the short run. (p. 12)

    reject a simple labor surplus theory; the negative coefficients estimated on the employment variables indicate that increased wage employment (at home or in the mines) from any given level of population has reduced traditional crop output, ceteris paribus.

  • Migrant wage income capitalised into rural production raises long-run productivity even as it lowers short-run output. (p. 13)

    it appears that wage income of migrants can enhance the working or fixed capital available to the traditional, rural sector, raising productivity in the longer run while in the short-run reducing output.

  • Lesotho is the exception, where heavy dependence on emigration coincided with neglected crops and erosion. (p. 13)

    Today it is utterly dependent on emigrant labor; crop production has been neglected and massive soil erosion has ensued

  • Malawi's 1974 recruiting suspension raised estate employment and lowered estate wages, but absorbed only part of the displaced miners. (p. 15)

    The result was almost an 80 percent rise in estate employment by 1978, and a 15 percent drop in the real wage of estate workers in the first two years after the cutoff.

  • Estates absorbed only half the lost mine jobs; the rest fell back on Customary Lands. (p. 15)

    But the additional employment on estates amounted to exactly half the decline in numbers previously at the Rand mines

  • Mine recruiting from Malawi never recovered after the ban was lifted, because South Africa had lost trust in Malawi as a source. (p. 16)

    but by then South Africa no longer trusted Malawi as a source and the number of mine workers from Malawi has never recovered

  • Forced relocation into the homelands swelled the pool of South African labour available to the mines. (p. 9)

    some 3.5 million black South Africans deemed not entitled to remain in White Areas have been forcibly relocated

Evidence ledger

Sources used

Source Kind Period How used
Annual time-series data 1946-1978 on mine employment by sending country (Botswana, Lesotho, Malawi, Mozambique, South Africa), mine wages, home wages, employment, crop output and values, cattle numbers, rainfall, fertilizer sales, tobacco grower numbers, estate acreage, construction output official statistical series 1946-1978 Supplies the consistent annual series for all variables used to estimate the migration, crop output, cattle and estate-labor equations.
National archives and libraries visited in Botswana, Lesotho and Malawi state archives 1946-1978 Archival and library holdings in the three countries were searched to locate the data series used in the model.
South African Department of Statistics and Meteorological Office data official statistical series 1946-1978 Provides South African series on mine and other wages, employment, crop values and rainfall for the homelands equations.
1970 censuses (de jure male population ages 18-35) combined with stock of men employed in South African mines population census c. 1970 Establishes the scale of mine recruitment relative to the young male population (over 80% Lesotho, about 50% Botswana, about 15% Mozambique).
Nyasaland announced annual recruiting quotas versus recorded new recruits official statistical series colonial period to 1974 Comparison of announced quota with recorded recruits identifies years in which quotas bound, which are excluded from the migration equation.
TEBA reports on excess South African applicants after 1978 chamber of mines or recruiting records post-1978 Supports truncating the South African mine labor equation at 1978 because of excess labor supply.
Vail and White (1980), Capitalism and Colonialism in Mozambique secondary literature compilation 1930-1960s Source for the 1942 circular, forced labor and competitive labor demand, and Portuguese attitudes, used to specify the Mozambique equations.
Portuguese 1942 circular and 1961 decrees on compulsory employment legislation gazette 1942-1961 Defines the forced registration period (dummy variable to 1961) used in the Mozambican migration and wage equations.
ILO (1962) Report Concerning the Observance by the Government of Portugal of the Abolition of Forced Labour Convention commission or official report 1950s-1962 Documents continued forced labor in Mozambique during the 1950s.
Secret clause and recruiting agreements between Portugal and South Africa secondary literature compilation mid-20th century Explains why Portuguese authorities allowed mine recruiting despite higher plantation labor costs (port traffic, recruiting tax, deferred pay held in gold).
Christiansen and Kydd (1983), Journal of Modern African Studies secondary literature compilation 1974-1977 Evidence that the 1974 suspension of Malawian recruiting was planned before the plane crash and that mine numbers never recovered after the 1977 lifting.
Lucas (1985a), "Mines and Migrants in South Africa" secondary literature compilation post-1974 Prior study by the author establishing that mining houses preferred domestic workers after 1974 and exercised monopsony power through centralized recruiting.
Case studies of migrant wage earners and farming (Kerven 1983; Cobbe 1982) secondary literature compilation 1980s Cited as corroborating the econometric finding that migrant wage earnings raise peasant crop and cattle productivity, and as evidence of Lesotho's soil erosion and neglected cropping.
Instrumental variables: wage rates of white miners in South Africa and mineral prices official statistical series 1946-1978 Used with principal components of exogenous variables as instruments for endogenous mine wage and outputs.
Author's data appendix (available on request) author survey or fieldwork 1946-1978 Itemizes sources and definitions of all variables, but is not printed in the paper.

New data

Author compiled consistent annual time-series (1946-1978) for Botswana, Lesotho, Malawi, Mozambique and South Africa homelands on mine employment, wages, domestic employment, crop output and value, cattle herds, rainfall, fertilizer and estate data, drawing on archives and statistical offices during a data-gathering trip. The series are not itemized in the paper; a detailed appendix is offered on request.

Inference

Quantification: Econometric (causal design).

A simultaneous-equations econometric model estimated on annual country time series (OLS, IV, AR and Fair-type corrections) covering mine labor supply, crop production, cattle accumulation, and estate labor demand and wages. Identification relies on capital-stock proxies built from accumulated wage and crop income, instruments from principal components, dummies for quota and forced-labor periods, and institutional narrative from secondary sources.

Key claims

Claim Basis Strength
Mine labor supply from each country responds positively to the expected wage differential between mines and home when quotas are not binding, so southern Africa was not an infinitely elastic labor reserve. (p. 321) Table 1 regressions on 1946-1978 country time series with mine wage, expected home wage, crop value and rainfall. Partial data
Migration to the mines reduces traditional crop output in the short run through labor withdrawal, rejecting a simple labor-surplus theory. (p. 323) Table 2 crop production regressions with negative coefficients on employment/population. Partial data
Capitalized migrant wage earnings raise crop productivity and cattle accumulation in the long run, except in Lesotho. (p. 323) Tables 2 and 3 coefficients on accumulated wages, built from a constructed capital stock series with assumed depreciation; corroborated by case studies. Partial data
Mine recruiting raised plantation labor costs and reduced estate employment in Mozambique and Malawi, creating conflicting interests between employers in sending countries and the mines. (p. 325) Table 4 estate labor demand and supply equations; Vail and White; quota history. Partial data
Ending forced labor in Mozambique reduced the supply of Mozambican miners to the Rand, since forced labor had pushed men toward mine contracts. (p. 323) Forced-registration dummy in Mozambique mine equation (Table 1) and Vail and White/ILO background. Partial data
The 1974 suspension of Malawian recruiting raised estate employment by almost 80 percent by 1978 and cut estate real wages, with Customary Lands absorbing displaced workers. (p. 326) Descriptive before/after comparisons of Malawian employment, wage and ADMARC purchase series, plus Christiansen and Kydd. Partial data
Roughly 3.5 million black South Africans were forcibly relocated to homelands after 1960, swelling the labor pool for the mines and prompting the mining houses to shift recruitment toward South African workers. (p. 320) Asserted background without cited source in the text; excess applicants reported by TEBA after 1978. Asserted

The author's caveats

The author notes that data are variegated across countries, that no capital stock series exists for peasant agriculture (so capitalized wage and crop income surrogates are used with an assumed depreciation rate), and that no time-series data exist on nonagricultural self-employment. Lesotho wage and crop data are missing for earlier years, Mozambique lacks subsistence crop output and construction employment (proxies used), and maize sales proxy production in Malawi. Collinearity between accumulated wages, crops and time is acknowledged, and the mechanism of productivity gains cannot be discerned.

Gaps in the evidence

The argument relies on constructed proxies (capitalized wages, expected wage differentials) and unitemized national series, so direct evidence on how remittances were invested (household surveys, farm-level data) is lacking. Archival TEBA and Chamber of Mines recruiting and deferred-pay records, plus household-level surveys of remittance use, would better test the long-run productivity mechanism and the forced-labor and quota effects; some of these existed at the time, such as TEBA records and Botswana and Lesotho surveys. Short samples (13 to 31 observations) and unreliable homeland data limit the identification of causal effects.

Replicability: Data poorly documented.