Migrant Labour and South African Economic Development¶
Jill Nattrass (1976). Migrant Labour and South African Economic Development. South African Journal of Economics, vol. 44, no. 1.
| Type | journal article |
| Scope | Economics and economic history of apartheid |
| Method | descriptive quantitative |
| Period studied | 1936–1970 |
| Regions | South Africa, Homeland rural areas (KwaZulu, Transkei, Ciskei, Lebowa, Bophuthatswana), White areas, Foreign labour-supplying areas |
| Themes | mining migrant labour, bantustans removals influx, labour markets wages, demography urbanisation, education human capital, inequality poverty unemployment, methods data sources, unions strikes |
| Citation key | nattrass1976a |
Argument¶
Nattrass argues that South African development has produced a large and growing system of oscillating (temporary) African labour migration, which she measures systematically for the first time. The system is selective by age and education, and it depends heavily on the Homelands. She claims the modern sector benefited from it, through savings on housing and downward pressure on wages, even though it carries inefficiencies and costs. Both the modern sector and the Homelands are now dependent on it through remittances and labour supply. Because market forces alone will not reverse this, she calls for heavy state intervention, including family housing, industrial decentralization, education transfers and a phase-out of the system over the long run. The argument rests on net out-migration estimated from 1970 census data by comparing a 'standard' population with actual rural populations.
Evidence and method¶
1970 Population Census Report 02-02-02 (Bantu age, occupation, industry, school standard, birthplace), with 1936, 1946 and 1960 census data for the time series. Also the Department of Bantu Education Annual Report 1971, Bantu Administration and Development reports, the Transkei labour report, Hansard on foreign migrants, the Marais Commission, Lombard and Van der Merwe (1972) and Wilson (1972). Chi-square tests are used on age and industry distributions.
Place in the debates¶
- cheap labour thesis / who benefits from migrant labour (see Cheap labour thesis and the function of the reserves). 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'.
- market forces vs institutions in creating the location imbalance of jobs and people (see Homelands, separate development, decentralisation and removals). Says the imbalance resulted largely from market forces but was strengthened by institutional elements, so the market alone cannot correct it and government intervention is needed.
- is migrant labour dysfunctional for capital / do employers prefer migrants (see Cheap labour thesis and the function of the reserves). 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.
- policy: abolish, phase out or reform the migrant system and homeland development strategy (see The mine migrant labour system: origins, recruitment, foreign labour and stabilisation). Realist reformist: the system cannot be abolished at once and should be phased out over decades. She favours family housing, decentralization, education subsidies and homeland-controlled labour supply.
Findings¶
-
Migrants were a large share of the work force in 1970. (p. 2)
In 1970 migrant workers in South Africa formed 28 per cent of the total male work force and 59 per cent of the African male work force enumerated in White areas.
-
The estimated stream of migrants is large and still understated. (p. 2)
the total number of migrants of 1 490 000 men and 260 000 women in Table 1, understates the size of the migration stream since some migrants choose to work in agriculture and rural industries.
-
Migration has grown rather than shrunk as development proceeded. (p. 2)
Far from a general reallocation of the population, in response to changing geographical demand patterns for labour, continuing development in South Africa has meant a continued growth in the size of the stream of temporary migrants.
-
The migrant stream grew at a compound annual rate of 3.1 per cent between 1936 and 1970. (p. 3)
Over the thirty-four year period covered by the estimates in Table 2 the number of migrants from rural areas grew at an average compound rate of 3,1 per cent per annum.
-
Migration drains educated men from the Homelands and lowers the returns on Homeland schooling. (p. 3)
Out-migration of such levels among the educated must drastically reduce the social rate of return on investment in education in the Homelands.
-
Migrants are younger and have shorter working lives than settled workers. (p. 4)
The average length of the working life in the modern sector of a migrant labourer is 19 years, whereas that of his stabilized colleague is 23 years, an increase of 20 per cent.
-
The migrant age profile is much younger than that of the stabilized work force. (p. 4)
Eighty-one per cent of the domestic migrant labour population is under forty years of age, whereas only 62 per cent of the stabilized work force is under forty.
-
The migrant system plausibly held African wages down. (p. 5)
It can also be argued that the system of temporary migration has held Black wages in South Africa at a lower level than would otherwise have been the case.
-
Housing savings to the modern sector are valued at about 4 per cent of 1970 GDP, based on stated assumptions. (p. 13)
the increased output resulting from the non-provision of housing for a stabilized work force in 1970 amounted to R393 000 000, which was 4 per cent of the gross domestic product in 1970 at 1963 prices.
-
Migration behaviour is remarkably uniform across the four major Homeland language groups. (p. 6)
There is a remarkable degree of consistency in the behaviour patterns of the age cohorts of migrants from the four major supplying language groups.
-
Apart from mining, employers show no marked preference for or against migrant labour. (p. 7)
apart from the mining industry, there seems to be no marked preferences for or prejudices against migrant labour.
-
KwaZulu depends heavily on migrant remittances. (p. 7)
these remittances in 1970 were the equivalent of forty-one per cent of KwaZulu's Gross Domestic Product, eighty-eight per cent of the area's subsistance sector income and one hundred and forty-two per cent of the areas estimated agricultural output.
Evidence ledger¶
Sources used¶
| Source | Kind | Period | How used |
|---|---|---|---|
| Population Census 1970, Report 02-02-02: Bantu Age, Occupation, Industry, School Standard, Birthplace | population census | 1970 | Basis for the standard-population estimate of migrants (Table 1), age/education/industry profiles, and ethnic-group migration rates (Tables 3-6). |
| Population Census 1960, Vol. 7, No. 1: Characteristics of the Population in Each Magisterial District and Economic Regions, Age, Marital Status and Home Language | population census | 1960 | Used to estimate the 1960 number of African temporary migrants from rural areas for the Table 2 time series. |
| Seventh Census of the Population of the Union of South Africa, Vol. II, Ages of the European, Coloured, Asiatic and Native Population, U.G.60/1950 | population census | 1936 and 1946 | Source for age data used to estimate migrants at the 1936 and 1946 census dates (Table 2). |
| Department of Bantu Education Annual Report 1971, RP31/1973 | commission or official report | 1971 | School enrolments by region used to compare educational profiles of those in and out of school (Table 3) and the age-structure estimates (Table 4). |
| Annual Report of the Department of Bantu Administration and Development 1971, RP 41/1973 | commission or official report | 1971 | Source for the industrial distribution of migrants from Bophuthatswana, Lebowa and the Ciskei in Table 6. |
| Transkei Government Department of Interior, Statistical Labour Report 1971 | official statistical series | 1971 | Source for the industrial distribution of Transkeian migrants in Table 6. |
| Reply by the Minister of Bantu Affairs, House of Assembly, 5 March 1971 (Hansard cols. 5,384-386), via SAIRR Survey of Race Relations 1971 | parliamentary debates | 1971 | Supplies the industrial distribution of foreign African migrants in Table 6. |
| South Africa 1974 (and 1975 edition), Department of Information, Pretoria | official statistical series | 1974-1975 | Provides the Department of Statistics migrant estimate for comparison and the GDP, subsistence and agricultural output figures behind the remittance estimates. |
| Lombard and Van der Merwe, Central Problems of the Economic Development of the Bantu Homelands (Finance and Trade Review, June 1972) | secondary literature compilation | 1967 estimates | Ratio of commuter to migrant income in 1967 used to estimate the number of homeland commuters added to the migrant total. |
| Francis Wilson, Migrant Labour in South Africa (SACC/Spro-Cas 1972) | secondary literature compilation | c.1970 | Wilson's estimate of 1 305 000 legal migrants used as a check on the reasonableness of the 1970 estimate. |
| Marais Commission, R.P./84 1970 | commission or official report | 1970 | Wage rates (low-wage sector at 43% of high-wage sector) used to estimate remittances as a share of KwaZulu GDP. |
New data¶
Author-constructed estimates of African oscillating migrants (1970 by age, sex and area; 1936/1946/1960/1970 time series), derived indirectly from census population structures using a standard-population method. Also derived tables of age, education and industry distributions and KwaZulu remittance estimates. Underlying calculations are not published.
Inference¶
Quantification: Statistical analysis.
Indirect demographic estimation: the national Bantu-language-group population structure is taken as a standard and deducted from the actual adult rural population to obtain net migrants. These estimates are compared across census dates and across ethnic groups, with chi-square tests of differences in age and industry distributions, followed by descriptive inference and policy recommendations.
Key claims¶
| Claim | Basis | Strength |
|---|---|---|
| About 1.49 million male and 260 000 female oscillating migrants were in the modern sector in 1970, so around 43% of economically active African men were migrants. (p. 2) | Standard-population estimate from the 1970 Census Report 02-02-02, compared with Wilson and Department of Statistics figures. | Partial data |
| The migrant stream has grown steadily since 1936, at 3.1% per annum, and the male migration rate rose from 25% to 34%. (p. 3) | Table 2, estimated from the 1936, 1946, 1960 and 1970 censuses. | Partial data |
| Migration is selective of the better educated: out-migration from Homelands reaches about 90% of the cohort with primary school plus four years of schooling. (p. 3) | Comparison of census educational profiles with school enrolments from the Bantu Education Annual Report 1971 (Table 3); the author notes the evidence is indirect. | Partial data |
| Migrants are younger and have shorter working lives (19 years versus 23) in the modern sector than stabilized workers. (p. 4) | Table 4 age distributions estimated from the 1970 census and the Bantu Education report; the author concedes the data are not wholly consistent. | Partial data |
| Migration behaviour is remarkably similar across the Xhosa, Zulu, Pedi and Tswana groups, but industrial destinations differ, with Transkeian and foreign migrants concentrated in mining and the low-wage sector. (p. 6) | Table 5 (1970 census) and Table 6 (RP41/1973, Transkei report, Hansard), supported by chi-square tests. | Partial data |
| The modern sector benefited from migrancy through avoided housing and social overhead costs, and migrancy kept African wages lower. (p. 5) | Hypothetical housing-cost calculation (footnote 32, assumed R125/R300 per migrant) and theoretical argument about wages and organization. | Illustrative or anecdotal |
| Remittances in 1970 equalled 41% of KwaZulu GDP and 142% of its agricultural output, making Homeland households heavily dependent on migrancy. (p. 7) | Preliminary estimate from migrant numbers, manufacturing wage rates, and South Africa 1974 and Marais Commission data. | Partial data |
The author's caveats¶
The author states that migration estimates are 'by nature somewhat unreliable owing to the lack of adequate statistical data'. She notes that Table 1 understates migrants, that Table 4 data are not wholly consistent, that the education evidence is indirect and overstates out-migration if Homeland schooling expanded faster, and that Table 6 relies on surrogates (the Pedi distribution stands in for the Zulu) and aggregated data. She calls the remittance figures preliminary. Full estimation details are available only on request.
Gaps in the evidence¶
The estimates rest on an indirect standard-population method with assumptions about demographic homogeneity, and have no direct validation from migrant surveys, labour bureau registers or Chamber of Mines records. Claims on wages, bargaining power and employer preferences are asserted rather than tested with wage or employment series. Fieldwork or household survey data on remittances, and pass-law and labour bureau registrations, could test the estimates; some of these existed in the 1970s, and archival versions are now accessible.
Replicability: Data poorly documented.