Father's employment and sons' stature: the long run effects of a positive regional employment shock in South Africa's mining industry¶
Martine Mariotti (2012). Father's employment and sons' stature: the long run effects of a positive regional employment shock in South Africa's mining industry. Stellenbosch Economic Working Papers 02/12. Department of Economics and Bureau for Economic Research, University of Stellenbosch.
| Type | working paper |
| Scope | Economics and economic history of apartheid |
| Method | econometric |
| Period studied | 1973–2008 |
| Regions | Transkei, Ciskei, Bophuthatswana, KwaZulu, Eastern Cape, homelands |
| Themes | mining migrant labour, bantustans removals influx, inequality poverty unemployment, labour markets wages, demography urbanisation, education human capital, methods data sources |
| Citation key | mariotti2012b |
Argument¶
The paper uses the 1974-77 collapse of foreign mine labour supply (Malawi, Mozambique) and the Chamber of Mines' resulting recruitment drive in four homelands as a natural experiment. A difference-in-difference comparison of adults in NIDS 2008 shows that African men born in the affected regions in 1976-77 are about 6 cm taller than other African men, which supports the foetal origins hypothesis and indicates severe malnutrition in the homelands. There is no effect on education or most health measures, apart from some improvement in long-term health. No effect appears for women, or for boys who were toddlers, small children or teenagers at the time. The paper isolates household income effects from regional public-service improvements, and gives individual-level evidence on living standards in the homelands under apartheid.
Evidence and method¶
National Income Dynamics Study Wave 1 (2008) adult sample, with measured height, birth location and year, education and health questions; 10,433 Africans in the restricted sample (4,098 men). Mine employment by origin from SAIRR annual surveys and Crush et al. (1991); Tomlinson Commission (1955) farm income data; Horrell (1969, 1973) and Wilson (1972) on homeland conditions, densities and remittances; Butler, Rotberg and Adams (1978) for population estimates.
Place in the debates¶
- Foetal origins hypothesis: do in utero nutrition and income shocks drive adult stature and health? (see Living standards, health and social policy). Supports it: only men exposed in utero in 1976-77 show large height gains, with no effect for earlier or later cohorts.
- Cheap labour thesis and migrant labour system (low mine wages justified by homeland subsistence) (see Living standards, health and social policy). Takes the standard account that foreign workers were preferred because they accepted low wages, and that homeland conditions had deteriorated. Argues even low mine wages meant a large rise in household income for destitute homeland families.
- Living standards of Africans under apartheid (data neglect and poverty in the homelands) (see Living standards, health and social policy). Contends living standards were very poor and unmeasured, and that the height gap with Whites reflects environment, not genetics.
Findings¶
-
Men born in the treated regions at the shock are about 6 cm taller than other African men. (p. 7)
Men who were born in the affected regions in 1976 and 1977 are on average 6 centimetres taller than other African men.
-
Foreign labour fell sharply while domestic recruitment roughly doubled between 1973 and 1977. (p. 4)
The number of foreign workers in the mining industry fell from 297, 000 in December 1973 to 189,000 at the end of 1977
-
The number of South Africans employed on the mines rose nearly threefold between 1974 and 1978. (p. 8)
The number of South Africans employed in the industry increased from 76 523 in 1974 to 228 109 by 1978, an increase of almost 200 %.
-
By 1973 foreign workers made up three quarters of the gold mining workforce. (p. 7)
By 1973, foreign workers made up 75 % of the gold mining workforce in South Africa.
-
Mine wages rose rapidly after the 1973 labour disputes, from a very low base. (p. 13)
In 1972 the monthly wage in the mining industry was R21. The monthly wage increased to R29 in 1973.
-
Homeland farm incomes were far below the poverty line (Tomlinson Commission, c.1953). (p. 12)
The Commission found that 12 % of the farmers made more than R120 a year and 44 % made less than R40.
-
Preferred estimates put the height gain at roughly 5.9-6.3 cm, with a large effect in 1976 in each treated region. (p. 23)
The results suggest that male heights of those born during the years of the shock increased by between 5.9 and 6.3 centimetres.
-
No effect on men born after the shock; boys born after in treated regions are shorter. (p. 25)
males born one to five years after the shock in the treated regions are 3.12 centimetres shorter than males born before the shock.
-
No significant impact on education; treatment coefficient is negative but insignificant. (p. 25)
I find no effect of the treatment on educational outcomes, the coefficient is
-
Only long-term health improves significantly for treated men. (p. 26)
I do find a positive and significant effect of treatment on long run health outcomes, males from the treatment group are less likely to be suffering from any long term health complaints.
-
The shock has no positive effect on women, who appear shorter in treated cohorts. (p. 23)
All specifications yield height decreases of between 1.8 and 3.7 centimetres for women born in the treated regions at the time of the shock.
-
Homelands held half of the African population on a small share of the land. (p. 4)
Their physical land mass constituted slightly more than 13 percent of the total South African land mass and yet they housed approximately 50 percent of the African population.
Evidence ledger¶
Sources used¶
| Source | Kind | Period | How used |
|---|---|---|---|
| National Income Dynamics Study (NIDS) Wave 1, 2008, adult sample | household or labour survey microdata | 2008 survey (released July 2009); respondents born 1941-1990 | Supplies heights, birth year and location, education and health measures for the difference-in-difference regressions on adult outcomes of Africans born in homeland districts. |
| NIDS derived household data file (province and district council of household) | household or labour survey microdata | 2008 | Supplements adult data to recover birth district for the 9,543 respondents who never moved, so the treatment variable can be built. |
| NIDS survey documentation and weighting guidance (SALDRU 2009) | other | 2009 | Guides the weighting specification used in the regressions, minus the stratification element. |
| South African Institute of Race Relations Annual Reports/Surveys (1969-1979), mining employment by origin | official statistical series | 1969-1979 | Documents the fall in foreign mine workers and rise in South African recruits (Tables 1-2), and mining wages 1972-1979, establishing the labour supply shock. |
| SAIRR Annual Surveys, mining wage series | official statistical series | 1972-1979 | Gives the monthly mine wage (R21 in 1972 to R141 in 1979) used to size the income increase from a mine job. |
| Tomlinson Commission, Summary of the Report of the Commission for the Socio-Economic Development of the Bantu Areas, UG 61/1955 | commission or official report | c.1953-1955 | Provides baseline farm incomes, livestock holdings, erosion and carrying capacity showing pre-shock poverty and overcrowding. |
| Crush, Jeeves and Yudelman (1991) mine employment data by homeland | secondary literature compilation | 1970-1978 | Timing and regional pattern of mine recruitment increases, and the numbers employed per region used to build the proportion-employed variable in Table 6 col. 5. |
| Butler, Rotberg and Adams (1978), homeland population and income estimates | secondary literature compilation | 1973 population estimates | Homeland population denominators for the mining-employment proportion variable; 1973 income range for Bophuthatswana and Kwa-Zulu. |
| Horrell (1969, 1973), SAIRR studies of African reserves and homelands | secondary literature compilation | 1969-1973 | Provides population densities, land quality, and share of wage given to homeland households to document overcrowding and size the income shock. |
| Wilson (1972) Labour in South African Gold Mines | secondary literature compilation | 1963-1972 | Source for remittance share (18%), family land pressure and cost-of-living poverty benchmark. |
| Mauder (1982) Atlas of Transkei population estimate | secondary literature compilation | 1982 | Population estimate (2.3 million) used for the ad hoc calculation of the share of Transkei households affected. |
| 1970 Population Census of South Africa (discussed as limited) | population census | 1970 | Described to show that apartheid-era census data on Africans were inadequate, motivating use of NIDS. |
| Abasebenzi workers' newsletter, 1973 | newspapers periodicals | 1973 | Cited for the Poverty Datum Line to show homeland farm incomes were far below subsistence. |
| Speeches of Paramount Chief Matanzima to Transkeian Legislative Assembly, 1970 and 1972 | parliamentary debates | 1970, 1972 | Evidence that homeland agricultural productivity stagnated and malnutrition was a concern. |
| Personal conversation with Francis Wilson, October 2011 | interviews oral history | October 2011 | Qualifies the remittance estimate by noting workers also brought wages home on leave. |
| Map data for homelands, provinces and district councils (Perry-Castañeda collection; Municipal Demarcation Board boundaries) | maps spatial data | apartheid-era and 2011 boundaries | Used to match 2011 district councils to former homeland areas and assign treatment status. |
New data¶
Constructed analysis sample from NIDS Wave 1 adults: 10,433 Africans (4,098 men, 6,335 women) with birth district assigned as treated or not by matching 53 district councils to former homelands; treatment is birth in an affected district in 1976-77. A regional mining-employment proportion variable (n=460) was also built from secondary figures.
Inference¶
Quantification: Econometric (causal design).
Difference-in-difference (OLS and ordered probit) comparing African men and women born in affected homeland districts in the shock years with those born in unaffected districts or other years, with controls for birth year, province, language, age and father's occupation. The shock is treated as quasi-exogenous because the Malawian and Mozambican withdrawals were unexpected and recruitment districts were set by policy. Placebo cohorts and several robustness specifications are examined. The income shock size rests on an ad hoc back-of-envelope calculation.
Key claims¶
| Claim | Basis | Strength |
|---|---|---|
| The foreign labour supply to the gold mines fell sharply in 1974-77 and was replaced by domestic recruits from four homelands, with domestic employment more than doubling. (p. 4) | SAIRR annual reports and survey tables (Table 2); Crush et al. for regional timing. | Systematic data |
| Roughly a quarter of households in the affected areas gained a substantial income increase (at least R14.22/month) relative to very low pre-shock farm incomes. (p. 14) | Ad hoc calculation using Transkei worker counts (Table 2), Mauder population estimate, Wilson remittance share and Tomlinson Commission income data. | Illustrative or anecdotal |
| African men born in affected homeland regions in 1976-77 are about 6 cm taller than other African men. (p. 23) | DiD OLS regressions on NIDS 2008 heights (Table 4 and Table 6), small treated sample (n=82). | Partial data |
| No height effect for men born before or after the shock, with a negative coefficient for cohorts born 1-5 years after; this supports an in utero (foetal origins) mechanism. (p. 25) | Table 7 placebo cohort regressions on NIDS heights. | Partial data |
| The shock had no effect on men's educational attainment or most health measures, but improved long-run health. (p. 26) | Ordered probit and OLS on NIDS education and health indices (Tables 8-9). | Partial data |
| No positive effect on women born during the shock; estimates are negative for height, and the author finds this troubling. (p. 23) | Tables 5 and 10-12, NIDS women's sample (n=137 treated). | Partial data |
| Homelands were severely poor and overcrowded before the shock, so mine wages, though low, were a large increase in household income. (p. 12) | Tomlinson Commission figures from 1953, Horrell, Wilson, Butler et al., and Matanzima speeches. | Partial data |
The author's caveats¶
The author notes small treated samples that limit power, and that the NIDS asks about birth location only for movers, so the household location file was used to fill gaps. She notes that father's occupation is too aggregated to identify mine work, that parental heights, income and education are unavailable, and that not all individuals in the treatment group were truly treated. Ciskei and Transkei cannot be separated, homeland data on earnings are unreliable, and the 1953 Tomlinson data are the latest for all homelands. Population estimates ignore growth, and the income shock rests on assumptions that cannot be verified. Sampling weights were simplified (stratification ignored), and wage rises elsewhere in the economy are acknowledged as confounders.
Gaps in the evidence¶
No direct records link individuals to a household member working in mines, so treatment is assigned only by birth district and cohort, and the 1976-77 and district choices are somewhat arbitrary. Mine recruitment records by magisterial district (TEBA/NRC) and homeland household income or nutrition surveys from the 1970s would test the income-channel and regional exposure more directly; some TEBA archives exist but are not used. Selective survival, migration out of homelands and measurement of height in adults born 30+ years earlier are not addressed with contemporaneous child-growth data, which was largely not collected.
Replicability: Data public and well described.