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, South Africa gold mines |
| Themes | mining migrant labour, bantustans removals influx, inequality poverty unemployment, labour markets wages, demography urbanisation, methods data sources |
| Citation key | mariotti2012a |
Argument¶
The paper treats the mid-1970s collapse of foreign (Malawian and Mozambican) labour supply to the gold mines, and the Chamber of Mines' resulting recruitment drive in four homelands, as a quasi-natural experiment in household income. Using a difference-in-difference design on the 2008 NIDS survey, it compares adults born in the affected homelands in 1976-77 with other African adults. African men born then are about 6 cm taller, which it reads as evidence of severe malnutrition and stunting in the homelands and as support for the foetal origins hypothesis. It finds no effect on men born before or after, on education, or on most health measures (long-term health aside), and no positive effect for women. The study also offers new individual-level evidence on living standards in the homelands, where apartheid-era data collection was poor.
Evidence and method¶
National Income Dynamics Study wave 1 (2008), with heights, birth year and birth district council (about 10,433 Africans: 4,098 men, 6,335 women). Mine employment by origin from SAIRR annual surveys 1969-78 and Crush et al. (1991). Tomlinson Commission (1955) farm incomes and Horrell (1969, 1973) homeland population densities. Wilson (1972) on remittances, and Mining wage series from SAIRR. Population estimates from Butler, Rotberg and Adams (1978) and Mauder (1982).
Place in the debates¶
- Fetal 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 born during the shock years show a height gain, with no effect on slightly older or younger cohorts.
- Intra-household allocation of male income (Duflo 2003) (see Living standards, health and social policy). Finds results consistent with male income gains not reaching existing children, and no benefit for girls born during the shock.
- Cheap-labour thesis / foreign migrant labour in the mines (see Cheap labour thesis and the function of the reserves). Describes the long-standing preference for low-wage foreign labour, which kept domestic Africans out of mining. Draws on Wilson, Leys and Crush et al. rather than adjudicating the debate.
- Standard of living under apartheid, as measured by biological indicators (see Living standards, health and social policy). Argues that heights show homeland poverty caused physical stunting that had previously gone unmeasured.
Findings¶
-
Foreign workers fell sharply as the mines lost Malawian and Mozambican supply, and domestic recruits rose. (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
-
Domestic African employment on the mines expanded almost threefold. (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 %.
-
Homelands were overcrowded and poor, and held half 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.
-
Mine wages for Africans rose sharply after the 1973 labour disputes, though they remained low. (p. 13)
Following labour disputes throughout 1973, the wage increased to R46 in 1974, R79 in 1975, R92 in 1976, R102 in 1977, R123 in 1978 and R141 in 1979
-
The preferred difference-in-difference result is a height gain of about 6 cm for men born in treated regions in the shock years. (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.
-
The male height effect is confined to the in-utero cohort; those born after the shock are shorter and earlier cohorts show no effect. (p. 25)
In column 1 I find that males born one to five years after the shock in the treated regions are 3.12 centimetres shorter than males born before the shock.
-
Only long-term health improves 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.
-
Women born in the treated regions at the shock are not taller; estimates are negative. (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.
-
The proportional-employment treatment variable gives a positive but insignificant height effect. (p. 24)
an increase of one percentage point in the proportion of workers in the mining sector leads to an increase in height of 0.68 centimetres
Evidence ledger¶
Sources used¶
| Source | Kind | Period | How used |
|---|---|---|---|
| National Income Dynamics Study (NIDS) Wave 1, 2008 adult sample and derived household data file | household or labour survey microdata | 2008 (Wave 1; released July 2009); birth cohorts 1941-1990 | Supplies adult height, birth year and district, education and health measures for the difference-in-difference regressions on the 1976-77 birth cohorts. |
| South African Institute of Race Relations Annual Reports/Surveys (1969-1979), mine employment tables | official statistical series | 1969-1978 | Gives the number of foreign and domestic African mineworkers by origin (Table 2), showing the fall in foreign labour and the rise in domestic recruits. |
| SAIRR Annual Surveys: mining monthly wages 1972-1979 | official statistical series | 1972-1979 | Shows the rise in African mining wages from R21 in 1972 to R141 in 1979, which is used to size the income shock. |
| Tomlinson Commission report (Summary of the Report of the Commission for the Socio-Economic Development of the Bantu Areas, UG 61/1955) | commission or official report | c. 1951-1955 | Provides baseline farm incomes, cattle holdings, erosion and carrying capacity in the homelands to show pre-shock poverty and the size of the income increase. |
| Crush, Jeeves and Yudelman 1991 (South Africa's Labour Empire) and Crush 1986: mine employment by homeland | secondary literature compilation | 1970-1978 | Dates and locates the employment increases by homeland and supplies the regional mining employment numbers for the continuous treatment variable in Table 6 column 5. |
| Butler, Rotberg and Adams 1978 (The Black Homelands of South Africa): homeland population and income estimates | secondary literature compilation | 1973 | Gives homeland population estimates (denominator for the proportion employed in mining) and 1973 income ranges. |
| Horrell (1969, 1973) SAIRR studies of African reserves and homelands | secondary literature compilation | 1969-1973 | Supplies population densities, land quality and remittance-share information for the homelands. |
| Wilson 1972 (Labour in South African Gold Mines) data on homeland conditions and remittances | secondary literature compilation | 1963-1972 | Provides the 18% official remittance share, overcrowding observations and recruitment patterns used to estimate household income gains. |
| Mauder 1982 (Atlas of Transkei) population estimate | secondary literature compilation | 1982 | Supplies the Transkei population estimate of 2.3 million used in the ad hoc calculation of the share of households affected. |
| 1970 Population Census of South Africa | population census | 1970 | Cited to show the poor apartheid-era data on Africans (different questionnaire, undercount), motivating the use of NIDS. |
| Abasebenzi workers' newsletter, 1973 | newspapers periodicals | 1973 | Cited for the Poverty Datum Line figure for a family to place homeland incomes in context. |
| Speeches of Paramount Chief Matanzima to the Transkeian Legislative Assembly, 1970 and 1972 | parliamentary debates | 1970, 1972 | Used qualitatively to show stagnant agricultural productivity and malnutrition in Transkei. |
| Municipal Demarcation Board boundary data and map collections (district councils 2011; homeland maps) | maps spatial data | 2011; apartheid-era homeland map | Used to match NIDS district councils of birth to former homeland boundaries and assign treatment status. |
| Personal conversation with Francis Wilson, October 2011 | interviews oral history | October 2011 | Supports the claim that workers also brought home part of their wages on leave beyond the official remittances. |
New data¶
Constructed a treatment indicator by matching NIDS Wave 1 respondents' district council of birth to the four affected former homelands, giving 10,433 Africans (4,098 men, 6,335 women), with regression samples of about 2,374 men and 3,865 women, and a small regional mining-employment proportion variable (N=460).
Inference¶
Quantification: Econometric (causal design).
Difference-in-difference OLS and ordered probit regressions comparing adults born in affected versus unaffected homeland districts in 1976-77 against other cohorts, with controls for birth year, province, age, language and father's occupation. The identifying assumption is that recruitment was exogenous to conditions in the homelands, and an ad hoc calculation of the income shock supports the foetal-origins interpretation.
Key claims¶
| Claim | Basis | Strength |
|---|---|---|
| Men born in affected homelands in 1976-77 are about 6 cm taller than other African men, supporting the foetal origins hypothesis and indicating severe stunting. (p. 7) | NIDS 2008 heights; DiD OLS regressions in Tables 4 and 6 (about 2,374 men) with a small treated N of 82. | Partial data |
| Mining recruitment of domestic workers rose sharply as foreign labour fell, in 1973-78. (p. 8) | SAIRR employment series (Table 2) and Crush et al. showing the timing and location of increases. | Systematic data |
| About 24% of households in the affected areas received a large income boost (at least R14.22 per month) against baseline farm incomes of R40-R120 a year. (p. 14) | An ad hoc calculation using Transkei worker numbers, Mauder's population estimate, Wilson's remittance share and Tomlinson Commission incomes; the author concedes the assumptions cannot be verified. | Illustrative or anecdotal |
| There is no height effect for men born before or after the shock, or who were toddlers or teenagers at the time, and no effect on education or most health measures; long-term health improved for men. (p. 25) | NIDS regressions in Tables 7-9 (OLS, ordered probit); only the long-run health coefficient is significant. | Partial data |
| Women born in the treated areas during the shock show no height gain and in fact appear shorter; there are no education or health effects. (p. 23) | NIDS regressions in Tables 5 and 10-12 (small treated N of about 137). | Partial data |
| The homelands were extremely poor and overcrowded before the shock, with few alternative employment opportunities. (p. 11) | Tomlinson Commission (1953-55 data), Horrell, Wilson and Matanzima's speeches, none later than 1973. | Partial data |
| The choice of recruiting homelands was independent of the outcome variables, and little migration into treated areas occurred, which supports exogeneity of treatment. (p. 6) | Argument from Leys (1975), Wilson (1972) and the pass laws; no test is run. | Asserted |
The author's caveats¶
The author notes that data collection in the homelands was unreliable and inadequate, and that the 1953 Tomlinson data is the latest for all homelands. The treated sample is small (82 men), which affects power, and stratification weights are ignored for conservative standard errors. Respondents who have never moved lack a birth location, so it is imputed from their household's location. Father's occupation is too aggregated to identify mining, treatment status is assigned from district councils, and Transkei and Ciskei cannot be separated. The income-shock calculation rests on unverifiable assumptions, and not everyone in the treatment group would actually have been treated.
Gaps in the evidence¶
The study has no direct measure of whether a respondent's father or household member worked on the mines, nor of household income or remittances in the 1970s, so the mechanism rests on an ad hoc calculation. Survival and migration selection of cohorts surviving to 2008 and residing in particular districts is not addressed, and the treated N is tiny, so a 6 cm effect is implausibly large. TEBA/Chamber of Mines recruitment records by district and individual-level mining employment histories existed at the time and could test the mechanism better. Contemporaneous nutrition or health surveys, such as clinic or school records, would give corroboration.
Replicability: Data public and well described.