The Demography of South Africa¶
Tukufu Zuberi; Amson Sibanda; Eric O. Udjo (eds.) (2005). The Demography of South Africa. A General Demography of Africa (series). M.E. Sharpe, Armonk, NY and London.
| Type | book |
| Scope | Wider apartheid literature |
| Method | mixed |
| Period studied | 1911–2000 |
| Regions | South Africa, former TBVC homelands, Southern African Development Community (SADC) labour-sending states |
| Themes | demography urbanisation, methods data sources, bantustans removals influx, mining migrant labour, labour markets wages, inequality poverty unemployment, education human capital, transition post apartheid, origins segregation |
| Citation key | zuberi2005 |
Argument¶
This edited volume uses the 1996 census, October Household Surveys and the partial vital registration system to map South Africa's demography and its racial inequalities after apartheid. Chapters on population statistics, fertility, mortality, HIV/AIDS, life tables, households, education and migration/employment share a theme: apartheid's racial classification, influx control, homelands, migrant labour and unequal social provision shaped both the demographic outcomes and the quality of the data used to measure them. Analysis is mostly descriptive and indirect-estimation work (Brass methods, Gompertz models, logistic regression). It is a demographic and data-quality contribution that offers context rather than economic argument.
Evidence and method¶
1996 South African Population Census (10% and 30% samples, full count); 1991 census micro-data; October Household Surveys 1995, 1997, 1998; 1970 census; birth and death registration and the population register; Stats SA life tables 1921-1996; archival and statutory sources on racial classification and censuses.
Place in the debates¶
- Race as biological category vs social construct in census/demographic practice (see Census, race classification and demographic data). Khalfani et al. treat racial classification as a state-built social construct that served stratification and distorted statistics. They nonetheless support retaining race categories post-1991 for redress and demographic analysis.
- Were apartheid-era population projections inflated (overestimating African fertility)? (see Census, race classification and demographic data). Udjo argues African fertility in the 1970s-80s was overestimated, partly for political reasons, which explains the lower-than-projected 1996 census count.
- Quality of 1991 census and size of undercount (see Census, race classification and demographic data). The editors argue that the 1991 census seriously undercounted Africans, and that demographic inflation of African males may have been shaped by political aims. They treat 1996 as the most complete enumeration.
- Extent of HIV/AIDS deaths and attribution of rising adult mortality (see Census, race classification and demographic data). Bah finds estimates sensitive to the cause-of-death lists used, with HIV/AIDS the leading cause of death by the 2002 sample. Udjo says the evidence does not support attributing rising child mortality mainly to HIV.
- Immigrants and nationals in the post-apartheid labour market (see Post-apartheid economic policy, governance and corruption). Zuberi and Sibanda find foreign-born and internal migrants more likely to be employed than South African-born nonmigrants, contradicting the view that immigrants are at a disadvantage.
Findings¶
-
Under the Union, early racial laws were designed to favour white labour while controlling Africans. (p. 9)
The first oppressive laws of the Union were designed to benefit European laborers, while segregating and controlling Africans.
-
The 1991 census counted barely half of Africans, and undercount adjustments were politically contested. (p. 15)
The CSS counted only 54 percent of the African population in the 1991 census enumeration.
-
Homeland 'independence' was largely symbolic, and homeland populations remained the labour force of the Republic. (p. 18)
The populations within the TBVC states and the SGT areas were predominantly African, and they continued to serve as the dominant labor force within the Republic of South Africa.
-
Homeland census results mattered financially because population size set customs revenue transfers. (p. 19)
The census results were used to determine how much money the Republic of South Africa paid these governments from customs revenues.
-
The 1996 census was far more complete than earlier ones. (p. 41)
Stats SA enumerated 93 percent of the population.
-
Fertility fell for all groups, with African TFR declining from about 5.4 in 1970 to 3.7 in 1996. (p. 55)
total fertility rates declined from about 5.4 in 1970 to about 3.7 in 1996 among Africans.
-
Apartheid-era projections overstated African fertility, partly for political reasons. (p. 57)
It would appear that the overestimation of fertility rates of Africans during the apartheid era was, at least in part, the result of the political environment.
-
Apartheid-era life tables were flawed by assuming complete death registration, so improvements in health and living standards were not reflected. (p. 213)
The source of these errors is largely in the faulty assumption of death registration being complete from 1936 onward.
-
Apartheid influx control, hostels and migrancy broke up African families and shaped household structure. (p. 221)
The whole economic system was, thus, set up to deliberately create unstable jobs, families, and households among the African population.
-
Household and mortality patterns reflect migrant-labour-driven absence of men from the former homelands. (p. 175)
men are "missing," a phenomenon attributed in the past to higher male mortality resulting from occupational hazards such as mining
-
Migrants and immigrants have better employment odds than nonmigrants; the recruiting history of the SAMS is documented. (p. 269)
The discovery of gold and diamond mines in South Africa in 1886 led to a huge demand for cheap labor, which could not be met locally.
Evidence ledger¶
Sources used¶
| Source | Kind | Period | How used |
|---|---|---|---|
| South African population censuses 1911-1996 (official census reports and questionnaires, racial classification tables) | population census | 1911-1996 | Traces the changing racial classifications and population sizes by race across censuses. |
| 1991 South African census micro-data and CSS weights, with HSRC aerial/survey adjustments (Report No. 03-01-26, Adjustment for Undercount) | population census | 1991 | Quantifies the share of the 1991 count from enumeration, survey, aerial photography and demographic inflation, and the resulting sex ratios. |
| 1996 South African Population Census (full count, 10% and 30% samples of micro-data; Report No. 03-01-18 undercount and 03-01-17 The Count and How It Was Done) | population census | 1996 | Main data base for fertility, mortality, orphanhood, household structure, education, migration and employment analyses across the volume. |
| 1970 population census and Report on Births 1964 to 1971 (birth registration) | population census | 1970 | Supplies 1970 fertility estimates for Africans (census births) and non-Africans (birth registration) as the starting point of the fertility trend. |
| October Household Surveys 1995, 1997, 1998 (national samples of 30,000/20,000 households) | household or labour survey microdata | 1995-1998 | Provides birth histories and mortality reports for P/F, Gompertz and orphanhood analyses. |
| 1991 South African census micro-data (full count) for household structure comparison | population census | 1991 | Compared with 1996 census to document changes in household size, headship and type by race. |
| Statistics South Africa vital statistics: Population Register (DHA) death data and Stats SA death notification database 1993-2000 | official statistical series | 1993-2000 | Basis for adjusting population register deaths and projecting HIV/AIDS-related deaths by age and sex. |
| Stats SA 15 percent sample of death notification microfilm rolls, Causes of Death in South Africa 1997-2001 (P0309.2) | official statistical series | 1997-April 2002 | Benchmark showing HIV/AIDS among leading causes of death, against which the projection methods are judged. |
| Official South African life tables 1921-1985 and 1985-94/1996 (Union Office of Census and Statistics 1923; Census and Statistics Office 1939; CSS 1987, Report 02-06-04; Stats SA 2000) | official statistical series | 1921-1996 | Appraised for method and completeness of death registration; published life expectancies and l(x) curves analysed. |
| Statutes and Government Gazettes (Population Registration Act 1950, Census/Statistics Acts, Births and Deaths Registration Acts 1923/1963/1992, Group Areas, Bantu Homelands Citizenship Act etc.) | legislation gazette | 1911-1998 | Documents the legal framework of racial classification, population registration and vital registration. |
| South African State Archives files (Moffat letter 1911; OCS radio broadcasts 1925; Statistical Council minutes 1921, SES files) | state archives | 1911-1925 | Used to show the aims of the census office and the Statistical Council's choice of four racial categories. |
| Newspaper and press coverage of the 1991 census (The Star, Transvaler, Bureau for Information press release) | newspapers periodicals | 1991 | Evidence of ANC opposition and the CSS defence of collecting racial data in 1991. |
| Interview with Wole Adegboyega, UNFPA census adviser (Dec 8, 1995) | interviews oral history | 1995 | Source for the rationale for relabelling the African category in the 1996 census. |
| Chimere-Dan (1993) review of fertility estimates; Sadie (1988) projections; Sibanda and Zuberi (1999) reconstructed census birth histories; Dorrington et al. (1999) | secondary literature compilation | 1970-1996 | Previous TFR estimates compared with the author's estimates to show overestimated African fertility in apartheid-era projections. |
| South Africa Demographic and Health Survey 1998 preliminary report | household or labour survey microdata | 1998 | Cited TFR of 2.9 and rising child mortality trend, used as a comparison. |
| HIV antenatal clinic and adult prevalence figures (Department of Health 1999; Shisana and Simbayi 2002) | official statistical series | 2000 | Regressed against provincial infant mortality estimates to test whether HIV explains the rise in infant mortality. |
| AIDSproj seroprevalence projection (Futures Group International) and CDC 1993 HIV definition / Stats SA coders' cause-of-death list | other | 1993-2000 | Supplies growth rate of AIDS cases and the two cause-of-death lists used to define HIV/AIDS-related deaths. |
| Lesotho adult male mortality (Timaeus 1984, WFS Scientific Reports no. 59) | secondary literature compilation | 1960s | Benchmark showing implausibly high African male mortality derived from the 1996 census. |
| US life tables and vital registration documents (Glover 1921, Greville 1967, NCHS publications, NOVS) | official statistical series | 1900-1992 | Comparative benchmark for how the US handled registration completeness in building life tables. |
| Southern African Migration Project and other literature on migrant labor, TEBA recruitment, SADC migration | secondary literature compilation | 1886-1999 | Context for migrant labour history and the argument about immigrant advantage in employment. |
Inference¶
Quantification: Econometric (causal design).
Descriptive demography and indirect estimation (Brass P/F, relational Gompertz, orphanhood and childhood mortality methods, growth balance), logistic and multinomial regression on 1996 census samples with covariates, cross-census comparison 1991-1996, polynomial projection of HIV deaths with ex-post/ex-ante adjustment, and a historical-documentary review of racial classification and vital registration.
Key claims¶
| Claim | Basis | Strength |
|---|---|---|
| The 1991 apartheid census badly undercounted Africans and relied on HSRC aerial/survey inflation and demographic adjustment, while the 1996 census was far more complete (93% enumerated). (p. 15) | 1991 micro-data weights, undercount reports, 1996 undercount report; Tables 1.3-1.5 | Systematic data |
| African fertility in the 1970s-80s was overestimated in apartheid-era projections while that of other groups was slightly underestimated; TFR fell from about 5.4 to 3.7 among Africans, 1970-96. (p. 57) | P/F and relational Gompertz fits to OHS 1995-98, 1996 census, 1970 census and birth registration compared with Chimere-Dan figures | Partial data |
| Child mortality among Africans is roughly twice that of other groups, and childhood mortality appears to have risen since the 1990s, but the data are error-prone and rise not mainly explained by HIV. (p. 110) | Indirect estimates from 1996 census children ever born/surviving; provincial regression of infant mortality on HIV prevalence | Partial data |
| HIV/AIDS and related causes account for a growing share of natural deaths among adults 15-49, rising to roughly a quarter (CDC list) or about 35% (Stats SA list) by 2000; HIV/AIDS ranked fourth among male and first among female causes in the 2002 sample. (p. 146) | Population register and Stats SA death data, projections, and the 15% sample of death certificates | Partial data |
| Pre-1985 official life tables assumed complete death registration, understated mortality and failed to show improvements in life expectancy, unlike US tables. (p. 213) | Published life tables 1921-85, life expectancy tables, l(x) curves, comparison with US practice and registration history | Partial data |
| Apartheid-era residence and labour controls shaped African household structure; after 1991 household sizes fell, one-person households rose sharply and African women are more likely to head households, net of covariates. (p. 228) | 1991 full count versus 1996 30% sample cross-tabulations and logistic regressions of female headship | Systematic data |
| Racial gaps in school completion are far larger than gender gaps, reflecting unequal apartheid-era resource provision, while enrollment gaps are small. (p. 263) | 10% sample of 1996 census, gap and ratio tables by province and race | Systematic data |
| Foreign-born immigrants and internal migrants have higher odds of employment than South African-born nonmigrants. (p. 287) | Logistic regressions on 30% sample of 1996 census, labour force aged 20-55 | Systematic data |
The author's caveats¶
Authors note age misreporting, omission of dead children, adoption and absentee effects in orphanhood data, poor quality of the 1997 OHS, truncated birth histories, undercount and TBVC exclusion in 1991, lack of cause-of-death data after 1996, cross-sectional census limits for migration and no information on past attributes or illegal migrants.
Gaps in the evidence¶
The volume rests almost wholly on the 1996 census and a few surveys, so claims about apartheid-era processes are inferred retrospectively rather than tested with contemporaneous data. Dedicated micro-data on African births and deaths before 1990, migrant-worker records or labour-market panels would better test causal links between apartheid controls and household, fertility or mortality outcomes; some archives existed then, others never recorded Africans. The 2001 census, available later, enables intercensal checks that the authors note but could not use.
Replicability: Data described but not public.