Long-term Effects of Apartheid's Homeland System on Social Capital in South Africa¶
Martin Abel (2014). Long-term Effects of Apartheid's Homeland System on Social Capital in South Africa. Harvard University (draft, with comments).
| Type | working paper |
| Scope | Wider apartheid literature |
| Method | econometric |
| Period studied | 1948–2010 |
| Regions | South Africa, former Bantustans (Ciskei, KwaZulu, Gazankulu, Venda, Lebowa, Bophuthatswana, Qwaqwa) |
| Themes | bantustans removals influx, politics ideology elections, demography urbanisation, comparative theory, methods data sources, transition post apartheid |
| Citation key | abel2014a |
Argument¶
The paper asks whether apartheid's forced removals to the homelands (about four million people) and its policies of ethnic division left lasting marks on social capital, meaning trust, national identity and trust in local authorities. It geocodes Afrobarometer respondents (2000-2010) and compares people in former homelands with other Africans. It also compares people near Surplus People Project resettlement camps with others inside the homelands. People in former homelands are less trusting of other South Africans and less proud of being South African, and the effect is mostly confined to those born before 1975, so these attitudes were apparently not passed on. Areas near resettlement camps show higher trust towards relatives, neighbours and both own and other ethnic groups. The author reads this as long-run learning from forced ethnic mixing.
Evidence and method¶
Pooled Afrobarometer South Africa survey waves 2000-2010 (about 2,400 respondents per wave, 92% geocoded). Surplus People Project 1983 data on 66 resettlement camps, with a camp list in Table 7. Secondary literature (Rogers 1980, Platzky and Walker 1985, Posel 2011, Thompson 2001) and newspaper quotations. NIDS 2008 for urban/rural trust comparisons.
Place in the debates¶
- Homelands as cheap labour reserves vs. other legacies of the Bantustan system (see Long-run legacy and persistence of apartheid). Acknowledges the cheap-labour-reserve function as the focus of the existing literature but deliberately studies other long-run effects on social capital and identity.
- Persistence of historical shocks and intergenerational transmission of trust (Nunn-Wantchekon, Guiso et al., Tabellini) (see Long-run legacy and persistence of apartheid). Finds attitudes shaped by removals only among those who experienced them, so no transmission to later cohorts. Resettlement camp areas show higher trust, which cuts against a simple persistence-of-mistrust story.
- Effectiveness of divide-and-rule ethnic engineering (see Long-run legacy and persistence of apartheid). Finds no higher own-group trust and no lower trust in other groups in homelands generally, so apartheid's ethnic division was not effective in the long run. National identity is weaker among older homeland residents.
- Pragmatists vs. purists within apartheid (segregation vs. labour needs) (see Origins, coherence and ideology of apartheid and Afrikaner nationalism). Adopts Posel's framing of tension between political supremacy and the economy's need for African labour, and treats the Bantustan system as a compromise between the two.
Findings¶
-
Estimates that around four million Africans were forcibly relocated to homelands in the 1960s-1980s. (p. 1)
apartheid regime forcefully relocated some four million Africans to rural homelands between 1960 and 1980
-
Notes that removal figures are only estimates, though most studies put them at four million or more. (p. 2)
most studies estimate that a total of at least four million people were removed to Bantustans over the next decades
-
Data check: SPP camp data cover roughly 78% of the displaced under an extrapolation assumption. (p. 10)
the SPP resettlement camp data would capture about 78% of the total number of 3.5 million people displaced
-
The main treatment definition is a 10km radius around SPP camps, covering a minority of the homeland population. (p. 10)
I use a 10km radius around camps which includes 23.5% of the homeland population
-
Trust results are mixed: lower trust in homelands overall, concentrated among older people, with variation within homelands. (p. 12)
overall people in former homeland areas seem to have lower levels of trust compared to people living in other areas
-
Crime fear and victimisation are higher in former homelands, driven by those who experienced removals. (p. 12)
People in Bantustans are both more afraid of crime and experience more theft and violence compared to those living outside former homeland areas
-
Resettlement camp areas are much more ethnically fragmented than other homeland areas, contrary to the regime's aim of homogeneous homelands. (p. 15)
The average fragmentation index for areas near resettlement camps is 0.294 compared to 0.079 of areas further away from the camps
-
People near camps trust other ethnic groups more than other homeland residents do. (p. 15)
people living in resettlement areas trust members from other ethnic groups more than those living in other homeland areas (10)
-
National identity is weaker in former homelands. (p. 17)
both of these measures of national identity are lower for people living in former homelands than in the rest of South Africa (11, 16)
-
Trust in chiefs and local government is not eroded in former homelands. (p. 17)
This is evidence against the hypothesis that living under the formal rule of tribal leaders led to a long-term erosion in the trust of local leaders.
-
Effects of removals appear confined to those who lived through them. (p. 20)
but that these values were not passed on to future generations
Evidence ledger¶
Sources used¶
| Source | Kind | Period | How used |
|---|---|---|---|
| Afrobarometer surveys, South Africa, six waves (2000-2010/11) | household or labour survey microdata | 2000-2010 (waves 1-6) | Main outcome data on trust, crime, national identity and institutional trust, regressed on homeland residence and distance to resettlement camps. |
| Surplus People Project (SPP) report, Forced Removals in South Africa, Volumes 1-5 (1983) and field research 1980-83 | secondary literature compilation | 1980-1983 | Source of locations and size of 66 resettlement camps, geocoded to compute respondents' distance to the nearest camp and define the 10km treatment. |
| Tomlinson Commission report (1954) | commission or official report | 1954 | Cited for the claim that homeland population exceeded what the areas could sustain. |
| National Income Dynamics Survey (NIDS) 2008 | household or labour survey microdata | 2008 | Used to argue trust levels towards neighbors and strangers are nearly identical in urban and rural areas, countering a density confound. |
| Newspaper reports: The Voice, Financial Mail, The Star, Daily Dispatch, Cape Times, The World (1968-1979) | newspapers periodicals | 1968-1979 | Anecdotal evidence on conditions in resettlement camps, crop declines, gang violence, minister statements and unrest against chiefs, mostly cited via Rogers (1980). |
| Rogers (1980), Divide and Rule: South Africa's Bantustans (quoting official statements, White Paper 1959, budget shares, election turnout) | secondary literature compilation | 1950s-1980 | Principal compiled source for descriptive claims on resettlement conflict, chiefs, budget dependence and low homeland election turnout. |
| Apartheid legislation: Group Areas Act 1950, Population Registration Act 1950, Bantu Authorities Act 1951, Bantu Resettlement Act 1954, Bantu Self-Government Act 1959, Bantu Homeland Constitution Act 1971, Second Bantu Laws Amendment Act 1974 | legislation gazette | 1950-1974 | Establishes the legal basis of the homeland system and chiefs' limited powers. |
| Estimates of total removals (SPP 1983; Unterhalter 1987; Mare 1980) | secondary literature compilation | 1960-1983 | Supports the headline figure of about four million people removed to Bantustans, and the 3.5 million displaced by SPP fieldwork used for coverage calculations. |
| Posel (2004) on post-apartheid migration patterns | secondary literature compilation | post-1994 | Supports the identification assumption of no large-scale selective migration out of homelands after apartheid. |
| Homeland boundary and resettlement camp geocoded map data (ArcGIS, Figures 1-3, 6) | maps spatial data | 1980-83 camps; 2000-2010 survey | Computes distance of respondents to nearest resettlement camp and creates maps of former homeland areas and kriged social capital outcomes. |
New data¶
Author geocoded 92% of Afrobarometer South Africa respondents and built a spatial dataset of 66 SPP resettlement camps (with relocated numbers where available), merging them with former homeland boundaries to compute respondent distance to camps, plus ethnic fragmentation indices by homeland.
Inference¶
Quantification: Econometric (causal design).
OLS regressions on pooled repeated cross-sections of Afrobarometer, comparing residents of former homelands to others (full country and 50km buffer), with age-cohort interactions (born after 1975) and within-homeland comparisons by distance (10km) to SPP resettlement camps with homeland fixed effects. Identification is by covariate control and spatial comparison, not a natural experiment; robustness by dropping homelands and kriged maps.
Key claims¶
| Claim | Basis | Strength |
|---|---|---|
| People in former homelands are less proud to be South African and less supportive of a united nation than people elsewhere. (p. 17) | Afrobarometer OLS, homeland dummy vs rest of SA and vs 50km buffer (Table 4 columns 11-12, 16-17). | Partial data |
| Living near former resettlement camps is associated with higher trust in own and other ethnic groups, relatives, neighbors and people in general. (p. 15) | Afrobarometer combined with SPP camp locations; 10km camp dummy with homeland fixed effects (Tables 2, 4, 6). | Partial data |
| Areas near resettlement camps are more ethnically fragmented than other homeland areas, which may explain higher inter-ethnic trust. (p. 15) | Fragmentation index from Afrobarometer language data: 0.294 within 15km of camps (N=820) vs 0.079 further away (N=1088). | Partial data |
| Perceived and actual crime are higher in former homelands, driven by older respondents. (p. 12) | Afrobarometer crime items (Table 3) regressed on homeland dummy and age interaction. | Partial data |
| Homeland rule under chiefs did not persistently erode trust in chiefs or local government. (p. 17) | Afrobarometer trust in chiefs and local government, homeland vs 50km buffer (Table 5); small samples for chiefs. | Partial data |
| Forced resettlement created conflict and hardship in homelands, e.g. Qwaqwa's population growth from 21,000 to 200,000 and endemic violence. (p. 4) | Secondary and newspaper accounts via Rogers (1980) and The Voice; no systematic data. | Illustrative or anecdotal |
The author's caveats¶
Author notes that the apartheid regime did not collect data on numbers removed, and that the accuracy and representativeness of SPP data is unclear. SPP did not do fieldwork in Transkei or in all homelands, and KwaZulu camps lack relocation numbers. Afrobarometer lacks migration data and the author cannot test whether resettlement sites had higher pre-removal trust. Extrapolates SPP coverage to about 78% of total displaced under an assumption of similar removals elsewhere. Notes ethnic fragmentation fluctuations may reflect sampling variation, and kriging is unreliable where observations are sparse.
Gaps in the evidence¶
Identification rests on cross-sectional comparisons with no pre-removal baseline for trust or social capital, so homeland and camp location effects cannot be separated from selection of sites and non-random camp placement. Individual-level data on whether respondents (or parents) were actually removed, and on post-1994 migration, would test the mechanisms and cohort claims better; some exists in later NIDS/census waves and SPP-type archival records. Many age-interaction and camp coefficients are insignificant with small samples, and conflict claims rely on anecdote rather than contemporaneous incident data.
Replicability: Data described but not public.