Race-Space Inequality in South Africa: A Study in Welfare Geography¶

Literature › Works

David M. Smith (1972). Race-Space Inequality in South Africa: A Study in Welfare Geography. Reprint of a report first published in South Africa; the book or series containing the reprint is not identified in the extract.


Type	book chapter
Scope	Economics and economic history of apartheid
Method	mixed
Period studied	1915–1973
Regions	South Africa, Witwatersrand, Pretoria, Durban, KwaZulu, Ciskei, Transkei, Namibia
Themes	inequality poverty unemployment, bantustans removals influx, labour markets wages, demography urbanisation, education human capital, politics ideology elections, mining migrant labour, comparative theory, business and state
Citation key	`smith1972`

Argument¶

Smith treats apartheid as a deliberate spatial design with two parts: an integrated economic space that supplies black labour, and a segregated social and political space (homelands and group areas). Because race and space coincide, racial inequality in income and welfare is also geographical inequality. Using welfare-economics concepts (social welfare function, Pareto frontier, equity criteria), he argues that inequalities cannot be justified by contribution, need or merit. The costs of separate development fall mainly on non-Whites and the benefits mainly on Whites. Growth alone will not narrow the gap, so reform must change the distributive preferences of the white-controlled social welfare function. The framing is normative and illustrative, not a data-driven test.

Evidence and method¶

Secondary and published statistics. Horrell's SAIRR Survey of Race Relations 1972 supplies wages by race and sector, infant mortality, pensions and education spending. Randall's Spro-cas volumes supply the racial income shares and Gini coefficients for 1936-67. Board, Davies and Fair (1970) and Fair (1965, 1972) supply the principal-components welfare surface by magisterial district. Other sources include Berghe, Biesheuvel, Horner, Best and Young on KwaZulu consolidation, Wilson on migrant labour, press reports, and maps of the homelands and the Pretoria area.

Place in the debates¶

Liberal vs revisionist on apartheid and growth / economic efficiency (see Liberal vs revisionist: was apartheid functional or dysfunctional for capitalist growth?). Sits between the two. He argues that separate development conflicts with economic efficiency (it encourages dispersal and creates barriers), but that white capitalists' dependence on cheap African labour means the conflict 'may not be as great as is sometimes supposed'. Growth will not by itself end racial inequality (citing Adam).
Cheap labour thesis (see Cheap labour thesis and the function of the reserves). Supports a cautious version. Industry's quick concession of wage increases in the 1973 strikes and high returns on investment suggest wages were below free-market levels. He notes that labour economics is contested (Horner) and that capital/labour substitution rates are not known.
Whether apartheid raised the welfare of non-Whites (Pareto-improvement defence) (see Liberal vs revisionist: was apartheid functional or dysfunctional for capitalist growth?). Allows that all races may be better off in absolute terms. He judges that most of the benefits accrue to Whites and most of the costs to non-Whites, and he rejects the 'generosity' reading of homeland expenditure as the necessary price of white survival.
Homeland viability and the logic of separate development (see Homelands, separate development, decentralisation and removals). Homelands cannot support their existing populations, let alone returning migrants. The urban African population's status as 'foreign' homeland citizens is a major flaw in the policy's logic.

Findings¶

Racial income inequality was very high and had not narrowed from the 1930s to 1967 (Gini 54.2 in 1967). (p. 14)

It appears from these data that there has been no appreciable change in the inter-racial distribution since the 1930s; indeed, since the middle of the 1950s it may have become slightly more unequal.
Summary statistic of racial inequality: the Gini coefficient across racial groups is very high. (p. 14)

The Gini coefficient is very high -- more than half way up the 0 to 100 scale.
Sectoral wage differentials by race in 1971 were large, with whites earning far more than Africans. (p. 12)

The superior position of the Whites in all cases is clear; they generally receive twice or three times the Coloured or Indian earnings, and over five times the
The white/non-white wage ratio was essentially unchanged over half a century of industrial transformation. (p. 23)

The ratio of white to non-white wages was almost exactly the same in 1960-61 as in 1915-16: 4.71 compared with 4.84 (Berghe, 1967, 305).
Industry's quick concession of wage increases during the 1973 strikes suggests African wages were below free-market levels. (p. 16)

the readiness with which industry conceded an estimated R70 M per annum in increases to its African labor early in 1973 in the face of strikes, suggests that wages are at a substantially lower level than they would be in a free market.
Apartheid's political aims were largely compatible with white capitalist interest in cheap African labour, so the conflict with economic efficiency was smaller than sometimes supposed. (p. 23)

the interest of the white political powers in maintaining racial domination is to a large extent compatible with White economic interests, which rely heavily on cheap African labor for their profits.
Poverty was concentrated among Africans, with small shares of Whites affected. (p. 12)
Social services were sharply unequal by race: infant mortality was far higher for Africans and Coloureds. (p. 13)

Infant mortality is estimated to be 140 per 1000 live births among Africans, 130 for Coloureds, 40 for Indians and 22 for Whites
Per capita education spending was highly unequal by race. (p. 13)

Per capita educational expenditures are about R280 for Whites, 81 for Indians, 73 for Coloureds and 17 for Africans
Urban relocations under group-area policies were extensive, falling mainly on Africans. (p. 18)

The number of Africans moved within the urban areas may exceed one million;
Overall cost-benefit judgement: the benefits of separate development go mainly to Whites and the costs mainly to non-Whites. (p. 20)

most of the benefits accrue to the Whites and most of the costs to the Non-Whites, the latter thus paying for the maintenance of their own position of disadvantage.
Reducing inequality requires changing distribution rather than relying on production growth. (p. 24)

Any serious attempt to eliminate spatial welfare inequity must concentrate on the distributive process rather than on the production side.

Evidence ledger¶

Sources used¶

Source	Kind	Period	How used
Statement by Prime Minister B. J. Vorster in Parliament, as reported in the Johannesburg Star, 3 April 1973	newspapers periodicals	1973	Supplies the epigraph, Vorster's claim that the only difference between white and black is geographical, which frames the paper's argument.
Map 'Southern Transvaal Land Use 1973', Urban and Regional Research Unit, University of the Witwatersrand	maps spatial data	1973	Source for Figure 2, which illustrates racially segregated residential areas and buffer zones in the Pretoria area.
Letter of R.A.M. Saloojee, Rand Daily Mail, 19 February 1973	newspapers periodicals	1973	Quoted as an example of Indian resident feeling about forced removal to Lenasia.
Board, Davies and Fair (1970), principal components welfare index by magisterial district	secondary literature compilation	circa 1960s	Supplies the only national welfare surface (Figure 4) showing high welfare in metropolitan cores and lowest welfare in the Ciskei, Transkei and Zululand.
Fair (1965) regional GDP per head estimates, 1959-60	secondary literature compilation	1959-60	Used to show GDP per head of about R20 in the low-welfare reserve areas against R300 nationally and R567 in the major core areas.
Fair (1970) correlations of per capita income with demographic variables at magisterial district level	secondary literature compilation	circa 1960s	Shows that the percentage of Whites is a weak predictor of income at district level, while the share of working-age Non-White males predicts it well.
Duncan (1970), The Plight of the Urban African, SAIRR Topical Talks 23	secondary literature compilation	1970	Cited for symptoms of deprivation (poverty, crime, commuting, health services) among Johannesburg's African families.
Poverty datum line estimates (Watts 1971; Horner 1972; Horrell 1973)	secondary literature compilation	circa 1970-72	Used to estimate the proportions of each race group living in poverty (about 2% of Whites, half the Coloureds, most rural Africans).
Horrell et al. (1973), A Survey of Race Relations in South Africa, 1972 (SAIRR), average monthly earnings by sector and race, 1971	secondary literature compilation	1971 (wages); 1972 survey	Source of Table 1 on wages by race and sector, and of data on infant mortality, pensions, education expenditure and removals.
Agricultural Census No. 43 of 1968-69, Department of Statistics, R.S.A.	agricultural census	1968-69	Source of the farm-worker wage figures in Table 1 (1969, cash and kind).
Randall (1972a), Power, Privilege and Poverty (Spro-cas), Appendix B Table 2: racial shares of population and income 1936-1967	secondary literature compilation	1936-1967	Source of Table 2 and Figure 5 (population and income shares by race and a Gini coefficient of inter-racial inequality).
Biesheuvel (1972), The Black-White Gap (SAIRR)	secondary literature compilation	1972	Cited for high returns on investment in South African industry, taken to reflect low labour cost, and for slower real wage growth among Blacks in the 1960s.
Best and Young (1972), Homeland Consolidation: The Case of Kwazulu, and Rand Daily Mail, 30 April 1973	secondary literature compilation	1968-1973	Supply the numbers of Africans to be resettled under KwaZulu consolidation and the number moved in the 'homeland' resettlement programme, 1968-71.
Rand Daily Mail, 30 April 1973 (resettlement of Africans 1968-71)	newspapers periodicals	April 1968 - June 1971	Establishes the figure of about 209,000 Africans moved under the homeland resettlement programme.
Fair and Shaffer (1964), Population Patterns and Policies in South Africa, 1951-1960, index of population dislocation	secondary literature compilation	1951-60	Used to argue that Africans suffered much more dislocation than other groups in urban relocation.
van den Berghe (1967), South Africa: A Study in Conflict, white to non-white wage ratios 1915-16 and 1960-61	secondary literature compilation	1915-16 and 1960-61	Supports the claim that the white to non-white wage ratio was virtually unchanged across industrialization (4.71 against 4.84).
Adam (1971b), attitudinal study of South Africa's power elite	secondary literature compilation	circa 1971	Cited for the finding that elite members who have travelled abroad tend to be more 'enlightened'.

Inference¶

Quantification: Descriptive tables.

A conceptual welfare-geography argument (welfare function, utility-space diagram) illustrated with descriptive tables and statistics drawn from other authors' compilations, chiefly SAIRR surveys. It uses comparison across race groups, time-series comparison of income shares 1936-67, and normative reasoning about equity. The author did no primary data collection and acknowledges that spatially disaggregated data are lacking.

Key claims¶

Claim	Basis	Strength
Inter-racial income distribution has not changed appreciably since the 1930s and may have become slightly more unequal since the mid-1950s; the racial Gini is about 54. (p. 14)	Table 2, built from Randall (1972a) data on population and income shares for 1936, 1946-7, 1956-7 and 1967.	Partial data
Whites earn two to three times Coloured or Indian earnings and over five times African earnings across major sectors (1971). (p. 12)	Table 1 from Horrell (1973) and the 1968-69 Agricultural Census, covering about 4.6M of 8.3M employees.	Systematic data
Welfare is highest in the metropolitan cores and lowest in the Ciskei, Transkei and Zululand (GDP per head about R20 against R300 nationally in 1959-60). (p. 9)	Board, Davies and Fair (1970) principal components welfare index and Fair (1965) GDP per head estimates.	Partial data
Poverty is concentrated among Africans and Coloureds: about 2% of Whites, half of Coloureds and most rural Africans are below the poverty datum line. (p. 12)	Poverty datum line estimates cited from Watts (1971), Horner (1972) and Horrell (1973); the figures are rough orders of magnitude.	Illustrative or anecdotal
The costs of implementing separate development (removals, resettlement) fall disproportionately on Non-Whites. (p. 18)	Resettlement counts from Best and Young (1972), the Rand Daily Mail and Horrell (1973), plus the Fair and Shaffer dislocation index.	Partial data
Industrialization and economic growth have not reduced inter-racial inequality; the white to non-white wage ratio was almost unchanged between 1915-16 and 1960-61. (p. 23)	Berghe (1967) wage ratios plus the Table 2 income shares.	Partial data
Black wages are below free-market levels, as shown by industry's ready concession of about R70M in increases after the 1973 strikes and by high returns on investment. (p. 16)	Inferred from a press-reported wage concession and Biesheuvel (1972); no wage or productivity data are presented.	Illustrative or anecdotal
Most benefits of separate development accrue to Whites and most costs to Non-Whites, so the inequalities are unjust. (p. 20)	Normative welfare-theoretic judgement resting on the descriptive figures above; no formal cost-benefit calculation, which the author says is extremely difficult.	Asserted

The author's caveats¶

The author states that data are available for only a few relevant variables and that comparative data for racial groups disaggregated geographically are virtually non-existent. Shortages of data make it impossible to derive intra-urban welfare surfaces, so aggregate race-group data are used to infer spatial differences. The wage figures are probably more representative of cities than rural areas. He states that the focus is not on precise empirical identification and that the interpretations are tentative.

Gaps in the evidence¶

The argument infers spatial inequality from national racial aggregates, and the poverty and welfare figures rest on secondary compilations with uncertain provenance. District-level or township-level data by race were needed, such as population census tabulations by magisterial district and racial group, municipal income and expenditure data, and household income surveys. Some of these existed in the Department of Statistics censuses and SAIRR surveys but were not used directly. Claims about cost-benefit and homeland transfers lack budget or removal-cost series, which official budget accounts could supply.

Replicability: Data poorly documented.