Unemployment in South Africa: A survey of research¶
Norman Bromberger (1978). Unemployment in South Africa: A survey of research. Social Dynamics: A journal of African studies 4(1). Routledge (Taylor & Francis).
| Type | journal article |
| Scope | Economics and economic history of apartheid |
| Method | synthesis review |
| Period studied | 1946–1977 |
| Regions | South Africa, African homelands/Reserves, White farming areas, Johannesburg/Reef, Durban, Pietermaritzburg, Cape Town |
| Themes | inequality poverty unemployment, labour markets wages, methods data sources, bantustans removals influx, agriculture land, manufacturing industrial policy, mining migrant labour, demography urbanisation |
| Citation key | bromberger1978 |
Argument¶
Bromberger surveys the 1976-78 research literature on African unemployment. He shows that estimates differ widely because of conceptual disputes (unemployment versus underemployment, subsistence-sector work, 'voluntary' non-employment) and weak data (census undercounts, no direct employment series for homeland agriculture or the informal sector). Despite this, all the studies agree that unemployment is high (over 10% on most estimates) and rising through the 1970s, and that high combined unemployment and underemployment is not new. He argues that structural unemployment cannot be explained by cyclical factors or by low investment. He points instead to labour-force growth, capital deepening and distorted factor prices, the political and structural neglect of homeland agriculture, and segregationist controls. He ends with a policy package borrowed from Third World research and an agenda for research, including the new Current Population Survey.
Evidence and method¶
Secondary survey of unpublished and published macro-estimates (Sadie, van der Merwe, Simkins, Loots, Knight) built from the 1960 and 1970 censuses, Department of Statistics data and Reserve Bank series. Also small-scale surveys (Simkins, Loots, Market Research Africa, Maree and Cornell, Desmond) and the first Current Population Survey release (Statistical News Release P.27, Oct 1977). Reserve Bank capital-stock and investment series and the Theron Commission report are also used.
Place in the debates¶
- How to define and measure unemployment versus underemployment in a partly subsistence economy (see Unemployment: causes, measurement and labour market institutions). Favours the Simkins/Knight 'full-time job equivalent' approach for capturing labour underutilisation. He notes that Sadie and van der Merwe resist it, and that the CPS definition is narrow and excludes discouraged workers and migrants resting in the Reserves.
- Level and trend of South African unemployment (estimates ranging from Sadie's 4-8% of male manpower to Simkins's c.20%) (see Unemployment: causes, measurement and labour market institutions). Treats the level as unresolved, but argues that all studies agree on a rising trend in the 1970s and on levels likely above 10%. He judges Sadie's census-anchored estimate 'fatally open' to the objection that the unemployed were undercounted in 1970.
- Voluntary or 'preferred' non-employment among Africans (see Unemployment: causes, measurement and labour market institutions). Argues that the possibility deserves serious attention, as choices made within constraints (influx control, low pay, deferred pay, labour allocation). He rejects the view that Africans simply work as little as possible and says the bulk of unemployment is involuntary.
- Causes of structural unemployment: capital intensity and factor-price distortion versus labour supply growth (see Unemployment: causes, measurement and labour market institutions). Rejects low capital accumulation as the cause. Treats capital deepening, imported technology, high African wages, subsidies, the colour bar and political incentives to mechanise as plausible but under-researched explanations.
- Functional neglect of homeland agriculture for the capitalist (mining) economy (see Homelands, separate development, decentralisation and removals). Reports the view of de Klerk and Bell that homeland neglect served the capitalist economy but says it is 'not entirely proven'. He calls for research on homeland economies and Lipton's 'two agricultures'.
Findings¶
-
The first Current Population Survey (Oct 1977) found African unemployment of 634,000 on a deliberately narrow official definition. (p. 3)
constitute a pool of 634 000 persons (334 000 males and 300 000 females)
-
The official definition excludes discouraged workers, so even the CPS figure of 12.4% understates joblessness. (p. 3)
a chronically unemployed person (in the normal sense of the term) who has given up actively looking for work during the preceding month is not counted among the unemployed but among the not economically active
-
About half the African population had de jure residence in the homelands, which complicates the measurement of unemployment and underemployment. (p. 4)
Sadie estimates 44,8% for 1960 and 52,6% for 1970
-
Sadie's low estimate of male African unemployment rests on the 1970 census count, which Loots shows was undercounted. (p. 7)
Sadie sets the range of male African unemployment for mid-1976 as 170 000 to 338 000 men
-
Van der Merwe's results imply that the African labour force was growing much faster than employment. (p. 8)
no less than 57% of the increment (in the African labour force) had not been provided with employment
-
Simkins's successive revisions show how assumptions drive estimates; the first series was far lower in 1961 than the third. (p. 9)
The first series ran from 8,9% in 1961 to 20,0% in 1976, the second raised the 1961 unemployment rate to 16,3% and lowered the 1976 rate to 14,6%
-
Simkins's final series shows massive and growing combined unemployment and underemployment, unchanged by the 1960s boom. (p. 9)
rising from 1,24 million in 1960 to 2,30 million in 1977
-
Loots's sensitivity analysis bounds the plausible 1977 rate at 10-15%. (p. 10)
in 1977 South Africa probably had an unemployment rate of between 10%-15% of the labour force
-
Unemployment could not be stabilised without implausibly fast growth. (p. 11)
a growth rate of national income of 5,3% would be required to stabilise the unemployment rate and of 6,7% to stabilise the number of the unemployed
-
Educated unemployment was not yet a major component of African unemployment. (p. 11)
dominant educational category among the unemployed in most of the small survey samples has been formed by those with more than four years of education but less than a Junior Certificate
-
High investment rates mean that the rate of capital accumulation does not explain unemployment; the nature of the investment does. (p. 13)
the real fixed capital stock of South Africa increased at an average annual rate of 5,7% per annum
-
Capital deepening is evident in the 1970s, which suggests labour-saving technical change. (p. 13)
the rise in the average capital-output ratio from 2,17 to 2,67 between mid-1971 and mid-1977
-
Political motives to avoid dependence on Black labour may encourage mechanisation, including on farms. (p. 14)
with his new combine harvester, a farmer would be able to reap the harvest helped only by his wife if it became necessary
Evidence ledger¶
Sources used¶
| Source | Kind | Period | How used |
|---|---|---|---|
| Current Population Survey (CPS) - Bantu, October 1977, Statistical News Release P.27 (31 January 1978) | household or labour survey microdata | October 1977 | Supplies the first direct official measure of African unemployment (634 000; 12,4% of the economically active), used to frame the problem and its definitions. |
| Population Census 1970 (economic activity and self-reported unemployment) | population census | 1970 | Anchors the Sadie and van der Merwe unemployment estimates (119 000 African males self-described as unemployed) and is criticised for classification deficiencies. |
| Population Censuses 1946, 1951, 1960 and 1970 (employment and labour force) | population census | 1946-1970 | Knight uses the census years to estimate residual labour and underemployment over the long run; Simkins revises the 1960 census female homeland agriculture figures. |
| Department of Statistics non-farm employment series | official statistical series | 1960s-1970s | Alternative source of non-farm employment figures that conflicts with census industrial classification and explains the differences among Simkins' estimates. |
| Agricultural Censuses (casual labour on White farms) | agricultural census | annual | Measures casual farm labour at a point in time each year, which limits employment estimates for White agriculture. |
| Registered unemployment figures / labour bureaux statistics | official statistical series | 1970s | Discussed as an unsatisfactory source of unemployment data because many do not register or do not qualify. |
| Theron Commission Report RP 38/1976 (Coloured unemployment, Table 5.2 participation rates) | commission or official report | 1960-1970 data, report 1976 | Cited on the unreliability of registered Coloured unemployment and on declining Coloured male participation rates and rising female rates. |
| Census enumerator instructions, 1970 (as quoted by Loots 1977) | secondary literature compilation | 1970 | Shows that the census undercounted the unemployed because enumerators classified self-reported unemployed rural men and others as employed. |
| Sadie (1977) estimates of homeland residence, manpower growth and male unemployment | secondary literature compilation | 1951-1976 | Provides the 1975-76 estimate of African male unemployment (170 000-338 000) and manpower growth rates, which the review summarises and critiques. |
| van der Merwe (1976, 1977) unemployment estimates 1970-1976, as tabulated in Loots (1977: Table 1) | secondary literature compilation | 1970-1976 | Source of annual increases in unemployment and the 'high' and 'low' estimates reproduced in Table 2. |
| Simkins (1976, 1977, 1978a) estimates of unemployment 1960-1977 (Table 20) | secondary literature compilation | 1960-1977 | Supplies the main time series of labour supply, employment and unemployment (18-22%) reproduced in Table 3 and the growth-rate thresholds. |
| Tomlinson Commission income norm (as used by Simkins) | commission or official report | 1950s | Used by Simkins as the income norm dividing homeland agricultural output value to derive full-time job equivalents. |
| Loots (1977) alternative estimates 1970-1976 | secondary literature compilation | 1970-1976 | Four estimates under varying agricultural employment assumptions (1976 rates of 18,6%, 14,2%, 11,8%, 10,2%) show sensitivity of unemployment levels. |
| Knight (1977) 'Is South Africa running out of unskilled labour?' / SALDRU Working Paper No. 7 | secondary literature compilation | 1946-1970 | Long-run estimates of residual labour and underemployment across census years. |
| Small-scale unemployment surveys: Market Research Africa (1977) African Omnireef Unemployment Study; Loots (1978); Simkins (1978c) urban and rural surveys | author survey or fieldwork | 1976-1978 | Provide evidence on age, education, migrant status, support and local unemployment rates (16%-42%). |
| Maree and Cornell (1978) Crossroads squatter sample survey; Maree (1977); Desmond (1978) Limehill | author survey or fieldwork | 1977-1978 | Evidence on informal sector involvement (about 10% of economically active males) and unemployment in resettlement and squatter areas. |
| South African Reserve Bank, Quarterly Bulletin (December 1977) and Annual Economic Report 1977; de Jager (1973) capital stock series | national accounts macro | 1956-1977 | Used for fixed investment ratios, capital stock growth, capital-output and capital-labour ratios to examine capital intensity. |
| Farm labour studies: Evans (1976) Viljoenskroon; van der Vliet and Bromberger (1977) Albany district | author survey or fieldwork | 1970s | Evidence of female casual farm workers having as much work as they want or reducing work, offered as indications of voluntary non-work. |
| Mining industry labour recruitment observations (gold mines 1975-77) | chamber of mines or recruiting records | 1975-1977 | Cited as a puzzle that gold mines faced contract-breaking and labour shortages while measured unemployment rose. |
Inference¶
Quantification: Descriptive tables.
Literature survey that compares and reconciles competing macro-estimates of unemployment (indirect labour force minus employment method) and small-scale surveys, discusses definitional and data problems, and then offers theoretical explanatory hypotheses (labour supply, capital accumulation, technology, urban-rural gap, education mismatch) without testing them.
Key claims¶
| Claim | Basis | Strength |
|---|---|---|
| The first CPS shows 12,4% of the African economically active population (634 000) unemployed in October 1977 under a narrow definition. (p. 13) | CPS October 1977, Statistical News Release P.27 (Table 1). | Systematic data |
| Sadie's and van der Merwe's estimates rest on the 1970 census self-classification, which undercounted the unemployed and is therefore no reliable anchor. (p. 18) | 1970 Census; enumerator instructions quoted by Loots (1977). | Partial data |
| Unemployment and underemployment, on Simkins' full-time job equivalent measure, were about 19% in the 1960s and above 20% in the 1970s, rising from 1,24 million to 2,30 million. (p. 19) | Simkins (1978a) indirect estimates from population, employment and output series. | Partial data |
| Estimates are highly sensitive to assumptions about agricultural employment, but all point to rates above 10%, more than a million unemployed and a rising trend in the 1970s. (p. 20) | Loots (1977) variations on Simkins' assumptions; comparison of all macro studies. | Partial data |
| Small surveys indicate unemployed people are mostly settled urban residents with limited education, supported by families, and that educated-youth unemployment is not yet significant. (p. 21) | Simkins (1978c), Loots (1978), Market Research Africa (1977) sample surveys. | Partial data |
| Low capital accumulation does not explain unemployment; capital deepening and distorted factor prices are more plausible, though untested. (p. 23) | Reserve Bank investment ratios, capital stock growth and capital-output/capital-labour ratios. | Illustrative or anecdotal |
| Structural unemployment is unlikely to fall because national income growth of 5,3% would be needed to stabilise the rate and 6,7% to stabilise the number unemployed. (p. 21) | Simkins (1978a: 41) model results. | Partial data |
| Some measured unemployment may be voluntary or preferred, given migrant labour options, low wages and influx controls, though the bulk is involuntary. (p. 22) | Sadie (1977), farm labour studies, Coloured participation rates and gold mine labour-shortage observations; mostly argument. | Illustrative or anecdotal |
The author's caveats¶
Bromberger stresses that the CPS definition is narrow and the first survey coincides with a severe recession; that census data suffer from underenumeration, infrequency and classification problems; that homeland agriculture, casual farm labour and the informal sector have no direct employment measures; that indirect series are smoothed by interpolation; and that small surveys do not establish the national rate.
Gaps in the evidence¶
The review rests almost entirely on secondary and working-paper estimates rather than original data, so the macro-estimates cannot be independently checked and their assumptions (homeland production norms, casual labour equivalents) remain unverified. The causal explanations in section 8 are hypotheses without any tested data on the unemployed, firm-level technology choices or migration histories. Direct data on hours worked, homeland production and informal sector activity would test the arguments better; the CPS monthly series would supply some of this, and later census, labour force and household survey microdata now exist.
Replicability: Data poorly documented.