African population, employment and incomes on farms outside the reserves, 1923-1969¶
Charles Simkins (1984). African population, employment and incomes on farms outside the reserves, 1923-1969. Second Carnegie Inquiry into Poverty and Development in Southern Africa, Carnegie Conference Paper No. 25 (Cape Town, 13-19 April 1984). SALDRU, University of Cape Town.
| Type | conference paper |
| Scope | Economics and economic history of apartheid |
| Method | descriptive quantitative |
| Period studied | 1923–1969 |
| Regions | South Africa (white farming areas), Eastern Cape/Ciskei, Natal/Zululand, Transvaal, Orange Free State |
| Themes | agriculture land, labour markets wages, bantustans removals influx, demography urbanisation, inequality poverty unemployment, methods data sources |
| Citation key | simkins1984 |
Argument¶
Simkins reconstructs the size, employment and incomes (cash, kind and own production) of Africans living on white-owned farms outside the reserves, 1923-1969. He does this by interpolating Population Census figures and combining them with Agricultural Census and other data. He argues that the Agricultural Censuses increasingly undercounted farm population after 1955, because Africans were increasingly in de facto occupation of unoccupied farms or parts of holdings. Farm employment, however, was measured reasonably well. Employment as a share of on-farm labour supply fell after 1955, and this helps explain why the labour-bureau and influx-control system lost its usefulness to farmers and gave way to expulsion and resettlement. Cross-sectional maps show a low-wage, low-cash 'backward' zone adjacent to the major reserves. Simkins uses these maps to qualify Lipton's 'two agricultures' thesis, proposing three overlapping agricultures instead.
Evidence and method¶
Six Population Censuses (1921-1970), Agricultural Censuses (1918-1969, including 1955, 1960 and 1965 population counts), Union Statistics for Fifty Years, the Report of the Native Farm Labour Committee 1937-39, the Natal Economics Department's 1936 estimate, 1980 Current Population Survey participation rates, and the 1959 Commission on European Occupancy of the Rural Areas. Magisterial-district cross-sections for 1955 and 1960 are also used.
Place in the debates¶
- Lipton's 'two agricultures' (efficient white vs backward black sector) (see Agricultural policy, productivity, farm size and land reform). Qualifies it: transitional belts are extensive and income ranges narrow, so he proposes three agricultures shading into one another (progressive capitalist, backward capitalist, reserve).
- Causes of the post-1955 shift from retention to removal of farm labour (state vs farmer pressure; Morris) (see Farm labour: mechanisation, labour shedding, wages and labour control). Accepts the argument that white farmer pressure helped bring in the labour-bureau/influx control system, but notes that labour demand levelled off within a decade, making it unnecessary to farmers.
- Reliability of farm-wage valuations in kind (see Data reliability and econometric method). Acknowledges that critics say farmer valuations are too high while others say many goods and services go uncounted; he uses farmer valuations and excludes the value of land and housing.
Findings¶
-
Farm employment as a share of on-farm labour supply fell sharply after 1955 because employment levelled off while the labour supply kept growing. (p. 9)
This reflects the levelling off of employment in absolute terms at a stage when labour supply continues to
-
The Agricultural Censuses recorded far fewer Africans on farms than the Population Censuses did in 1955, 1960 and 1965, which Simkins attributes mainly to de facto African occupation outside white supervision. (p. 3)
The Agricultural Census also counted the total African farm pop..llation in 1955, 1960 and 1965 and found 2119435, 2144085 and 1776952 persons respectively.
-
The number of holdings fell far faster than the area held, which is consistent with growing African occupation of the land on farms. (p. 5)
the number of holdings fell fran a peak of 106 220 in 1959 to 91 886 ten years later
-
The pre-war cash-and-kind income estimate for 1937 is validated by a contemporary University of Natal estimate. (p. 12)
yields a total incane in cash and kirrl- in 1937 of £9.66 million, which agrees very well with an estimate of £9.73 million made for .1936 by the Department of Economics at the University of Natal.
-
Own agricultural production became a much smaller share of African farm income, falling from about 28% in 1929 to about 10% in 1969 as wage incomes rose faster. (p. 15)
the share of production in 1969 was aJ:out 10%, canpared with the 28% of forty years earlier
-
Production per head of Africans on farms stayed at a roughly stable fraction of production per head in the reserves. (p. 15)
the former fluctuated without trerrl at about 59% of the latter
-
Regional income per worker and the cash share of income were correlated in 1955, with low-wage districts paying less in cash. (p. 18)
the lONer the incane per head, the lewer ten:l.s to be the proportion of inCCITe paid in cash
-
1955 was a turning point for farm Africans, as for reserve residents, because of the labour bureau system and modern influx control. (p. 21)
1955 was a turning point for Africans on White farms just as it was for Africans in the reserves
-
Within a decade of the new system, farmers' demand for labour had levelled off, making the retention machinery unnecessary and shifting policy toward expulsion. (p. 21)
danand for labour had started to level off in absolute tenns, rendering the systen unnecessary for farmers
-
State resettlement under the 1936 Land Act's anti-squatting provisions drove emigration from the backward farm areas near the reserves in the 1960s. (p. 23)
the resettlE!nent of fann residents to the reserves by the state in terms of the anti-squatting provisions of the 1936 Lard Act
-
The spatial pattern of incomes supports a three-agriculture typology rather than a sharp white/black dichotomy. (p. 23)
Perhaps it would be best to distinguish three agricultures in South Africa which shade into one another
Evidence ledger¶
Sources used¶
| Source | Kind | Period | How used |
|---|---|---|---|
| Population Censuses (1921, 1936, 1946, 1951, 1960, 1970), African farm population | population census | 1921-1970 | Six census observations of the African farm population are the basis for the population estimates, interpolated annually 1923-1969. |
| Statistical News Release P11.1 of 12.2.1969 (1960 census farm population) | official statistical series | 1960 | Source of the 1960 African farm population figure in Table 1. |
| Unpublished tabulation from the 1970 census C-returns | population census | 1970 | Source of the 1970 farm population figure in Table 1. |
| Agricultural Censuses (1934 onward; 1955, 1960, 1965, 1969 and others), population, employment, production, incomes, casual workers | agricultural census | 1934-1969 | Supplies African farm population counts for comparison, regular and domestic employment, production, and cash and kind incomes, including magisterial district data for 1955 and 1960. |
| Agricultural Census 1955 (Table B1) and holding definitions | agricultural census | 1955 | Provides the definition of a 'holding' and evidence of boys aged 10-16 engaged in farm work, which justifies the adjusted participation rate. |
| Current Population Survey, March 1980 (rural African labour force participation rates) | household or labour survey microdata | March 1980 | Source of the age-specific participation rates (Table 2) used to turn population estimates into a labour supply curve. |
| Simkins, 'Distribution of the African population of South Africa by age, sex and region-type 1960, 1970 and 1980', SALDRU Working Paper No. 32 | secondary literature compilation | 1960-1980 | Supplies the 1980 rural age distribution used to compute average participation rates. |
| Simkins, 'The Context of African Unemployment' (mimeo), Table 12 | secondary literature compilation | 1960-1980 | Source of the Table 2 participation rates, and the reference for census participation rates being too high. |
| Simkins, 'Agricultural Production in the African Reserves of South Africa, 1918-1969', DSRG Working Paper No. 10 | secondary literature compilation | 1918-1969 | Companion study whose methods and valuation at 1946 producer prices are reused here, and which supplies the reserve production per head comparison and the reserve population growth rates. |
| Report of the Commission of Enquiry into European Occupancy of the Rural Areas, Pretoria, 1959 (incl. Chapter 4, South African Police evidence) | commission or official report | 1959 (data for 1946-1954) | Used for analysis of the rising African/White ratio and for estimates of Africans on unoccupied farms, which the author uses to explain the census undercount. |
| South African Statistics 1978, p. 9.4 (number and area of holdings) | official statistical series | 1959-1969 | Shows the fall in the number of holdings against a small fall in area held, which suggests the growth of African de facto occupation. |
| Union Statistics for Fifty Years, p. G-3 | official statistical series | 1918-1957 | Source of regular and domestic African farm employment to 1957 in Table 3. |
| Report of the Native Farm Labour Committee 1937-39, Pretoria, 1939 | commission or official report | 1937-1939 | Documents the farm labour shortage and supplies illustrative family incomes in cash and kind, from which the 1937 baseline income per head is built. |
| Native Economic Commission Report 1930-32 (wage tabulations) | commission or official report | c.1930 | Cited through the Native Farm Labour Committee, which warrants treating per capita cash and kind incomes as constant from 1929 to 1939. |
| Agricultural Census 1937 | agricultural census | 1937 | Provincial shares of persons working on White farms are used to weight provincial incomes into the average of £4.33 per head. |
| University of Natal Department of Economics estimate, 'The National Income and the Non-European', in Hellman (ed.), Handbook of Race Relations, 1949 | secondary literature compilation | 1936 | Independent estimate (£9.73m for 1936) used to check the author's 1937 total income figure. |
| Consumer price index (used to convert incomes to 1946 prices) | official statistical series | 1937-1969 | Converts cash and kind incomes to constant 1946 prices. |
| Summary Report on Agricultural and Pastoral Production 1936-37 to 1938-39 and 1946-47 to 1953-54 | official statistical series | 1936-1954 | Clarifies which payments in kind farmers' valuations include, so the author can avoid double-counting with production. |
| Agricultural Census 1962 (separate tabulations on grazing and arable land) | agricultural census | 1962 | Shows that the value of land made available to Africans is excluded from in-kind income, so production can be added to earnings without double-counting. |
| South African Institute of Race Relations, Farm Labour in the Orange Free State, 1939; Neethling, SABRA 1954; R.H. Smith, Native Farm Labour in Natal, SAJE 1941 | secondary literature compilation | 1939-1954 | Used to discuss over- and under-valuation of in-kind income and to support the cash-share and wage relationship. |
| 1960 Population Census, magisterial district populations, with Western Cape African labour replacement line (1963) | population census | 1960 | Used with 1960 Agricultural Census employment to build the cross-sectional employment as a percentage of labour supply by district (Figure 3), and to compute growth rates 1960-70 (Figure 4). |
New data¶
Annual estimates of the African farm population, labour supply, regular and domestic employment, cash and kind income, agricultural production and income per head for farms outside the reserves, 1923-1969 (interpolated where years are missing). Also cross-sectional magisterial district series for 1955 and 1960 income and employment, and 1960 per capita agricultural income combining reserve and farm data.
Inference¶
Quantification: Descriptive tables.
Time series built by interpolating census observations and applying assumed participation rates, checked against a cross-sectional comparison by magisterial district shown on maps. Discrepancies between the Population and Agricultural Censuses are explained by inference, then used to read regional patterns of labour supply, wages and de facto African occupation. There is no formal statistical testing.
Key claims¶
| Claim | Basis | Strength |
|---|---|---|
| Agricultural Census farm population counts fall far below Population Census estimates after 1955, which the author explains partly by growing African de facto occupation of farms with no resident White, Coloured or Asian holder. (p. 6) | Comparison of Population Census and Agricultural Census counts for 1955, 1960 and 1965, plus the Commission of Enquiry and the fall in holdings versus area held. The author notes the account is not fully confirmed. | Partial data |
| Regular and domestic employment was 90% or more of the on-farm labour supply from 1936 to 1956, then dropped below 80% for most of the 1960s as employment levelled off while labour supply kept growing. (p. 9) | Table 3 employment series (Union Statistics for Fifty Years and Agricultural Censuses) set against labour supply derived from census population and assumed participation rates (45% male, 20% female). | Partial data |
| The 1960 Agricultural Census did not significantly undercount regular workers and domestic servants, so any undercount of population was concentrated among the old and the young. (p. 10) | 1960 cross-section by magisterial district showing many districts with employment at 81% or more of labour supply, an inference from the census data alone. | Partial data |
| Per capita cash and kind income on farms was roughly constant in the 1930s, rose during the war and early 1950s, fell substantially over 1955-65 and rose again in the late 1960s, while the share of income from own production fell from 28% to about 10%. (p. 15) | Table 4, built from the 1937 Native Farm Labour Committee baseline, Agricultural Census wage and production data, interpolation, and an assumption of constant incomes from 1929 to 1939. | Partial data |
| Regions with plentiful labour supply (Eastern Cape, north-western Natal, parts of central and northern Transvaal) are the low-wage, low cash-proportion regions near major reserves, whereas the rest of the Cape and the southern Transvaal are progressive, high-wage regions. (p. 21) | 1955 and 1960 district-level maps (Figures 3, 6 and 7), read by eye for correlations. | Partial data |
| Some parts of the White rural areas gave Africans higher agricultural incomes per capita than the reserves, so the 'two agricultures' thesis should be replaced by three overlapping agricultures (progressive capitalist, backward capitalist, reserve). (p. 23) | Figure 8, combining farm wages and own production with reserve production by magisterial district in 1960 prices. | Partial data |
| 1955 was a turning point because the labour bureau system and influx control slowed emigration, but farm labour demand levelled off within a decade, and 1960s emigration from backward areas was driven partly by state resettlement under the 1936 Land Act. (p. 21) | Population growth rates in Table 1, the 1960-70 growth rate map (Figure 4) and argument drawn from Morris. No direct data on removals. | Illustrative or anecdotal |
The author's caveats¶
The author notes that the Agricultural Census population counts fall 25-46% below the Population Census estimates, and that the latter are probably slight undercounts. He also notes that the 1921 farm population is assumed, that the 1960 and 1970 census participation rates are too high, that farmer valuations of in-kind income may be biased, and that the casual worker residence split is unknown. Other limits are the lack of 1946 and 1948-51 incomes, interpolation of missing years, the lack of income estimates for 1923-28, and the unrecorded African-controlled sector after the mid-1950s. The explanation of the post-1955 gap is acknowledged to be inadequately confirmed.
Gaps in the evidence¶
The central explanation, de facto African control of unoccupied farms, is inferred from the gap between the two censuses and is never tested against direct evidence such as farm-level surveys, magistrates' or police records, or Commission files. Wage and income levels rest on a single 1937-39 benchmark, assumed constant for ten years, and on farmer-reported in-kind valuations, so household budget surveys or wage records would test them better. The 1980 participation rates applied to the whole period, and the 17% casual-worker adjustment, rest on little evidence, though census microdata and later survey data now exist to check them.
Replicability: Data described but not public.