Recent Statistics in Agriculture¶
Delia Hendrie (1976). Recent Statistics in Agriculture. SALDRU Working Paper No. 2. Southern Africa Labour and Development Research Unit (SALDRU), University of Cape Town.
| Type | working paper |
| Scope | Economics and economic history of apartheid |
| Method | descriptive quantitative |
| Period studied | 1950–1974 |
| Regions | South Africa (white farming areas), Cape, Natal, Transvaal, Orange Free State, 60 economic regions |
| Themes | agriculture land, labour markets wages, methods data sources |
| Citation key | hendrie1976 |
Argument¶
A statistical compendium rather than an argued study. It reproduces and reorganises official agricultural census data from the Department of Statistics: farming units, production, employment and earnings, by race, province and economic region. The short introduction makes three points. Agriculture's share of GDP fell even as output volume rose. The number of farming units declined steadily. Farm employment of regular, casual and domestic workers fell after 1969. Hendrie stresses that the census covers only Asian, Coloured and White holders, so the homelands are excluded. She argues that its employment and earnings data are of uncertain reliability, especially the implausible swings in casual labour counts. She calls for better, clearer census questions and richer data (age, sex, education, farm population) to support research on farm labour.
Evidence and method¶
Department of Statistics, Reports on Agricultural and Pastoral Production (Agricultural Censuses No. 39, 43, 44, 46), covering 1964/65, 1968/69, 1970/71 and 1972/73. Statistical News Release of 15 April 1976 (1973/74 census). South African Reserve Bank national accounts. Statistical Survey accompanying the 1976-77 Budget Speech. Union Statistics for 50 years. Department of Agricultural Economics and Marketing marketing data, used as a check on production figures.
Place in the debates¶
- Reliability of official farm labour statistics (agricultural census) (see Data reliability and econometric method). Sceptical. Casual employment figures fluctuate implausibly, likely because farmers misread the 'casual labourer' category. Production returns conflict with marketing data. Employment and earnings cannot be independently checked.
- Farm labour shedding / decline of agricultural employment in the 1970s (see Farm labour: mechanisation, labour shedding, wages and labour control). Documents the post-1969 decline in regular, casual and domestic farm employment and the fall in farming units, but offers no causal explanation. A follow-up analytical paper is promised.
Findings¶
-
The agricultural census covers only farms operated by Asian, Coloured and White holders; African farming in the reserves is excluded. (p. 3)
These statistics relate therefore to the farming activities of Asian, Coloured and White holders only.
-
Agriculture, forestry and fishing's share of GDP declined between 1950 and 1974 even as the volume index of agricultural production rose. (p. 3)
Table 1 shows the declining contribution of agriculture, forestry and fi shi ng, to the Gros~ Domestic Product
-
The number of farming units fell steadily from 111,586 in 1955 to 79,855 in 1974, indicating farm consolidation. (p. 3)
reveals a steady dec rease in the number of farmi ng units - from 111 586 in 1955 to 79 855 in 1974.
-
Between 1969 and 1974 regular farm employees fell 14%, casual employees 19% and domestics 15%. (p. 3)
During the period 1969-1974, the number of regular employee~ has decreased by 14%, the number of casual employees by 19%, and the number of domestics by 15%
-
Census production figures understate output relative to marketing data; maize marketed exceeded reported production by 22% in 1970/71. (p. 3)
for example the 1970/71 agricultural census found that the quantity of maize marketed was 22% greater than the production figures submitted by
-
There is generally no independent way to check the reliability of the census data. (p. 3)
In most cases it is impossible to gauge the reliability of the census data, due to the lack of any other data relating to farming activities.
-
Racial wage gaps among regular farm employees were huge. In the Transvaal, Africans averaged R12.86 a month in 1973 (up from R8.82 in 1965) while Whites averaged R276.17, and African regular employment fell 10.4% over 1969-73. (p. 35)
TRANSVAAL 287 434 289 071 259 151 0,6 -10,4 8,82 11 ,54 12,86
-
Earnings figures include cash and payments in kind but exclude free housing. Monthly averages divide annual wages by August employment, which interacts with the casual-labour count problems. (p. 38)
To obtain the average monthly wage, total wages (cash and kind) paid during the twelve mOl)ths ended 28 February or 30 June of any year, were divided b~ employment as at 31st August
-
Key demographic data on farm workers are not collected, which limits research on the farm labour force. (p. 4)
Further data on the age and sex o{ employees, their educational level, the total population on the' farms
Evidence ledger¶
Sources used¶
| Source | Kind | Period | How used |
|---|---|---|---|
| Report on Agricultural and Pastoral Production 1972/73, Agricultural Census No. 46, Report No. 06-01-10 | agricultural census | 1972/73 (31 August 1973) | Source for farming units by size group, employment by race and region, and average monthly wage deviations (Tables 3, 4, 6, 9-11). |
| Report on Agricultural and Pastoral Production 1970/71, Agricultural Census No. 44, Report No. 06-01-08 | agricultural census | 1970/71 | Source for 1971 crop, fruit tree and livestock totals (Table 5), farming unit counts and employment series. |
| Report on Agricultural and Pastoral Production 1964/65, Agricultural Census No. 39 | agricultural census | 1964/65 | Source for 1965 employment and earnings by race and province in Table 8. |
| Report on Agricultural and Pastoral Production 1968/69, Agricultural Census No. 43 | agricultural census | 1968/69 | Source for 1969 employment and earnings by race and province in Table 8. |
| Statistical News Release, 15 April 1976 (Agricultural Census 1973/74) | official statistical series | 1973/74 | Supplies the latest (1974) counts of farming units and farm employees. |
| Union Statistics for 50 years | official statistical series | 1955-1960 and earlier | Source of the 1955 and 1960 counts of farming units in Table 2. |
| South African Statistics 1974 | official statistical series | 1961, 1965 | Used to cross-check Table 7 employment figures, revealing unexplained discrepancies. |
| South African Reserve Bank, A Statistical Presentation of South Africa's national accounts for the period 1946 to 1970 | national accounts macro | 1946-1970 | Source of agriculture's percentage contribution to GDP, 1950-1970 (Table 1). |
| Statistical Survey in connection with the Budget Speech 1976-77, W.P. 8-'76 | national accounts macro | 1970-1974 | Source of the 1971-74 GDP share figures for agriculture, forestry and fishing in Table 1. |
| Department of Agricultural Economics and Marketing maize marketing estimates | official statistical series | 1970/71 | Used as an external check on census production figures; reveals a 22% maize discrepancy. |
| Department of Statistics investigation into census maize discrepancy | commission or official report | 1970/71 | Cited to explain the maize discrepancy as farmers' misreading of 'production season'. |
| Statistics Act and census regulations (compulsory returns) | legislation gazette | n.d. | Establishes that census returns are compulsory and which racial groups of holders are covered. |
Inference¶
Quantification: Descriptive tables.
A statistical compendium with brief commentary. It presents time series (1950-74, 1960-74) and cross-sections by economic region (1971, 1972/73) drawn from official census reports. Inference is limited to reading trends and checking reliability against marketing data and series discrepancies. There is no modelling.
Key claims¶
| Claim | Basis | Strength |
|---|---|---|
| Agriculture's share of GDP fell sharply from 1950 to 1974 even as the volume of agricultural production rose. (p. 3) | Table 1, drawn from the SARB national accounts and the 1976-77 Budget Statistical Survey. | Systematic data |
| The number of farming units declined steadily from 1955 to 1974, a fall of about 28%. (p. 3) | Table 2, using Union Statistics for 50 years and the Agricultural Census reports 44, 46 and the 1973/74 release. | Systematic data |
| Farm employment fell steadily after 1969 across all categories (regular -14%, casual -19%, domestics -15%), indicating shrinking employment opportunities in farming. (p. 3) | Table 7, from Census Nos. 44, 46 and the 1976 news release. | Partial data |
| The census data on casual employment are unreliable, since the implausible swings (592,000 to 770,000 to 328,000 to 779,000) probably reflect differing interpretations of 'casual labourers'. (p. 4) | Internal inspection of the Table 7 series; the author notes the interpretation explanation 'has been suggested'. | Illustrative or anecdotal |
| The census production figures can contain significant error; the 1970/71 census maize marketing figure was 22% above farmers' reported production. (p. 3) | Comparison with Department of Agricultural Economics and Marketing marketing data and an investigation by the Department of Statistics. | Partial data |
| The census covers only Asian, Coloured and White holders, so statistics relate only to their farming activities (African-held land, i.e. the reserves, is excluded). (p. 3) | The census description in the introduction and the structure of the census regulations. | Asserted |
| Production patterns and farm sizes vary strongly by economic region (e.g. average farm area ranges from 38 ha in region 37 to 5,521 ha in region 11). (p. 11) | Tables 3, 4 and 6, from Census No. 46 and No. 44 data. | Systematic data |
The author's caveats¶
The author states that the reliability of the census data cannot be gauged in most cases because no other data on farming exist. Only production can be checked against marketing data, and employment and earnings cannot be checked at all. Casual employment shows implausible swings. The 1972/73 census gave incomplete production coverage by region, so the tables mix 1970/71 and 1972/73. Pre-1955 data are not comparable because of a change in the definition of a farming unit. The census excludes African holders. Some earnings entries are flagged as errors (zero wage with positive employment), and some figures are footnoted as apparently erroneous.
Gaps in the evidence¶
The paper is a data presentation and tests no argument. It would need data on the age and sex of employees, their education, and the total farm population, which the author notes are absent. Data on African-occupied land and on labour tenants and squatters would complete the picture of farm employment. A farm-level survey or the census microdata could test the reliability of the employment and wage categories, but these were not available then. Later census reports and the 1970 population census would allow cross-checks.
Replicability: Data public and well described.