The incomes of farm workers and their families: A study of maize farms in the Western Transvaal

Literature › Works

Michael de Klerk (1984). The incomes of farm workers and their families: A study of maize farms in the Western Transvaal. Second Carnegie Inquiry into Poverty and Development in Southern Africa, Carnegie Conference Paper No. 28. SALDRU, University of Cape Town.

Type	conference paper
Scope	Economics and economic history of apartheid
Method	mixed
Period studied	1980–1981
Regions	Western Transvaal, Coligny, Delareyville, Koster, Lichtenburg, Schweizer-Reneke, Wolmaransstad
Themes	agriculture land, labour markets wages, inequality poverty unemployment, methods data sources
Citation key	klerk1984a

Argument

Using a 1982 survey of 61 maize farms (data for the 1980-81 agricultural year), the paper estimates the all-inclusive incomes of farm workers' families and compares them with the Minimum Level of Living (MLL). Average family income (R220 a month) is above the MLL requirement (R188), but incomes are very uneven across the year because bonuses and seasonal pay are lumpy. Most families (about 85 percent) therefore fell below the MLL for part of the year, typically about 6 months, so rural poverty is chiefly seasonal. Family income and months of poverty depend mainly on the permanent wage rate and on the number of permanent workers per family. Regression results link higher mechanization to fewer permanent workers per family and so to lower family incomes and more months of poverty, despite rising wage rates. The author stresses that the sample is probably biased upward, so the picture is better than the regional average.

Evidence and method

Survey of 61 maize farms in six Western Transvaal magisterial districts, selected non-randomly through co-operatives and stratified by area and location. It was done in Feb-Mar 1982 and covers 1980-81. Farmer interviews on wages, rations, bonuses, seasonal teams and housing; a small informal Transkei interview survey of seasonal workers on four of the farms as a check on class bias. Supplementary sources: the 1978 Agricultural Census, the 1980 Population Census, the Department of Agriculture's maize production cost surveys (Fenyes 1983), Bureau of Market Research MLL estimates (Aug 1981), Simkins' household-size estimate, and the author's 1983 MA thesis.

Place in the debates

• Causes and consequences of farm mechanization for farm labour (Habakkuk-type labour-scarcity versus push/pull explanations) (see Farm labour: mechanisation, labour shedding, wages and labour control). Finds that causation between mechanization and loss of permanent workers per family ran in both directions (an urban labour pull and a farm push). It argues that mechanization raised wages but lowered family incomes and intensified seasonal poverty.
• Family subsistence theory of wages (do farmers depress wages by hiring many workers per household?) (see Farm labour: mechanisation, labour shedding, wages and labour control). Rejects it for this sample: there was effectively no relationship between workers per family and the permanent wage rate.
• Measurement of rural poverty against the urban MLL (see Data reliability and econometric method). Adopts the MLL, adjusted to exclude housing, transport, taxes and the like, and argues that average monthly income is misleading. Poverty should be assessed seasonally by comparing the regular cash-plus-food flow with the MLL.

Findings

• The sample's mean permanent wage (R1777 p.a. all-found, excluding housing) was far above Department of Agriculture estimates, so the survey is biased upward. (p. 24)

  Nevertheless, by comparison to other estimates a mean of R1777 p.a. is very high.

• The author warns that the sample presents a better-than-average picture of farm workers' conditions. (p. 25)

  What is most important is to bear in mind that the picture represented by the sample is probably substantially more favourable than average.

• Permanent wages were mostly in kind and in deferred bonuses; regular cash was a minor share. Food was almost half the total and cash and bonus about a quarter each. (p. 60)

  Food contributed almost one half of the total, and regular cash wages and annual payments about one quarter each.

• The post-harvest bonus was large and varied widely across farms, creating dependence on loans from farmers during the year. (p. 28)

  The average value of the post-harvest bonus was R502, with a standard deviation of R367.

• Seasonal maize harvest work shifted from 'external' workers in black rural areas to 'internal' workers resident on white farms, who were paid less, as harvesting was mechanized in the 1970s. (p. 32)

  In 1968, 'internal' workers occupied only 9 percent of such jobs. By 1981, this had risen to 48 percent.

• Farm domestic workers were among the lowest-paid workers in the economy, and their wages were treated as a supplement to the permanent worker's income. (p. 38)

  What is clear is that domestic workers on farms are amongst the most poorly paid - perhaps the most poorly paid - groups of workers in the economy.

• Mean all-inclusive family income was R220 a month, above the R188 MLL, but more than half of families earned less than R200. (p. 40)

  Within the range of R86-R450, the mean income.was R220 per month (standard deviation R79).

• Despite the above-MLL average, most families were below the MLL for part of the year, so rural poverty is largely seasonal. (p. 61)

  it is most likely that the great majOrity of families - probably about 85 percent - received sub-MLL incomes for several months of the year - typically about 6 months.

• Each additional permanent worker in a household added much more to family income than a wage increase did, and cut the months of poverty sharply. (p. 52)

  On the other hand, each extra permanent worker in a household added R14l8 p.a. to family income and reduced the number of months below the MLL by 4,12.

• Greater mechanization was associated with more months of sub-MLL income, because fewer permanent workers per family outweighed rising wages. (p. 53)

  The net result was an intensification rather than an amelioration of seasonal poverty.

• Farmers did not appear to hire many permanent workers per household in order to depress wages. (p. 54)

  farmers did not consciously try to employ as many permanent workers as possible from each household so as to try to depress their wage rate

• Skills and schooling raised permanent wage rates, and growing skills may ease farm labour organization. (p. 58)

  It is significant that the average permanent wage rate on farms where some or all workers were reported to have these skills was R1982 p.a. as against R1419 p.a.

Evidence ledger

Sources used

Source	Kind	Period	How used
Author's survey of 61 maize farms in six Western Transvaal magisterial districts (Coligny, Delareyville, Koster, Lichtenburg, Schweizer-Reneke, Wolmaransstad), 1980-81 agricultural year	author survey or fieldwork	February-March 1982 (data for 1980-81 agricultural year; some farms back to 1968)	Primary source for permanent, seasonal and domestic wage rates, worker numbers per family, food and housing, and thus family income estimates and the poverty analysis.
Smaller informal survey of seasonal workers in the Transkei (workers from four of the surveyed farms)	interviews oral history	c. 1982	Cross-check on farmers' wage reports to assess class-bias/non-sampling error.
Department of Agriculture survey of the cost of maize production, 1981, Western Transvaal (Fenyes, Statistiese Bylae tot die Verslag oor die Benutting en Vergoeding van Plaasarbeid in die Suid-Afrikaanse Landbou, 1983, vols I and III)	official statistical series	1977-78 to 1980-81	Benchmark for permanent wage levels, cash component, bonus and food values, against which the author's sample is shown to be biased upward.
Department of Agriculture postal records of 103 Western Transvaal maize farmers	official statistical series	1981	Second comparator for average permanent wage (R1164 p.a.).
RSA Department of Statistics, Census of Agricultural and Pastoral Production, 1978, Report 06-01-14 (Tables 3.2, 3.4, 4.1)	agricultural census	1978 (year ended 31 August 1978)	Number of farms in the districts, and total cash and ration wages and number of domestic workers, used to estimate domestic wages.
RSA Department of Statistics, Agricultural Census No. 49, 1976, Report on Agricultural and Pastoral Production, part 4, Table 3	agricultural census	1976	Population distribution of farms by size to assess over-representation of large farms in the sample.
RSA Department of Statistics, Population Census 1980, Geographical Distribution of Population, Sample Tabulations, Report 02-80-01, Table 4.1	population census	1980	District shares of rural black population to test the sample's representativeness of families.
Bureau of Market Research, The Minimum and Supplemented Living Levels of Non-Whites ..., Report no. 92, August 1981 (Table VE-Bl)	commission or official report	August 1981	Provides the Minimum Level of Living (MLL) requirements for a family of 6, adjusted to derive the poverty line against which farm family incomes are compared.
Simkins, C.E.W., 'On Generating Synthetic Samples of Black Households' (SALDRU), based on Department of Statistics Current Population Survey tabulations	secondary literature compilation	unspecified	Supplies assumed family size of 6 for farm households since no family size data was collected.
de Klerk, M.A. thesis, 'Technological Change and Employment in South African Agriculture: maize harvesting in the Western Transvaal, 1968-1981' (UCT, 1983) and Carnegie Conference Paper no. 27 (1984)	secondary literature compilation	1968-1981	Source for the shift in seasonal jobs from external to internal workers, and for mechanization/wage causality claims.
Secondary estimates of gleaning wastage rate (Hunt, Farm Power and Machinery Management, 1977; van Wyk, Die Ekonomiese Aspekte by die Oes en Lewering van Mielies, 1970)	secondary literature compilation	1970-1977	Bound the unknown wastage rate (2.9 to 30 percent) needed to convert crop-share pay to daily wages.
Price quotations from Noord-Wes Kooperasie and livestock agents Slabbert, Verster and Malherbe (personal communications)	company records or reports	c. 1981	Prices used to value food rations (maize meal, milk, meat) in kind wages.
Farmer interviews and co-operative member registers (three co-operatives) used for sampling	interviews oral history	1982	Sampling frame (non-random, stratified by size and location) and farmer reports on workers' wages, food, housing, leave and schooling.

New data

Survey of 61 maize farms (about 2 percent of farms) in six Western Transvaal districts, Feb-March 1982, covering wages, worker numbers, in-kind payments and some conditions for 1980-81; plus a small informal Transkei interview survey of seasonal workers from four of the farms. Non-random, stratified sample.

Inference

Quantification: Statistical analysis.

Descriptive tabulation of the author's non-random farm survey, with family income constructed from wage-rate estimates and assumed family size, compared against the Bureau of Market Research's urban Minimum Level of Living (adjusted) and against Department of Agriculture wage benchmarks; seasonal poverty bounded by best- and worst-case scenarios; bivariate correlations/linear regressions across 49-54 farms.

Key claims

Claim	Basis	Strength
Average permanent worker wage in the sample (R1777 p.a., about R148 p.m.) is well above Department of Agriculture estimates, so the sample is biased upward. (p. 19)	Author's survey compared with Fenyes/Department of Agriculture 1981 maize cost survey and postal records; district and farm-size reweighting.	Partial data
Average family income (R220 p.m.) exceeds the adjusted MLL (R188,46), yet 44 percent of sample families fell below it on average monthly income. (p. 38)	Constructed family income from survey wages and worker numbers per family compared with adjusted BMR MLL for a family of six.	Partial data
Poverty on white maize farms is seasonal: about 85 percent of families, typically for about 6 months, received sub-MLL income. (p. 43)	Calculated best-case/worst-case months below MLL from income timing, with no data on expenditure patterns.	Illustrative or anecdotal
Family income and months below the MLL depend chiefly on the permanent wage rate and the number of permanent workers per family; families always above the MLL owed this mainly to more permanent workers. (p. 48)	Correlation coefficients (Table 19) across 49-54 farms and descriptive profiles of consistently above/below-MLL families.	Partial data
Greater mechanization (fewer permanent workers per 1000 ha) was associated with fewer permanent workers per family, lower family income and more months of sub-MLL income, despite possibly higher wage rates. (p. 47)	Weak correlations (r72 = -0,27 significant, r71 = 0,25 near significant) plus earlier work by the author; causation argued by inference.	Partial data
Domestic workers on farms are among the most poorly paid groups in the economy (about R24 p.m. in 1981). (p. 33)	1978 Agricultural Census wage bill/employment for six districts inflated by the 54 percent rise in permanent wages (Department of Agriculture).	Partial data
Skills and schooling raise permanent wages: farms where some workers had driving or maintenance qualifications paid R1982 p.a. against R1419 p.a. on others. (p. 52)	Simple comparison of farm mean wages in the survey, with farmers' reports on qualifications.	Illustrative or anecdotal

The author's caveats

The survey was designed for mechanization and employment, so family size, composition and expenditure data were not collected. The sample is non-random and biased towards large farms and over-represented districts, so wages and incomes are biased upward. Class bias in farmer-only responses was only partly checked. Off-farm income and remittances were assumed to be zero, valuation of in-kind wages is approximate, the seasonal wastage rate is assumed, and the family definition was never formally set.

Gaps in the evidence

The argument needs worker-reported household data on family size, composition, off-farm income, remittances, pensions and monthly expenditure, rather than farmer reports and assumed values. A random sample or the full co-operative/census frame would test representativeness, and a measured gleaning wastage rate would firm up seasonal wages. Dated farm-level payroll records or a rural household survey (the 1982-85 Carnegie/SALDRU material or later rural surveys) could test the seasonal poverty and mechanization claims, which rest on weak correlations and assumed month-by-month income timing.

Replicability: Data described but not public.