A Descriptive Analysis Of Employment Trends In South African Agriculture¶
Nick Vink; Johann Kirsten (2001). A Descriptive Analysis Of Employment Trends In South African Agriculture. Working paper 2001-08, Department of Agricultural Economics, Extension and Rural Development. University of Pretoria.
| Type | working paper |
| Scope | Economics and economic history of apartheid |
| Method | synthesis review |
| Period studied | 1946–2010 |
| Regions | South Africa, Western Cape, KwaZulu-Natal, Northern Province |
| Themes | agriculture land, labour markets wages, manufacturing industrial policy, macro fiscal finance debt, transition post apartheid, methods data sources |
| Citation key | vink2001 |
Argument¶
The paper documents the long secular decline in South African farm employment and attributes it largely to state policy that favoured capital-intensive, labour-saving technology (cheap credit, tax concessions, skewed R&D), leading to labour shedding after the Second World War and especially from the late 1960s. It then uses six regional and commodity case studies to ask whether land reform, deregulation and other policy shifts could raise employment. It finds that without Lipton's 'four reforms' there is little scope for job growth, and that the main gains come from fruit exports. It concludes that only a wider export drive in KwaZulu-Natal and Mpumalanga could make agriculture a major creator of jobs.
Evidence and method¶
Development Bank of Southern Africa standardised employment series (1946-1996); Central Statistical Service agricultural censuses and surveys; October Household Survey full-time-equivalent data (Klasen and Woolard 1999); Townsend's (1997) econometric PhD study; programming and macro models of the Western Cape and maize sector (Van Zyl, Vink); farm-household case studies by Eckert, De Klerk, Conradie, Lyne and Ortmann, and Kirsten; industry projections (Kleynhans et al 1997); OLS trend lines.
Place in the debates¶
- Causes of farm mechanization and labour shedding (Habakkuk-type factor-price thesis) (see Farm labour: mechanisation, labour shedding, wages and labour control). Relies on Townsend (1997): a close relationship between the machinery/labour ratio and relative factor prices after 1950. Policy distortions (credit, tax, R&D allocation) drove labour-saving technical change, and the removal of subsidies in the 1980s partly reversed it.
- Land reform/small farms and rural employment (Lipton's farm size-efficiency argument) (see Agricultural policy, productivity, farm size and land reform). Qualified support for Lipton: land reform alone yields little or negative net employment (multiplier losses, risk in semi-arid areas), and works only with the 'four reforms'.
- Labour legislation and farm job creation (see Farm labour: mechanisation, labour shedding, wages and labour control). Suggests, mostly on anecdotal evidence, that labour and tenure laws raise farm labour costs; admits the employment effect is unclear because the cost of capital has also risen.
- Deregulation and agriculture's role in development (Mellor's three questions) (see Agricultural policy, productivity, farm size and land reform). Favours deregulation and export orientation; judges that agricultural productivity growth has not been matched by demand expansion, so agriculture is unlikely to be a major job creator unless exports grow.
Findings¶
-
Farm employment fell steadily in 1985-1993, with job losses among both regular and casual workers. (p. 4)
the sector shed about 150 000 regular employees between 1985 and 1993, and about 120 000 casual and seasonal workers between 1985 and 1992.
-
Tractorisation raised farm employment after the Second World War, but grain-harvest mechanisation from the late 1960s reduced it, especially seasonal employment. (p. 6)
Employment (both permanent and seasonal) increased with the introduction of tractors in the period after the Second World War, then declined with the introduction of mechanised harvesting of grain crops from the late 1960s.
-
Townsend finds that relative factor prices closely explain the machinery-labour ratio after 1950, and that the removal of favourable credit and tax policies shifted farms back towards labour. (p. 9)
After the abolition of favourable credit and tax policies labour use increased as it was substituted for expensive capital.
-
State policy skewed R&D towards labour-saving technology and drove labour shedding up to the mid-1980s, which was inappropriate for a labour-surplus economy. (p. 10)
government policies skewed the allocation of R&D expenditure towards labour saving technical change, which was hardly appropriate for a labour-surplus economy.
-
Modelled Western Cape deregulation and land reform scenarios all show positive employment effects, because resources shift from wheat to more labour-intensive dairy and wool. (p. 12)
The employment effect of all ten policy scenarios analysed was positive, as is shown in the Table 4 below.
-
Moving land from commercial wheat farms to small farms is likely to cost jobs once upstream and downstream effects are counted. (p. 13)
The resulting low input mode of production will have a negative effect on employment opportunities, with an estimated loss of 10,8 to 14,9 jobs per 1000 hectare allocated to the small farms.
-
In KwaZulu-Natal, land redistribution to small farms yields only a modest net employment gain, mostly at the expense of hired workers. (p. 17)
land redistribution is unlikely to produce a significant net increase in the number of employment opportunities.
-
In semi-arid Northern Province, rural reform adds little net employment once existing hired work and risk are included. (p. 17)
net livelihoods created through a process of rural restructuring would only increase by 1,1% if the quality of the livelihoods created is compromised.
-
Aggregating the case studies, the authors find little scope for farm job growth without complementary reforms, apart from fruit exports. (p. 18)
What the case studies confirm is that, in the absence of the ‘four reforms’, there is little scope for increasing employment in agriculture.
-
Agriculture could become a major employer only through a broad export drive in KwaZulu-Natal and Mpumalanga. (p. 25)
The only way in which agriculture can conceivably become a major creator of employment opportunities for the country as a whole would be through a wider and deeper export drive
Evidence ledger¶
Sources used¶
| Source | Kind | Period | How used |
|---|---|---|---|
| Development Bank of Southern Africa, Standardised Employment Series of South Africa (formal sector and agricultural employment, 1946-1996) | official statistical series | 1946-1996 | Establishes the long-term secular decline in farm employment and in agriculture's share of formal employment (Figures 1 and 2), including a fitted exponential trend. |
| Central Statistical Service (CSS) farm employment statistics, various years (regular workers, domestic workers on farms, working owners) | official statistical series | 1918-1993 (trend line fitted 1958-2010) | Supplies the regular employment series (Figure 3) and Table 1 figures used to show the trend of regular farm employment and a linear projection. |
| Agricultural Census / Agricultural Survey (casual labour cash income and employment) | agricultural census | 1958-1990s (limited data points) | Source for casual employment series and the full-time-equivalent (FTE) conversion (Figure 4) used to show casual labour trends. |
| CSS, 1993 (Agricultural Census 1993 provincial data) and DBSA data files | agricultural census | 1993 (value added 1991) | Provides the provincial composition of farm labour, value added and value added per hectare (Table 2). |
| 1988 Agricultural Census data (used as the base for the Western Cape model) | agricultural census | 1988 | Base data for the programming model of Western Cape agriculture that gives the employment effects in Table 4. |
| October Household Survey 1994-1997 (CSS Statistical Release P0317.10, 1998) | household or labour survey microdata | 1994-1997 | Alternative employment data showing agricultural employment possibly stable or rising; the CSS release shows a drop that the authors suspect reflects exclusion of part-time and seasonal work. |
| Klasen and Woolard (1999) full-time-equivalent conversion of October Household Survey employment | secondary literature compilation | 1994-1995 | Used to show that agricultural employment (including forestry and fisheries) did not continue to decline in 1994-95. |
| Townsend (1997) econometric study of policy distortions and agricultural performance (PhD thesis) | secondary literature compilation | c. 1950-mid-1980s | Relied on for econometric evidence that relative factor prices and policy distortions drove machinery-using, labour-shedding technical change. |
| Van Zyl (1989a; 1989b) model of the South African maize industry | secondary literature compilation | late 1980s | Model results on how marketing liberalisation and land shifts to small farms affect maize-sector employment. |
| Western Cape agricultural programming model (Vink et al 1996; Van Zyl et al 1997) | secondary literature compilation | base year 1988 | Ten policy scenarios yield employment effects of +6.9% to +10.8% (Table 4) and underpin case study 1. |
| Eckert (1996) small-farm wheat option study, Western Cape | secondary literature compilation | 1996 | Estimates net employment effects of reallocating wheat land to small farms, including upstream and downstream industries. |
| Kleynhans et al 1997, Agrifutura Project projections for Western Cape agricultural industries to 2010 | secondary literature compilation | 1996-2010 | Basis for the extrapolated employment trends in the fruit industry (Table 5), including the 28% growth figure. |
| De Klerk (1996) and Conradie et al (1996) studies of small-scale pear and apple production | secondary literature compilation | 1996 | Used to conclude that small-scale deciduous fruit farming is feasible but creates few new livelihoods. |
| Vink and Van Zyl (1997) water policy model results, Western Cape | secondary literature compilation | 1997 | Simulated employment effects of water-availability reductions and tariff changes (Table 6). |
| Lyne and Ortmann (1996) farm household models for KwaZulu-Natal commercial farmland | secondary literature compilation | 1996 | Provides the 6.8% net increase in farm employment under redistribution of land to small farms. |
| Kirsten (1996) case study of a magisterial district in semi-arid Northern Province | secondary literature compilation | 1996 | Gives the estimated livelihoods potential (+10.8%, only +1.1% net of hired work, negative at higher income targets). |
| Industrial Development Corporation, Trade for growth (IDC, 1997) | commission or official report | 1997 | Cited for the persistence of declining manufacturing employment, used to argue about labour-law rigidities. |
| Marketing of Agricultural Products Act No 47 of 1997 | legislation gazette | 1997 | Cited as the legislative basis of agricultural marketing deregulation. |
| Labour legislation: Labour Relations Act, Basic Conditions of Employment Act, Land Reform (Labour Tenants) Act, Extension of Security of Tenure Act | legislation gazette | 1990s | Discussed as possibly raising farm labour costs and constraining employment, supported only by anecdotal evidence. |
| Employment intensity of agriculture by sector (Table 3, hectares per permanent employee) | other | Not stated | Shows horticulture is most labour intensive and animal production least, supporting the case-study mechanisms. |
| Regional land and agricultural resource comparisons (South Africa vs Angola, Mozambique, Tanzania, Zambia; land potential by province) | other | Not stated | Used to argue that South Africa has limited horizontal expansion potential for crop and horticulture production. |
| Thirtle, Sartorius von Bach and Van Zyl (1993) total factor productivity estimates, 1947-1991 | secondary literature compilation | 1947-1991 | Evidence for healthy productivity growth in agriculture. |
Inference¶
Quantification: Descriptive tables.
Descriptive time-series trends from official employment series with simple OLS trend lines. This is followed by synthesis of six model-based case studies, from programming and farm household models, whose micro results are extrapolated to national level and summed in a 'heroic' table. Final sections are qualitative discussion of resource base, policy and trade environment.
Key claims¶
| Claim | Basis | Strength |
|---|---|---|
| Farm employment has declined secularly since the 1940s, with about 150,000 regular workers shed 1985-1993 and about 120,000 casual workers shed 1985-1992. (p. 4) | DBSA Standardised Employment Series and CSS/Agricultural Census figures in Figures 1-3 and Table 1. | Systematic data |
| Post-war mechanisation and policy distortions favouring capital (credit, tax concessions, R&D) drove labour shedding up to the mid-1980s. (p. 10) | Townsend's (1997) econometric study, relayed second-hand, plus the shape of the series for regular and seasonal workers. | Partial data |
| Employment in Western Cape agriculture would rise by roughly 7-11% under market liberalisation, with or without land reform, a prediction the authors say deregulation has partly borne out. (p. 12) | Programming model based on 1988 Agricultural Census data (Table 4), with informal validation against post-deregulation experience. | Partial data |
| Land redistribution to small farms yields little or negative net employment gain in wheat, KwaZulu-Natal commercial farms and semi-arid Northern Province. (p. 17) | Eckert (1996), Lyne and Ortmann (1996) and Kirsten (1996), farm-level and household models reported second-hand. | Partial data |
| Fruit export industries in the Western Cape are the main source of potential employment growth, about 28% by 2010. (p. 14) | Kleynhans et al (1997) industry projections, assuming constant labour and yield per hectare (Table 5). | Partial data |
| Without the 'four reforms', there is little scope for increasing agricultural employment, and only a wider export drive in KwaZulu-Natal and Mpumalanga could make agriculture a major employment creator. (p. 18) | Extrapolated case-study results summarised in Table 7, which the authors call a 'heroic' summary, plus qualitative resource-base discussion. | Illustrative or anecdotal |
| Reduced water availability for irrigation would sharply cut Western Cape farm employment (-3.6% to -31.2%), whereas tariff increases have little net effect. (p. 16) | Vink and Van Zyl (1997) model simulations (Table 6). | Partial data |
The author's caveats¶
The authors note that the casual/seasonal category is notoriously hard to measure, so the 1993 rise may be measurement error. October Household Survey data are not comparable with Table 1 and may exclude part-time and seasonal work. The FTE trend rests on few data points and should be treated with circumspection. Only four pages of the 1993 Agricultural Census were available in 1997. The Western Cape model treats horticulture rudimentarily. Table 7 is called a 'simplified and heroic summary' whose extrapolation is uncertain.
Gaps in the evidence¶
The argument relies on simulation models and second-hand case-study results with little direct econometric evidence on how labour costs, land reform or legislation affect farm employment. Farm-level panel data, full agricultural census series (the 1993 census, later surveys) and OHS/Labour Force Survey microdata would allow direct tests; most of these now exist. The claims about labour legislation are anecdotal and need firm-level wage and employment data before and after the 1990s reforms.
Replicability: Data poorly documented.