Demythifying Contract Farming: Evidence from Rural South Africa

Abstract

This paper intends to contribute to the ongoing debate about whether and how restructured agri-food markets can provide viable market opportunities for small-scale farmers in South Africa. It analyses contract farming from the small-scale farmer perspective to better understand the implications for small-scale farmers of contractual arrangements with processing and/or marketing firms.
The paper, based on empirical research conducted in the Limpopo Province of South Africa using a combination of qualitative and econometric analyses, argues that contract farming is not a panacea for small-scale farmers. On the one hand, contract farming improves agricultural production for contract farmers who benefit from increased incomes, enables better access to services and resources, and creates new opportunities to participate in markets. However, on the other hand, the results show that contract farming remains limited and mostly involves the already better-off, who have benefitted from specific development paths and public support. This case study shows that contract farming in itself does not appear to provide an efficient means of reducing poverty, nor does it provide an institutional tool through which to improve rural livelihoods. It does, therefore, not represent a tool appropriate for the majority of small farmers or for redressing the historical imbalances in the South African agricultural sector.

Full text

This article was downloaded by: [Cirad-Dist Bib Lavalette]
On: 01 February 2013, At: 08:46
Publisher: Routledge
Informa Ltd Registered in England and Wales Registered Number:
1072954 Registered office: Mortimer House, 37-41 Mortimer Street,
London W1T 3JH, UK
Agrekon: Agricultural
Economics Research, Policy
and Practice in Southern
Africa
Publication details, including instructions for
authors and subscription information:
http://www.tandfonline.com/loi/ragr20
Demythifying contract
farming: Evidence from
rural South Africa
S. Freguin-Gresh , M. d'Haese & W. Anseeuw
Version of record first published: 01 Feb 2013.
To cite this article: S. Freguin-Gresh , M. d'Haese & W. Anseeuw (2012):
Demythifying contract farming: Evidence from rural South Africa, Agrekon:
Agricultural Economics Research, Policy and Practice in Southern Africa, 51:3,
24-51
To link to this article: http://dx.doi.org/10.1080/03031853.2012.749567
PLEASE SCROLL DOWN FOR ARTICLE
Full terms and conditions of use: http://www.tandfonline.com/page/
terms-and-conditions
This article may be used for research, teaching, and private study
purposes. Any substantial or systematic reproduction, redistribution,
reselling, loan, sub-licensing, systematic supply, or distribution in any
form to anyone is expressly forbidden.

The publisher does not give any warranty express or implied or make
any representation that the contents will be complete or accurate or
up to date. The accuracy of any instructions, formulae, and drug doses
should be independently verified with primary sources. The publisher
shall not be liable for any loss, actions, claims, proceedings, demand, or
costs or damages whatsoever or howsoever caused arising directly or
indirectly in connection with or arising out of the use of this material.
Downloaded by [Cirad-Dist Bib Lavalette] at 08:46 01 February 2013

Agrekon
Vol. 51 (3) 2012  ISSN Print 0303-1853/Online 2078-0400
© Agricultural Economics Association of South Africa pp 24–51
DOI: 10.1080/03031853.2012.749567
DEMYTHIFYING CONTRACT FARMING:
EVIDENCE FROM RURAL
SOUTH AFRICA
S. Freguin-Gresh*, M. d’Haese** and W. Anseeuw***
ABSTRACT
This paper intends to contribute to the ongoing debate about whether and how restructured
agri-food markets can provide viable market opportunities for small-scale farmers in South
Africa. It analyses contract farming from the small-scale farmer perspective to better understand
the implications for small-scale farmers of contractual arrangements with processing and/or
marketing firms.
The paper, based on empirical research conducted in the Limpopo Province of South Africa
using a combination of qualitative and econometric analyses, argues that contract farming is
not a panacea for small-scale farmers. On the one hand, contract farming improves agricultural
production for contract farmers who benefit from increased incomes, enables better access to
services and resources, and creates new opportunities to participate in markets. However, on
the other hand, the results show that contract farming remains limited and mostly involves the
already better-off, who have benefitted from specific development paths and public support.
This case study shows that contract farming in itself does not appear to provide an efficient
means of reducing poverty, nor does it provide an institutional tool through which to improve
rural livelihoods. It does, therefore, not represent a tool appropriate for the majority of small
farmers or for redressing the historical imbalances in the South African agricultural sector.
Keywords: contract farming, small-scale agriculture, poverty, South Africa
1 INTRODUCTION
Poverty levels in South Africa remain high and concentrated, both socio-
economically and geographically, with the previously disadvantaged communities
within the rural areas particularly affected (Pauw & Mncube, 2007).1 In this con-
* CIRAD, UMR ART-Dev & Universidad Centro Americana, Instituto de Investigacion
y Desarrollo Nitlapan, Apartado A 242 Campus de la UCA, Managua, Nicaragua
(E-mail: freguin@cirad.fr).
** Department of Agricultural Economics, Ghent University, Coupure links 653, 9000 Gent,
Belgium.
*** CIRAD, UMR ART-Dev & University of Pretoria, Post-Graduate School of Agriculture and
Rural development , 0002 Pretoria, South Africa.
Downloaded by [Cirad-Dist Bib Lavalette] at 08:46 01 February 2013

25
Demythifying contract farming: Evidence from rural South Africa
text, agriculture, and particularly the small-scale sector, is considered as one of
the main economic sectors to revitalise the rural economy (National Planning
Committee, 2010). As such, the government has implemented several policies to
support the small-scale sector (DAFF, 2010). These initiatives have, however, had
only limited success (Anseeuw, 2004; Perret HWDO, 2005).
It is in this context that the development of integrated value chains with the
emergence of new private actors, partly a result of the global restructuring of
markets and the deregulation of the agricultural sector, is seen as an opportunity
for small-scale farmers. Indeed, the contractual arrangements accompanying these
evolutions represent a possibility for small-scale farmers’ integration into modern
value chains. Contract farming is generally considered as an attractive mechanism
for integrating poorer farmers into the open-market economy (Glover, 1984;
Key & Runsten, 1999; Poulton et al., 2010) and, subsequently, for increasing
production and farm income. But, as mentioned by Poulton et al. (2010), it may be
selective, excluding the poor and subjecting them to high risks and agribusiness
normalisation, while failing to increase incomes due to unequal bargaining power,
with the farmers losing out.
This paper intends to contribute to the ongoing debate about whether and how
restructured agri-food markets can provide viable market opportunities for small-
scale farmers in South Africa. Based on empirical research into contract farming
in the Limpopo Province of South Africa, the paper analyses the determinants of
participation in contracts and estimates their implications in terms of livelihoods.
The South African case is particularly interesting. On the one hand, due to the
historical imbalances in access to land and secure tenure regimes, input and output
markets, infrastructures and quality control systems, small-scale farmers face high
transaction costs. Moreover, the differences in scale of trade between the small-
scale farmers and agribusinesses put the farmers at a disadvantaged bargaining
position (D’Haese & Van Huylenbroeck, 2005). On the other hand, the well-
developed value-chains and the public supports available, in particular through
land reform and positive action programmes, represent incentives and should
contribute to a conducive market and production environment (DAFF, 2010).
,QWKH¿UVWVHFWLRQDOLWHUDWXUHUHYLHZDQGEDFNJURXQGLQIRUPDWLRQRQFRQWUDFW
IDUPLQJ DUH SUHVHQWHG ZLWK D VSHFL¿F IRFXV RQ 6RXWK$IULFD 7KLV LV IROORZHG
in the second section, by the methodology used in the research study. Through a
combination of qualitative and quantitative approaches, we will establish a typol-
RJ\RIIDUPKRXVHKROGVDQGDQDO\VHWKHVLJQL¿FDQFHDQGGHWHUPLQDQWVRIFRQWUDFW
farming and of farm incomes. The results of these are presented in section three.
This will then lead, in section four, to wider conclusions on the role of agricultural
contracts in general and in revitalising South Africa’s previously marginalised
small-scale farming sector in particular.
Downloaded by [Cirad-Dist Bib Lavalette] at 08:46 01 February 2013

26
S. Freguin-Gresh, M. d’Haese and W. Anseeuw
 %$&.*5281'21&2175$&7)$50,1*
 &RQWUDFWIDUPLQJDQGW\SHVRI
DJULFXOWXUDOFRQWUDFWV
&RQWUDFWIDUPLQJFDQEHGH¿QHGDVDQDJUHHPHQWEHWZHHQDIDUPHUDQGDEX\HU
ranging from simple oral arrangements to formal written documents, in which
SDUWLHVUHVSHFWLYHO\FRPPLWWRVHOODQGEX\VSHFL¿FYROXPHVRUDFUHDJHVXQGHUSUH
established conditions (Glover, 1984; Minot, 1986). The buyer can be a local or a
transnational agribusiness (processor, exporter, retail outlet or shipper), a private
plantation, a parastatal with its own production, or local merchants (greengrocers,
wholesalers, hawkers, brokers, etc.).
In institutional economics, contract farming is described as a hybrid agreement
that positions itself between the two extremes of the institutional arrangement
spectrum, namely spot markets and market integration. In spot markets, products
are sold and bought immediately, at a price set during the transaction and with no
LQYROYHPHQWRI WKHEX\HU LQWKHSURGXFWLRQRULQ WKHGH¿QLWLRQ RIWKHFRQGLWLRQV
of the transaction. At the other extreme, full vertical integration implies that the
buyer controls all stages of production, processing and distribution throughout the
value chain. Between these poles, contract farming allows the buyer a measure of
control (decision-making rights) over production without formally engaging in
farming activities (Grosh, 1994; Ménard, 2005). The allocation of risk depends
on the terms of the contracts. As such, contract farming provides a response to
market failures with respect to inputs, credit, insurance, information and outputs,
by reducing the associated transaction costs, monitoring, transfer of goods and
bargaining and enforcement (Key & Runsten, 1999; Poulton et al., 2010).
The literature on contract farming differentiates between three classic types
of contracts according to their main objectives, the transfer of decision-making
rights and the shifting of risk from the farmer to the buyer (Key & Runsten, 1999;
Mighell & Jones, 1963; Minot, 1986). 0DUNHWVSHFL¿FDWLRQFRQWUDFWV refer to pre-
harvest agreements that engage a buyer in providing a market outlet to a farmer
under pre-established conditions often related to price, quantity, quality and timing.
Thus, the farmer delegates a part of the risk to the buyer, while keeping control
RYHUSURGXFWLRQ%RWKWKHIDUPHUDQGWKHEX\HUEHQH¿WIURPWKHSULFHSUHPLXPRQ
WKHTXDOLW\DQGWKHVWDELOLW\LQWKHÀRZRIVXSSO\RISURGXFWVWRVSHFL¿HGPDUNHWV
0DQDJHPHQWSURYLGLQJ SURGXFWLRQPDQDJHPHQW FRQWUDFWV are similar to
marketing contracts. These contracts, however, delegate some of the farmer’s
control over the production process to the buyer. In terms of these contracts the
DGRSWLRQRIVSHFL¿FIDUPLQJSUDFWLFHVODQGSUHSDUDWLRQSODQWLQJGDWHVVHHGOLQJV
fertilizer application rates and dates, etc.) or the choice of post-harvest manage-
ment practices will come under the technical supervision of the buyer to attain
Downloaded by [Cirad-Dist Bib Lavalette] at 08:46 01 February 2013

27
Demythifying contract farming: Evidence from rural South Africa
higher quality and to control the timing of output. The buyer recoups the costs of
extension from the proceeds of marketing a higher-quality product according to
the timing of demand.
Finally, UHVRXUFHSURYLGLQJFRQWUDFWV are the closest arrangement to full ver-
tical integration and require not only that the buyer provide a market outlet to
the farmer but also that he delivers input packages on credit and corresponding
technical assistance in its use. These result in the buyer having major control over
production with the contract shifting most decision-making rights and risks to the
buyer (Figure 1).
Figure 1: Typology of contracts
Source: Adapted by the authors from Mighell & Jones (1963) and Minot (1986)
Contract farming may overcome certain constraints small-scale farmers are
typically faced with in developing countries, such as access to resources (inputs,
services, and information) and markets. In South Africa, black small farmers
struggle with access to resources as a result of the marginalisation which occurred
through discriminatory policies under apartheid (Eastwood et al., 2006). Today,
about 1.2 million black small farmers occupy 18 per cent of the farm land (13
per cent of the communal land in the former homelands and 5 per cent of land
redistributed through land reform). These farmers mainly engage in family-based,
Downloaded by [Cirad-Dist Bib Lavalette] at 08:46 01 February 2013

28
S. Freguin-Gresh, M. d’Haese and W. Anseeuw
subsistence agriculture (compared to around 40,000 large-scale farmers, owning 82
per cent of the privately owned agricultural land, characterised by highly intensive
farming activities and producing 95 per cent of the country’s marketed agricultural
output). Most of them lack access to resources (land, water, infrastructure, and
credit facilities). Consequently, contract farming could facilitate their access to
information, technical assistance, credit and inputs, reduce the uncertainty around
marketing their products, improve their integration into modern value-chains and,
consequently, increase their farm incomes (Glover, 1984; Key & Runsten, 1999;
Minot, 1986; World Bank, 2007), as well as providing institutional mechanisms
WRDGGUHVVWKHGLI¿FXOWLHVWKH\IDFH+RZHYHUFRQWUDFWIDUPLQJFRXOGSRWHQWLDOO\
also lead to increasing market segmentation and exclusion (Little & Watts, 1994;
Porter & Phillips-Howard, 1997a, 1997b; Poulton et al., 2010; Vorley et al., 2007)
or remain limited in terms of the number of farmers involved, limiting the overall
impact (Losch et al., 2010).
 'HYHORSPHQWDQGVFDOHRIFRQWUDFWIDUPLQJ
LQ6RXWK$IULFD
Contract farming has a long history in South African agriculture (Karaan, 1999;
Kirsten & Sartorius, 2002a, 2002b; Louw et al., 2006; Porter & Phillips-Howard,
1997a, 1997b; Ortmann & King 2010). The historical development of agricultural
contracts can be traced to the early 1900s in relation to the emergence of
FRRSHUDWLYHV2UWPDQQ.LQJ7KLVKLJKOLJKWVWKHVLJQL¿FDQW
role of contracts in the development of white commercial agriculture. Indeed,
XQWLO WKH V ZKLWH IDUPHUV EHQH¿WHG IURP LQSXW VXSSOLHV VHHGV IHUWLOL]HUV
chemicals and credit), marketing facilities through various boards and service
provision (storage, transport) organised by the cooperatives through various
institutional arrangements. Due to discriminatory policies, black farmers located
in the homelands had only limited, if any, access to most domestic markets, and
consequently to their contracts. They commercialised their outputs – when they
were able to generate surpluses – on informal local markets within the homelands.
Contract farming in South Africa has been analyzsd for small-scale farmers
LQ WUDGLWLRQDO YDOXH FKDLQV VXFK DV WHD VXJDUFDQH WLPEHU WREDFFRÀRZHUV DQG
beverages (Kirsten & Sartorius, 2002b). These schemes appear, however, to be
PRUHWKHUHVXOWRIVSHFL¿FLQLWLDWLYHVRIODUJH HVWDWH DFWRUV GXULQJ WKH DSDUWKHLG
and pre-liberalisation period, than what McMichael and Myhre (1991) call the
“restructuring of agrifood markets”. Quantifying the scale of contract farming,
ZKDWHYHUWKHOHYHORIDQDO\VLVLVH[WUHPHO\GLI¿FXOW:KHUHDVVRPHDXWKRUVDWWHPSW
to estimate the scope of its development in Africa (Grosh, 1994; Little & Watts,
1994), most studies in this respect focus on its impact on the farm and household
(e.g. Bellemare, 2010a, 2010b; Maertens & Swinnen, 2009). In a study on contract
Downloaded by [Cirad-Dist Bib Lavalette] at 08:46 01 February 2013

29
Demythifying contract farming: Evidence from rural South Africa
farming in South Africa, Vermeulen et al. (2008) estimated its scale in the fresh
produce sector and assessed the level of participation of small-scale farmers. This
research shows that almost 80 per cent of the volumes of fruits and vegetables
supplying the South African processing industry (21 per cent of the production)
was exchanged through contracts and that between 70 per cent and 100 per cent of
the produce sold in supermarkets was supplied under contract, while the meat and
egg sectors favoured full vertical integration (Vermeulen et al., 2008). This study
DOVRVKRZHG WKDWRQO\SHUFHQW RIWKHFRQWUDFWVLGHQWL¿HGLQYROYHG VPDOOVFDOH
IDUPHUVZLWKIHZVXSSOLHUV7KHVHUHVXOWVFRQ¿UPWKH¿QGLQJVRISUHYLRXVVWXGLHV
which show that contract farming, for fresh produce in particular, usually implies
a small number of producers, and very few small farmers (Table 1).
Table 1: Extent of contract farming in South Africa for selected commodities
6XEVHFWRU RIIDUPHUVXQGHU
FRQWUDFWV
RIVPDOOVFDOHDQG
HPHUJLQJIDUPHUV
XQGHUFRQWUDFWV
RI6RXWK$IULFDQ
IDUPHUVXQGHU
FRQWUDFWV
Sugar cane 16 045 14 445 small-scale
growers (8% of
sugarcane production)
+ 385 emerging
growers
1.2%
Timber 50 000 15 000 4%
Cotton 3 000 – 0.2%
Processed fruits,
snack and nuts
2 709 209 0.2%
All fresh fruits and
vegetables
3 430 278 0.3%
Processed
vegetables
350 87 ns
Sources: authors’ compilation from Vermeulen 2008; FAO 2004; South African Sugar Association 2011; FAO Expert
Consultation on Contract farming in Africa, Johannesburg, South Africa (4–7 May 2009).
* The Department of Agriculture, Forestry and Fisheries’ annual report 2009/10 estimates 40 000 commercial farm units
in 2007 and 1.2 million small-scale farms in the former homelands.
3 METHODOLOGY
Measuring the contribution of agricultural contracts to farmers’ welfare levels is
methodologically challenging. Contracts are often not randomly distributed among
households (Barrett et al., 2010) and their access needs to be instrumented in a
regression analysis of their impact on welfare (Bellemare, 2010b). In this paper, a
three-step selection model is developed, addressing the determinants of access to
contracts, as well as the decision to market produce, while estimating the impact
Downloaded by [Cirad-Dist Bib Lavalette] at 08:46 01 February 2013

30
S. Freguin-Gresh, M. d’Haese and W. Anseeuw
of contracts on income and not on welfare. Secondly, the econometric model is
entwined with a qualitative approach that studies a typology of households and
their development paths. This enables us to draw more informed conclusions on
the reasons why some farmers are being excluded from markets and/or contract
farming. The combination of standard econometric models with qualitative
analyses is particularly original and relevant.
 6LWHVHOHFWLRQDQGVDPSOH
Data for this study was collected from farm households in the Greater Tzaneen
Municipality in the Limpopo Province2 (Figure 2). Within the greater municipality,
the survey was conducted in the region of Nwa’Mitwa which includes the
settlements of Mandlakhazi, Mbekwani, Nwa’Mitwa, Nwadjaheni and Babanana
as well as the private farms surrounding the community (Jaffray, Welverwacht,
Taganashoek, La Dauphine, Duplex and Uitzoek).
Figure 2: Location of the study area
The selection of the study area was based on the following criteria:
1. Agro-climatic conditions and importance of agriculture
The area is characterised by relatively good agro-climatic conditions with a
tropical/semi-arid climate (average temperature of 25°C, with annual rainfall of
between 500 and 700 mm) and by relatively homogeneous soils (alluvial and sandy
Downloaded by [Cirad-Dist Bib Lavalette] at 08:46 01 February 2013

31
Demythifying contract farming: Evidence from rural South Africa
RQWKHWRSDQG ÀDW DUHDV FOD\ VRLOVLQWKHORZHUO\LQJ DUHDV 'HVSLWH DGLVWLQFW
dry season, crop production is possible throughout the year under irrigation. As a
result, agriculture is well developed and is the most important economic activity
in the Mopani District where the study area is located.3
2. Historical background, land characteristics and poverty prevalence
The southern part of Nwa’Mitwa is located on tribal lands that were part of the
Gazankulu Homeland. Agriculture is representative of the prevailing situation
in the former homelands, consisting of small plots with communal land tenure,
cropping and livestock systems based on traditional practices, farming activities
dedicated to family consumption, high demographic and land pressure, etc. Most
households face a daily struggle with poverty, particularly due to the high average
age of the population as well as general under- and- unemployment. The northern
part consists of commercial farms. This sub-region is subject to land claims and
some governmental projects have been implemented.
3. Market proximity, off-farm jobs and contract farming opportunities
The proximity to commercial farms both located in the study region and in the Great
Letaba River Valley, one of the leading regions for fruit and vegetable production,
has enhanced the development of agricultural wage labour opportunities and the
presence of contract farming. The population of the study area is estimated at around
16 000 households (Municipal Demarcation Board, 2006) of which, according to
RXU¿QGLQJV LDERXW DUHLQYROYHGLQDJULFXOWXUHLQFOXGLQJOLYHVWRFNDQG
self-consumed cropping activities in gardens and (ii) 82 are private land owners.
A probability sample was not possible because exact lists of community members
were unavailable. It was consequently decided to conduct:
(a) 110 surveys using a questionnaire among a random group of respondents geo-
graphically spread over the area, which allowed, considering the available time
and resources, us to capture the existing diversity and to provide the necessary
information for the establishment of a typology. The survey was random but
the number of commercial and contract farmers was purposefully higher. Four
cases were deleted from the analysis due to missing data, resulting in a data set
of 106 valid cases.
EFRPSOHPHQWDU\LQWHUYLHZVDPRQJVHOHFWHGKRXVHKROGVLGHQWL¿HGGXULQJWKH
¿UVWLQWHUYLHZVLQRUGHUWREHWWHUXQGHUVWDQGWKHLUVSHFL¿FWUDMHFWRULHV
(c) 36 complementary surveys among farmers engaged in contracts and also
LGHQWL¿HGLQ¿UVWURXQGRILQWHUYLHZV
(d) 239 additional short surveys to check how representative the results were in
relation to the size of the population in the region.
Downloaded by [Cirad-Dist Bib Lavalette] at 08:46 01 February 2013

32
S. Freguin-Gresh, M. d’Haese and W. Anseeuw
The dataset is rich and representative regarding the diversity of farming systems
and contracts in the study area. Firstly, it combines quantitative and qualitative
data. Secondly, as we will show in the analysis, it covers a whole range of farming
systems ranging from less developed and subsistence small-scale systems to more
advanced large-scale and commercial ventures. Thirdly, it includes farmers with
and without contracts and assesses the type of agreements they are involved with.
Finally, the dataset enables comparison across farm households with different
land tenure regimes and different levels of governmental support, including
EHQH¿FLDULHVIURPODQGUHIRUPSURJUDPPHV
 4XDOLWDWLYHGDWDDQDO\VLV
Two methodological approaches were used to analyse agriculture and rural
livelihoods. Firstly, an agrarian systems diagnostic approach was implemented in
order to characterise the agro-ecological, technical, historical and socio-economic
IDFWRUV LQÀXHQFLQJ WKH WUDQVIRUPDWLRQ RI WKH UXUDO HQYLURQPHQW &RFKHW 
Devienne, 2004; Dufumier, 1996). Then agricultural practices were analysed (such
as combinations of crops and livestock and their productivity levels) which were
subsequently linked to the asset endowment and households’ development paths
to establish a consistent typology. Secondly, a livelihood approach was applied
WRHDFKLGHQWL¿HGW\SHRIKRXVHKROGDLPLQJDWXQGHUVWDQGLQJWKHFRPELQDWLRQRI
activities and income sources.
 4XDQWLWDWLYHDQDO\VLV
The econometric approach aimed at analysing the determinants of the uptake of
contracts and contract farming’s contribution to the households’ farm income.
7KHDQDO\VLVZDVGRQH LQ WKUHH VWDJHV ,QWKH¿UVWLQVWDQFHD SURELW PRGHO ZDV
XVHG WR DQDO\VH WKH XSWDNH RI FRQWUDFWV 7KH VHFRQG VWHS WKH ¿UVW PRGHO RI D
Heckman model) was a probit analysis of factors determining whether or not
farmers commercialise their agricultural produce (if not, they are considered
to be subsistence farmers). The third step (the second model of the Heckman
model) was a regression analysis of the determinants of farm income for farmers
participating in markets. The independent variables included the probabilities of
KDYLQJDFRQWUDFWVDYHGUHVXOWVRIWKH¿UVWSURELWPRGHODQGWKHLQYHUVHPLOOVUDWLR
that was saved from the second probit model. The estimation of the probit model
as well as the Heckman procedure (models two and three were jointly estimated)
was done using Stata. A description of the Heckman model is given in Appendix 1.
The rationale behind this approach is the latent endogeneity and selection-bias
issues around the uptake of contracts and commercialisation of on-farm income.
While market orientation, as opposed to the choice of subsistence production, is
intrinsically linked to farm income, we suspected selection biases and corrected
Downloaded by [Cirad-Dist Bib Lavalette] at 08:46 01 February 2013

33
Demythifying contract farming: Evidence from rural South Africa
these by estimating a Heckman model (Greene, 2000). We also suspected that the
uptake of contracts would be endogenous to farm income as a buyer’s decision to
procure from a certain region and within that region from a number of sellers, is
not random (Barrett HWDO, 2010; Bellemare, 2010b). Because the analysis seeks
to test whether contract farming increases farm income, we needed to instrument
the uptake of contracts to address these potential endogeneity problems while
controlling for the decision to sell agricultural produce. Following Wollni and
Zeller (2007), we introduce the probabilities obtained from the probit model on
uptake in the farm income model. It is worth noting that farmers who had contracts
were marketing their produce, while not all farmers who were marketing had
contracts. Contract uptake is instrumented by the importance of off-farm income in
the total household income.4 Furthermore, there is no theoretical reason to assume
a causal relationship between the share of income from off-farm sources and
agricultural income. Likewise, the likelihood to market produce is instrumented
E\ WKH OHYHO RI WUDQVIHUV KLJK VRFLDO JUDQWV DQG UHPLWWDQFHV PD\ LQÀXHQFH WKH
likehood that farmers commercialise, but they are exogenous and arguably not
necessarily causally linked to farm income.
7KH LQGHSHQGHQW YDULDEOHV VHOHFWHG IRU WKH ¿UVW SURELW PRGHO DUH VLPLODU WR
those commonly used in the literature for similar models (Bolwig et al., 2009;
Miyata et al., 2009; Wollni et al., 2010; Wollni & Zeller, 2007) namely (a)
household characteristics (household size, age and gender of the head); (b) assets
endowments (land area, size of the cattle herd) and (c) share of off-farm income
in total income.
The model for household i then becomes:
iiiiii arealandJHQGHUsquaredDJHDJHizehouseholds& __ 543210
DDDDDD
iii uIDUPoffLQFRPHsharecattle  ___
76
DD
(1)
with Ci being a dummy variable (0/1) indicating if the household has a contract or
not, and ui being the error term. The probability of Ci=1 is calculated by a probit
PRGHO VLPLODU WR WKH ¿UVW VWHS RI WKH +HFNPDQ PRGHO JLYHQ LQ HTXDWLRQ  LQ
Appendix 1.
Independent variables for the second probit model were similar to exogenous
KRXVHKROG FKDUDFWHULVWLFV LQ WKH ¿UVW PRGHO KRXVHKROG VL]H DJH DQG JHQGHU RI
the head), supplemented with the level of education of the head and income from
transfers. As mentioned above, we consider the two latter variables as relevant
because these additional income sources will determine the choice of farming
V\VWHPZLWKRXWGLUHFWO\LQÀXHQFLQJWKHIDUPLQFRPH
The model is for household i:
Downloaded by [Cirad-Dist Bib Lavalette] at 08:46 01 February 2013

34
S. Freguin-Gresh, M. d’Haese and W. Anseeuw
iii squaredDJHDJHizehouseholds  _
3210
EEE
E
ii educationJHQGHU  54
EE
iiii vedracesUHPLWJUDQWVsocial  inftan_ 876
EEE
(2)
with viEHLQJWKHHUURUWHUP7KLVIXQFWLRQLVWKH¿UVWVWHSRID+HFNPDQVHOHFWLRQ
model as explained in Annex 1.
Agricultural income is regressed against (a) the household structure (age,
gender), (b) farm characteristics (number of active people on the farm, land, access
to irrigation and land tenure regimes) and (c) the probability of access to contracts
The model then becomes for household i:
iiiiii tenureprivatelandJHQGHUDJHPHPEHUVactive
Y
 543210 __
J
J
J
J
J
J
iiii Z,0L&edraesbeneficariUHIRUPland  9876 inf__
J
J
J
J
with IMi being the inverse mills ratio/lambda, and Zi the error term.
The second step of the Heckman procedures is expressed by equation (1) and
solved as presented in Annex 1, for those households participating in the market. It
should be noted that due to the relatively low number of cases in this second step,
the number of variables was limited and the results were interpreted with care.
However, the outcome of the econometric models is compared and triangulated
with the qualitative data analysis.
4 RESULTS
 +RXVHKROGW\SRORJ\
Of the 106 households interviewed, seven types of farming systems were
LGHQWL¿HG 7DEOH  )RXU FULWHULD ZHUH XVHG L WKH VL]H DQG WKH W\SH RI ODQG
(garden, communal area, private land); (ii) the commercialisation of agricultural
products, including the buyers (spot markets, local merchants, supermarkets,
processors, export agents) and the destination of the products when sold (local,
domestic, export markets); (iii) the combination of crops and livestock (share of
staple food, of fruits and vegetables, of livestock in the farm production value) and
asset endowment (equipments, irrigation systems etc.) and (iv) access to and the
importance of off-farm income.
7KH W\SRORJ\ HQDEOHV WKH LGHQWL¿FDWLRQ RI GLIIHUHQW KRXVHKROG VWUDWHJLHV
DQGWKHTXDQWL¿FDWLRQ RIWKHGLVWULEXWLRQ RIFRQWUDFWVZLWKLQWKH VWXG\DUHD7KH
typology presented in this article includes all households interviewed, including
DJURXSVLJQL¿FDQWLQWHUPVRIQXPEHURIKRXVHKROGVEXWZKLFKLVQRWHQJDJHG
in the commercialisation of its production. It was decided to keep this group to
DGGUHVVWKHLVVXHRI VFDOH DQG VLJQL¿FDQFH RIFRQWUDFWVZLWKLQWKH VWXG\ UHJLRQ
The other six types enable the analysis of the conditions of engagement and entry
barriers regarding contractual agreements.
Downloaded by [Cirad-Dist Bib Lavalette] at 08:46 01 February 2013

35
Demythifying contract farming: Evidence from rural South Africa
Table 2: Typology of farming households in the Greater Tzaneen Municipality
Q RIWKHVDPSOH RI++LQDUHD
Micro-farmers cultivating residential gardens
for self-consumption, depending on off-farm
activities, remittances and social grants
15 14.2 53.8
Subsistence small-scale farmers depending on
off-farm income, combining staples for self-
consumption and vegetables for local markets
28 26.4 39.6
Small-scale producers of staples and F&V for
local markets, depending on off-farm activities
and social grants
6 5.7 4
Medium-scale producers specialised in
vegetable production for the local and
domestic markets
24 22.6 1
Emerging industrial chicken and vegetable
producers
16 15.1 0.2
Extensive commercial farmers, producers of
fruits mainly for the domestic market
11 10.4 1
Intensive commercial producers of fruits
and vegetables for the domestic and export
markets
6 5.7 0.4
7RWDO 106 100 100
Notes: *Number of detailed questionnaires to a random group of respondents allowing for the capturing of the diversity
of household types. **Based on the results of the 239 short interviews conducted and being representative of the
population in the study area.
4.1.1 Micro-farmers cultivating residential gardens for
self-consumption, depending on off-farm incomes
(n=15)
The micro-farmer group is very heterogeneous in terms of livelihoods. These
mostly female-headed households implement survival strategies to cope with very
low incomes and take any opportunity that allows them to improve their livelihoods
such as small irregular jobs in the service sector in the community, casual agricultural
labour, social grants and remittances (with the younger, active population often
having migrated). However, despite engaging in these diverse activities, these
households seldom succeed in generating an income above the poverty line. In
terms of farming, they have no or very poor access to land and consequently rely
only on the cultivation of residential gardens. They combine starchy staples and
vegetable production for family consumption, producing only during the rainy
season due to a lack of access to irrigation water. As a result, farming activities are
limited and contribute only marginally to this group’s subsistence (21 per cent of
Downloaded by [Cirad-Dist Bib Lavalette] at 08:46 01 February 2013

36
S. Freguin-Gresh, M. d’Haese and W. Anseeuw
JOREDOLQFRPH:LWKLQVXI¿FLHQWDFFHVVWRDVVHWV DQGLQVXI¿FLHQWSURGXFWLRQWKLV
group is mainly excluded from markets and farming provides only a basis for food
security.
4.1.2 Subsistence small-scale farmers depending on
off-farm income, combining staples for
self-consumption and vegetables for
local markets (n=28)
The second group consists of couples composed of a retired person and an active
person engaged in a permanent activity (small business). Social grants or off-farm
income are invested in an irrigation system which provides water for domestic
use, for the sale of drinking water to the community and for irrigation purposes
(manual only). Unlike the micro-farmer group, these households have access to
a plot of arable land in the communal lands (on average one hectare), allowing
them to cultivate staples and vegetables and to keep an orchard (usually mangoes).
They may also have a small herd of cattle that grazes on communal land and may
fatten pigs. Their produce is sold on spot markets within the community or to local
PHUFKDQWV7KH IDUPLQJ DFWLYLW\ DOORZVWKHPWRKDYHIRRG DQG D¿QDQFLDOEDVLV
but, for most of them, farming is not considered to be a productive activity (17
per cent of total income). They furthermore have no investment capacity and rely
heavily on social grants for their subsistence.
4.1.3 Small-scale producers of staples and fruits and
vegetables for local markets, depending on
off-farm activities and social grants (n=6)
The third group combines small-scale farming (28 per cent of global income),
off-farm activities (taxi, small business) and social grants, which represent a
VLJQL¿FDQWSDUWRIWKHLULQFRPH2IIIDUPLQFRPHLQYHVWHGLQDQLUULJDWLRQ
system, has allowed them to develop marketable year-round vegetable production.
These households have access to a plot in the communal area (average size 1.8
hectares) which they cultivate in addition to a garden. Products are sold on spot
markets or to local merchants. For this group, farming is a productive activity and
the basis of their livelihoods; they would like to develop their farming activity
if their constraints can be overcome (limited access to resources, lack of credit,
GLI¿FXOWLHVLQFROOHFWLQJDQGWUDQVSRUWLQJWKHLUSURGXFWVWRPDUNHWVHWF
Downloaded by [Cirad-Dist Bib Lavalette] at 08:46 01 February 2013

37
Demythifying contract farming: Evidence from rural South Africa
4.1.4 Medium-scale producers specialised in vegetable
production for the local and domestic markets
(n=24)
These households are better endowed (largest plots in communal lands) and due
WR PRUH HI¿FLHQW LQIUDVWUXFWXUH SULYDWH ERUHKROHV LUULJDWLRQ V\VWHPV WUDFWRUV
and private vehicles) they have been able to develop a marketable vegetable
production (up to three cycles per year) which they sell to local merchants, to fresh
produce markets or under formal contracts to supermarkets (organic production-
PDQDJHPHQW FRQWUDFWV), to processors or restaurants (PDUNHWLQJ FRQWUDFWV). As
a result of successful but expensive practices, farming has become the pillar of
their livelihood (67 per cent of total income), the rest being non-farm sources.
)DUPLQJ LV D SUR¿WDEOH PHDQV RI H[LVWHQFH WR WKHP EXW ZLWKRXW WKH VXSSRUW RU
WKHRSSRUWXQLWLHVIURPZKLFKWKH\EHQH¿WWHGRQWKHEDVLVRIWKHLUSHUVRQDOVRFLDO
networks in order to access production factors and market opportunities, or both,
they would not have been able to develop this activity.
4.1.5 Emerging industrial chicken and vegetable
producers (n=16)
These households are specialised in intensive vegetable production under PDQDJH-
PHQWSURGXFWLRQFRQWUDFWV with agribusinesses and/or industrial broiler production
under UHVRXUFHSURYLGLQJFRQWUDFWV with a local agribusiness (Bushvalley). Agri-
culture is their only economic activity. The viability and the sustainability of this
specialised and intensive but expensive farming system is questionable, both in
agro-ecological as well as economic terms. It is not clear if the farmers engaged
in this production system would have had the means to invest and renew their
equipment and to develop an economically sustainable activity without massive
external (mostly governmental) support.
4.1.6 Extensive commercial farmers, producers of fruits
mainly for the domestic market (n=11)
([WHQVLYHFRPPHUFLDOIDUPHUVFRPELQHDQLQGHSHQGHQWRUDTXDOL¿HGSHUPDQHQW
off-farm activity and managerial farm requiring numerous workers. They are well
equipped, with an operational irrigation system and a tractor. They have mostly
specialised in extensive mango production (low use of inputs and workforce).
Mangoes are usually harvested green to be delivered to local processors or sold
ripe to merchants, fresh produce markets or exporting agents. Some of them have
contracts (PDUNHWLQJ) which are usually verbally concluded.
Downloaded by [Cirad-Dist Bib Lavalette] at 08:46 01 February 2013

38
S. Freguin-Gresh, M. d’Haese and W. Anseeuw
4.1.7 Intensive commercial producers of fruits and
vegetables for the domestic and export
markets (n=6)
The households of this last group are specialised in managerial farming. They
have developed an intensive (in terms of labour, capital and inputs) production of
fruits under irrigation. They own tractors, greenhouses, warehouses and a packing
unit to satisfy the requirements and standards of their buyers (local merchants,
processors, fresh produce markets and exporting agents) with whom half of
them have various types of contracts (mostly PDUNHWLQJ but also production-
PDQDJHPHQWFRQWUDFWV). Their activities are concentrated on large areas of private
land. They combine their production activities with extensive cattle breeding on
private pastures.
Table 2 also presents the distribution of the six types, based on a shorter survey
conducted among a larger number of farmers in the study area (n=239). The
resulting distribution differs from the one based on the long survey. The micro-
IDUPHUVVHHPWREHWKHODUJHVWJURXSIROORZHGE\VXEVLVWHQFHVPDOOVFDOH
IDUPHUV)URPWKHIDUPHUVLQWKHDQDO\VLVWKDWSDUWLFLSDWHGLQWKHORQJ
LQWHUYLHZ PHGLXPVFDOH IDUPHUV  HPHUJLQJ LQGXVWULDO FKLFNHQ SURGXFHUV
DQGLQWHQVLYHDQGH[WHQVLYHFRPPHUFLDOIDUPHUVVHHPWREH
overrepresented.
 ([WHQWDQGFKDUDFWHULVWLFVRIFRQWUDFWIDUPHUV
The extent of contract farming in the region remains limited with only 36 farmers
(34 per cent of the interviewees) having at least one contract, whether verbal or
LQZULWLQJ)ROORZLQJRXU¿QGLQJVLQWKHVKRUWVXUYH\FRQWUDFWIDUPHUVUHSUHVHQW
less than 1.2 per cent of the total number of farm households in the region. The
limited extent of contract farming can be explained by the fact that, as shown by
the typology analysis, the majority (54 per cent of the interviewees) are micro or
small-scale subsistence-oriented farmers. As they do not participate in markets
WKH\DUHDOVRH[FOXGHGIURPFRQWUDFWIDUPLQJDVFRQ¿UPHGLQ7DEOHZKLFKOLQNV
the farm household typology with participation in markets, contracts and the
destination of the agricultural produce.
Downloaded by [Cirad-Dist Bib Lavalette] at 08:46 01 February 2013

39
Demythifying contract farming: Evidence from rural South Africa
Table 3: Buyers, contracts, and types of households
7\SHVRI
EX\HUV
HDFK
W\SH
KDYHDW
OHDVWRQH
FRQWUDFW
7\SHVRIDJULFXOWXUDOFRQWUDFWV
Micro-farmers No sale 0 No contract
Subsistence small-scale No sale 0 No contract
Spot markets
Small-scale producers Spot markets 0 No contract
Local
merchants
Medium-scale producers Local
merchants
57 No contract
Fresh produce
markets
(FPM)
No contract
Restaurants Informal marketing agreements
Supermarkets No contract or formal organic
production-management contracts
Processors Informal marketing agreements or
formal production-management
contracts
Emerging farmers Local
merchants
100 No contract or informal
agreements
Road-side
stalls
Informal marketing agreements
FPM No contract or informal marketing
agreements
Supermarkets Formal organic production-
management contracts
Processors Informal agreements or formal
production-management contracts
Extensive commercial
farmers
Local
merchants
27 No contract
FPM No contract or informal marketing
agreements
Processors Informal gentlemen’s agreements
Exporting
agents
Formal market-speci±cation
contracts
Downloaded by [Cirad-Dist Bib Lavalette] at 08:46 01 February 2013

40
S. Freguin-Gresh, M. d’Haese and W. Anseeuw
Intensive commercial
farmers
Local
merchants
50 No contract or informal
gentlemen’s agreements
FPM No contract or informal
gentlemen’s marketing agreements
Processors Informal agreements or formal
marketing or production-
management contracts
Exporting
agents
Marketing contracts
Tables 4 and 5 compare the mean values of the major characteristics of contract
IDUPHUV DQG QRQFRQWUDFW IDUPHUV 7KH WZR JURXSV GLIIHU VLJQL¿FDQWO\ LQ WKHLU
demographic characteristics, asset endowment and income structure. Contract
farmers seem to have smaller households (with fewer older members and more
children) and are led by younger heads (mostly male); they have better access
to land, mostly under private tenure or in land reform schemes and all of them
have access to irrigation (through various types of irrigation systems), with larger
cattle herds and higher agricultural incomes. Among the group of non-contracted
IDUPHUVZHFRXQWPRUHIHPDOHKHDGHGKRXVHKROGVZLWKVLJQL¿FDQWO\ORZHUOHYHOV
of education, less land (consisting mainly of gardens) and a greater dependency
on social grants.
Table 4: Comparison of means between contract and non-contract farmers
(standard deviation in brackets)
&RQWUDFWHGIDUPHUV
(n=36)
1RQFRQWUDFWHG
IDUPHUVQ 
7VWDW
Household size (nb) 2.56
(1.92)
5.17
(3.11)
5.338***
Share children in household (%) 14.41
(17.76)
6.88
(18.02)
2.049**
Age of household head (years) 50.42
(11.84)
57.41
(14.47)
2.500**
Available land (ha) 98.04
(106.43)
20.17
(63.60)
–4.035***
Cultivated land (ha) 36.58
(63.76)
10.69
(30.98)
–2.300***
Share cultivated (%) 41
(33)
92
(22)
8.197***
Number of cattle 19.03
(43.88)
4.26
(13.28)
–1.974*
Downloaded by [Cirad-Dist Bib Lavalette] at 08:46 01 February 2013

41
Demythifying contract farming: Evidence from rural South Africa
Agricultural income (R/year) 2,095,423
(3,122,966)
143,956
(600,139)
–3.714***
Remittances received (R/year) 2,666
(5,928)
4,670
(14,609)
0.789
Social grants received (R/year) 4,369
(6,991)
8,314
(7,714)
2.572**
Total income (R/year) 2,115,694
(3,115,194)
171,919
(601,174)
–3.708***
Share of income from off-farm
source (%)
16.5
(28.5)
63.7
(33.9)
7.568***
Note: *** signi±cant at 1%, ** signi±cant at 5%, * signi±cant at 10%
Table 5: Comparison of shares (%) of determinants between contracted and
non-contracted farmers
&RQWUDFWHGIDUPHUV
(n=36)
1RQFRQWUDFWHG
IDUPHUVQ 
&KLVTXDUH
stat
Gender household head (% male) 83 47 12.915***
Access to off-farm income (% yes) 56 89 14.795***
Education above primary level (% yes) 94 33 36.276***
Rain-fed production (% yes) 11 50 14.460***
Private land tenure (%) 50 17
Communal land tenure (%) 14 10
Gardens (%) 0 63
Land reform (%) 36 0
Garden + communal land (%) 0 10 60.893***
Note: *** signi±cant at 1%, ** signi±cant at 5%, * signi±cant at 10%
a Chi-square test for land reform versus contract farming
In the probit model we explain participation using several variables which are
expected to determine households’ ability to engage in contracts. The probit model
LVKLJKO\VLJQL¿FDQWDQGSUHGLFWVDOPRVWSHUFHQWRIWKHREVHUYHGHQJDJHPHQWLQ
contracts (Table 6). The model supports the conclusions of the typology analysis
DQG¿QGVWKDWODQGVL]HLQFUHDVHVWKHSUREDELOLW\WRHQJDJHLQFRQWUDFWVZKLOHWKH
share of off-farm income in the total income decreases this probability. The model
FRQ¿UPVWKDWIDUPHUV ZLWK ORZODQGHQGRZPHQWV PLFURVXEVLVWHQFH DQGVPDOO
scale farmers) are more likely to be excluded from contract farming.
Downloaded by [Cirad-Dist Bib Lavalette] at 08:46 01 February 2013

42
S. Freguin-Gresh, M. d’Haese and W. Anseeuw
Table 6: Probit model of uptake of contracts (0: no contract; 1: contract)
&RHI±FLHQW 5REXVWVWGHUURU ]VWDW
Household size –0.247 0.086 –2.87***
Age of household head (yrs) 0.201 0.091 2.21**
Age squared (yrs²) –0.002 0.001 –2.40**
Gender of household head (1:male/2:female) –0.248 0.352 –0.70
Share off-farm income (%) –0.012 0.007 –1.68*
Log land size (log ha+1) 0.282 0.138 2.04**
Cattle (number of heads) 0.001 0.005 0.19
Constant –3.870 2.229 –1.74*
N= 106
Wald chi²=36.67
Pseudo R²= 0.519
Correctly classi±ed 87.5%
The combined quantitative and econometric analysis thus assists in classifying
WKHFRQWUDFWIDUPHUVLQWRWZRFDWHJRULHV7KH¿UVWJURXSRIKRXVHKROGVFRQVLVWVRI
large-scale commercial farmers. The analysis of their trajectories (see Figure 3)
shows that they access contracts as preferred suppliers of processors, supermarkets
and exporting agents as a result of a good asset endowment base (large cultivated
DUHDV RQ SULYDWH ODQG HI¿FLHQW LUULJDWLRQ V\VWHPV DQG HTXLSPHQW 7KH\ KDYH
succeeded in equipping the farm and becoming, relatively speaking, highly
productive due to the public support they received during apartheid (subsidised
interest rates, tax concessions and price support combined with strong institutional
arrangements with cooperatives). After liberalisation, the deregulation of markets
DQGWKHGLVPDQWOLQJRISDUDVWDWDOVRQO\WKHODUJHVWDQGWKHPRUHHI¿FLHQWIDUPHUV
in this group were able to meet the required volumes and quality (norms and safety
standards) and succeeded in remaining preferred suppliers.
The second group, which is the target group of the project, consists of smaller-
scale farmers located on both private (redistributed) land and communal areas.
The analysis of their trajectories shows convergent trends. Most of them accessed
support measures as a result of social networks established before the end of
apartheid (for example, access to land and support as public workers or as decision
PDNHUV LQ WKH KRPHODQGV 2WKHUV EHQH¿WWHG IURP UHFHQW SXEOLF VXSSRUW LQ WKH
context of Broad-Based Black Economic Empowerment for Agriculture and other
DI¿UPDWLYHDFWLRQSURJUDPPHV
Downloaded by [Cirad-Dist Bib Lavalette] at 08:46 01 February 2013

43
Demythifying contract farming: Evidence from rural South Africa
Figure 3: Trajectories of households per types of farming systems
Source: Authors
The trajectories of the emerging vegetable and broiler producers are good examples
of how some small-scale farmers succeeded in concluding contracts in the study
area (Figure 3). Seven individual farmers received productive farmland in 2002
through the Land Reform’s Settlement Land and Acquisition Grant programme,
complemented with loans from the Land Bank. After acquisition their farms
collapsed due to, amongst other factors, inadequate experience or institutional
support. In 2005, the Limpopo Department of Agriculture intervened with a
multi-million rand poultry project as part of the agri-BEE drive, including the
construction of environmentally controlled houses with a carrying capacity each
of 40 000 broilers per 32 day cycle (Anjuère & Boche, 2009; Business Report,
2QHRIWKHFRQGLWLRQVIRUWKHIDUPHUVWREHQH¿WIURPWKHLQIUDVWUXFWXUHZDV
WKHFRQFOXVLRQRID¿YH\HDUUHVRXUFHSURYLGLQJFRQWUDFW which the Department of
Agriculture signed with the processor, Bushvalley. Broiler production effectively
started in 2007. The contract terms include the following: Bushvalley is to provide
farmers with all the required inputs (poultry equipment, one-day-old healthy
broiler-type chicks from a local hatchery, high quality balanced foodstuffs, poultry
litter, disease prevention/control medicines, etc.) and a strict monitoring plan for
feeding and managing the chickens. All costs (input, transport and processing) are
to be deducted from the amount producers are paid; the price is not pre-established
Downloaded by [Cirad-Dist Bib Lavalette] at 08:46 01 February 2013

44
S. Freguin-Gresh, M. d’Haese and W. Anseeuw
and depends on the weight of the chickens. In addition to the provision of technical
assistance, quality and hygiene inspections are to be conducted to ensure the
highest quality of the products delivered (Anjuère & Boche, 2009; Limpopo
3URYLQFLDO*RYHUQPHQW7KHSURMHFWOHDGHUVDOVREHQH¿WWHGIURPLQWHQVLYH
training regarding poultry production in environmentally controlled facilities,
provided through the DoA’s comprehensive agricultural support programmes.
 'HWHUPLQDQWVRIDJULFXOWXUDOLQFRPH
A Heckman model was used in order to test whether contracts could improve
OLYHOLKRRGV 7KH ¿UVW VWHS RI WKH PRGHO 7DEOH  LV D SURELW PRGHO RI WKH
determinants of the likelihood that the farmers commercialise their produce or not.
Indeed, it is obvious that if a farmer is not able to generate marketable output, he
will not be able to participate in any arrangement; testing if contract farming is a
GHWHUPLQDQWRIWKHIDUPLQFRPHLVLQWKDWFDVHLUUHOHYDQW7KH¿UVWVWHSRIWKHPRGHO
thus shows that an education above primary level is an important determinant of
commercialisation. Furthermore, households that receive larger social grants and
do not have access to irrigation infrastructure are less likely to commercialise any
agricultural produce, which also corresponds with the results of the qualitative
analysis.
Table 7: Probit model of commercialising agricultural produce (0: no sales; 1: sales)
6WHSRI+HFNPDQPRGHO &RHI±FLHQW 6WGHUURU ]VWDW
Household size 0.052 0.063 0.83
Age of household head (yrs) 0.062 0.144 0.43
Age squared (yrs²) –0.0005 –0.001 –0.39
Gender of household head (1:male/2:female) –0.638 0.556 –1.15
Secondary education or more (1:yes) 1.615 0.537 3.00***
Log social grants (R) –0.293 0.168 –1.74*
Log remittances (R) 0.044 0.050 0.87
Rain-fed production (1:yes) –0.926 0.459 –2.01**
Constant 1.524 4.138 0.37
The second step of the model is a regression of the determinants of farm income
7DEOH 7KH ODWWHUVHHPVWREHSRVLWLYHO\LQÀXHQFHGE\WKHQXPEHURIDFWLYH
members on the farm, the level of specialisation in agriculture, the size of the
cultivated land, and whether this land was acquired under private tenure or land
UHIRUP7KHVHGHWHUPLQDQWVDUHDOO³FODVVLF´VWUXFWXUDOYDULDEOHVLQÀXHQFLQJIDUP
income, as widely commented on in the literature. Further literature (Eaton &
Downloaded by [Cirad-Dist Bib Lavalette] at 08:46 01 February 2013

45
Demythifying contract farming: Evidence from rural South Africa
Shepherd, 2001; FAO, 2005; IFAD, 2003; Little & Watts, 1994; World Bank,
FRQ¿UPVWKDWFRQWUDFWVSRVLWLYHO\LQÀXHQFHIDUPLQFRPH
Table 8: Regression results for determinants of agricultural income
(dependent variable: log of income from agriculture)
6WHSRI+HFNPDQPRGHO &RHI±FLHQW 5REXVWVWGHUURU ]VWDW
Household members active on farm 0.607 0.222 2.73***
Age of household head (yrs) –0.012 0.013 –0.94
Gender of household head (1:male/2:female) 0.482 0.402 1.20
Probability of contract 2.016 0.881 2.29**
Log land size cultivated (log ha+1) 0.535 0.176 3.04***
Private land tenure (1:yes) 1.141 0.638 1.79*
Land reform bene±ciaries (1:yes) 1.796 0.615 2.92***
Rain-fed production (1:yes) –0.203 0.500 –0.41
Constant 7.623 1.000 7.62***
Inverse mills – Lambda –0.597 0.560 –1.07
n= 106
Censored obs = 31
Uncensored obs = 75
Wald ch²=130.65
Rho=-0.443
Sigma= 1.348
7KHDQDO\VLVRI WKH +HFNPDQ PRGHOFRQ¿UPVWKDWIDUPHUV ZKR JHQHUDWHKLJKHU
farm incomes are also better endowed with regard to capital and production factors
(land, private tenure, equipment, etc.). The qualitative analysis (Figure 3 and Table
3) suggests that these farmers engage in contracts for different reasons: (i) they are
PRUHFRQFHUQHGZLWKHQVXULQJVWDELOLW\LQWKHÀRZRISURGXFHWRVSHFL¿HGPDUNHWV
rather than with accessing production factors, credit and other inputs as they
DOUHDG\KDYHDFFHVVWRWKHVHWKLVLVVSHFL¿FDOO\WKHFDVHIRUFRPPHUFLDOIDUPHUV
who participate in either informal agreements or formal PDUNHWLQJ FRQWUDFWV;
or (ii) contract farming is ‘part of the deal’ for gaining access to development
programmes (agri-BEE, public-private partnerships, land reform programmes,
other public programmes for rural development, small-scale enterprises and
agriculture, etc.), as illustrated by the emerging farmers. In this case, contract
farming is a sine qua non condition for the farming system which is largely based
on public support.
Downloaded by [Cirad-Dist Bib Lavalette] at 08:46 01 February 2013

46
S. Freguin-Gresh, M. d’Haese and W. Anseeuw
5 CONCLUSION
The objective of this paper was to contribute to the ongoing debate about whether
and how contract farming can provide viable market opportunities for small-scale
farmers in South Africa.
The results show, however, that contract farming is not a panacea for small-
scale farmers. They highlight that contracts mostly involve the already well-
RIIZKR DUH HLWKHU ODUJHVFDOHPDQDJHULDOFRPPHUFLDO IDUPHUV KDYLQJ EHQH¿WHG
from public support during apartheid and which enabled them to become highly
productive, well-equipped and well-inserted in output markets, or medium-scale
IDUPHUVZKR KDYHEHQH¿WHGIURPFDVHVSHFL¿FSXEOLF SURJUDPPHVDQGRUVRFLDO
RU SROLWLFDO FRQQHFWLRQV %RWK WKH TXDOLWDWLYH DQG TXDQWLWDWLYH DQDO\VHV FRQ¿UP
the existence of entry barriers for small-scale farmers in concluding contracts
(production and commercialisation scales, education and asset endowments such
DV DFFHVV WR ODQG DQG LUULJDWLRQ $V VXFK DOWKRXJK VLJQL¿FDQW FKDQJHV KDYH
occurred in South African agriculture, small-scale farmers with limited access to
DVVHWVDQGZKRUHO\PDLQO\RQGLYHUVL¿HGLQFRPHVSDUWWLPHZRUNVRFLDOJUDQWV
etc.) to sustain their livelihoods, remain excluded and thus often marginalised.
Hence, the results support the concerns around the inclusivity of contract farming
posed elsewhere in the literature (Losch et al., 2010; Poulton et al., 2010; Vorley
et al., 2007, amongst others).
The results lead us to emphasise the need for disaggregated analyses of farm-
ing systems, of their different structures, their diverse roles and, subsequently, of
their different needs, policy measures and support systems. They also question
the effectiveness of contract farming as a “mythic” tool to integrate small-scale
farmers into restructured markets and modern value-chains and to reduce rural
poverty and inequality. The heavy support and subsidised measures necessary
to make contract farming work for the poor cast further doubt on the viability
of these instruments. Rethinking instruments is a must in order to support the
different types and roles of agriculture in the context of social dependency and
exclusion of productive income-generating activities affecting the rural poor.
This is particularly the case in South Africa, marked by its agrarian history. It
applies, however, also to other developing countries where the majority of farmers
are engaged in agriculture solely for subsistence, within a complex and strongly
GLYHUVL¿HGOLYHOLKRRGV\VWHP
NOTES
1 Poverty rate indices for South Africa vary according to sources and methodologies: the
World Bank (2011) estimates that 22 per cent of the population lives on less than R524 per
month whereas the National Planning Commission (2010) estimates that 49 per cent of the
population lives on less than R283 per month.
Downloaded by [Cirad-Dist Bib Lavalette] at 08:46 01 February 2013

47
Demythifying contract farming: Evidence from rural South Africa
2 Limpopo Province is located in the north of South Africa. It is a typical developing area,
characterised by the export of primary products and the import of manufactured goods and
services. It is one of the poorest regions of South Africa.
3 Mainly citrus and subtropical fruits are produced in the area for both domestic and export
markets. The area is also known for its tomatoes (constituting approximately 60 per cent of
tomato production in South Africa).
4 We assume that off-farm income is exogenous to farm income at household level because it
FRQVLVWVPDLQO\RIZHOIDUHJUDQWVDQGUHPLWWDQFHV,QWKHFDVHRIFURVV¿QDQFLQJOLYHOLKRRG
GLYHUVL¿FDWLRQ LQFUHDVLQJ WKH VKDUH RI RIIIDUP LQFRPH LQ WRWDO LQFRPH PD\ LQFUHDVH
agricultural income. It is equally possible, however, that farm income may decrease as the
VKDUHRI RIIIDUPLQFRPH LQFUHDVHVLQ WKRVHLQVWDQFHV ZKHUHWKH RIIIDUPLQFRPH VDWLV¿HV
the household’s utility levels.
REFERENCES
Anjuère, M. & Boche, M., 2009. Diagnostic agraire des coteaux de la rivière Nwanedzi,
province du Limpopo, Afrique du Sud, $JULFXOWXUHFRPSDUpHHW'pYHORSSHPHQWUXUDO. Paris:
AgroParisTech/University of Pretoria/CIRAD/NAMC.
Anseeuw, W., 2004. La reconversion professionnelle vers l’agriculture marchande et politiques
publiques. Le cas des mineurs du Northern Cape. Grenoble, France: Universite Pierre
Mendes.
Barrett, C.B., 2008. Smallholder market participation: Concepts and evidence from eastern and
southern Africa. )RRG3ROLF\ 33: 299–317.
Barrett, C.B., Bachke, M.E., Bellemare, M.F., Michelson, H.C., Narayanan, S. & Walker, T.F.,
2010. Smallholder participation in agricultural value chains: Comparative evidence from three
continents, in: Version, W.P. (Ed.). Available online at: http://ssrn.com/abstract=1733942
(accessed January 2011).
Bellemare, M.F., 2010a. Agricultural extension and imperfect supervision in contract farming:
Evidence from Madagascar. $JULFXOWXUDO(FRQRPLFV 41: 507–517.
Bellemare, M.F., 2010b. As you sow, so shall you reap: The welfare impacts of contract farming,
in: MPRA Paper No. 23638, P.J. (Ed.). Munich Personal RePEc Archive. Available online at
http://mpra.ub.uni-muenchen.de/23638/ (accessed January 2011).
Bolwig, S., Gibbon, P. & Jones, S., 2009. The economics of smallholder organic contract farming
in tropical Africa. :RUOG'HYHORSPHQW 37: 1094–1104.
Business Report, 2010. Chicken house needs funding to survive, %XVLQHVV5HSRUW
Cochet, H. & Devienne, S., 2004. Comprendre l’agriculture d’une région agricole: question de
méthode sur l’analyse en termes de systèmes de production, in: SFER (Ed.), /HVV\VWqPHVGH
SURGXFWLRQDJULFROHSHUIRUPDQFHVpYROXWLRQHWSHUVSHFWLYH. Lille: SFER.
D’Haese, M. & Van Huylenbroeck, G., 2005. The rise of supermarkets, friend or foe of rural
development. Evidence of consumption data in the Transkei area, South Africa. )RRG3ROLF\
30: 97–113.
Dufumier, M., 1996. /HVSURMHWVGHGpYHORSSHPHQWDJULFROH0DQXHOG¶H[SHUWLVH. Paris: Karthala.
Eastwood, R., Kirsten, J. & Lipton, M., 2006. Premature deagriculturalisation? Land inequality
and rural dependency in Limpopo province, South Africa. 7KH -RXUQDO RI 'HYHORSPHQW
Studies 42: 1325–1349.
Downloaded by [Cirad-Dist Bib Lavalette] at 08:46 01 February 2013

48
S. Freguin-Gresh, M. d’Haese and W. Anseeuw
Eaton, C. & Shepherd, A., 2001. Contract farming – Partnership for growth. )$2$JULFXOWXUDO
Services Bulletin 145.
FAO, 2005. 0DUFKpV GHV SURGXLWV GH EDVH WHQGDQFHV PRQGLDOHV UHWRPEpHVORFDOHV. Zurich:
FAO.
Glover, D.J., 1984. Contract farming and smallholder outgrower schemes in less-developed
countries. :RUOG'HYHORSPHQW 12: 1143–1157.
Greene, W.H., 2000. (FRQRPHWULF$QDO\VLV. 4th Ed. New Jersey: Prentice Hall.
Grosh, B., 1994. Contract farming: An application of the new institutional economics. Journal of
$IULFDQ(FRQRPLHV 3: 230–261.
IFAD, 2003. 3URPRWLQJPDUNHWDFFHVV IRU WKH UXUDO SRRU LQRUGHUWR DFKLHYH WKH0LOOHQQLXP
'HYHORSPHQW*RDOV. Rome: IFAD.
Karaan, A., 1999. Bridging the small-big divide: A transaction cost approach to enterprise
modeling for mussel mariculture in Saldanha Bay. $JUHNRQ 38: 680–692.
Key, N. & Runsten, D., 1999. Contract farming, smallholders, and rural development in Latin
$PHULFD7KH RUJDQL]DWLRQRI DJURSRFHVVLJQ ¿UPVDQG WKHVFDOHRI RXWJURZHUSURGXFWLRQ
:RUOG'HYHORSPHQW 27: 381–401.
Kirsten, J. & Sartorius, K., 2002a. Can small-scale farmers be linked to agribusiness? The timber
experience. $JUHNRQ 41: 295–325.
Kirsten, J. & Sartorius, K., 2002b. Linking agribusiness and small-scale farmers in developing
countries: Is there a new role for contract farming? 'HYHORSPHQW6RXWKHUQ$IULFD 17: 503–
529.
Limpopo Provincial Government, 2007. Limpopo MEX for Agriculture, MME Dikeledi Magadzi
to hand-over poultry projects.
Little, P. & Watts, M., 1994. /LYLQJXQGHUFRQWUDFW&RQWUDFWIDUPLQJDQGDJUDULDQWUDQVIRUPDWLRQ
LQ6XE6DKDUDQ$IULFD Madison: University of Wisconsin Press.
Losch, B., Fréguin-Gresh, S. & White, E., 2010. 6WUXFWXUDO GLPHQVLRQV RI JOREDOL]DWLRQ RQ
DJULFXOWXUHDQG UXUDO GHYHORSPHQW6\QWKHVLVUHSRUW. RuralStruc program, working paper.
Washington DC: World Bank.
Louw, A., Chikazunga, D., Jordaan, D. & Biénabé, E., 2006. 5HVWUXFWXULQJIRRGPDUNHWVLQ6RXWK
$IULFD'\QDPLFVLQFRQWH[WRIWKHWRPDWRVXEVHFWRU. Working Paper. Regoverning markets.
Department of Agricultural Economics, Extension and Rural Development, University of
Pretoria.
Maertens, M. & Swinnen, J.F.M., 2009. Trade, standards, and poverty: Evidence from Senegal.
:RUOG'HYHORSPHQW 37: 161–178.
McMichael, P. & Myhre, D., 1991. Global regulation vs. the nation-state: Agrofood restructuring
and the new politics of capital. &DSLWDODQG&ODVV 43: 83–105.
Ménard, C., 2005. A new institutional approach to organization, in: Ménard, C.A.S. (Ed.), 00
+DQGERRNRI1HZ,QVWLWXWLRQDO(FRQRPLFV. Dordrecht, the Netherlands: Springer: 281–318.
Mighell, R.L. & Jones, L.A., 1963. 9HUWLFDO&RRUGLQDWLRQLQ$JULFXOWXUH. Agricultural Economic
Report 19. U.S. Department of Agriculture.
Minot, N.W., 1986. &RQWUDFWIDUPLQJDQGLWVHIIHFWVRQVPDOOIDUPHUVLQOHVVGHYHORSHGFRXQWULHV.
MSU international development papers. Department of Agricultural Economics, Michigan
State University.
Downloaded by [Cirad-Dist Bib Lavalette] at 08:46 01 February 2013

49
Demythifying contract farming: Evidence from rural South Africa
Miyata, S., Minot, N. & Hu, D., 2009. Impact of contract farming on income: linking small
farmers, packers and supermarkets in China. :RUOG'HYHORSPHQW 37: 1781–1790.
0XQLFLSDO 'HPDUFDWLRQ %RDUG  0XQLFLSDO 3UR¿OHV $YDLODEOH RQOLQH DW KWWSZZZ
GHPDUFDWLRQRUJ]D0XQLF3UR¿OH2QOLQHLQGH[DVS[
DAFF (Department of Agriculture, Forestry and Fisheries), 2010. 7KH6WUDWHJLF 3ODQ IRU WKH
'HSDUWPHQW RI $JULFXOWXUH )RUHVWU\ DQG )LVKHULHV. Pretoria: Department of Agriculture,
Forestry and Fisheries.
National Planning Commission, 2010. 'HYHORSPHQW,QGLFDWRUV . Pretoria: The Presidency
of the Republic of South Africa.
Ortmann, G. & King, R.P., 2006. 6PDOOVFDOHIDUPHUVLQ6RXWK$IULFD&DQDJULFXOWXUDOFRRSHUD-
WLYHVIDFLOLWDWHDFFHVVWRLQSXWDQGSURGXFWPDUNHWV" Staff Paper series 06–4Department of
Applied Economics, College of Agricultural Food and Environmental Sciences, University
of Minnesota.
Ortmann, G.F. & King, R.P., 2007. Agricultural Cooperatives I: History, Theory and Problems.
$JUHNRQ 46: 40–68.
Ortmann G.F. & King R.P., 2010. Research on agri-food supply chain in Southern Africa
involving small-scale farmers: Current status and future possibilities. $JUHNRQ 49:397–417.
Pauw, K. & Mncube, L., 2007. 7KHLPSDFW RI JURZWK DQG UHGLVWULEXWLRQ RQ SRYHUW\ DQG LQ
equality in South Africa. International Poverty Center. Country Study 7. Development Policy
Research Unit, University of Cape Town.
Perret, S., Anseeuw, W. & Mathebula, N., 2005. 3RYHUW\DQGOLYHOLKRRGV LQUXUDO6RXWK$IULFD
,QYHVWLJDWLQJGLYHUVLW\DQGG\QDPLFVRIOLYHOLKRRGV Case studies in Limpopo. Unpublished
Project report 05/01. University of Pretoria, Kellogg’s Foundation.
Porter, G. & Phillips-Howard, K., 1997a. Comparing contracts: An evaluation of contract
farming schemes in Africa. :RUOG'HYHORSPHQW25: 227–238.
Porter, G. & Phillips-Howard, K., 1997b. Contract farming in South Africa: A case study from
KwaZulu-Natal. -RXUQDORIWKH*HRJUDSKLFDO$VVRFLDWLRQ 82: 1–38.
Poulton, C., Dorward, A. & Kydd, J., 2010. The future of small farms: New directions for
services, institutions and intermediation. :RUOG'HYHORSPHQW 38: 1413–1428.
Vermeulen, H., Kirsten, J.K. & Sartorius, K., 2008. Contracting arrangements in agribusiness
procurement practices in South Africa. $JUHNRQ 47: 198–221.
Vorley, B., Fearne, A. & Ray, D., 2007. 5HJRYHUQLQJPDUNHWV$SODFHIRUVPDOOVFDOHSURGXFHUV
LQPRGHUQDJULIRRGFKDLQV" London: Gower.
Wollni, M., Lee, D.R. & Thies, J.E., 2010. Conservation agriculture, organic marketing, and
collective action in the Honduran hillsides. $JULFXOWXUDO(FRQRPLFV 41: 373–384.
:ROOQL0=HOOHU0'RIDUPHUVEHQH¿W IURPSDUWLFLSDWLQJ LQVSHFLDOW\ PDUNHWVDQG
cooperatives? The case of coffee marketing in Costa Rica. $JULFXOWXUDO(FRQRPLFV 37: 243–
248.
World Bank, 2007. :RUOG'HYHORSPHQW5HSRUW$JULFXOWXUHIRU'HYHORSPHQW. Washington
DC: World Bank.
World Bank, 2011. World Bank Data by Country. Available online at: http://data.worldbank.org/
country/south-africa
Downloaded by [Cirad-Dist Bib Lavalette] at 08:46 01 February 2013

50
S. Freguin-Gresh, M. d’Haese and W. Anseeuw
$33(1',;'(6&5,37,212)7+(+(&.0$102'(/
7KHHPSLULFDO VSHFL¿FDWLRQRIPDUNHW SDUWLFLSDWLRQDQGLQFRPH IURPDJULFXOWXUH
problem takes the following form:
iiii MXY
HGE
 (1)
where Yi is agricultural income, Xi denotes the explanatory variables that
LQÀXHQFHDJULFXOWXUDOLQFRPHDQGMi is the dummy variable indicating the market
participation decision (binary choice). In case of self-selection, the least square
estimator dwill be biased, i.e. overestimating the effect of market participation.
The solution lies in incorporating a standard treatment effects model, which
extends the programme participation model (1) with (Greene, 2000; Maddala,
1983):
.0otherwise,0if1 *
*
!

iii
iii
MMM
u:M
J
(2)
In reality, Mi
* is unobservable, whereas Mi is the dichotomous parameter of
PDUNHWSDUWLFLSDWLRQLQÀXHQFHGE\WKHVHWRIH[SODQDWRU\YDULDEOHV:i) as directly
measured in the survey. Hence, the probability of market participation (2) can be
GH¿QHGDV

)(-1)0(Proband
)()(Prob0Prob)1(Prob *
ii
iiiii
:M
::uMM
J
JJ
)
) ! ! (3)
wherein ) is a cumulative probability distribution for u (Maddala, 1983). The
distributional form of ) can either be a normal or logistic distribution. Since the
issue of which probability distribution to use is unresolved (Greene, 2000), we
assumed the distributional form of ) as the cumulative normal (with f representing
the standard normal pdf) and as such the problem can be solved as a probit model.
Finally, the estimation of the linear regression model (1) will be:
>
@
>
@

  iiiiiii :XMEXMyE
JOUVGEHGE
H
and11
>@


»
¼
º
«
¬
ª
)


i
i
iii :
:
XMyE
J
JI
UVE
H
1
0(4)
The factor (LQVHUWIRUPXODKHUHLVGH¿QHGDVWKHLQYHUVH0LOOV¶UDWLRODPEGD$V
previously indicated, the estimation of this inverse Mills’ ratio provides an answer
Downloaded by [Cirad-Dist Bib Lavalette] at 08:46 01 February 2013

51
Demythifying contract farming: Evidence from rural South Africa
to the issue of self-selection bias or whether the estimator of market participation
in the linear regression is biased. This is shown by the expected difference in
agricultural income between market participants and non-participants derived
from (4) as:
>@>@

»
¼
º
«
¬
ª
))

 
ii
i
Eiiii MYEMYE 1
01
I
UVG
(5)
Downloaded by [Cirad-Dist Bib Lavalette] at 08:46 01 February 2013