Southern African Regional Poverty Network (SARPN) SARPN thematic photo
Country analysis > Mozambique Last update: 2020-11-27  

Interlinked transactions in cash cropping economies:
The determinants of farmer participation and performance in the Zambezi River Valley of Mozambique

Rui Benfica1, David Tschirley, and Duncan Boughton

Department of Agricultural Economics, Michigan State University

September 2006

SARPN acknowledges the African Economic Research Consortium (AERC) as the source of this document:
[Download complete version - 379Kb ~ 2 min (16 pages)]     [ Share with a friend  ]


This paper investigates the determinants of participation and performance of tobacco contract farmers, and the effects of participation on overall crop and household incomes in the Zambezi Valley of Mozambique. We test the existence of threshold effects in land holdings and educational attainment to identify the types of farmers that benefit. Several results stand out. First, participation in the schemes is driven by factor endowments, asset ownership and alternative income opportunities, and very little by demographic factors. Second, we find no returns to education in tobacco; this result is consistent with previous research in Mozambique but surprising in an agronomically demanding crop like tobacco. Third, there appear to be economies of scale in tobacco production, perhaps through more efficient use of hired labor. If true, tobacco could drive greater economic differentiation through the growth of “emergent” or commercial smallholder households – something that has been conspicuously lacking in Mozambique to date. Fourth, farmers without wage income are more likely to grow tobacco; since other research shows that wage labor has driven most income growth in Mozambique over the past six years, tobacco could be inequality reducing. Tobacco growers also hire much more labor than non-growers, contributing to second-round inequality reducing effects. Further analysis, preferably in a general equilibrium framework, is needed to understand how the simultaneous forces of economic differentiation and spreading of economic benefits will affect income distribution. Potential adverse environmental impacts also deserve far more attention than they have received to date.


Contract farming is a pervasive institutional arrangement for cash cropping throughout the developing world. The persistence of this approach derives from two factors. First, smallholder farmers’ low shadow wage rates give them substantial cost of production advantages over larger farmers, especially on crops requiring high labor input. Second, small farmers often are cash constrained and have poor access to input and credit markets. As a result processors, needing raw material to amortize fixed investments, provide these farmers with inputs (and possibly other services) on credit, and attempt to recover that credit upon purchase of the output.

These arrangements have been analyzed at length for at least two decades (Minot, 1986; Glover and Kusterer, 1990; Little and Watts, 1994). However, empirical assessments based on detailed household level data and controlling for possible selection bias in participation, are relatively rare2. In this paper, we focus on tobacco contract farming schemes in the Zambezi Valley of Mozambique. We develop two versions of sample selection models (Heckman, 1979; Greene, 2003) to investigate the determinants of farmer participation, the determinants of net income from tobacco once in the scheme, and the effect of participation on overall crop income and total household income. A key contribution of this paper is its investigation of threshold effects of education and land holdings; rather than focusing on the average effect of participation, we ask what type of farmer benefits from participation.

Section two of the paper describes the study area and sampling procedures. Section three compares participants and non-participants, while Section four presents our conceptual and empirical models, and results. We close with a discussion of policy implications.

View presentation - 84Kb < 1min

  1. I acknowledge important contributions to this output provided by David L. Tschirley and Duncan H. Boughton, Michigan State University. They are recognized as co-authors for publication purposes. Important feedback was also provided by Channing Arndt, Purdue University.
  2. See Warning and Key (2002) for a recent example with a much smaller sample size and fewer independent variables than we use here. See also Jayne et al (2004) for application in a panel data set.

Octoplus Information Solutions Top of page | Home | Contact SARPN | Disclaimer