In much of rural Africa, high transaction costs limit farmers’ market participation and thus their potential for income growth. Transaction costs can affect not only whether a farmer sells product but also whether sales occur at the farm gate on at a market. If production behavior is related to a chosen sales location, then analysis of interventions can be improved by explicit consideration of the decision of where to sell. This paper develops a double-selection model that explains consumption and production decisions by semi-subsistence farmers who first decide whether to be a seller and then whether to sell at the farm gate or at an off-farm location before deciding on production and consumption. The study tests the validity of this dual-criteria model against a single criterion model in which a grower first decides to be a seller and then decides production, consumption and sales location simultaneously. Dual-criteria and single-criterion models are compared while correcting inconsistency in estimations due to violation of homoskedasticity and normality assumptions in selection equations. The results suggest that the dual-criteria model provides more information than the single-criterion model using a sample of cassava producer in Benin.
A heteroskedastic double selection model
International Food Policy Research Institute (IFPRI)