Abstract
This study presents a quasi-experimental analysis of the impact of FairTrade certification on the commercial performance of coffee farmers in Tanzania. In doing so the study emphasises the importance of a well-contextualised theory of change as a basis for evaluation design. It also stresses the value of qualitative methods to control for selection bias. Based on a longitudinal (pseudo-panel) dataset comprising both certified and conventional farmers, it shows that FairTrade certification introduced a disincentive to farmers’ commercialisation. We explain this counterintuitive conclusion on the basis of the ‘cooperative life cycle’ theory developed by US agribusiness scholars.
Acknowledgement
This study benefited from the logistical support and in-depth local knowledge of Eliudi Doto and Kaminjoge Mhamerd.
Notes
1. The information provided in this first paragraph was retrieved from FairTrade’s official website (www.fairtrade.net).
2. The name of the association involved in this study is omitted in respect of confidentiality agreements.
3. The selected district is a major production site for Arabica coffee in Tanzania (no Robusta coffee is grown in this district). The topography of the district includes mountains, valleys and plateaus, ranging from 800 to 1900 m above sea level. The micro-climate of the district is described as temperate to cool tropical climate with a unimodal rainfall pattern (December–April). The average annual rainfall is 1224 mm, ranging from 1000 to 1600 mm. Coffee harvesting is done during the dry season, from August to November. The average minimum and maximum temperatures for the district are 19 and 23°C, respectively.
4. It is important to stress that the authors of this study did not have any influence on the sampling process. The 10 cooperatives considered in this study were selected by the association’s officers, who stated they had no clearance from the central administration to disclose additional information and data regarding the other member-cooperatives operating in the district. As such, the representativeness of the sample used in this study at the district and association levels could not be established.