Determinants of adoption of genetically modified maize by smallholders in KwaZulu-Natal, South Africa

Abstract

Previous research on small-scale farmers in KwaZulu-Natal, South Africa indicates that certain genetically modified maize seed types improve production efficiencies and increase net returns (Regier 2012). Yet despite the substantiated advantages, not all farmers have adopted genetically modified maize. The purpose of this research is to identify the determinants of adopting certain types of genetically modified maize over traditional or conventional hybrid maize for 184 small-holders in two villages in KwaZulu-Natal, South Africa. Previous adoption studies use socioeconomic characteristics of the farmer as well as farm-level production characteristics to determine the probability that a farmer will implement an improved agricultural technology. While many studies employ a binomial approach to adoption, this study tests the probability of adopting three different GM varieties—the insect resistant Bt maize, the herbicide tolerant Roundup Ready® maize, and the stacked trait BR maize. Furthermore, the model is enhanced by farmers’ open-ended explanations of their perceptions on genetically modified maize and of the major production constraints they face. Following results from previous adoption studies, this research tests three hypotheses in a three different model structures. The first hypothesis tests whether farmers are more likely to adopt if they have greater financial means to cover higher expected production costs. This is tested by variables measuring off-farm employment and expected production costs. The second hypothesis tests whether farmers with less labor availability are more likely to choose maize with the herbicide tolerant technology, either the Roundup Ready® or stacked BR maize, which reduce the need for weeding. The final hypothesis is whether there are differences in the determinants of adoption that differentiate GM adopters into three distinct categories. These hypotheses are tested in three model structures that test the binary probability of adopting GM maize over non-GM, the probabilities of adopting each maize variety separately, and the intensity of adoption. The first finding is that many non-adopters have greater access to income and are more likely to sell a portion of their yield than are many farmers who adopted, especially in comparison to those who plant RR maize. Also, BR farmers are more likely to report input expenses as a major constraint in their adoption decision. Results for the second hypothesis show that those who planted either RR or BR maize did in fact have less family labor available, used less total labor, and used a greater proportion of family to hired labor. Finally, there are differences in the determinants for geographic site, education, self-sufficiency in maize supply, number of family members working off-farm, and whether households planned to sell any of their maize yields. This indicates that adoption should be considered according to each genetically modified trait.