The International Consortium on Agricultural Biotechnology Research (ICABR)
Smallholder Adoption and
Economic Impacts of Bt Cotton
in Makhathini flats,
Despite the adverse reaction to genetically modified (GM) crops in Europe, there is a broad consensus that biotechnology may be particularly important for developing countries. Indeed, the recent International Fund for Agricultural Development report makes a strong case that effective use of biotechnology will be essential to the alleviation of rural poverty in the developing countries for the foreseeable future.
Herbicide and insecticide tolerant traits account for more than 85% of the types of GM crops grown world-wide. Insect resistance has also been a popular target for the GM companies. Here, the focus has primarily been on the transfer of a set of genes controlling production of a natural insecticide in a bacterium called Bacillus thuringiensis (or Bt) to crops. The Bt-toxin acts specifically on Lepidoptera (including bollworm in cotton, stem borers in maize), and is harmless to all other insect species.
Currently, the majority of commercial GM crop releases have been in the USA, Canada and some countries in South America. Indeed, the USA, Canada and Argentina account for 99% of the GM crops area in the world. Outside these areas, GM crop release on a commercial scale has been limited. In Africa, for example, commercial scale release of Bt cotton and maize is only taking place in South Africa.
The Genetic Modified Organism Act, passed in 1997 and implemented in 1999, paved the way for the introduction and commercialisation of GM crops in South Africa. Approximately 3,000 hectares of Bt-maize were planted in 1998 and up to 50,000 hectares of GM maize has been planted in 1999. Bt-cotton is grown mostly in the Northern Province with some in KwaZulu-Natal and the Free State. Cotton accounts for 1% of total South African agricultural production, generating approximately US$ 50 million per annum. 100,000 hectares are grown by 1,530 commercial farmers and 3,000 small-scale farmers, mostly grown under dryland conditions.
Since 1998, smallholder farmers in one of the lower potential cotton areas of South Africa have been adopting a genetically modified cottonseed variety (NuCOTN 37-B with Bollgard). The area is Makhathini Flats in KwaZulu-Natal province, where rural households cultivate land allocated to them by their tribal chiefs. Landholders face uncertain tenure arrangements without guaranteed ownership succession within the family. Good quality cropping land is scarce, unfenced, and under threat from livestock that devastate crops due to the communal grazing systems. Labour is also a problem in the rural areas, and because of male migration to the towns, the labour source of farmers is comprised of children, elderly people and female labourers.
The survey was conducted in November 2000 in Makhathini Flats, with a sample of one hundred smallholder farmers. It concentrates on the adoption of Delta Pinelands GM cotton variety (Bollgard) and examines the impact on yields, gross margins and technical efficiency. The stratified sample is of forty non-Bt cotton growers and sixty Bt cotton growers. By the 1999/2000 season, 12% of the 4,000 farmers in the region have adopted the new GM cotton seed. Data was obtained for the 1998/1999 and the 1999/2000-seasons.
The results show that the average farm size was six hectares, but 62% of the farms were less than five hectares. 73% of farmers owned livestock; the majority had a predominance of female labour and 25% had non-farm income sources.
Agronomic problems: 57% of farmers considered pests to be their biggest agronomic problem and 62% of these categorised the bollworm as the major pest. Too much rain was a distant second at 24%, followed by weeds, which was ranked highest by only 11%.
Non-agronomic problems: 82% of farmers sited lack of credit as their major non-agronomic constraint, as compared with 14% who rated land scarcity as their biggest problem and only 4% whose greatest difficulty was lack of labour.
The adopters of Bt cotton were more experienced than the non-adopters and had larger farms. However, this is not a matter of adopter characteristics; rather, it is these farmers who are more likely to be granted credit. The key factors affecting adoption were thus the availability of credit, or other means of purchasing inputs, such as non-farm income. There really seem to be no technical reasons to prevent all the smallholders from adopting, if only sufficient credit can be made available.
Bt cotton gave higher yields per hectare than the other varieties. Adopters averaged 475 kg/ha in the first season, as compared with 457 kg/ha for non-adopters. In the second season, the adopters produced 425 kg/ha while the non-adopters slumped to only 304 kg/ha, which is a 40% yield advantage. Bt cotton gave higher yields per kg of seed in both seasons. Adopters averaged 50 kg of cotton per kg of seed in the first season, as compared with 37 kg for non-adopters. In the second season, the adopters produced 44 kg while the non-adopters only got 23 kg for each kg of seed. Bt cotton increased seed cost in both seasons.
Adopters spending averaged R197 per hectare in the first season, as compared with R119 for non-adopters. In the second season, the adopters spent R214 as compared with R127 for the non-adopters. This was offset by the fact that Bt cotton reduced chemical cost in both seasons. Adopters averaged R93 per hectare in the first season, as compared with R132 for non-adopters. In the second season, the adopters spent R83 as compared with R129 for the non-adopters. Higher yields increased output value, at the expense of increased seed cost, which were offset by lower chemical cost. The average gross margins (output value minus the cost of intermediate inputs) for Bt cotton was R781 in the first season as compared with R791 for the non-adopters. The second season revealed that the adopters gross margins averaged R677 as compared with R428 for the non-adopters, an advantage of 35%.
Both yields and gross margins are useful, but they are partial measures of efficiency, which fail to take account of major inputs such as labour. Thus, they are supplemented by efficiency frontiers, which consider the efficiency with which all inputs are converted into outputs, using only the more reliable input and output quantity data and avoiding prices, which are less well recorded or simply non-existent.
Both deterministic and stochastic frontiers were used, the first applying programming techniques and the second relying on econometric estimation. In either case, the results confirm the farm accounting results, showing that the Bt cotton adopters were considerably more efficient than those who used the non-Bt varieties.
The programming approach showed that the adopters, average efficiency level was 62% in the first season, as compared with 45% for non-adopters, giving an advantage of 38%. In the second season the adopters were 66% efficient, while the non-adopters were 46% efficient, so the adopters advantage rose to 43%. The stochastic frontier approach showed that the adopters average efficiency level was 88% in the first season, as compared with 66% for non-adopters, giving an advantage of 33%. In the second season the adopters were 74% efficient, while the non-adopters were 48% efficient, so the adopters advantage rose to 54%.
Finally, there is no evidence that the better off farmers gained more than the less well off: indeed, income inequality was slightly reduced.