Performance Results and Characteristics of Adopters of Genetically Engineered Soybeans in Delaware

Agricultural and Resource Economics Review, Oct 2004 by Bernard, John C, Pesek, John D Jr, Fan, Chunbo

With regard to the human capital variables, two of the education dummy variables-HiSchool and College-were significant at the 1 % level. These results revealed that farmers with high school or college education had substantially higher yields than those without a high school degree. The remaining significant variable was the dummy for Kent County, where yields were higher than those in the base New Castle County. While this result was expected, the lack of significance for Sussex County, hypothesized to be the most productive, was not.

For the remainder of the variables, the lack of significance for farm size (Acres), partly also a proxy for farm management, and narrow row spacing (Narrow) were also initial surprises. The Acres coefficient did, however, conform to the notion that the GE technology should be scale-neutral. Narrow row spacing, which intuitively would seem to imply higher yields per acre, was likely not significant simply due to the fact that the vast majority of farmers had adopted this technique. Finally, the relatively low R2 value (0.2124) suggested a number of factors explaining the variation in yields were not captured in the model.

Turning to the results of the weed control costs model presented in table 4, the percentage of soy acreage to GE soybeans was significant, with the expected negative sign, at the 5% level. Here, the coefficient suggests an extra percentage to the GE variety reduces control costs by about 50. This translates into a savings of approximately $5 per acre for a farm planting only GE soybeans. In contrast to the yields model, farm size (Acres) was significant, with the expected negative sign. Regardless of the use of GE soybeans, economies of scale would be expected with the larger farms, as is confirmed by the model's findings. High school education was significant at the 10% level, with an unexpected positive sign. Neither the county dummies nor narrow row spacing were found to influence weed control costs. As with the yield model result, where Narrow also had been strongly hypothesized to be significant, the lack of significance of this variable in the weed control costs model was the most surprising. Again, however, this finding may simply be explained by farmers' prior adoption and use of narrow row spacing.

Conclusion

While findings from earlier studies based mostly on controlled field experiments showed questionable benefits from adopting GE soybeans, survey results from adopting Delaware farmers revealed higher yield and lower weed control costs. Some of the differences in these findings are likely the result of the fact that adopters tend to be larger scale, better managed farms-aspects not captured by field experiments. Indeed, farm size was shown to be one of only two variables significant in the diffusion process. The other significant variable, use of a computer for farm management, further reinforces the hypothesis that better managed farms adopt more readily and benefit more from the change. Thus the adoption of the new GE crop technologies, despite their apparent scale-neutrality, may advance the agricultural treadmill process toward fewer and larger farms.