The International Consortium on Agricultural Biotechnology Research (ICABR)
Estimating Farm-Level Effects of Adopting Biotechnology
The rapid adoption of biotech crops in the United States in recent years largely reflects the benefits of potential increases in crop yields and pest control cost savings from this technology. Estimates of these farm-level effects induced by biotech adoption not only determine benefits to producers, but also affect the distribution of benefits among the stakeholders.
However, estimates of the farm-level effects differ significantly, depending on data sources. For example, a recent article by Falck-Zepeda, Traxler, and Nelson (hereafter FTN) in Agribusiness (2000) shows that adopters yields for 1997 herbicide-tolerant soybeans were 13.0% higher than nonadopters in the Corn Belt based on the USDA Agricultural Resource Management Studies (ARMS) survey. In contrast, Moschini, et al in another article of Agribusiness (2000) assumed no yield difference, which is more in line with what most analysts believe. Clearly, mean differentials between adopters and nonadopters from the ARMS survey reflect the combined effect of biotechnology and other factors. The effects on pest control costs differ even more dramatically across data sources than the impacts on crop yields.
Variations among various estimates call for a careful evaluation of these estimates and efforts to find ways to reconcile these differences. In the case of Bt cotton, a private data source EMD (Enhanced Market Data) was used in studies of 1996 and 1997 Bt cotton by FTN (AJAE May 2000; Agribusiness, 2000). This data source, based on a survey of consultants, provided pairwise comparisons of cotton yield and pest control cost differences between Bt and non-Bt cotton fields. ARMS survey data, to be comparable, require further analysis to isolate the effect of biotechnology from other factors.
Accordingly, the main purpose of this paper is to compare the farm-level effects, at the regional level, of adopting biotechnology obtained from various data sources. The analysis focuses on 1997 herbicide-tolerant soybeans and Bt cotton in the U.S. These data sources include: (1) the ARMS survey; (2) the private EMD data; and (3) the elasticity-based estimates obtained by excluding the effects of factors other than biotechnology through econometric analysis of the ARMS survey data (Fernandez-Cornejo, et al, 1999). Yield and pesticide use elasticities from this adoption-impact model are further analyzed in this study to show yield and per-acre pest control cost differentials between adopters and nonadopters.
Considerable differences exist in terms of the farm-level effect on crop yields from various data sources and across production regions. For example, while adopters 1997 soybean yields are shown to have been 14.2% higher than nonadopters in the ERS Heartland region based on the ARMS survey, the elasticity-based estimate indicates only a 3% higher yield for adopters. Similarly, Bt cotton adopters' yields are estimated to have been about 12% higher in the Southern Seaboard based on the ARMS survey and EMD data, but the increase reached 21% based on the elasticity-based estimate.
In general, there were savings in pest control costs for herbicide-tolerant soybean adopters. However, the savings were considerably higher based on the ARMS survey than those based on the elasticity-adjusted estimate. For example, adopters' savings were as high as 31% in the Heartland based on the ARMS survey, but were only 0.3% based on the elasticity-based estimate. This, in part, is attributed to an increase in the use of glyphosates and total herbicides for adopters in this region. Similarly, while savings in pest control costs for Bt cotton adopters averaged about 5-7% in the Southern Seaboard based on the ARMS survey and the elasticity-based estimate, the saving reached as high as 60% based on EMD data.