Agriculture Reference
In-Depth Information
While mean comparisons are illustrative, definitive conclusions about
relative yields are possible only if the data are generated under experimental
settings where factors other than adoption are controlled for by making them
as similar as possible (Fernandez-Cornejo and McBride, 2002; NRC, 2010).
17
This is not the case with survey data.
18
Bt use is not random. Surveyed farmers
are not randomly assigned to a treatment group (adopters) and a control group
(nonadopters). Consequently, adopters and nonadopters may be systematically
different from one another (for example, in terms of management ability). If
these differences affect both farm performance and Bt adoption, they will
confound the analysis (Fernandez-Cornejo and McBride, 2002; Fernandez-
Cornejo et al., 2002). This self-selection
19
biases the statistical results unless it
is corrected (Greene, 1997). Fernandez-Cornejo and Wechsler (2012) specified
an econometric model to estimate the impact of adoption that accounts for
self-selection. Using this model, they found that a 10-percent increase in the
probability of adopting Bt corn was associated with a 1.7-percent increase in
yields in 2005, and in a new ERS analysis using 2010 survey data, they found
a 2.3-percent increase in yields (table 6). Using a similar econometric method
to analyze cotton data, ERS researchers found that a 10-percent increase in the
probability of adopting Bt cotton was associated with a 2.1-percent increase in
yields in 1997 (Fernandez-Cornejo and McBride, 2002).
The effect of HT seeds on yields is mixed.
The evidence on the impact of
HT seeds on soybean, corn, and cotton yields is mixed (table 4). Several
researchers found no significant difference between the yields of adopters and
nonadopters of HT; some found that HT adopters had higher yields, while
others found that adopters had lower yields. For instance, an ERS study found
that a 10-percent increase in the adoption of HT cotton led to a 1.7-percent
increase in cotton yields. HT soybean adoption was associated with a
statistically significant, but small, increase in yields: a 10-percent increase in
the probability of adopting HT soybeans was associated with a 0.3-percent
increase in yields (Fernandez-Cornejo and McBride, 2002).
ARMS results show that HT soybean yields were 5 bushels per acre (3
percent) higher than conventional soybean yields in 2006 (but only
significantly different at the 10-percent level) (table 7). In the case of corn,
ARMS results show that HT corn yields were similar to those of conventional
corn in 2010. However, unlike soybeans, the majority of corn (and cotton)
producers in recent years use seed with stacked traits (figures 9 and 10).
Multiple stacked traits make evaluating the effect of individual GE traits on
yields and profitability more complicated.
