A new study published in PLoS ONE by Wilhelm Klümper and Matin Qaim surveyed the literature on the farm-level effects of GMO adoption. They conducted a Meta analysis - a type of quantitative literature review - covering 147 previous studies.
What did they find?
This is a a decent study which summarizes what most people who follow the literature already know. There will, no doubt, be attempts by the anti-GMO crowd to discredit the study. However, Qaim is a productive, well known agricultural economist. He’s published in the best outlets in agricultural and development economics, and even in journals like Science and Nature Biotechnology.
Like any Meta analysis, the study isn't perfect, and is only as good as the studies being reviewed. A few criticisms. The analysis didn't much differentiate between insecticides (the use of which has almost certainly fallen) and herbicides (total use is probably up, but because GM producers have switched to less toxic herbicides the total toxicity is likely down). Also, some of the underlying studies may not have done a good job separating yield gains from traditional hybrid breeding from gains conveyed by biotechnology per se, so the yield gains attributable to GM may be a bit overstated (e.g., see this study). What I’m saying here is that corn/soybean yields probably would have increased regardless of whether GM was adopted, so you have to “back out” the increase attributable to GM; some studies do that well, others don’t. Finally, as you can see in their figures, there is a lot of heterogeneity across studies; the mean effects for yields and profit are positive, but some studies show negative.
In any event, this is a good study that re-confirms my own reading of the literature.