Online Forum March-April 2013

I would like to thank Dr. Falck-Zepeda, (IFPRI) for the paper by IFPRI’s Gruere and Sun on factors thought to increase cotton yields in India. I encourage panel members to read this non referred paper, noting that color or dashed lines in the figure would have greatly aided in interpreting the results. The Gruere and Sun analysis is standard and based on the best available data, albeit very rough and incomplete in nature. While such aggregate analyses may provide insights, they provide no biological explanation of the issues and do not fully separate the effects of different factors on yield and profitability. More detailed analyses are required beyond such panel data.

Hence, Dr. Solleiro's is absolutely right in the sense that unless the field trials of GMOs are allowed, no tangible decisions can be taken by the regulators. Such data could be used to construct a weather driven system models that explicitly capture the biology of the crop and pests, and then can be included as the production function in the bio-economic analysis at local and regional levels. This would allow separation of the effects biotic (pests and plant growth) and abiotic factors on yield and hence profitability at the field or larger region. Unfortunately, IPR considerations preclude such research on GMOs unless permitted by corporate interests. Such restrictions occur in the USA as indicated by a 2009 letter from twenty-six leading university entomologists from the corn-belt to the Environmental Protection Agency protesting the restricted access to GMO seeds for experimental purposes. They wrote... "No truly independent research can be legally conducted on many critical questions involving these (GMO) crops".
Additionally, SEC also goes beyond yield and profitability, and are, to my understanding, not directly related to the objective and scope to the Protocol.
I note that field trials and system models have very limited value in assessing possible adverse effects of GMOs on conservation and sustainable use on biological diversity as the Protocol mandates, and even less for the value of biological diversity to indigenous and local communities (i.e., text of the Protocol). Adequate implementation of SEC will require more comprehensive and precautionary approaches.

Last, plotting the 1995-2004 data on suicides in Maharashtra State in India shows an increase of 3.82 per year per 100,000 of population above the relatively constant national rate. Approximately 52 per 100,000 suicides occurred in 2004 in Maharashtra State that in a population of 2.1 million translates to 1,092 suicides and possibly more. Poor farmer bankruptcy in the face of uncertain rains in Maharashtra State were compounded by the adverse effects of the pesticide treadmill before 2002, with the additional costs of Bt cotton after 2002 being a contributing factor (i.e., the aggregate analysis of Guillaume et al. 2008 aside).

Indeed the paper by Gruere and Sun is not an externally peer reviewed paper. IFPRI Discussion papers  undergo an internal peer review process before publication. However, we can find many citations in the peer review literature that are more or less in agreement with the findings by Gruere and Sun.

Jonas Kathage and Matin Qaim. Economic impacts and impact dynamics of Bt (Bacillus thuringiensis) cotton in India PNAS 2012 109 (29) 11652-11656; published ahead of print July 2, 2012, doi:10.1073/pnas.1203647109

Vijesh V. Krishnaa and Matin Qaim. Bt cotton and sustainability of pesticide reductions in India. Agricultural Systems, Volume 107, March 2012, Pages 47–55.

Shahzad Kouser and Matin Qaim. Impact of Bt cotton on pesticide poisoning in smallholder agriculture: A panel data analysis. Ecological Economics. Volume 70, Issue 11, 15 September 2011, Pages 2105–2113

Carl E. Pray, Latha Nagarajan, Jikun Huang, Ruifa Hu, Bharat Ramaswami (2011), Chapter 4 The Impact of Bt Cotton and the Potential Impact of Biotechnology on Other Crops in China and India, in Colin A. Carter, GianCarlo Moschini, Ian Sheldon (ed.) Genetically Modified Food and Global Welfare (Frontiers of Economics and Globalization, Volume 10), Emerald Group Publishing Limited, pp.83-114.

This leads us to the issue of understanding of the definition of "decision under import" and the scope of Article 26, and how do they apply for practical decision making. As the two articles below (Stone 2012 and Herring and Rao 2012) describe the situation in India with Bt cotton, where there are multiple of issues related to formal socioeconomic assessments and the institutional issues related to their implementation which raise lots of questions about how will an assessment contribute to a better decision making process.

In the specific case of debate on India’s Bt cotton, a vast majority of the peer reviewed literature points to an overall positive experience –but this does not necessarily mean that every single farmer in India won from the adoption of Bt cotton- against the grey literature who either paints a picture of “extreme failure” to one of “remarkable success”. The question then becomes whom to believe, whose evidence is more credible. Here lies one of the major issues for decision by the parties in terms of deciding on a decision making process, how will decisions makers decide in this situation, where there is no consensus of impact and where the likelihood of having a consensus is quite remote. We are after all, debating whether the Green Revolution benefited or not farmers in developing countries, 40+ years after it happened.

We have to be practical about this issue as Dr. Jose Luis Solleiro has pointed out in this forum.


Ronald J Herring, N Chandrasekhara Rao. On the ‘Failure of Bt Cotton’ Analysing a Decade of Experience. SPECIAL ARTICLE Economic & Political Weekly, May 5, 2012 vol xlviI no 1.

Glenn D. Stone. Constructing Facts: Bt Cotton Narratives in India. September 22, 2012 vol xlviI no 38. Economic & Political Weekly.

I would question the proposal that only "peer review" literature shoul be taken into account when we treat SEC.
Why not include for instance medical , producer of residents testimony affected by the transgenic production model?
In several countries where it has been introduced GE crops in a massive fashion, and where there are already serious social  problems due to GE crops; there is little capacity for social and economic research. So my question is: what happen with the victims affected model?

I propose using peer review as one of the criteria for judging quality of evidence, but it is not the only criteria. I have proposed also maintaining scientific rigor, replicability, and using elements of best practice to conduct risk and now socioeconomic assessments.

As experience in the medical, pharmaceutical and pesticide regulatory processes have shown, we need to uphold the highest standard possible to judge the impact of an intervention – a medicine or an LMO – as there are many factors that can confound the relationship between cause and effect.

We do not have the luxury of conducting double blind studies which are the gold standard in the approval of pharmaceuticals, we can only mimic some of these through quasi-experiments and through statistical methods, which are relatively early in the development.

Carefully collected anecdotes can be quite powerful in helping identify issues and hypotheses, but if you really want to establish causality -and in some cases even correlation- you have to test this using a rigorous approach(es) to evaluation. This of course opens the discussion on the many limitations that we have in terms of socioeconomic methods to assess these issues.

These limitations are more evident when we are talking about using a socioeconomic assessment as part of a process that will lead to a regulatory decision. This in practice means doing a lot of modeling and projections based mostly on assumptions and perhaps some baseline data, as there is no technology on the ground that we can measure. In the later case we have to be quite vigilant as assumptions can be misused and thus lead the analyst to the wrong conclusions.

Perhaps if you could document your statements with specific cases of socioeconomic impacts, it would make it easier to understand the examples given.

Dr. Jose Falck-Zepeda asked that I use reviewed citation and case studies. Ignoring socioeconomic considerations for convenience, the decision to import must be made on the basis of need, and in the case of GMO field crops on an ecological understanding of the ecological and socio-economic bases of the pest problem that is to be solved. By way of preamble, we should recognize that economic analyses of panel data provide insights into the contribution of factors to a problem addressed by the data – they tell us little about the underpinning root problem. I will illustrate the consequences of this inherent conflict by using industrial cotton in California where Bt cotton has been adopted in the southern desert areas and only marginally in the Great Central Valley, where during the 1960s and early 1970s the common entomological wisdom was that a small plant bug, Lygus hesperus Knight was causing heavy losses and variable yields, and heavy spraying was begun to control it.

Professor Robert van den Bosch was not a cotton expert, but upon examined the problem his intuition told him something was deeply wrong and with colleague, Louis Falcon,  asked something radical for its time - What would happen if no sprays were applied in Central Valley cotton? Most cotton experts thought it would be a disaster. To test this, they embedded the chemical treatments in a section of unsprayed cotton (640 acres, a mile square block of cotton) to see the benefits, if any, of pesticide use. What they found was astonishing - the very large area of untreated cotton had far fewer pests, and had higher yield of the same quality as any of the expensive chemical treatments (Falcon, van den Bosch et al. 1971). The experts called the results an aberration, so they repeated the experiment three more years with the same result. They had found that the pesticides were causing pest resurgence and outbreaks of highly damaging secondary pests that had been kept in check by natural enemies, and of course resistance to insecticides was developing in the background (see Gutierrez, A. P., L. A. Falcon, W. B. Loew, P. Leipzig and R. van den Bosch. 1974. An analysis of cotton production in California: A model for Acala cotton and the efficiency of defoliators on its yields. Env. Ent. 4(1): 125-136). When I told agriculture economist Professor Uri Regev that Central Valley cotton farmers had been spending money to lose money, he said "... that cannot be ...."  Farmers simply didn't know the root issues and were following industry and extension service recommendations. Regev would later call this the first documented case of market failure in pest control.

So as a thought experiment, we might ask what would have happened if Bt cotton had been available at the time, before these experiments were made and had been introduced to the Central Valley of California to solve the pest problem? Farmers would have reduced pesticide use because Bt cotton was protecting their crop, and natural control of cotton pests would have given the van den Bosch-Falcon results with Bt cotton providing the highest yields. (Bt cotton does not control most of the pests). Of course, adoption of Bt cotton would have occurred over time as farmers got off the pesticide treadmill and pest pressures would have been reduced regionally.  Of course analysis of economic panel data collected during this transition period would have shown increased yield, reduced pesticide use and decreases in farmer worker poisonings, etc. as benefits of Bt cotton. These are the results for Indian cotton in the article listed by Dr. Jose Falck-Zepeda in his reply to my prior posting. These analyses would have begged the question of whether Bt cotton was needed in the first place.  Hence, what is needed in Indian (and elsewhere) cotton are van den den Bosch-Falcon field experiments. (We note that organic cotton is grown in parts of India.) A limiting step, however, is that one needs industry permission to use the GMO seed for experimental purposes, and of course there is the question of funding for such work.  Systems models for cotton and cotton pests exist that could be parameterized using appropriate data from such experiments (Gutierrez, A. P and S. Ponsard, 2006. Physiologically based model of Bt cotton-pest interactions: I. Pink bollworm: resistance, refuges and risk. Ecological Modelling 191:346-359.Gutierrez, A. P., J.J. Adamcyzk Jr. and S. Ponsard.  2006. A Physiologically based model of Bt cotton-pest interactions: II. bollworm-defoliator-natural enemy interactions. Ecological Modelling 191: 360-382). These models could be used as the production function in economic analyses. Last, the connection of such models to bio-economic theory is found in Regev, U., A. P. Gutierrez, S. J. Schreiber, and D. Zilberman. 1998. Biological and Economic Foundations of Renewable Resource Exploitation. Ecological Economics, 26: 227-242).

Similar results accrued in understanding  the ecological basis of the pest problem in green revolution rice (e.g., the work of Peter Kenmore).




Similar results accrued in understanding of the ecological basis of the pest problem in green revolution rice (e.g., the work of Peter Kenmore).

A couple of observations based on my and other colleagues’ experience assessing socioeconomic considerations. Assuming that the inclusion of socioeconomics considerations in a country’s decision making process for import decisions is done via an assessment -of whatever intensity or complexity- this would necessarily imply an assessment or claim of potential gains or costs. In this situation, there is no technology on the ground and thus we cannot measure or estimate realized gains. The only situation where one can estimate realized gains is after the technology has been released and used by farmers.

This is an important distinction especially as to which of the two situations is relevant, will have an impact on the method(s) that may be available to conduct an assessment. When estimating potential gains (before release) the number of methods is limited, compared to those that may be used to estimate realized gains. This is just the nature of data availability.

If decisions on import, be it for FFPs or containment/confined field trials or for general release, are based on need then it is important indeed to understand how the technology works. Bt cotton is quite specific in its ability to control insects. The initial releases controlled only specific budworm/bollworm in cotton. With later releases, control ability was expanded to a slightly broader set of lepidopterans and now potentially to other insects. Need arises if the target pest is of economic importance and only in this situation it makes sense for a farmer to adopt. An aside, we will have to clearly separate and describe need for containment/confinement as a specific event may not be of economic importance in a country but it may be used to derive via research another one which may be, thus the need may arise for a separate decision making process.

So perhaps a better example of need in the USA would have been the Southeast USA as compared to California. Bt cotton was introduced in the Southeast USA in 1996, after two years of devastating crop losses to the point where the extension service officers in Alabama were saying that if Bt cotton were not to be introduced in 1996, it would be quite hard to justify planting cotton in the state. The need indeed was realized and there was 88-90% adoption in Alabama the first year of release. An estimation of the economic gains from the introduction of Bt cotton in the USA can be found in Falck-Zepeda, Traxler and Nelson 2000. I repeated this study with a much better dataset from the USDA-ARMS survey and it pretty much yielded the same results (See  Price, Lin, Falck-Zepeda and Fernandez0-Cornejo 2003).

By the way, the issue of secondary pests such as sucker insects (Lygus spp. and others) becoming of economic importance was proposed by entomologist in Alabama and other places as a potential impact of Bt cotton at the time of original adoption and thus were recommending strategies to monitor and manage these in the field.

References
Falck-Zepeda, J. B., G. Traxler and R.G. Nelson. “Surplus Distribution from the Introduction of a Biotechnology Innovation American Journal of Agricultural Economics. 82 (May 2000):360-369.

Gregory, K. Price, W. Lin, J. B. Falck-Zepeda, and J. Fernandez-Cornejo. “Size and Distribution of Market Benefits From Adopting Biotech Crops.” 2003. USDA-ERS Technical Bulletin No. (TB1906) 44 pp

The claim that …”The only situation where one can estimate realized gains is after the technology has been released and used by farmers…” is partially correct, as the development and use of physiologically based demographic models (PBDMs) of crop systems allows prospective assessments. Such models describe the weather driven biology of the interacting species rendering the model independent of time or place – they require weather data and initial conditions. The biology can be studied without implementing the GMO technology – one only needs to parameterize the PBDM. We use such models to assess the potential distribution and impact of invasive species under current and future climate change scenarios – to see where they can invade. Standard models used for this purpose are similar to multivariate economic models that make implicit economic, ecological, and mathematical assumptions about the system being studied, but lack mechanistic biological underpinnings. PBDMs are explicit formulation of the biology and have mechanistic underpinnings, and in bio-economics can be used as the production function. This approach allows relatively good assessments of the ecological bases of a pest problem and to assess if the technology can resolves it. 

To correct an impression, in central California the initial late 1960s concern was about plant bugs (Lygus), but with the onset of ecological disruption by pesticides, bollworms, budworms and defoliators soon became considerably much more important than the original target pest, and according to the local extension agents threatened to destroy the industry. Bt cotton in southern California also quickly established to protect against pink bollworm, but only after ecological disruption made budworm and bollworms the principle pests, which by the way have now been largely forgotten. I don’t know the situation in Alabama, but because extension agents express opinions about a pest situation and the need for the Bt technology doesn’t mean it was needed. Rarely would extension agents have the data or capacity to evaluate such problem in a dynamic context – it is more likely ecologically grey research. The analyses cited that demonstrated the benefits of Bt cotton in the SE USA appear to be post hoc analyses much like the ones done for Indian Bt cotton that estimate effects given the situation wherein the data were measured. van den Bosch–Falcon field experiments are needed to demonstrate the need for a technology (India, Alabama and elsewhere), and PBDMs models are very useful for explaining the underlying bases of the problem.

Last, rather than take up any more forum space and time, anyone seeking relevant reprints may request them at casas.global@berkeley.edu.