To Bt Or Not To Bt? Risk and Uncertainty Considerations in Technology Assessment is the title of a new working paper from IGIDR by Sarthak Gaurav and myself (Srijit Mishra). I give below the abstract and some extracts from the Introduction and Conclusion, but without the references.
Abstract
The acreage under the transgenic Bt (Bacillus thuringiensis) cotton seeds
in India has risen significantly since its legalization in the year 2002. Discussions
on the advantages from the technology have focused on increments in
productivity and income, without much analysis on risk. We point out that
claims on productivity gains seem to be misplaced, as appropriate
counterfactuals do not exist for the same hybrids. In this article we analyse production
costs and crop incomes in drought years to test a simplistic theory of risk
based on first principles. We employ a mixed-methods framework to draw
inferences by combining data from two cross-sectional surveys in Gujarat (Saurashtra
and Southern-Plains) and Maharashtra (Western Vidarbha) for the period 2009-10
and compare it with unit-level data for the corresponding regions from a
nationally representative sample for the period 2002-03. Empirical evidence, though
limited, brings out the problem of how a high cost technology could be
associated with higher risks and may be dominated by traditional alternatives
under certain conditions. Ethnographic accounts from the field provide
qualitative support to our understanding of potential risks and uncertainties
associated with the new technology.
Introduction
It is observed that in less than a decade since
its legal introduction in 2002, there has been a remarkable diffusion of Bt
cotton technology in India, with the
genetically modified seeds being used to cultivate nearly 90 per cent of the
total area under cotton in 2010. During
this period, the yields and overall production has increased on an average, and
India, which is the second largest producer of cotton in the world, became a
net exporter of cotton from being a net importer. Should we accept these trends
as evident of a successful adoption of the technology? Or, is there something
more than what meets the eye – like the dominance of the QWERTY technology for
keyboard layouts? In other words, can the
increasing acreage under Bt cotton be solely attributed to inherent benefits of
the transgenic seed technology or there are at work other mechanisms that demand
explanation? More importantly, are we missing out on critical dimensions of assessing
the technology amidst the ‘deceptive rhetoric, spin, and soundbite science
portraying the wonders or horrors of the new technology’
propounded by the competing discourses of the pro and anti Bt camps?
...
For a scientific evaluation of the hypothesized superiority of Bt cotton seeds in the causal sense, the use of these seeds should be the differentiating factor in attainment of higher yields and profitability in cotton production, ceteris paribus. Also, it should be borne in mind that in the absence of appropriate counterfactuals and adequate controls, any evidence that rejects the null hypothesis of Bt seeds not being superior to their non Bt counterparts, does not necessarily imply the superiority of Bt seeds, and vice versa. In the case of Bt cotton, the co-existence of scores of varieties with Bt genes and non Bt hybrids complicates the separability of the effects of the Bt strain. Unfortunately, the Bt cotton discourse has failed to acknowledge this fundamental logic while attempting to establish the positive or negative impacts.
In this article, we argue that an improved understanding of risk and uncertainty is central to the assessment of Bt cotton technology by analyzing outcomes in two drought periods, across two cross-sections. First, we use unit-level data from the Situation Assessment Survey of Farmers from the 59th round of the National Sample Survey (NSS), for the period 2002-03 across specific regions of Gujarat (Saurashtra and Southern Plains) and Maharashtra (Western Vidarbha). Then we employ data from our surveys and ethnographic inquiries in the similar agro-ecological zones in the two states for the period 2009-10. Furthermore, the Gujarat sample provides interesting comparisons across Bt and non Bt growers and ethnographic findings from the field in Wardha district of Vidarbha allow us to draw critical inferences on the Bt technology.
By taking up the analysis in two drought periods – one at a time when the legal (official) commercial transgenic varieties were introduced, and the other at a time when the technological diffusion can be assumed to have played out well, the findings of this article attempt to sensitize Bt cotton analysis to issues of risk and technological uncertainty. We believe, this article has two important contributions to improving our understanding of the Bt cotton debate. First, we argue for the need to consider the fundamental aspect of risk in the evaluation of the impact of Bt seeds. This assumes significance at a time of crisis in the Indian agriculture and spate of farmers’ suicides among predominantly cotton growers.
Second, our study is a contribution in mixed-methods research paradigm. In this context, we introduce a methodological innovation that combines quantitative and qualitative findings from the field with secondary data from a nationally representative sample that provides a rough baseline approximation in the context of Bt cotton evaluation for the two major cotton growing regions of the country, namely, Saurashtra and Southern Plains in Gujarat and Western Vidarbha in Maharashtra. It is quite likely that some farmers in the nationally representative sample would have already used illegal Bt seeds Nevertheless, the analysis can be safely assumed to be representative of non Bt usage scenario and hence a valid benchmark for conditions of cotton production in the relevant regions. For the current purpose we compare and contrast results from the NSS survey for the crop season of 2002-03 with that of some location specific and somewhat independent surveys in the above-mentioned two cotton growing regions during 2009-10.
...
Conclusion
In this article we argue that the perceived benefits and costs of introduction of the genetically modified seed technology, needs to be re-evaluated from the perspective that by being a substantial addition to cost in the farmers input vector, it also has risk considerations that need to be analyzed. Its dynamic interaction with a host of other factors like changing pest ecology, price risks, multi crop systems and market dynamics needs attention.
Using data from the Situation Assessment Survey of Farmers of the NSS 59th round, for a drought period 2002-03, we derive baseline scenarios for regions in Gujarat and Maharashtra. Data from the field and ethnographic evidence in similar regions provide changes in conditions of cotton production in the drought year 2009-10. First, we show that Bt technology is first order stochastically dominated by non Bt technology for our sample of cotton growers in Gujarat and the Bt technology adds substantial input costs to the production system.
Second, findings from the village level study in Maharashtra provide evidence of technology uncertainty, rampant deskilling and sub-optimal production decision making by farmers with limited information. Widespread institutional failures in the market for seeds emerge as a major concern in the farmers’ adjustment process and demands immediate attention.
Last but not the least, we provide the first evidence on the riskiness of Bt cotton technology by employing a simple risk framework (Mishra 2008) which takes the ratio of costs with that of net returns from cultivation of a specific crop. In this theoretical framework Bt technology can be both risk increasing or risk decreasing given the actual conditions under which the production occurs, but empirical evidence presented in this analysis indicate that Bt cotton has been risk increasing.
Even though our datasets are limited in terms of external validity and we employed a mixed-methods framework to address our core research question, we get substantive qualitative and quantitative confirmation on the Bt technology being associated with increasing risk in a drought year. This perspective is critical for the technology evaluation because the farmers stand to lose more in bad crop years than they would have in the case of traditional varieties. Moreover, if the net gains from Bt (at higher expected yields) are offset by the high intensity losses (frequent or otherwise), the yield advantage promised by Bt seeds should be taken with a pinch of salt because the absolute returns are so low that an increase in returns in such years would not lead to an increase in savings that can be used to compensate in a bad year. The riskiness of this technology becomes even more important from the fact that around 65 per cent of India’s cotton is produced in rainfed conditions and 35 per cent is irrigated.
We would like to emphasize that this article is not an addition to the existing literature on the productivity or socio-economic impact of Bt cotton technology, rather it is an attempt to introduce a new perspective in evaluating the technology by analyzing farm level data from a nationally representative sample and field level data from an ongoing panel data based study in one of the most challenging socio-economic and institutional contexts where the merit of the technology could be put to test. The contribution of our paper is that it introduces an alternative perspective on how the inherent riskiness of a so-called innovation like Bt cotton should be identified, irrespective of its potential benefits or observed merits across different parts of the country and among different groups of farmers. We demonstrate how a neglect of the simple economics of crop production and the associated risk factors can play out against the farmers who make a conscious decision to adopt a particular technology given multiple constraints, with the hope that it will pay off well.
There are certain methodological issues in Bt cotton evaluation that have not been dealt with in this paper but could be taken up as future research. Bt hybrids outperforming the earlier non Bt varieties should not be taken as a positive increment in yield on account of Bt alone, and the separability of effects of the hybrid impacts from Bt traits in a seed should be strived at. Variability of yield and other dimensions like price risks need to be assessed rigorously. Moreover, Bt technology could also induce moral hazard in farming practices as the farmer would expect the superior technology to be causing higher yields. An analysis of the dynamic performance of the technology over an appropriate period of time across different agroclimatic conditions could throw more light and also help us study the associated effects on the ecology and environment. Despite these limitations, our analysis does suggest that the proposition – to Bt or not to Bt; demands more non reductionist analyses that take cognizance of the localized variations and institutional realities.
In this article, we argue that an improved understanding of risk and uncertainty is central to the assessment of Bt cotton technology by analyzing outcomes in two drought periods, across two cross-sections. First, we use unit-level data from the Situation Assessment Survey of Farmers from the 59th round of the National Sample Survey (NSS), for the period 2002-03 across specific regions of Gujarat (Saurashtra and Southern Plains) and Maharashtra (Western Vidarbha). Then we employ data from our surveys and ethnographic inquiries in the similar agro-ecological zones in the two states for the period 2009-10. Furthermore, the Gujarat sample provides interesting comparisons across Bt and non Bt growers and ethnographic findings from the field in Wardha district of Vidarbha allow us to draw critical inferences on the Bt technology.
By taking up the analysis in two drought periods – one at a time when the legal (official) commercial transgenic varieties were introduced, and the other at a time when the technological diffusion can be assumed to have played out well, the findings of this article attempt to sensitize Bt cotton analysis to issues of risk and technological uncertainty. We believe, this article has two important contributions to improving our understanding of the Bt cotton debate. First, we argue for the need to consider the fundamental aspect of risk in the evaluation of the impact of Bt seeds. This assumes significance at a time of crisis in the Indian agriculture and spate of farmers’ suicides among predominantly cotton growers.
Second, our study is a contribution in mixed-methods research paradigm. In this context, we introduce a methodological innovation that combines quantitative and qualitative findings from the field with secondary data from a nationally representative sample that provides a rough baseline approximation in the context of Bt cotton evaluation for the two major cotton growing regions of the country, namely, Saurashtra and Southern Plains in Gujarat and Western Vidarbha in Maharashtra. It is quite likely that some farmers in the nationally representative sample would have already used illegal Bt seeds Nevertheless, the analysis can be safely assumed to be representative of non Bt usage scenario and hence a valid benchmark for conditions of cotton production in the relevant regions. For the current purpose we compare and contrast results from the NSS survey for the crop season of 2002-03 with that of some location specific and somewhat independent surveys in the above-mentioned two cotton growing regions during 2009-10.
...
Conclusion
In this article we argue that the perceived benefits and costs of introduction of the genetically modified seed technology, needs to be re-evaluated from the perspective that by being a substantial addition to cost in the farmers input vector, it also has risk considerations that need to be analyzed. Its dynamic interaction with a host of other factors like changing pest ecology, price risks, multi crop systems and market dynamics needs attention.
Using data from the Situation Assessment Survey of Farmers of the NSS 59th round, for a drought period 2002-03, we derive baseline scenarios for regions in Gujarat and Maharashtra. Data from the field and ethnographic evidence in similar regions provide changes in conditions of cotton production in the drought year 2009-10. First, we show that Bt technology is first order stochastically dominated by non Bt technology for our sample of cotton growers in Gujarat and the Bt technology adds substantial input costs to the production system.
Second, findings from the village level study in Maharashtra provide evidence of technology uncertainty, rampant deskilling and sub-optimal production decision making by farmers with limited information. Widespread institutional failures in the market for seeds emerge as a major concern in the farmers’ adjustment process and demands immediate attention.
Last but not the least, we provide the first evidence on the riskiness of Bt cotton technology by employing a simple risk framework (Mishra 2008) which takes the ratio of costs with that of net returns from cultivation of a specific crop. In this theoretical framework Bt technology can be both risk increasing or risk decreasing given the actual conditions under which the production occurs, but empirical evidence presented in this analysis indicate that Bt cotton has been risk increasing.
Even though our datasets are limited in terms of external validity and we employed a mixed-methods framework to address our core research question, we get substantive qualitative and quantitative confirmation on the Bt technology being associated with increasing risk in a drought year. This perspective is critical for the technology evaluation because the farmers stand to lose more in bad crop years than they would have in the case of traditional varieties. Moreover, if the net gains from Bt (at higher expected yields) are offset by the high intensity losses (frequent or otherwise), the yield advantage promised by Bt seeds should be taken with a pinch of salt because the absolute returns are so low that an increase in returns in such years would not lead to an increase in savings that can be used to compensate in a bad year. The riskiness of this technology becomes even more important from the fact that around 65 per cent of India’s cotton is produced in rainfed conditions and 35 per cent is irrigated.
We would like to emphasize that this article is not an addition to the existing literature on the productivity or socio-economic impact of Bt cotton technology, rather it is an attempt to introduce a new perspective in evaluating the technology by analyzing farm level data from a nationally representative sample and field level data from an ongoing panel data based study in one of the most challenging socio-economic and institutional contexts where the merit of the technology could be put to test. The contribution of our paper is that it introduces an alternative perspective on how the inherent riskiness of a so-called innovation like Bt cotton should be identified, irrespective of its potential benefits or observed merits across different parts of the country and among different groups of farmers. We demonstrate how a neglect of the simple economics of crop production and the associated risk factors can play out against the farmers who make a conscious decision to adopt a particular technology given multiple constraints, with the hope that it will pay off well.
There are certain methodological issues in Bt cotton evaluation that have not been dealt with in this paper but could be taken up as future research. Bt hybrids outperforming the earlier non Bt varieties should not be taken as a positive increment in yield on account of Bt alone, and the separability of effects of the hybrid impacts from Bt traits in a seed should be strived at. Variability of yield and other dimensions like price risks need to be assessed rigorously. Moreover, Bt technology could also induce moral hazard in farming practices as the farmer would expect the superior technology to be causing higher yields. An analysis of the dynamic performance of the technology over an appropriate period of time across different agroclimatic conditions could throw more light and also help us study the associated effects on the ecology and environment. Despite these limitations, our analysis does suggest that the proposition – to Bt or not to Bt; demands more non reductionist analyses that take cognizance of the localized variations and institutional realities.
Those interested, may read the full paper, To Bt Or Not To Bt? Risk and Uncertainty Considerations in Technology Assessment.
