More on GMO wheat

A couple weeks ago, the St. Louis Post-Dispatch ran a story on GMO wheat that's been picked up by several other news outlets.

This is a controversial topic, even among some farm groups, and it is one I've touched on several times in the past (e.g., see here or here).

In the most recent article, I'm identified as a "a supporter of GMO wheat."  I can understand why the reporter would write that, but I think it is better to say that I'm a supporter of reasonable regulation and producer choice.  

It wouldn't bother me if a seed company or University put out a GMO wheat variety that didn't pass the market test (that no one wanted to buy).  But, what is troubling is the position that wheat producers cannot have access to a perfectly safe technology while canola, soy, and corn producers can.  Yes, there are some complicated trade issues involved, and there are fears about market power, but I see little reason these issues can't be sorted out in the marketplace, as it has with these other commodities.

By "reasonable regulation", what I mean is that we've created this strange climate around GMOs that both make the regulatory costs of introducing a new variety quite high and raise the hackles of some of our trading partners.  But, I'm not sure it's a very reasonable climate.  For example, I'll note that some of my excellent colleagues at Oklahoma State have released a new wheat variety that is herbicide resistant but that is not, technically, a GMO.  However, as would be the case with a GMO, producers are not allowed to save and replant the seed because the variety is protected by a patent.  In short, the wheat breeders have delivered almost everything one would expect in a GMO, except it isn't technically a GMO.  

What that tells me is that it's often silly to focus on the tool (i.e., whether a certain genetic technique was used) rather than the outcome.  But, as the example also shows, when artificial barriers to innovation and trade are introduced, entrepreneurs will find a way around it if the demand is there.  

P.S. On the topic of wheat genetics, note the recent article forthcoming in the American Journal of Agricultural Economics by Jessie Tack, Andy Barkley, and Lanier Nalley.  They show that yield potentials have been increasing steadily over time, but the gap between potential yield and actual farm yield has also been increasing over time.  They attribute the gap to on-farm management/production decisions.

Taleb on GMOs

Russ Roberts, the host of one of my favorite podcats, Econ Talk, recently had Nassim Taleb on to talk about Taleb's recent paper urging the use of the precautionary principle with regard to GMOs. I've previously commented on Taleb's paper but thought I'd add some thoughts about this recent interview.

First, Taleb's argument.  He argues that GMOs are risky and that the risks are not well understood. The risk are of the sort, he argues, that could (eventually) lead to catastrophic outcomes - i.e., they are not localized risks.  Moreover, he argues, the risks are asymmetric: the possible bad outcomes are much worse than the possible good.  He argues that with each additional new trait, there is only a small perceived increase in risk,  but that eventually - because of the "fat tails" of the risk distribution - it is inevitable that a really bad outcome will occur.  As a result, he argues, we shouldn't use GMOs  It's the cautious, wise thing to do in his assessment.

Taleb is obviously a smart guy and he makes some valid points.  In particular, I think one of the key take-aways for those of us who see value in using biotechnology - is that we want to make sure that risks are of the sort that they don't result in epidemics; that is, how can we diversify or put a stop-gap on our risk exposure?

Despite these insights, I think there are a number of challenges with Taleb's take on this topic.  First, if you listen to the interview Taleb takes great lengths to make an appeal to authority.  He establishes a hierarchy with geneticits/biologists at the bottom, statisticians above them, and then risk experts at the top.  In his assessment, the biologist is not in a position to judge risks or causal claims, it is the statistician who knows how to do this (forget the fact not all statisticians agree about the burden of proof - the whole frequentist vs. Bayesian approach is the most obvious example), and then the risk theorist is at the top (Taleb himself being the risk expert).  As such, he essentially tries to inoculate himself from any criticism, saying that a biologist can't criticize the risk theorist (Russ Roberts calls him out on this at one point, but Taleb quickly dismissed the point).  

To make his point, Taleb invokes the carpenter fallacy.  The carpenter makes the roulette wheel, but if we wanted to know our chances of hitting on on the number 22 after 150 plays, would we ask the carpenter?  No, according to Taleb, we'd ask the probability expert.  But, there is a huge assumption being made here: that the probability expert knows how the wheel is made - how many possible numbers there are, whether the wheel is balanced, whether the slots in the wheel are of equal size, etc., etc.  In short, a good probability theorist needs to know everything the carpenter knows and more.

Unfortunately, I don't see much evidence that Taleb has spent much time trying to understand GMOs or modern agriculture, and as such it is hard to take his probability judgments in this domain seriously.  For example, one of his examples is that the Irish potato famine, which he argues was caused by a lack genetic diversity.  However, there are good reasons to believe that British politics were the key contributing factor to the famine.  Moreover, what he doesn't seem to get with regard to modern GMOs is that a GMO isn't a variety.  A particular trait - say herbicide resistance - is introduced into many, many varieties in different parts of the country and the world. Moreover, not all herbicide-resistant crops are resistant to the same herbicide.        

Going further, herbicide residence can be "naturally" bred into plants.  There are rice and wheat varieties on the market that are not GMOs but that are herbicide resistant.   Why are these not risky but the GMOs are?  My original comment about Taleb's paper is that he didn't focus on marginal risks; he treats GMOs as a separate class without looking at how plant breeding is done in real-life agriculture: is he also against hybrids? Mutagenesis? Cisgenics? Marker Assisted Breeding?  These all have risks that are on par with GMOs, so I'm curious why only the focus on one particular technique?

As I've sad many times, a GMO isn't a single thing - it is many, many things, each with different benefits and risks.  A hammer can be used to bash heads or build skyscrapers.  It is just a tool.  Same with biotechnology.  What I would prefer to see from Taleb and others is a careful discussion of a particular trait they find worrying along with a careful articulation of why that particular trait is likely to lead to a particular global harm - and what it is about the tool of biotechnology that is particularly worrisome for that trait vs. other breeding techniques?  If everyone was planting the same variety of the same herbicide resistant crop (and we only had one herbicide that we knew worked), I'd be concerned, but that not what's happening.

As it is, we simply have Taleb making an appeal authority and to unknowability.  We can't know what bad things might happen, so Taleb says we should be cautious.  But no evidence of risk is not evidence of an eventual black swan.  Indeed, it seems to me it is a recipe for stagnation.  And it isn't at all clear to me the downside risks are always greater than the upside.  When penicillin was first discovered, the risks were unknown but the benefits were (ultimately) immense.  A precautionary approach might have shelved antibiotics, but we took the chance and with great effect to our life expectancy.  Life is full of risk.  The answer isn't to hole ourselves up in the closet, but rather to think about ways of taking insurance against possible risks while venturing out into the world.

I'll conclude with one last thought.  Taleb makes reference to the Hayek bottom-up vs. top-down planning.  He says GMOs are the top-down sort.  I'm not so sure.  Real life farmers and people have to be willing to buy varieties that have the GMO traits.  No one is forcing that outcome.  It is true that competition will limit - to some extent - the diversity of plants and genetics that are observed because some plants aren't tasty or aren't high enough yielding.  But most plant breeders keep all kinds of "ancient" varieties precisely for the purpose of trying to breed in new traits to today's varieties (and folks working on synthetic biology are creating their own, new strands of DNA, creating new diversity).  Geography also increases diversity.  Iowa grows a lot of corn, Oklahoma doesn't because it isn't our comparative advantage.  I see little reason to believe that a single GMO variety will perform well in all locations.  So, yes individual companies are planning and creating new varieties, but it is all our local knowledge of what works in our places and conditions that determine whether particular genetics offered by a particular company are used.  We do not have a seed czar or a DNA czar.  

DNA labels

Last week's release of the Food Demand Survey (FooDS) has probably generated as much attention as anything I've put out on this blog.  From my perspective, its hard to know why - I've had longer, more insightful posts, and I've shown similar results in the past (e.g., that people desire avoiding ethylene as much as GMOs).  Part of that just goes to show what a few serendipitous retweets will do for web traffic.  

The issue garnering the attention is the question I added to this month's FooDS asking people (N=1,015; nationwide; demographically weighted to match US population) whether they supported or opposed 10 different food policies.  I was mainly interested in how people responded to the question on mandatory calorie labels on restaurant menus, because that issue has been in the news much of late given the release of the FDA's final rule (and because of the prior research I've worked on with Brenna Ellison on the topic).  But, I often think asking about a single issue in isolation, without much context, isn't particularly insightful.  What's often more interesting are the relative comparisons against other policy issues (or changes over time, which is the main focus of many of the FooDS questions).  Thus, the questions on GMOs, and the question that sparked the most interest: mandatory labels on foods containing DNA.

A few comments.  I do not interpret answers to these sorts of questions as necessarily reflecting some sort of deeply held beliefs, but rather they often represent quick, gut reactions.  As the ballot initiatives in California, Colorado, Oregon, and Washington have shown, mandatory GMO labels initially poll at very high levels (at levels similar to what we found in the recent survey), but in all four of those states, labeling failed to garner 50% of the vote.  Clearly, many people's views about mandatory GMO labeling are not fixed constructs, but are (at least at this point) somewhat malleable and are open to education and persuasion.  As such, I do not believe polls of this sort provide "the" answer on whether a policy should pass/fail, but rather provide initial insights for where the conversation will begin.  

There also seemed to be some insinuation that I "tricked" respondents or that the question was leading.  I have a hard time seeing it (were 61% of respondents also led to say that they opposed sugar soda taxes?).  The order of the items was randomized across respondents.  As some commentators have pointed out, the question on DNA labels probably could have been better worded.  It's worded as saying "mandatory labels on foods containing DNA."  So, let's say that you know a lot of foodstuff contains DNA and you want labels on, say, nutritional content, then it could be that you'd say "support" not because you find DNA worrisome but because you want nutritional labels.  I doubt that's how most people interpreted the question, but it's a possibility.  There is ample evidence that the public doesn't understand much about genetics.  For example, back in 1999 in a paper in Science, Gaskell et al. asked true/false questions of the sort, "Ordinary tomatoes do not contain genes while genetically modified tomatoes do."  This question has been repeated in many subsequent surveys, and it is often found that many people (incorrectly) say "true".  

I'd be careful about saying this means that people are not smart enough to make their own food decisions (I've written extensively on the topic of paternalism).  I think it mainly implies people don't much need to think about such issues (i.e., they don't have an incentive to think carefully about the issue).  Nonetheless, as I've noted about Kaheman's work, it should make us wary of the availability entrepreneur.  

Finally, I'll noted that I've seen several thoughtful comments on my latest survey.  See, for example, Ilya Somin at the Washington Post or Robbie Gonzalez at i09.        

What is a GMO?

Andrew Pollack had an interesting article in the New York Times yesterday that was ostensibly about companies using various techniques to get around regulations surrounding GMOs.  But, that's not what I think the key point of his article is.  Rather, it emphases exactly how hard it is to define what a "GMO" is and it underscores the lack of precision by opponents of the technology who use the term (and no I'm not referring to the folks shown in the Jimmy Kimmel segment that didn't even know what the acronym stands for).  The article also appropriately raises the issue of the costs and entry barriers that exist with the current regulatory regime (a topic I've previously touched on).  

Here's what I wrote about the definitional difficulties a couple months ago in the Milken Review:   

“Genetic engineering involves the transfer of a gene (or multiple genes) from one species to
another through synthetic means. Just because the process occurs in a lab, it doesn’t follow that the resulting seeds couldn’t have been produced by “natural” means. . . . Resistance to certain herbicides, for example, can also be attained, albeit at a slower rate, via traditional plant breeding. Indeed, many strains of rice grown today are conventionally bred to be resistant to herbicides. Traditional plant breeding requires the breeder to find wild or unusual cultivars that display the trait of interest and repeatedly crossbreed them with a commercial variety until getting an offspring that is similar to the original commercial variety yet exhibits the desired trait.

Genetic engineering, by contrast, attempts to speed up the process by moving only those genes of interest into the commercial variety. Sometimes these genes come from wild
variants of the same species (using so-called cisgenic technology) or from entirely different species (using transgenic technology) [Pollack’s NYT article seems to mainly be about gene editing - turning on and off genes already present in a plant]. As the comparison of cisgenic and transgenic technologies suggest, the dividing line between what is and what is not genetically engineered is fuzzy and somewhat arbitrary: Transgenic is often considered genetic engineering, whereas cisgenic is not, despite the fact that both approaches use the same methods and differ only in the origin of the genes transferred.

Some of the unusual cultivars used in aforementioned conventional crossbreeds are created by mutagenesis – that is, exposing seeds to radiation or to chemicals in hopes of random, beneficial, mutations. This approach has been used for more than half a century and is not considered genetic engineering, nor is it regulated as such. In fact, certified organic
seeds can arise from varieties produced via mutagenesis.”

Then, later in the same article . . .

“Ultimately, it must be recognized that genetically engineered foods are not a single “thing.” To broadly claim that they cause harm lacks precision (not to mention evidence). One needs to tie a specific genetic alteration to a specific type of harm. It is possible to imagine genetic modifications that could trigger allergies (the purely hypothetical example of inserting a peanut gene into corn comes to mind). But most of the commercially used applications on the market today are not of this sort, and new GE crops that were couldn’t pass regulatory muster. ”

Some plant researchers from UC Davis were quoted in Pollack's article as saying,

“the regulatory framework had become “obsolete and an obstacle to the development of new agricultural products.””

GMOs and the Precautionary Principle

Much has been written about Nassim Taleb's coauthored paper arguing that the precautionary principle dictates that we should avoid GMOs.  Given the prominence of the author and his willingness to berate detractors, the paper has received more attention than the ideas in the paper merit.  

This piece by Stuart Hayashi raises an excellent point.  The issue shouldn't be about the presence of risk per se but about risk on the margin.  How much riskier (or less risky) are GMOs compared to other techniques?  As Hayashi point's out, Taleb's argument is akin to running an experiment without the control.  There is an implicit assumption that using GMOs (the experiment) are unambiguously riskier than not (the control).

Hayashi's post had an summary description of Taleb's main argument, which also shows how the same sort of logic can be used to argue that GMOs should be adopted.  I've taken Hayashi's description of Taleb's argument and replaced a few of Hayashi's words with my own in brackets:

"The argument is as follows. If we talk about the risk [likelihood] of a GMO doing damage [creating benefits] on any one particular day, it seems that that risk [potential benefit] is minuscule. But what is the statistical risk[likelihood] of a GMO inflicting harm [being created that creates enormous benefits] one day . . . eventually? As time advances, that risk[likelihood] of a GMO eventually causing turmoil [great good] increases exponentially . . . Therefore, the argument concludes that as long as transgenic technology is employed, it is inevitable that one day, something devastating [wonderful] concerning GMOs will occur. Therefore, the one method whereby we can guard [help secure] ourselves against this otherwise-impending harm [benefit] is to avoid [promote] usage of genetic engineering"

Even if Taleb's argument is right, it must also be right for any number of other possible risks we face: from say, new diseases that come about from interacting with domesticated or wild animals, from risks of using alternative energy sources (curiously, Taleb says nuclear energy is excluded because "the nature of these risks has been extensively studied"), from risks of potentially electing a warmongering despot, from risks of developing robots and artificial intelligence, from risks of comets passing by the earth, from risks of returning from space travel, risks from conventional plant breeding, etc, etc.  It is not an argument that is anyway unique to risks from GMOs.  Maybe Taleb is right and we are all just sitting around waiting for some worldwide disaster.  Even if that's true, I seriously doubt it will be the GMOs that get us.