Thursday, June 4, 2015

Debunking that Freakonomics "hit piece" on local food movements

This is particularly disappointing to see, as I have in general loved the material that Dubner and Leavitt put out in their books, but I suppose I should take some comfort in the fact that this piece is by Steve Sexton, a PhD student at UC Berkeley. Other sites, like Mother Jones, have already reacted to Sexton's intellectually thin gruel, but I don't think they did him justice.

Sexton essentially sets out to argue that "locavore" food strategies--favoring local producers over far-distant corporate farms--are both more costly in terms of food prices and detrimental to the environment:
My conservative estimates are that under the pseudo-locavore system, corn acreage increases 27 percent or 22 million acres, and soybean acres increase 18 percent or 14 million acres. Fertilizer use would increase at least 35 percent for corn, and 54 percent for soybeans, while fuel use would climb 23 percent and 34 percent, for corn and soybeans, respectively. Chemical demand would grow 23 percent and 20 percent for the two crops, respectively. 
In order to maintain current output levels for 40 major field crops and vegetables, a locavore-like production system would require an additional 60 million acres of cropland, 2.7 million tons more fertilizer, and 50 million pounds more chemicals. The land-use changes and increases in demand for carbon-intensive inputs would have profound impacts on the carbon footprint of our food, destroy habitat and worsen environmental pollution.
This sounds incredibly damning, until you realize that the "conservative" in Sexton's "conservative estimate" is functionally more of an ideological label than a statistical descriptor.

Since this piece has been getting such enormous traction at sites like "scibabe" [she calls it a "great read"], it's actually important to devote the time to fisking it, and to revealing the nature of Sexton's "conservative estimates."

First, consider this:
It is difficult to estimate the impact of a truly locavore farming system because crop production data don’t exist for crops that have not historically been grown in various regions. However, we can imagine what a “pseudo-locavore” farming system would look like—one in which each state that presently produces a crop commercially must grow a share proportional to its population relative to all producers of the crop. I have estimated the costs of such a system in terms of land and chemical demand. 
Ok, we can only hope that Sexton does better in his eventual dissertation than the first link suggests, because the definition of "truly locavore" comes from a Wikapedia page entitled "Local Food" that (A) carries this disclaimer at the top:
This article is written like a personal reflection or opinion essay that states the Wikipedia editor's particular feelings about a topic, rather than the opinions of experts.
... and (B) the only definition of "locavore" refers to individuals, not production systems, and is itself sourced back to a 2006 Time magazine article that never actually provides a working definition in the sense that Sexton uses it.

Forget about "pseudo-locavore"because, uh, Sexton just made it up.  It's a synonym for "straw man," as a matter of fact.

Here's the key paragraph where he palms the card:
It is difficult to estimate the impact of a truly locavore farming system because crop production data don’t exist for crops that have not historically been grown in various regions. However, we can imagine what a “pseudo-locavore” farming system would look like—one in which each state that presently produces a crop commercially must grow a share proportional to its population relative to all producers of the crop. I have estimated the costs of such a system in terms of land and chemical demand.
Notice here the definition, which is in bold.  Got that?  The implementation of Sexton's imagined "pseudo-locavore" system REQUIRES each state to produce commercial crops in proportion to its population.  In other words, Sexton is saying, "I'm going to require (by law, presumably) my pseudo-locavore system to deliver EXACTLY THE SAME DIET that is currently being produced, and then compare efficiencies."

He's not going to examine any system that any advocates of changing food production and/or distribution have ever put forward; no, he's going to create his own clown-car, Gestapo-mandated system, attribute it to them, and then tear it apart.  Sort of.  Turns out he doesn't even do that "tearing apart" thing particularly well.

Here's where you find his "estimates."  Let's note at the outset that his "analysis" is published by the Gianni Foundation of Agricultural Economics, and lacks any footnotes or source citation.  You can't actually reproduce Sexton's numbers because (A) he never tells you where he got the underlying stats, or (B) what statistical process he used to massage them in order to develop his argument.  He also attributes multiple arguments to "locavores" without ever bothering to quote anybody, name anybody, or provide any material for additional reading beyond other Gianni Foundation publications.

Again, let's hope he does better on his dissertation.

His assumptions, even if one spots him the lack of documenting his evidence, are wonderful:
These assumptions reallocate production so that each state produces an average “diet” for each if its residents. Because of data limitations, production is reallocated in this analysis for each crop only over those states for which a complete set of data exists. For instance, yield data for a given crop do not exist for states that are not currently producing that crop, so it is impossible to determine input demands.
This is truly amazing, as Sexton fails to tell us how many states this would entail leaving out, which states he uses, or how the develops the regions that he will mention in his following paragraph. Moreover, Sexton decides that the total impact of his "pseudo-locavore" system can be modeled through only four crops:  corn, soybeans, oats, and milk.  Leave aside that any such system that doesn't include wheat production or produce couldn't possibly pretend to bear the weight that Sexton wants his model to carry, he also leaves out such niceties under input demands as the impact of the existing Federal milk price support system or the percentage of soybeans utilized in the production of ethanol and therefore grown (and presumably modeled in his system) but not consumed as food or fodder.

Sexton palms many other cards in his presentation, including
These assumptions reallocate production so that each state produces an average “diet” for each if [sic] its residents. ... 
Using the regional mean production costs and state-level data on yield ... 
If a national price for inputs is assumed, these input cost changes can be interpreted as changes in input demand ...  
Table 2 reports the states that gain the most farmland under local production and those that lose the most, in absolute terms. Extrapolating this change across the 2.26 billion acres of farmland in the United States, the agricultural land base would grow by 214.8 million acres— an area twice the size of California. ...
None of these assumptions is either sourced or quantified, nor is there any validation for his methodology.  For all any skeptical reader could know, Sexton reaches up into his head and picks out numbers that sound good.

And he has to stumble past a number of problems in his conception without trying to draw attention to them.  Consider milk:
Notably, however, results for milk suggest that production costs decrease under the “pseudo-locavore” scenario, and purchased feed is substituted for grazing and feed produced in the dairy farm. The changes in feed consumption suggest carbon savings relative to the status quo, but the increased number of cows would induce more carbon emissions. Because of the way data for milk are reported, the change in head of cattle accounts for efficiency differences across states, where as input costs do not.
Actually, Sexton hasn't proven at any point that more cows would be required, nor does he do more than thinly swipe at the impact of Federal price controls on input costs.

Or there's this:
Large monocropped farms are more dependent than small polycrop farms on synthetic fertilizers and tilling operations to restore soil nutrients. They also face heightened pest pressure because they provide a consistent environment for breeding of crop-specific pests. Higher pest pressure increases demand for chemical damage control agents. Disposal of farm residues, like animal waste, also becomes a significant environmental challenge on industrial farms. The direct environmental costs of large-scale agriculture are clearly non-trivial. What is unclear, however, is whether the environmental benefits of small, poly-cropped farms outweigh the loss of efficiencies that are equally well-documented to accompany the increasing scale of production. 
In this case the card may not be where you think it is.  Sexton admits the comparative environmental damage of large monocropped farms, but then then says that it is "unclear" whether the environmental costs are offset by the environmental benefits of small poly-cropped farms.  It's that UNCLEAR where Sexton palms the card.  In an entire essay in which (without benefit of source or methodology citations) he has been arguing exactly that IT IS CLEAR (and even more forcefully in his Freakonomics one-off) that large-scale farming is economically and environmentally beneficial in a comparative sense, he hides a hedge that threatens to invalidate his entire argument.

I think the card just dropped on the floor.

Here's one of my other favorites:
For instance, agricultural economists have rejected the notion that farm policy is to blame for the obesity epidemic in America. While policy has made grains relatively cheap, it has also made sugar more expensive. 
Yep, it's made sugar more expensive in order to create a demand market for high-fructose corn syrup. Which, of course, has NOTHING to do with the obesity epidemic in America.

Here's my favorite card. Remember that Sexton's straw man "pseudo-locavore" system is designed to examine the costs of delivering exactly the same diet that the average person eats now, which makes this not only disingenuous but intellectually dishonest:
Would a local food system improve American diets? In two key respects, the likely answer is no. First, as this analysis has shown, a local food system would greatly increase the costs of food production by imposing constraints on the efficient allocation of resources. The monetary costs of increased input demands from forsaken gains from trade and scale economies will directly bear on consumer welfare by increasing the costs of food. Research shows that as incomes rise, fresh produce as a share of diets increases. Therefore, given that locavorism would effectively make consumers poorer by increasing the cost of food, it is hard to see how local production improves diets or health outcomes. 
What Sexton knows (and we know he knows it because of the few sources he cites for his Freakonomics article) is that local food advocates (A) make arguments about the quality and freshness of food; and (B) have, as a part of their strategy, CHANGING both the kinds and qualities that people consume.  In other words, he has created a system that locavores don't support, analyzed it to find it wanting, and then applied to it to other than the objectives the local food movement is trying to achieve.

All the while pretending he's actually doing serious research.

Finally, he ends with the assertion that local food advocates--not lack of population control, not climate change, not governmental policies, not trade regulations not market manipulations--but local food advocates (think of the woman in the denim skirt ahead of you in line at the farmer's market) will cause mass starvation and/or the clear-cutting of the rain forests in order to avoid mass starvation by 2050.

And, yes, he really does say that:
If mass starvation is to be avoided in the current century, then we must either forsake natural land, including tropical forests, or renew our commitment to crop science ["crop science" is here used as a surrogate for current monocrop corporate agriculture]. 
Ironically, I'm an agnostic on the whole locavore thing, from a whole variety of perspectives.  But when I read about a piece of garbage "research" being floated as somehow authoritative when it is essentially one large assertion backed up by ... more assertions ... then I can't resist the urge to point it out.

No comments:

Post a Comment