Corn-Based Supermodels
Mildly Disagreeing With OurWorldInData
I. Introduction
Here’s an idea for you: If you want your country to escape poverty, get really, really good at growing corn.
Or wheat. Or potatoes. Or basically any other standard agricultural grain. After you’ve done that, set the economy to simmer and set a timer for a few years. Once the timer goes off, take the lid off the economy and, hey presto, you’ve got a fully grown manufacturing sector and now you're off to the races with modern economic growth.
When I pitch this to people, it tends to raise some eyebrows. Partially because of the incredibly mixed metaphors about cooking, racing, and ending poverty. Partially because it’s not especially obvious why getting really good at making food is going to make you really good at making all of the stuff that isn’t food.
To the first charge, I offer no defense other than that this is my blog and I will mangle the English language as much as I damn well please.
As to the second, the mechanism basically goes like this:
II. Farmville
Imagine a lone town in the middle of nowhere (let’s call it Farmville) that doesn’t interact with any other towns. In fact, the people of Farmville haven’t even made contact with the rest of civilization. Life in Farmville sucks pretty bad. Every single resident has to perform backbreaking labor in the fields just to make enough food to avoid starvation.
The residents of Farmville would prefer if life wasn’t like this. Ideally, only some of them would work on the farms making enough corn and wheat for everyone to eat and then everyone else would do something else. A few years ago, Farmer Jack drew up plans for a widget factory (everyone loves widgets); maybe the rest of them could work making widgets? Unfortunately, they can’t switch to making widgets because then there wouldn’t be enough food for everyone and half of Farmville would starve to death. Bummer.
So, Farmville continues along in this rut until, one day, Farmer Jill gathers everyone together and announces that she’s invented something called the tractor. This new invention will double the output per person for everyone who works in agriculture. The Farmville citizenry is overjoyed; this means they will be able to send people to work in Jack’s widget factory while still feeding everyone!
From here, life in Farmville gets a lot better. The people working in the widget factory invent a whole bunch of mechanical things (trucks, machines, fertilizer, etc) that help them both with making more widgets and also make people better at farming. This frees up even more people to leave the farm and go make widgets and farming tools and so on and so on. Eventually, maybe Farmville decides that they would actually like to be able to eat more food than just enough to subsist, so they send a few people back to the farm from the widget factory to make a surplus of food. But, after enough time, the people over at Jack’s widget factory make so many mechanical things that make farming so much easier that only one person in Farmville needs to tend the farms and they still produce more food than Farmville could possibly eat if they tried. The residents of Farmville are now much better off than before, all of these new inventions mean that, on top of having widgets (something they didn’t have before), they now also have even more food per person than they started out with! It turns out improving agricultural output with a tractor was really useful.
Now, let’s say that you don’t think Farmville will totally solve the problem of food production. Maybe you think there is a hard cap to how much technology can improve the corn-to-person ratio. Even still, the exact final composition of what percent of Farmville is going to be working on the farm isn’t that important; it’s going to depend a lot on exactly how many widgets or bushels of wheat a person can produce in an day and what weight Farmville puts on having wheat vs. widgets. And no matter what, Farmville is still better off than they were before.
The point of our story is that until Farmville saw improvements in agricultural production they were stuck producing only food, because to be able to make anything else you need to be able to eat first. It’s only after you solve what’s been creatively termed “The Food Problem'' that you are able to spend your time doing other sorts of productive behavior1. Unfortunately, it's all that other, non-food related stuff like manufacturing and science that leads to the exponential growth we’ve come to expect from modern economies.
So, to get your country out of poverty, you need to get really good at growing corn.
Technically, we are discussing a slightly modified version of a specific “Structural Change Growth Model” proposed by T.W. Schultz in 1953 2, but I prefer Corn-based Supermodel.
Is this mostly so I could use the graphic at the start? Yes.
But, it’s also because structural change growth model is a really boring term that doesn’t really capture what is going on in the way that ℂ𝕆ℝℕ-𝔹𝔸𝕊𝔼𝔻 𝕊𝕌ℙ𝔼ℝ𝕄𝕆𝔻𝔼𝕃 does.
The “Supermodels” part of my term I’ve stolen from Hatcher and Bailey’s Modeling the Middle Ages (though I can’t guarantee it originates there).3 They use supermodels as a catch-all term for the grand models of economic and social development that get floated to try and explain vast swathes of human history. I think this sort of broad thesis about development probably gets to make it into that illustrious group, even if it isn’t trying to explain as long of a time-period as, say, a Malthusian model of all pre-modern growth.
III. Some Empirical Evidence
For the most part, I think the corn-based supermodel is a pretty good heuristic for how a country can kickstart economic growth. It tells a convincing story about why economic growth was so slow for thousands of years and then suddenly took off and, more importantly, it matches the data we have pretty well. Here are the charts of GDP and cereal yields in the UK from 1270 until now4:
It’s not just that total output of grain is correlated with GDP growth, here’s another chart from Herrendorf, Rogerson, and Velentinyi (2011) showing how the share of employment in agriculture and agriculture as a percentage of GDP has declined across a series of developed countries even while total agricultural production has gone up (implying that it is specifically agricultural productivity that has changed)5:
Building off of this data about how currently rich countries got rich, it stands to reason that countries nowadays should experience the same results if they cranked up their food production.
That’s basically the thesis of this piece put out by OurWorldInData6. The author argues that one of the keys to solving poverty in subsaharan Africa is finding ways to significantly improve the efficiency of labor in agriculture. I think this is right. Well… I think this is almost right.
Here’s what the author, Hannah Ritchie, says about agricultural productivity and structural transformation:
“To escape poverty, farmers need to increase labor productivity – to produce more food per hour worked. It is a deep societal problem when most of the population works in farming and gets little money in return. The farmers’ families are unable to get a good education; improve healthcare; and to free up labor so that their children can become teachers or build new industries outside of agriculture.”
…..
Some countries within Sub-Saharan Africa generate as little as half of this regional average. This makes it impossible for families to escape poverty. Most are smallholder farms. They need family members to work and contribute. They also often cannot afford to invest in education or other opportunities that might allow them to move into industries with higher wages. Without increasing labor productivity, most of the population will have to continue working in agriculture.
I read this as making two claims. First, that improving labor productivity in agriculture is sufficient for encouraging structural transformation. Second, that improving labor productivity is necessary for structural transformation
These claims definitely make sense under the story we were telling about Farmville.
More food (from productivity increases) → More people doing other stuff → Productivity improvements → Economic development.
The question is if Farmville is the right story in the first place for understanding development in the 21st century. That is, is modern sub-saharan Africa similar to Farmville?
In some ways, this is a clear yes.
Here’s a heat map of daily caloric deficits worldwide. It’s pretty clear that sub-saharan Africa would benefit substantially from being able to produce more food7.
So, both Farmville and our real world countries of interest are struggling to produce even a subsistence level of food. They share more similarities than that. Remember how Farmville had a widget factory that would have been a much more efficient use of their time, it’s just that they couldn’t afford to switch people to working in it? Developing countries seem to have a similar problem.
Gollin, Lagakos, and Waugh (2014) find that workers in developing countries who are working in agriculture earn substantially less and are less productive than those that are working in other sectors8. This is really weird, we would expect workers who knew that they could earn a much higher wage in another industry to switch over to take advantage of that extra income. (Note: this difference in wage probably isn’t because of higher skills/human capital, Gollin et al attempt to account for this in their measurement).
A potential explanation for this is that workers are in some way trapped in the agricultural sector, perhaps because of constraints on food availability.
So far, it looks like our Farmville model is doing pretty well, the real world has similar scenarios of countries struggling to produce food and being unable to take advantage of higher productivity sectors that they would like to switch into. But, as I’ve been building up to, I think there are a couple of places where the story of Farmville and the story of the real world come apart
IV. Farmville No More
First, Trade. Farmville is an isolationist society with no contact with the outside world, something most developing countries aren’t. This means we have to consider the possibility of engaging with the outside world.
Imagine that 10 miles from Farmville is the farming town of Stardew Valley which, being much more efficient at faming than Farmville, is producing more than enough food to feed both towns (and has no interest in making widgets). Let’s also assume that a Farmville worker will generate more value per hour building widgets rather than farming corn. There is no reason to try and improve Farmville’s agriculture now! They should just immediately switch to making widgets and sell them to Stardew Valley in exchange for food. They will receive all of the benefits of industrialization while being able to purchase lower cost food from elsewhere. There would be no point in increasing agricultural output unless it would exceed Stardew Valley’s rate. So, in this model, agricultural productivity isn’t necessary for industrialization.
Of course, this also isn’t a perfect representation of the real world. Trade involves transportation costs, regulations and tariffs, and complex markets. But, the general point that agricultural production doesn’t necessarily need to be done domestically stands. For real world examples of this, the WTO maintains a list of developing countries that are net importers of foods that can be found here.9
Furthermore, the opposite problem can emerge, where trade incentivizes moving workers into agriculture instead of out of it. In Farmville, past a certain point of food production the demand for food largely disappears as the small domestic market is saturated. That is, each person in Farmville is only going to want to buy so much corn with their widgets. Under open trade, the price for agricultural goods is set by the global demand for food rather than locally, so prices shouldn’t decrease once everyone local has purchased enough. This means that a country won’t necessarily see a lower and lower price per corn cob for each additional one they produce (in economic terminology the market is closer to perfect competition where prices are taken as given).
Why does this matter? Because if this was the case, then productivity improvements to agriculture would actually increase the returns to agriculture relative to manufacturing (because production per hour is going up while prices remain constant), incentivizing workers to move from industry to agriculture.
Bustos et al. (2016) is a paper that provides super interesting evidence of this sort of phenomenon10. They find that the introduction of a certain productivity enhancing technology in Brazil (new techniques that allowed more plantings per year) which raised the marginal product of labor (how much an additional worker would produce) pulled more people into the agricultural sector and out of manufacturing.11
So, productivity improvements both aren’t strictly necessary for structural transformation and aren’t strictly sufficient to cause to structural transformation. Instead, the results of productivity improvements are a function of the background conditions of a society and vary based on location.
The second place agricultural productivity improvements can fail to translate into industrialization is if the labor supplied to a sector is inelastic. By inelastic, I mean that the amount of labor supplied is not very sensitive to changes in various factors. So, where labor is allocated in the economy will stay the same even if productivity changes mean that it would be more efficiently used elsewhere.
Eberhardt and Vollrath (2018) find that elasticity of agricultural labor varies across countries and that differences in labor elasticity explain a fair amount of how countries respond to productivity increase (higher elasticity means you shift more people out of agriculture)12.
Here’s a sort of weird example about how elasticity affects the labor supply after a productivity change to help you get your head around it.
Imagine Susy, a Farmville resident, gathers up all 50 residents and announces
“Alright everyone, I’ve figured out the solution to our food problem. I used my summoning circle to talk to Cthulhu and he said that if we all gather at the stroke of midnight and ritually sacrifice a goat he will increase our crop yield by ten-fold. But, he said it was really important that all 50 of us do it, otherwise Farmville will be consumed by a writhing, unknowable mass of eldritch tentacles.”
This is admittedly an odd example, but, arguably, Cthulhu ritual sacrifice is a totally inelastic form of agricultural productivity improvement. Even though food per capita will go through the roof from Susy’s innovation, you can’t transfer anyone to a different sector because then Cthulhu will eat you all. Thus, no structural transformation can occur.
I think that a more real world version of this sort of variation in elasticity is in the size of farms. Ritchie points out in their piece that a majority of farmers are located on Smallholder farms, <2 hectare farms usually farmed by a single family13. I think these types of owner-operator farms are an example of where low elasticity may be problematic for structural transformation.
That is, when farms are really small, it seems likely to me that productivity improvements may result in less transformation in the allocation of labor. There are a couple reasons for this. First is that these families own a very specific type of capital that they are invested in and therefore are going to have higher frictions compared to if they were workers on a commercialized farm. When you own capital (livestock, land, etc) there are higher costs to transferring industries than if you are a wage laborer on a larger farm.
Furthermore, when you have a small number employees (just immediate family) on a farm, you have fewer options regarding how much labor you can select. A large farm with lots of employees can simply reduce how much labor they hire, releasing some % of employees into the industrial sector as productivity improves. However, at the other extreme with just one person, you don’t have the option to fine tune what percent of your current labor you want to keep. You just select whether to spend your time working on the farm or not (effectively either 0 or 100% of labor supply)14 . This inability to fine tune labor is going to hamper the release of labor to industry, as smallholders won’t be able to release labor piece by piece.
Another reason smallholders may be inelastic is that we might expect elasticity to vary with the ease of migration. If you can’t afford to get to manufacturing jobs, it doesn’t matter how much they pay, you will still be stuck in rural areas. To that extent, elasticity is also a result of background policy choices in investment in infrastructure and redistribution, so we have another reason productivity improvements by themselves aren’t sufficienct for change15.
And migration isn’t the only way infrastructure makes smallholders less elastic. It’s pretty well documented that smallholders are less integrated into the market for agriculture and have fewer opportunities to sell their goods.16 This would also reduce the structural effects of productivity improvements, as the resulting income benefits of higher productivity would be smaller for smallholders as they won’t be able to offload products at as high rates as they otherwise could.
I should be clear that I’m not really that familiar with the literature on labor elasticity in developing country agriculture, so take my proposed mechanisms with a big grain of salt. But, there does seem to be decent evidence that smallholder farmers are less responsive to changes in economic conditions than other farmers and that they have a lower labor elasticity generally17. So, we do have some evidence that labor elasticity likely dependent on the underlying makeup of the agrarian sector.
V. Conclusion
What should we make of the fact that agricultural productivity isn't necessarily a silver bullet? Probably not too much. I’m fairly convinced that no one factor is the silver bullet for development and, furthermore, something doesn’t need to cause development for it to be very, very good. Even if food isn’t going to always build you a manufacturing sector, producing more food is going to raise the incomes of farmers and make food cheaper. Those are important things to consider! The World Bank thinks improvements in agriculture have around 2.5X the impact on poverty that improvements in other sectors do. That’s a really significant way of improving the world, even if it doesn’t necessarily result in compositional changes in the labor force.
I also don’t want to sound like I don’t think that agricultural productivity has any effect on structural transformation. It definitely did historically and it definitely still does now. The takeaway here should be that it is neither a necessary nor sufficient condition for industrialization, not that it isn’t desirable.
Finally, I should make it clear that I think the OurWorldInData piece is actually very good. My disagreement is A. very minor and B. possibly incorrect. The author is both much smarter than me and also an actual expert on the subject matter (I have shamelessly stolen their graphics throughout this post). I am a 22 year old with an internet connection whose only professional experience with agriculture is getting paid by my grandad to pick tomatoes part time one summer. Evaluate my critiques with that in mind.
Footnotes:
Gollin, D., Parente, S. L., & Rogerson, R. (2007). The food problem and the evolution of international income levels. Journal of Monetary Economics, 54(4), 1230–1255.
https://www.economicsonline.co.uk/global_economics/structural_change_theory.html/
Hatcher, John, and Mark Bailey. Modelling the Middle Ages: The History and Theory of England’s Economic Development. Oxford ; New York: Oxford University Press, 2001.
https://ourworldindata.org/grapher/cereal-yields-uk
Berthold Herrendorf, Richard Rogerson, Ákos Valentinyi, Chapter 6 - Growth and Structural Transformation, Editor(s): Philippe Aghion, Steven N. Durlauf, Handbook of Economic Growth, Elsevier, Volume 2, 2014, Pages 855-941, ISSN 1574-0684, ISBN 9780444535467, https://doi.org/10.1016/B978-0-444-53540-5.00006-9.
https://ourworldindata.org/africa-yields-problem
https://ourworldindata.org/hunger-and-undernourishment
Gollin, Douglas, David Lagakos, and Michael E. Waugh. “The Agricultural Productivity Gap*.” The Quarterly Journal of Economics 129, no. 2 (May 1, 2014): 939–93. https://doi.org/10.1093/qje/qjt056.
https://docs.wto.org/dol2fe/Pages/SS/directdoc.aspx?filename=q:/G/AG/5R11.pdf&Open=True
Bustos, Paula, Bruno Caprettini, and Jacopo Ponticelli. “Agricultural Productivity and Structural Transformation: Evidence from Brazil.” American Economic Review 106, no. 6 (June 1, 2016): 1320–65. https://doi.org/10.1257/aer.20131061.
I should note, this finding is specifically for productivity improvements that increased the marginal product of labor. Productivity improvements that decreased the marginal product of labor (i.e. labor saving enhancements) were found to have the opposite effect.
Eberhardt, Markus, and Dietrich Vollrath. “The Effect of Agricultural Technology on the Speed of Development.” World Development 109 (September 2018): 483–96. https://doi.org/10.1016/j.worlddev.2016.03.017.
Note: They are specifically looking at how elasticity varies with climate, but I think the general result applies.
https://ourworldindata.org/farm-size
This is a simplification. Workers can and often do split their time between sectors both within seasons and seasonally. The point is just that rigidities emerge as farm size decreases that make optimal allocation of labor much harder.
I should note that the author of the Our World in Data piece directly acknowledges this point. My disagreement is that she frames it exclusively as a barrier to productivity improving, while I think it is also a barrier to productivity improvements mattering.
“World Bank. 2007. World Development Report 2008 : Agriculture for Development. Washington, DC. © World Bank. https://openknowledge.worldbank.org/handle/10986/5990 License: CC BY 3.0 IGO.”
Merfeld, Joshua D. (2020) : Smallholders, Market Failures, and Agricultural Production: Evidence from India, IZA Discussion Papers, No. 13682, Institute of Labor Economics (IZA), Bonn
and
Collier, Paul, and Stefan Dercon. “African Agriculture in 50 Years: Smallholders in a Rapidly Changing World?” World Development 63 (November 2014): 92–101. https://doi.org/10.1016/j.worlddev.2013.10.001.