Can we solve the industrial revolution?

May 25, 2025

Summary: The question “wherefore Neolithic revolution” is as important as “wherefore Industrial revolution”. But it turns out some Italian guy answered the first question, and it was really easy. Does this imply anything about the second question?

Economists grab every field we can; we expand and gnaw and grasp. We eat fields like public policy or epidemiology, and make methodological improvements to some math-y field. Every once in a while, however, we touch on a question which rhymes with one of our great motivating questions. We find a big question about the organization of commerce and society, with a perfectly econometrics-shaped hole in it, somewhere we hadn’t yet looked.

Here’s a question: why did we let Jared Diamond get away with it? I don’t only mean economists here: I mean everyone who took a history survey course at a liberal arts college in the past two decades. The man was an ornithologist!

It took economics a long time to claim the neolithic revolution, and it shouldn’t have. The word “revolution” is right there in the name, and the parallel to the industrial revolution is strong. The growth economists arrived first, studying the step change from hunter-gatherer groups to farming.Oded Galor, Michael Kremer. Then the political economists arrived, interested in the first states: how the concentrated and exercised power.Example 1, 2, 3. In my opinion, this could be regarded as a separate revolution. There were 5,000 years between the first farmers and the first states. The invention of the state was just as important for how people lived, consumed, and produced as the invention of agriculture. This is the type of question on which James Scott is strong.

All of these are fun and exciting questions, well-positioned to nerd-snipe an economist in their field. However, they are not the central question of the Neolithic revolution. That is: after 300,000 years of being human, why did seven independent groups of us invent farming within 2,000 years of each other? This is a question for the economic historians, not the growth economists or political economists. The problem is that the economic historians have been distracted by 1780s Britain.

This is a far more interesting question than “why did Britain industrialize first.”

The Neolithic revolution was so crazily overdetermined. If it hadn’t happened in Mesopotamia or the Yangtze river valley or central Mexico, it would have happened in the Mississippi valley or Papua New Guinea. If it hadn’t happened 12,000 years ago, it would have happened 10,000 years ago.

It is also ex nihilo. There are fun puzzles of pre-Neolithic humans (tool use, language, inter-hominid competition), but they’re not (yet) puzzles for economists. Economics remains not only a human science, but a post-agricultural human science.Looking through my Zotero for counterexamples, I found a few papers from the 1980s which asked whether pigeons followed standard axioms in microeconomics. These were in the QJE and JPE. The door on animal-economics seems to have shut after that. Possibly an example of economists treading on another field’s toes, and deciding it wasn’t worth it? All of economics takes the features of the Neolithic revolution – farming, specialization of labor, agglomerations – for granted. It is the genesis.

Look: take my word for it that “Wherefore Neolithic?” is the more interesting question. Then I have good news and I have bad news. The good news is that we have The Answer. The bad news is that it was super simple.

The problem of the Neolithic

Andrea Matranga is a professor at the University of Turin. He only has one publication in an economics journal, and it provides The Answer. This paper has so much explanatory power that economists can just be done with the question. If someone had answered the same question about the industrial revolution with the same amount of explanatory power, economic historians would shit themselves.

Before the Neolithic revolution, humans lived in small groups of hunter-gatherers, following herds and herbs to productive land. When a group got too large to forage the territory around their campsite, young men would split off and form a new group. This was the entirety of how society and commerce, such as it was, organized itself. A Mesopotamian group of people in 15,000 BP was recognizably organized like a troop of chimpanzees today, or the Lakota 500 years ago. Where food is plenty, you stop to eat; when food becomes scarce, you move with it.

The human “evolutionary environment” is not a useful geographic concept. By 10,000 BC, in our dispersed hunter-gatherer groups, we ranged from Tierra del Fuego and Tasmania to the Arctic tundra; geographically, we were widely capable. But all of this evolution and dispersal had taken place in the same climatic era, the Pleistocene. We were made by that Ice Age. In those days, men were men, the earth was less tilted, its orbit more circular than elliptical, and the longest days of the year occurred at aphelion, when the earth is furthest from the sun. All of this combined to provide a stable northern hemisphere, with all of the rough edges of December and July sanded away.

Then the Ice Age ended, abruptly, 12,000 years ago. Our orbit became more elliptical, because Saturn and Jupiter gave us a tug. Our tilt became more pronounced, because of torque from our equatorial bulge. And the timing of our tilt and our ellipse fell out of whack. Now, the long northern days came while we were closer to the sun, meaning hotter summers, and the short northern days came at aphelion, meaning cooler winters. So: seasons.Here’s a little animation Claude made for me. The scale is exagerrated. Note that, in the Post-Neolithic period, the northern hemisphere is tilted further away from the sun at Aphelion, when we’re furthest from the sun.

Confronted with the loss of a stable climate, groups of hunter-gatherers starved. In temperate zones (think between 30◦N and 40◦N, e.g., Iraq, central China, the Mediterranean Nile, North Carolina, northern Mexico) summers were more lush than ever, but winters were freezing and harsh. There was less game to hunt, and fewer roots to gather. Closer to the equator (think between 10◦ and 20◦ on either side, e.g., Peru, the Niger delta, Ethiopia, PNG), winters didn’t freeze, but they became starkly dry. As the geological record shows, and as Matranga establishes from climate data, so many humans had never before lived in seasonal climes.

This isn’t a new observation; the change in orbital behavior and tilt was identified by Milutin Milanković in the 1920s. To make this an interesting paper, Matranga had to establish that seasons caused farming. He was not the first scientist to try to explain the cause of the neolithic; he wasn’t even the first to point to climate. Jared Diamond points to shorter growth seasons favoring cereals; older scholarship blames drier climates for worsening returns to hunting and gathering. Some point to Malthusian pressures: at the end of the Ice Age, populations rose, and when temperatures fell again, these larger populations were compressed into smaller areas. Matranga takes no truck:

A running issue with these climatic theories is that while they fit the archaeological sequences for the locations they describe, they have not been able to explain why agriculture did not emerge earlier in other locations were subject to similar conditions. The last Ice Age lacked neither warm conditions, nor dry ones, nor climatic improvements followed by rapid reversals, and yet agriculture was not invented.

Any cause one claims for the Neolithic has to explain the near-simultaneous, independent development of farming in diverse locations. It can’t be dryness; there was dryness before. It can’t be warmth; there was warmth before. It can’t be population cycles; populations have always boomed and busted. To explain the Neolithic, you need something that hadn’t been anywhere, and now was everywhere. Seasonality gave us this means. The motive is what makes this an economics paper.

Think like a Bronze Age hunter-gatherer: now that your winter is cold and/or dry, you will starve. There’s a fun literature based on interpreting health and diet from millennia-old bones; Matranga uses these to back-out welfare effects of the Neolithic revolution.Economics papers have gotten much longer over the past decades and this is why. This would have been a great paper without the welfare effects. But it’s cool that he added them. While groups were nomads, hunting and gathering, their bones showed Harris lines. Like the rings in a tree’s trunk, Harris lines appear when a young person has gone through a starvation period, and then is re-fed, prompting rapid catch-up growth. As winters grew worse, these periods of starvation became longer and more severe.

The economic response to starving is to store food. Summers were as productive as ever, and there was more food than you could eat. Starving in the winter concentrates the mind wonderfully in the summer. And this is what kiboshed nomadism. Storing food is a very effective means of avoiding starvation in a harsh winter, but it also means the group must stay close to its stores. The group’s travelling radius shrinks, increasing the amount of food it has to extract from a smaller piece of land, and increasing its knowledge about that land and its autochthonous plants. This means farming.

Matranga goes much further. He assesses each of seven autochthonous agricultural societies for the sequence in which they became sedentary, began storing food, and began farming. He evaluates alternative incentives to seasons for storage. He tests other channels, finding further support: proximity to mountains delayed the adoption of agriculture, because nomads can avoid seasonal effects by moving uphill in warm weather and downhill in cold weather. Nomads in flat terrain had no options but starve or store. Matranga’s evaluation of the bone record is deep, focusing on starvation patterns, changing heights, and calcium deposits.

But all of that is overkill.He had good reason for the overkill, as this was his job market paper. He had already answered the question! No one ever has to write another paper about why the Neolithic happened in those places at that time, ever again. No further research is needed. With a time machine, perhaps, or greater archeological evidence, it would be interesting to explain the individual episodes, but the great question of independent and autochthonous farming has been solved.

And I’m underwhelmed. Like the invention of farming, this paper was overdetermined. A convergence of archeological data, new climatic data, and geospatial tools made this a straightforward hypothesis to test, and there’s no efficient market of ideas: Everything of importance has been said before by somebody who did not discover it. I would expect, somewhere, in some 1910s manuscript, the idea “seasonality made food storage necessary, which enabled agriculture” has been written down before. Matranga gets all the credit for this discovery. I’m just surprised that such an important question had such a simple answer.

Why can’t everything be this simple?

But “important questions have complex answers” is a founding myth of economics. It took me a long time to realize that faculty saw seminars are entertainment: just waiting for the twist, the shock, the surprising result and unanticipated mechanism. Seasonality is unanticipated, but it’s also simple in that it is sufficient. An economic historian shudders at the thought that the Neolithic revolution had a single cause: because we have given up on finding a single cause for the Industrial revolution.

In academic economics, “Economic History” often means “History of the Industrial Revolution.” This isn’t just Eurocentrism; there’s something about curves in graphs which captures the imagination of an economic historian. There’s nothing as sexy as 19th century British textiles.

Despite its proximity, the Industrial revolution is more amorphous than the Neolithic. We mostly agree on the facts: the Industrial revolution started in Britain, sometime around the the late 1700s, and involved the economy. That extensional definition (“The thing that began in the British economy around the late 1700s”) is sometimes the best we have.

As for what caused this amorphous thing, there are so many potential explanations. I’m not trying to convince you of any particular story, so I’ll sprint through them:

First, there are social and institutional changes. During the plague years of the 1300s and 1400s, many workers died, and wages for the remaining workers soared. As the population began to climb in the late 1500s, wages fell to normal levels. There were other changes to the labor supply, such as the official end to feudalism after the English Civil War. Innovations in agricultural productivity reduced labor demand in rural areas; alongside already falling agricultural wages, this pushed labor into cities, increasing agglomeration effects.

Others point to inflation, as the supply of gold and silver in Europe spiked with imports from the Americas, although others hold the Price revolution was due earlier. Access to Atlantic trade itself may have been causal. Also prominent are patents: legal monopoly rents provided a strong incentive for manufacturers to innovate and increase productivity. Secure property rights, also following the English Civil War and Glorious Revolution, probably played a role in securing investment. Some point to the abolition of internal tariffs during the Hundred Years’ War. This created a larger internal market than e.g., France or the Low Countries. A larger market created more competition and incentive to innovate.What’s the conventional citation for this one? I couldn’t find it in my Zotero.

There are plenty of cultural explanations. Weber’s Protestant work ethic is most famous; others point to a secular shift to later marriage, leaving more room for human capital accumulation while young. A more educated population is more likely to invent industrial technologies. A more fundamental cause may be European “scientific empiricism” and general mechanical literacy, which might in turn have fallen out of a very specific form of institutional nobility.

A final category of cause rhymes with the Neolithic explanation, but geology instead of astrophysics: easy access to cheap energy, namely, the shallow coal lodes of Wales and north-western England.

I’m not throwing my weight behind any of these theories. If I had to choose, easy coal+falling wages seems to have a lot of explanatory power. The three lessons you should take away from that list are that economists have put a lot of effort into finding an answer, there isn’t a clearly winning answer, and whatever answer there is is probably multi-causal. None of these three things are true of the Neolithic.

What would an equivalent “answer” look like for the Industrial revolution? And why haven’t we found it yet? I see two possibilities: First, maybe Matranga, and I, are wrong, and seasonality is not a necessary and sufficient cause of the Neolithic. Second, maybe the Neolithic is just an easier question.

I really like Matranga’s answer; I find his methods and his model compelling, and climatic seasonality is sufficient for me to understand the decline of nomadism and rise of agriculture. I’ve never met the man, but he’s probably handsome, good with kids, and knows the shopkeeper by name. However, the downside of answering a less-explored question is that there is less criticism: a paper in a top economics journal claiming to “solve” the industrial revolution would get picked apart in blogs, tweets, NYT comments, other journals. Matranga’s paper was peer reviewed, of course, and has gotten some coverage — but nowhere near the scrutiny of more mainstream economic history topics.

If Matranga’s answer is wrong or incomplete, I would guess it’s wrong or incomplete in missing some other necessary cause. The only direct response from an economist is a recent job market paper out of Brown University which claims that his explanation is incomplete. Specifically, that the extinction of large mammal species is a better explanation of the eventual development of farming. I don’t buy the mechanisms in the extinction paper at all, and I think this argument is a weaker contribution than Matranga’s seasonality.Briefly: Extinction is endogenous: the reason wild cattle and some elephants went extinct is because humans ate them too fast! We coexisted with these mammals for hundreds of thousands of years before hunting them to extinction; something must have changed. What could it have been? Probably seasonality. Slightly more provocatively, probably sedentarism! It’s possible they’re both right: seasonality caused sedentarism, and sedentarism only causes agriculture if large mammals are present to hunt to extinction. I remain fairly sceptical that the large mammals were necessary or sufficient. A more general story of resource depletion (of large mammals, and any other food resource) seems to do just as well.

Greater scrutiny may come from outside the field. Anthropologists and archeologists haven’t interacted much with Matranga’s results, besides a few generally positive tweets. He uses data on seven autochthonous inventions of agriculture; maybe we’ll find a new site which is in the mountains, or exists in a highly non-seasonal area. Maybe I’ll check back on this literature in ten years and find a thriving debate about the causes of the Neolithic.

But even with some caution, “Wherefore Neolithic” just looks easier than “Wherefore Industrialization.” Most importantly, we have many independent observations of the Neolithic occurring, so we can hold different factors like altitude, latitude, and biome stable and look at common factors to identify the single cause. The industrial revolution only started once; what we have instead is a dataset of places and times where it didn’t start. We can reduce our confidence in single causes.E.g., coal is easily available in the Ruhr Valley, and there was no revolution in Germany; European wages fell during Charlemagne’s reign, but there was no revolution then; etc. But a minuscule sample size makes any explanation harder to justify.

Also, the Industrial revolution seems not to have a single cause. While seasonality was sufficient for the Neolithic, the explanation for the Industrial revolution will be far more complex: some weighted subset of the above factors might have caused it, or some factors which no historians have pointed to yet.

This is a sample size problem. Each of the seven cases Matranga includes is a sharp test of just the one seasonality hypothesis. The Industrial revolution is almost the inverse: just one case (Britain, 1780), but a dozen hypotheses (coal, wages, patents, Parliament, Price revolution).This is just the claim that you need n ≫ k to avoid overfitting. n = 7 is weak, but ~fine for linear regression.

Why was there only one Industrial revolution? The many Neolithics were spread over millennia; why couldn’t China have undergone an Industrial revolution after Britain? The answer is that industry diffuses much faster than agriculture: China didn’t have to have their own revolution because they could just borrow Britain’s. So if we could go back and stop diffusion, maybe we’d get a bigger sample size.

Good things come in threes?

Every blog post needs an AI section nowadays, and this one comes ready-built. Among the many competing definitions of “Artificial General Intelligence”, one which has been gaining ground is “Transformative AI”. The approach TAI takes is to define AGI by the outcome: we’ll know we have TAI when the effects are of similar scale to what happened in the Neolithic or Industrial Revolutions.

This always sounded a bit weird to me. The Neolithic and Industrial revolutions are very different! And if you’re including those, there are other things which could fit too – the first states came into being about halfway between the two revolutions, and had similarly huge effects.

Given what we know about the causes and effects of the Neolithic and Industrial revolutions, which one will AI be like? Or: how will AI be like the Neolithic, and how will it be like the Industrial revolution? We spend a lot more time thinking about the Industrial revolution than the Neolithic, and those similarities will come to mind more easily. But just like how we missed a simple and important answer to “Wherefore Neolithic?” for so long, we might be missing other important insights from the Neolithic.