Far Side of the Constraint by Arata Hoshino

Essay #01

One Constraint

Arata Hoshino ·

Listen to this essay
0:00 / 9:50

In 1915, about 93 million acres of American cropland grew feed for horses and mules. By 1960, the figure was 5 million. In between, the tractor arrived on the farm and the car on the street, and motor vehicle registrations rose from 468,500 in 1910 to almost 27 million in 1930.

Land was the scarcest input of that age, and when one source of power replaced another, land was also the very first thing released. Whenever an argument about how a society should be run seems to have no end in sight, I always remind myself of that fact.

The argument with no end in sight today, especially in advanced economies, is about three problems. Stagnant economies, collapsing birth rates, and wealth that no longer reaches households are usually filed under economics, culture, and politics. Treated separately, they call for solutions that contradict each other. Raising the birth rate calls for large-scale fiscal spending, AI-led growth calls for attracting investment on a large scale, and reducing inequality calls for heavy taxes on wealth. Under a government that tries all three at once, owners move their capital abroad to escape the tax (capital flight), and the hoped-for investment falls through along with it. So none of the solutions ends up working. My view is that all three are symptoms of one constraint, a stall in "a civilization's capacity to build," broadly understood.

The figure below shows, in American data, the constraint and two of its symptoms, one above another, in the order they appeared. Energy use per person peaked in 1978–79 and has since fallen by close to a fifth. The labor share (the part of income paid to workers) swung up and down for decades, then fell steeply after 2000 and has not recovered. Births per woman (the number of children a woman would have over her lifetime at that year's rates), the slowest of the three to move, sit at the bottom.

The constraint first, then two symptoms

Primary energy per American, million Btu200250300350peak in 1978 and 1979, never regainedLabor’s share of US GDP, %556065up and down for decades, then a steep fall after 2000Births per American woman1.52.02.53.03.519501960197019801990200020102020replacement, 2.1falling since 2007, to record lows

Energy use per person stopped growing first, at the end of the 1970s. Labor’s share of income swung up and down for decades, then fell steeply after 2000 and has not recovered. Births have fallen since 2007, to record lows. The fall of the 1960s came with the end of the baby boom.

Source: US Energy Information Administration, Monthly Energy Review, Table 1.7 (2025 provisional); Penn World Table 10.01 via FRED; World Bank.

Arguments about institutions end when technology disrupts their premises

Arguments about institutions are in fact seldom settled by one side prevailing. More often, revolutions in energy and machines change the conditions an argument rested on, and it fades away without anyone quite noticing. This essay looks at that idea through two historical cases (the horses, incidentally, make a third).

In 1798, Thomas Malthus wrote that there was a ceiling on how many people the island of Britain could feed, because, left unchecked, population grows faster than the food the land can produce. On top of that, he argued that the ceiling was already a bottleneck in his own day. Britain then spent half a century wrangling over that ceiling, in disputes over the Corn Laws, the Poor Laws, rent, and wages. Politically, the matter was settled the moment Parliament repealed the Corn Laws in 1846, while the ceiling itself was raised by the spread of coal.

Now, to see how much coal mattered, try replacing it with trees. Say you had tried to match the heat of the coal that England and Wales dug in 1850 with firewood. It would have taken 48.1 million acres of woodland (assuming the trees were cut no faster than the woods grew back), one and a half times the land area of England and Wales. However keen a forester you were, sooner or later you would be left with nothing but bare hills.

Malthus's arithmetic of population and the economy was right on its own premise, that food output is tied to land and depends on sunlight alone. But once Britain could burn coal, sunlight stored up long ago (coal is the remains of ancient plants carbonized underground), the land limit on which his whole argument rested was gone, and the argument lost its premise.

Japan's argument was never settled

Japan went through a similar change in the last years of the Tokugawa shogunate, in the 1860s, when the country was still arguing over whether to expel the foreigners or open the ports. While most people were caught up in that argument, the leaders of the Meiji Restoration put their effort into bringing in the most advanced technology of the day, the steam engine and the telegraph, and found that once the machines were in place, the terms of the argument changed.

The dispute over opening the country was never settled. At some point after the engines of foreign steamships had been taken apart and copied, telegraph wires strung, and the first railways laid, it quietly stopped being a question at all.

The horses left, and the land opened up

The horses make the same sequence plain across a whole country's farmland. Outside American farms, the number of horses fell from 3.45 million in 1910 to 380,000 in 1925, a drop of no less than 89 percent in fifteen years. On the farms, the tractor took over the field work. Cropland feeding horses and mules fell from its 1915 peak of about 93 million acres to 65 million by 1930 and to 5 million by 1960. The land released added up to roughly 88 million acres.

In all three cases, the revolutions in energy and machines came first. Each time a premise collapsed, an argument that had seemed endless faded away.

The UAE builds the power first

I grew up in Japan and, through those years, watched my country's population and economy shrink. I now live in the UAE, and to my eye, the practical order of building here looks much like the industrial-age cases above. Broadly, the UAE puts supply in place first. It builds power plants and computing capacity, brings in a large workforce, and settles the institutional questions afterward. Advanced economies today, Europe and Japan among them, do it the other way around.

They have treated energy and labor as costs and planned supply to follow demand. By contrast, the UAE's Barakah nuclear plant (5.6 gigawatts) and Al Dhafra solar plant (2 gigawatts) were built in the opposite order. The capacity to be procured was fixed first, and only then did the auction choose the developer and set the price (Al Dhafra's winning bid was 1.35 US cents per kilowatt-hour, the lowest published anywhere at the time). So here the price of a kilowatt-hour matters less than how many gigawatts can be drawn, for how many years, without anyone having to think about the meter.

If you check the price before you switch on a power-hungry appliance, energy is a cost where you live, and plans get made around its scarcity.

Which premises of today's three arguments can be disrupted?

Today's three arguments look as endless as the Corn Law debates once did. Will they, like the old arguments, fade away with neither side prevailing, as revolutions in energy and machines change their premises?

If so, what should we be asking? My thesis, Far Side of the Constraint, frames that broad capacity to build as the product of three terms. The first is institutions (the pull that draws in talent and investment, including fast permitting and light regulation). The second is effective energy abundance (energy close to free and expected to stay that cheap). The third is labor input (once cheap human labor, and today, more and more, the reach of automation). In a product, a term near zero keeps the result small however strong the other two are. The practical question is which of the three is missing and how to supply it. I would ask readers who invest to check which of the three terms is moving now.

As for the horses, they never lost a debate. The cropland that fed them shrank from about 93 million acres to 5 million because a new source of power arrived, and land, the scarcest input of that age, was the first thing released. When an argument about institutions looks endless, look first at what the new energy and the new machines are about to release.

For readers: Which argument around you looks endless right now?

Contact
Email · LinkedIn · X
Newsletter
LinkedIn newsletter · Substack
Podcast
Spotify · YouTube
Read the investment thesis

Notes

  1. Horses and land: Alan L. Olmstead and Paul W. Rhode, "Reshaping the Landscape: The Impact and Diffusion of the Tractor in American Agriculture, 1910–1960," Journal of Economic History 61(3), 2001, 663–698. Horses outside farms, 3.45 million in 1910, 2.13 million in 1920, and 380,000 in 1925, from note 16 at p. 670, citing USDA figures compiled by C. L. Harlan. Cropland for horses and mules, about 93 million acres in 1915 (79 million for animals on farms), 65 million by 1930, and 5 million by 1960, at pp. 664–665; the authors put the land released at roughly two thirds of all cropland harvested in 1920 within the territory of the Louisiana Purchase, with some 80 million acres of pasture besides. Motor vehicle registrations: Federal Highway Administration, Highway Statistics Summary to 1995, Table MV-200, 1997 (468,500 in 1910 and 26,749,853 in 1930). Land as the first input released: Far Side of the Constraint, Premise.
  2. The three problems, their treatment as separate questions, and the prescriptions that contradict each other: Far Side of the Constraint, Premise. The single constraint: Investment Thesis, abstract.
  3. Figure: US Energy Information Administration, Monthly Energy Review, Table 1.7 (207 million Btu per person in 1949, a joint peak of 350 in 1978 and 1979, and 282 in 2025, provisional); the labor share from Penn World Table 10.01 via FRED (series LABSHPUSA156NRUG; 64.9 percent at its 1970 peak, 64.0 in 2001, and 56.8 in 2023); births per woman from the World Bank. The fall in energy use per person: Investment Thesis, section 03. The fall in the labor share across most countries and industries since the early 1980s: Investment Thesis, section 07 (Loukas Karabarbounis and Brent Neiman, "The Global Decline of the Labor Share," Quarterly Journal of Economics 129(1), 2014, 61–103). Fertility as the slowest-moving symptom: Far Side of the Constraint, Premise.
  4. Malthus: T. R. Malthus, An Essay on the Principle of Population, as it affects the Future Improvement of Society, London: J. Johnson, 1798, chapters 1 and 2; he argued that the constraint was already operating, not that a collision lay ahead. The woodland equivalence is E. A. Wrigley, Energy and the English Industrial Revolution, Cambridge University Press, 2010, p. 99, assuming an acre of woodland on a sustained yield gives the heat of a ton of coal (48.1 million acres in 1850, 150 percent of the land surface); the production totals and land surface are those of England and Wales. The repeal of 1846 was a real political event, and the claim here is that the premises of the argument were replaced, not that physics did the work of politics. How much weight coal deserves remains debated, and Gregory Clark and David Jacks ("Coal and the Industrial Revolution, 1700–1869," European Review of Economic History 11(1), 2007, 39–72) and Alexander Tepper and Karol Jan Borowiecki ("Accounting for Breakout in Britain: The Industrial Revolution through a Malthusian Lens," Federal Reserve Bank of New York Staff Report 639, 2013) estimate a smaller contribution.
  5. Britain and late Tokugawa Japan as cases of arguments that faded, and the Gulf following the same order: Far Side of the Constraint, Premise.
  6. Barakah: four APR1400 units, 5.6 GW design capacity, all in commercial operation from September 2024 (Emirates Nuclear Energy Corporation). Al Dhafra: 2 GW nameplate, awarded in April 2020 at 1.35 US cents per kWh, then the lowest published price in the world (Abu Dhabi National Energy Company, press release, 22 December 2020). Energy treated as a cost, and capacity fixed before the auction: Far Side of the Constraint, Premise and its note 2.
  7. The three arguments likely to fade, the practical question, and the investor's first question: Far Side of the Constraint, Premise. The equation and the definitions of its terms: Investment Thesis, section 01.