What happened to the hand weavers?

Talk it over with a neighbour before we move on
?
raw cotton
fibre
yarn
cloth
clothes
Clean
Spin
Weave
Sew
done by machine
done by hand
In the 1770s and 1780s, machines took over spinning, and yarn became far cheaper. Weaving was still done by hand, on a loom at home.

What do you expect happened to hand weavers: their numbers, and their pay? And what happens next?
All slides, one per page
© 2026 Dr. Jesse Roberts

Every machine moved the bottleneck

Lancashire cotton, 1780s to 1860: each click is one turn of the chain
raw cotton
fibre
yarn
cloth
clothes
Clean
Clean
Spin
Weave
Weave
Sew
Sew
done by machine
done by hand
1790s
The weavers boom
Cheap yarn piles up at the weavers' door. Demand for hand weavers, and their pay, soar.
1815–30
The power loom
Weaving is the bottleneck now, so it is the next target. Pay per hand-woven piece: 3s → 1s 4d.
1824–30
The self-acting mule
Mule spinners are the best-paid, best-organised workers. After their 1824 strike, owners back a machine to run without them.
1793–1860
The cotton gin
Upstream, Lancashire's demand for raw cotton spreads slavery: 0.7 million enslaved people in the US in 1790, 4 million in 1860.
1850s
The sewing machine
Downstream, cheap cloth made sewing by hand a growing trade, until it too was mechanised.
Yarn: Clark (2004). Weavers: Allen (2016); piece rates via Spartacus Educational. Mule: Lazonick (1979). Enslaved population: US Census, via Mississippi History Now.

Why Britain first?

France and India had spinners too. Why were the machines built in Britain?
?
Place
Wages
Energy
Did machines pay?
Britain
High
Coal, cheap
Yes
Low Countries
High
Peat, pricier
Less so
France
Lower
Wood, expensive
No
India, China
Low
Wood, expensive
No
Why were British wages high?
1348
The Black Death kills a third to half of Europe.
After
Fewer workers, so wages rise. Later marriage and fewer births keep them up.
1700s
A British worker costs a lot next to coal and a machine.
Robert Allen: machines came first where labour was expensive and energy was cheap.
Allen, The British Industrial Revolution in Global Perspective (2009); Voigtländer & Voth, AER 2013. 

Cheap cloth

The second-order effects of fabric automation
−91%
cost of a pound of cotton yarn, 1784 to 1832
3s → 1s 4d
a hand weaver's pay per 28-yard piece, 1815 to 1830
~50%
of British exports were cotton goods by the 1830s
0.7M → 4M
enslaved people in the US, 1790 to 1860
Gained
  • Consumers: washable cotton shirts, underwear and bedding for ordinary households
  • Machine-builders and engineers: a new skilled trade
  • Mill owners: early profits, until competition cut prices
Paid
  • Hand weavers: pay collapsed within one working life
  • India's weavers: priced out of a world market they had led
  • Mill towns: crowded, unhealthy cities for two generations
  • Enslaved people: forced to grow the raw cotton
New jobs
  • Women and children: mill work that needed no apprenticeship, often better paid than the farm or domestic work open to them
  • The catch: long days and dangerous machines, which led to the Factory Act of 1833
The gains were permanent; many costs landed far from Manchester, on people with no say.
Clark (2004); Harley (2010); Allen (2009, 2016); Galbi (1997) on child labour; US Census via Mississippi History Now. India: the role of colonial tariffs is debated.

Declining Birth Rates

Observed data only, 1950 to 2023: no projections
Births per woman
012345619501970199020102.1: replacementUS 1.62UK 1.56Japan 1.21Italy 1.20South Korea 0.72
Already shrinking: 2023 population as % of its peak
Bulgariapeak 198875%Romaniapeak 199083%Hungarypeak 197991%Greecepeak 201091%Japanpeak 200997%Italypeak 201398%
The UK and the US are still growing, largely through immigration. In Eastern Europe, emigration added to low birth rates.
UN World Population Prospects 2024, estimates 1950–2023 (wpp2024 data package). Below 2.1 since: UK 1973, Japan 1974, Italy 1976, South Korea 1984, US 1972.

How many people? 1950 to 2100

Talk it over with a neighbour before we move on
?
0501001502001950200020502100today · UN projection →Nigeria 199United States 121India 102Japan 62China 44
Population, 2025 = 100 for each country. Solid: estimates. Dashed: UN medium projection.
Japan and China are already shrinking; Nigeria will roughly double. Why does a population stop growing, then fall?
The usual pattern as countries get rich
  • Children survive, so families need fewer
  • In cities, children cost money instead of earning it
  • Women's education and paid work
  • Pensions replace children in old age
  • Reliable contraception
Nearly every country that has grown rich has seen births fall below replacement: the plateau comes with prosperity.
UN World Population Prospects 2024 (wpp2024 data package): estimates to 2023, projection after.

Who is over 65?

As population growth slows, the population ages
Population (2025 = 100)0100200Nigeria 199World 123US 121Japan 62South Korea 42Share of adults (18+) who are 65 and over0%20%40%60%South Korea 51%Japan 44%US 34%World 30%Nigeria 16%1950200020502100today · UN projection →
Japan
Population peaked in 2009. Adults over 65: 8% in 1950, 35% in 2023, about 1 in 3.
South Korea
Growth stopped around 2020. Adults over 65: 10% in 2000, 22% in 2023, about 44% by 2050.
Nigeria
Still growing fast. Adults over 65: 6% today.
Why
Fewer births make each new generation smaller than the one ageing above it.
As population growth slows, a larger share of adults is over 65.
UN World Population Prospects 2024 (wpp2024 data package): ages 65+ over ages 18+; estimates to 2023, medium-variant projection after. Solid: estimates. Dashed: projection.

When would people have to retire?

Talk it over with a neighbour before we move on
?
5060708090195020002050210065today · UN projection →Retirement age (years) that keeps retirees at 1 in 6 adults
South Korea 88Japan 86US 79World 75Nigeria 65
Suppose a country wants to keep 1 in 6 adults retired, no more. At what age would people have to retire?
Japan
About 60 in 1975, 77 today, 86 by 2100.
South Korea
68 today, 81 by 2050.
United States
69 today, 74 by 2050.
To keep 1 in 6 adults retired, Japan's retirement age would be 77 today.
UN World Population Prospects 2024 (wpp2024 data package), single years of age; estimates to 2023, medium-variant projection after. Adults = aged 18 and over.

Who pays for the retirees?

Talk it over with a neighbour first
?
Pensions are paid from taxes on wages, and the number of workers is shrinking. Where can the money come from?
Option 1
Work longer
Raise the retirement age.
Option 2
Bring in workers
Immigration from younger countries.
Evidence: Japan: a record 2.57 million foreign workers.
Cost: Takes talent from the countries that need it most.
Option 3
Tax the machines
Tax capital and automation, not only wages.
Cost: Capital can move abroad, and the tax may slow automation.
Option 4
Own the production
A public fund owns automated production and sells to young countries.
Evidence: Norway's fund: NOK 3.8 million per person. Japan earns ¥42 trillion a year abroad.
Every option moves the cost somewhere: to older workers, other countries, capital, or the young.
Norway: Government Pension Fund Global, end of 2025 (Ministry of Finance). Japan: primary income ¥42.28 trillion, FY2025 (Ministry of Finance); foreign workers, MHLW (Jan 2026).

The other side of the ledger: running out of workers

Everything so far assumed workers are plentiful and machines threaten them
04080120203020502070100: one person over 65 per working-age adult31.368.4103.3South Korea: people 65+ per 100 aged 15–64 (projection)
~2084
world population peaks, at about 10.3 billion
>50%
of countries already have fertility below replacement (2.1)
0.80
births per woman in South Korea in 2025
2080
the year the world has more people over 65 than children under 18
Over a forty-year career, workers, not jobs, may become the scarce thing.
UN World Population Prospects 2024; Statistics Korea (2025 births; dependency projection, Nov 2026).

Does automation help when workers run short?

Talk it over with a neighbour first
?
A country's workforce is shrinking and ageing. Do robots and AI rescue it? What do they fix, and what do they leave broken?
The evidence: ageing alone explains close to half of the differences between countries in robot adoption. The fastest-ageing countries (Japan, Korea, Germany, Italy) adopt and invent the most automation.
Output
Yes
Fewer workers, each producing more, can keep living standards rising.
Pensions
Not automatically
Pensions are paid from taxes on wages. If income shifts to capital, output rises while the tax base shrinks.
Care
As a helper
Sensors, lifting aids and AI paperwork help carers. Japanese homes that adopted robots hired more care workers, not fewer.
Wages
Capped
Scarcity lifts pay for complement work. For tasks a machine can do, its cost is the ceiling.
The robots that cost jobs in the robot towns fill jobs no one is there to do in an ageing country.
Acemoglu & Restrepo, “Demographics and Automation”, Review of Economic Studies (2022). Care: Eggleston et al., Stanford (2026).

Same shortage, different answers

Robots, migrants, or a smaller economy: most countries mix all three
South Korea1,066Germany469Japan452United States323Industrial robots per 10,000 manufacturing workers
South Korea · robots
Fertility 0.80; the world's highest robot density. Oldest OECD country by 2045 (projected).
Japan · robots and migrants
30% are 65+; a record 2.57M foreign workers. Most care homes use monitoring or lifting robots, and those homes hired more carers.
China · robots at scale
Shrinking population; 54% of the world's new industrial robots in 2024.
United States · migrants
Younger, and leans more on immigration than on robots.
Sub-Saharan Africa · the young
87% of world population growth to 2100; needs about 2M new jobs a month by 2040.
East Asia got rich doing the world's factory work. If that work is automated instead of sent abroad, poorer countries lose the first rung: Rodrik's premature deindustrialisation.
Automation answers an ageing country's shortage, and may pull the factory ladder up behind it.
IFR World Robotics 2025; Statistics Korea; Japan MHLW (Jan 2026); Eggleston et al., Stanford (2026); Lam, Leibbrandt & Allen (GLM|LIC); Rodrik (2016).

What if one country froze automation?

No new machines and no AI from today. Existing machines stay.
?
What happens in 5 years? In 30? After its population peaks? Who is protected, and who pays?
Years 1–5
Protected: workers in exposed jobs keep them.
Cost: AI-made imports and services bought from abroad compete anyway, unless the border closes.
Years 5–30
Protected: wages in protected jobs hold.
Cost: 1 point slower growth a year leaves neighbours 35% richer after 30 years. Firms and talent leave.
After its population peak
Protected: scarce workers earn more.
Cost: No machines to fill the gap: output and pensions fall, unless it imports workers.
History: Elizabeth I refused the stocking frame in 1589 to protect hand knitters; it went to France and came back. Tokugawa Japan shut itself off until 1853, then modernised in a rush.
A freeze buys time, then the technology arrives on someone else's terms.
Scenario reasoning, not a forecast. 1.01³⁰ ≈ 1.35. Stocking frame: Christ's College, Cambridge; Belper North Mill.

What if every country froze it, together?

No leak, no flight, no arms race, on paper
?
Who wins before the world's population peaks? Who pays after it?
Before the peak: labour gains bargaining power
  • Young workers keep the first rung of the ladder
  • Shrinking workforces lift wages: Engels' pause in reverse
  • Factory work stays a route out of poverty for poorer countries
  • AI data centres become stranded assets: a market shock
After the peak: fewer workers, more retirees
  • Korea: 103 people over 65 per 100 of working age by 2070 (projected)
  • Output per person falls unless people work longer
  • Care absorbs a growing share of every workforce
  • Fewer people, fewer new ideas: growth itself slows (Jones, 2022)
Can it be enforced? AI is software already on millions of computers. Checking a ban means watching all computing, and every rival has a reason to cheat. The ozone treaty worked because a few firms made the chemicals and substitutes existed.
A global freeze moves the cost from today's workers to tomorrow's retirees.
Scenario reasoning, not a forecast. Statistics Korea projection (Nov 2026); Jones, “The End of Economic Growth?”, American Economic Review (2022).

Build for our future selves

The timescales differ, but not by much: South Korea already has both problems at once.
?
Working-age population already past its peak: Japan 1996 · Germany 1996 · Italy 2010 · China 2014 · South Korea 2018
Now
Don't ignore the people the transition hits. The hand weavers paid for a change that made everyone else richer.
Next
Don't plan on young workers forever. Build the care, health and production technology an older country will need, and decide who owns what it makes.
How might policy encourage innovation for the future, which requires a present incentive, and not kill the present labor market?