The Income Floor

The Scale of Displacement

Forty-Seven Percent

In 2013, economists Carl Benedikt Frey and Michael Osborne at Oxford published a study that has since become one of the most cited — and contested — papers in labour economics. Their conclusion: 47% of US occupations were at high risk of automation within 10 to 20 years. The number was so large that critics immediately questioned its methodology. Surely, they argued, jobs don't simply disappear — they transform. Technology creates as much as it destroys.

The critics were not entirely wrong. But the subsequent decade made Frey and Osborne look prescient in the wrong direction: they may have underestimated the speed. McKinsey's 2025 assessment found that the technology available today could automate approximately 57% of current US work hours — not in some future decade, but now, if deployed at scale. The constraint is no longer capability; it is adoption. Goldman Sachs estimates that up to 300 million full-time jobs globally are in the impact zone of AI automation. The World Economic Forum projects the displacement of 92 million jobs by 2030, offset by the creation of 170 million new ones — a net gain on paper, and a social catastrophe in practice if the transitions are not managed.

The gap is the problem. New jobs do not appear in the same towns, require the same skills, or serve the same people whose old jobs disappear. A truck driver in rural Ohio is not naturally redeployed as a prompt engineer in San Francisco. The math of net job creation has always obscured the human arithmetic of displacement — the individuals and communities who fall through the gap between the world that existed and the one that doesn't yet.

The industries most exposed are not the ones that often dominate the conversation. Administrative and office support, where 46% of tasks are already automatable; manufacturing, at 45%; customer service at 41%. These are not marginal occupations. They are the backbone of middle-income employment across the developed world — and much of the developing world, where cheap labour once served as a rung on the industrialisation ladder. That ladder is being pulled up.

Robots do not simply displace workers in existing industries. As Oxford Economics projects, 20 million global manufacturing jobs could be replaced by robotic automation by 2030. But the deeper disruption is coming from AI's capacity to enter knowledge work: legal drafting, financial analysis, diagnostic medicine, software development. These are domains that previously seemed insulated by their cognitive complexity. The question is not whether the disruption will come. The question is what society does while it does.

The First Experiment

Speenhamland, 1795

The magistrates of Berkshire were not economists or philosophers. They were landowners responding to an emergency. The French Revolutionary Wars had disrupted grain imports; the harvest of 1795 had been poor; bread prices had surged beyond what agricultural labourers' wages could reach. People were hungry. The Speenhamland System, named after the village where the meeting was held, was their improvised answer.

The mechanism was simple. A sliding scale linked the subsistence payment to the price of bread and the size of the recipient's family. If a man's wages fell below the threshold for the current bread price, the parish made up the difference. For the first time, something like an income floor existed in English law — not as a right, but as a relief measure. It spread quickly across southern England.

The consequences were not what the magistrates intended. Because the parish would top up any wage to the subsistence level, employers discovered they could pay below subsistence, and the public poor rate would subsidise the difference. Farmworkers' wages stagnated, then fell, because there was no longer a market incentive to pay them fairly. The system that was meant to help the poor ended up transferring wealth from taxpayers to landowners, with the labourers caught in the mechanism as its beneficiaries in name and its victims in practice. What was designed as a floor became a ceiling.

The Speenhamland System was abolished in 1834, replaced by the New Poor Law — a punitive reform premised on the idea that poverty was a moral failure and relief should be made as unpleasant as possible. The workhouse became its instrument. It was, in retrospect, a social catastrophe dressed as fiscal discipline, and it took decades of progressive reform to undo its worst effects.

But the underlying problem Speenhamland addressed was real: a technological and agricultural shift so rapid that existing labour markets could not absorb displaced workers. In the early 19th century, that shift was mechanical threshing and enclosure. In the early 21st, it is machine learning and robotics. The parallels are imperfect but instructive: then as now, the core question was whether society could build a floor beneath the people the market had stopped needing.

The Modern Pilots

Finland, Stockton, Kenya

Between 2017 and 2023, three significant experiments tested a version of the Speenhamland impulse in the modern world, with rigorous controls, randomised designs, and enough scale to generate meaningful data. Their results challenge almost everything the political mainstream has assumed about what happens when you give people money unconditionally.

Finland's experiment ran from 2017 to 2018. The government randomly selected 2,000 unemployed people aged 25 to 58 and paid them €560 per month for two years, with no conditions attached. They compared outcomes to a matched control group of 175,000 people who received standard unemployment benefits. The employment results were essentially flat: the basic income neither increased nor reduced the likelihood of finding work, with the treated group working approximately six more days in the second year, a difference too small to be definitively attributed to the payment. But the wellbeing results were striking. Recipients reported substantially less depression (22.3% experienced at least two weeks of depression in the past year, compared to 32.4% in the control group), greater life satisfaction, more trust in institutions, and greater confidence in their own futures. The income floor, it turned out, did not make people idle. It made them less anxious.

Stockton, California ran a smaller but methodologically rigorous trial between 2019 and 2021 (the SEED programme, led by then-mayor Michael Tubbs). 125 residents selected from census tracts at or below the city's median household income received $500 per month for two years, unconditionally. The results confounded the standard objections. Full-time employment among recipients increased by 12 percentage points compared to the control group. Recipients moved from part-time to full-time work at more than twice the rate of those who received nothing. Income volatility fell. At the start of the programme, only one in four participants could cover a $400 unexpected expense. One year in, more than half could. Anxiety and depression fell measurably. The money, it turned out, gave people the stability to take the risks that lead to better employment.

Kenya's GiveDirectly experiment is the world's largest and longest basic income study, launched in 2016 across 195 villages involving approximately 23,000 individuals. The study compared three approaches: a 12-year guaranteed income of $22.50 per month; a 2-year version of the same amount; and a one-time lump sum of $500. The results yielded a finding that no economist had clearly predicted: the duration of the guarantee matters as much as the amount. Recipients who knew they would receive payments for 12 years outperformed those receiving a 2-year guarantee on nearly every measure, despite receiving the same amount of money in the first two years. The long horizon allowed them to plan, save, invest, and take entrepreneurial risks that a 2-year window did not justify. The security of the floor was not just financial — it was cognitive. It freed up mental space previously consumed by economic precarity.

Taken together, the three experiments suggest a coherent picture. A guaranteed income floor does not produce idleness. It produces the conditions for purposeful risk-taking. The threat of utter destitution, it turns out, is not a good motivator for economic participation. It is a good motivator for paralysis.

The Question After Work

What Fills the Void

The deepest challenge of a post-labour economy is not economic. Economists will eventually work out a mechanism for distributing the surplus that machines generate. The real question is anthropological: what does a human life mean when it is not organized around work?

For most of recorded history, labour was the primary structure through which people understood their identity, their status, their social belonging, and their purpose. The craftsman, the farmer, the merchant, the soldier. These were not just jobs. They were answers to the question of who a person was. The industrial revolution accelerated this fusion, tying wages to worth, hours to value, and employment to dignity in ways that were not natural but that came to feel inevitable. The Protestant work ethic, as Weber described it, made toil sacred: a form of virtue that distinguished the worthy from the idle, the deserving from the dependent.

UBI, at its most ambitious, does not merely solve an income problem. It proposes a renegotiation of that entire moral framework. If you are paid to exist, if your material security is decoupled from your economic output, then the question of how you spend your time becomes genuinely open. The data from Finland and Stockton suggests that most people, freed from the desperate scramble for survival, do not become passive. They pursue education, caregiving, creative work, civic engagement, the kinds of activity that markets have always undervalued because they don't generate measurable GDP.

But there is a harder version of the question. Not every person displaced by automation is a frustrated artist or an underutilised intellectual waiting for the chance to flourish. Some are people who found meaning specifically in the discipline and structure of physical work — in the satisfaction of making things, moving things, fixing things. A society that writes off that kind of meaning as expendable is not being enlightened. It is being careless.

The honest answer is that we do not yet know what the post-labour human looks like at scale. We have glimpses from the UBI pilots — more stability, less anxiety, more risk-taking, more community investment. We have the Speenhamland warning that a poorly designed floor can become a mechanism of exploitation. And we have the deeper warning of history: technological transitions that happen faster than social institutions can adapt have always produced violence, upheaval, and suffering, regardless of how much net wealth they eventually create.

The income floor is not a utopia. It is a commitment: however rapidly the machines advance, the human beings displaced by them will not be simply abandoned. That their fall will have a bottom. Whether we build that bottom wisely enough to avoid the Speenhamland trap is, perhaps, the defining policy question of the next twenty years.