Essay #05
The distribution of ownership
On 4 July 2026, every baby born in the United States since the start of 2025 became an investor. Under a law passed the year before, the federal government pays 1,000 dollars into an account for each of them, invested in a low-cost fund that tracks the American stock market, and on that day the accounts went live. The sums are small, but the idea behind them is large, and I think it is the beginning of an answer to what will be the political question of the 2030s.
My view is optimistic about supply and pessimistic about distribution. Put precisely: somewhere in the 2030s, redistribution by taxation will be tried, and it will fail. What can work instead is ownership.
What is breaking. Modern economies distribute wealth through tasks. Work is cut into tasks, and income is paid for performing them. Do X and you are paid Y. That one line is the whole of how a modern economy shares out what it makes.
The line began to fail before AI. The labor share of income has fallen since the early 1980s across most countries and industries, and about half of the fall is explained by machines and computers getting cheaper and firms replacing people with them. Automation always has two effects. It hands human tasks to machines, which lowers the labor share, and it creates new tasks for humans, which raises it. In the United States of recent years the first accelerated and the second weakened.
In the occupations AI reaches most, employment of workers aged 22 to 25 has fallen relative to other age groups while employment of experienced workers has held. It came through a halt in hiring, not through layoffs. The measured gap stood at 19 percent as of June 2026. That finding is recent and contested; a study of the same question in Danish administrative records attributes only a small fraction of the decline it measures to AI. What is not contested is where the break sits: at the entrance, at the young end.
Why the obvious remedies do not hold. The political demand for redistribution is rising. In November 2025, candidates campaigning as democratic socialists won the mayoralties of New York and Seattle, and Gallup found favorable views of capitalism among Americans down to 54 percent.
Wealth taxes and cash transfers have been tried, and they fail for a duller reason than the old worry that high rates stop people working. The wealth being taxed slips away. Owners get exemptions written in, have assets valued low, or hide them, long before any of them moves house.
Figure 6-2 The graveyard of the wealth tax
| Country | Abolished | Note |
|---|---|---|
| Austria | 1994 | |
| Denmark | 1997 | |
| Germany | 1997 | |
| Netherlands | 2001 | |
| Finland | 2006 | |
| Iceland | 2006 | Abolished 2006; reintroduced 2010–14 |
| Luxembourg | 2006 | |
| Sweden | 2007 | |
| France | 2018 | Abolished 2018; replaced by a property-only tax |
| Norway | still in force | |
| Spain | still in force | suspended 2008, reinstated 2011 |
| Switzerland | still in force |
Each bar is one country's individual net wealth tax. Black bars are still in force.
Source: Perret (2021), Fiscal Studies; OECD (2018), The Role and Design of Net Wealth Taxes in the OECD.
Around 1990, twelve OECD countries taxed net wealth. Three still do. France's wealth solidarity tax was cut back in 2018 to a tax on real estate alone, after decades in which the taxable wealth drifted out of reach. Where wealth and its owners can move, taxing it country by country has not been sustained.
Cash transfers have the best experimental evidence available, and it points the same way. A thousand people were given a thousand dollars a month for three years and compared with two thousand who received fifty. Recipients worked slightly less, were slightly less likely to be employed, and the gains in stress and food security faded after the first year. Cash buys autonomy and immediate peace of mind. A cheque does not change who owns the capital, and ownership is the variable that is moving.
The design. So move the ground of distribution from tasks to ownership. If the labor share is falling for structural reasons, households can hold their share of income only by owning a piece of the capital that is replacing them. Give every citizen a stake in automated capital from birth. Every citizen becomes an investor before becoming a worker.
The earliest versions already exist. The American accounts described at the top of this essay cover every child born between 2025 and 2028. Connecticut's baby bonds do the same at state level. Alaska has paid every resident a dividend from oil for four decades with no measurable effect on employment, and Norway's sovereign fund holds roughly 380,000 dollars per citizen. The proposal here is to take that design to the scale the coming decades will require.
Nor is it a new kind of institution. Broad middle classes have always been made by turning citizens into owners. The Homestead Act moved about a tenth of American territory into private hands. Postwar Japan moved 1.74 million hectares to 4.75 million tenant households and cut the share of farmland worked by tenants from nearly half to a tenth. If the nineteenth century distributed land and the twentieth distributed home ownership and pensions, what the twenty-first has to distribute is a stake in automated capital.
Where would the capital come from? Not mainly from taxing wealth that moves; that would rebuild the failure the design is meant to escape. It would come from three places: public seed money paid at birth and left to compound; a share of the new capital itself, taken by the state as equity in the energy, computing and automation plant it permits and connects, as Norway did with its oil, and placed in a citizens' fund rather than the treasury; and voluntary contributions from firms and wealthy individuals, whose motive need not be charity, since the firms that own the plant will need customers who can pay.
The same technology also changes who can own a business. Owning one used to mean hiring people and carrying responsibility for their lives, and that fixed cost set the smallest size at which a firm could survive. AI and robotics shrink that minimum, so more people can own businesses. But if the threshold falls with no design for spreading ownership, what multiplies is people who run a business and own no capital.
Why the hurry. Institutions and population run on different clocks. Institutions of distribution can be made by legislation, so once the decision is taken they move within a few years. Population cannot be legislated into place: a child born now is a customer from the first day, a taxpayer in twenty years and a founder, if ever, in forty. To institutionalize distribution in the 2030s is to buy the customers, the founders and the taxpayers of the 2050s now. Legislation can settle where income comes from, but it cannot shorten the time a child takes to grow up, which is why the design has to start now.
Where this ends, I think, is with investors paid for the use of what they funded rather than for owning the company that built it. Once it is cheap to record that someone actually benefited from a thing, the payment can go straight to whoever funded the thing, without passing through whoever built it.
The three terms, institutions, energy and labor input, now sit in different countries under different owners. Where to stand among them is the subject of the thesis itself.