Chapter 6 of 13

Locking It In

About 23 minutes to read

Fifteenth of the month

On 15 July 2021, money began to arrive in the bank accounts of tens of millions of American families. It came from the federal government, and it came every month, on or around the fifteenth: up to $300 for each child under six and $250 for each child between six and seventeen. The families did not have to apply for a job programme or prove that they were looking for work. Families with no earnings at all, who had been largely shut out of the old child tax credit because it was paid only against income tax owed, now received the full amount.

The payments came from a temporary expansion of the child tax credit, passed in March 2021 as part of a pandemic relief law, the American Rescue Plan. Its effect was measured with unusual precision. According to the Census Bureau's Supplemental Poverty Measure, which counts tax credits and other benefits as income, the share of American children living in poverty fell from 9.7 per cent in 2020 to 5.2 per cent in 2021, the lowest figure ever recorded. Census researchers attributed more than two million of the children lifted above the poverty line that year to the expanded credit alone. Surveys found that families spent the money mostly on food, rent, utilities, clothing and school costs.

The expansion was for one year. In December 2021 the last monthly payment went out. Attempts to extend it failed in the Senate, where the governing party's majority was a single vote and one of its own senators would not support the cost. In 2022 the child poverty rate more than doubled, to 12.4 per cent. About five million more children were poor than the year before. Nothing had changed about the children, their parents or the economy that could explain the difference. A gain had been won, measured, celebrated — and not locked in.

The first book's fifth force was the locking-in of gains into law and lasting institutions. It found that breakouts which were not locked in were reversed, often within a generation: the reforms of Uru-inimgina of Lagash, the Gracchi's land reforms, the Tudor laws against enclosure. Chapters 2 to 5 of this book have asked whether the other four forces will be available over the next fifty years. This chapter asks the question that decides whether any of them matter: since the financial crisis of 2008, which gains have lasted, which have not, and why?

Why some things last

Political scientists have studied why some reforms endure and others are quickly undone, and the most useful answer begins with an observation about the American and British welfare states in the 1980s.

Ronald Reagan and Margaret Thatcher both came to power determined to cut back the state. Both did a great deal: they cut taxes on the rich, weakened unions, privatised industries and reduced some benefits. But the political scientist Paul Pierson, in Dismantling the Welfare State? (1994), found that the core programmes of the welfare state — public pensions, health care, the large universal benefits — survived both governments largely intact. His explanation was that mature social programmes create their own defenders. Once millions of people depend on a pension or a health service, those people become a constituency that any government attacks at its peril. As the political scientist E. E. Schattschneider had put it half a century earlier, new policies create a new politics.

Britain's National Health Service is the clearest case. Founded in 1948, it has survived every government since, including those most hostile to public provision. Nigel Lawson, Thatcher's chancellor and no friend of the public sector, wrote in his memoirs that the NHS was "the closest thing the English people have to a religion". It has been underfunded, reorganised and partly opened to private contractors. It has never been abolished, and no serious party has proposed abolishing it.

The same mechanism protected the Affordable Care Act in the United States. Passed in 2010, it extended health insurance to some twenty million people. The party that opposed it made its repeal a central promise for seven years, and in 2017 it controlled the presidency and both houses of Congress. In the early hours of 28 July 2017, the Senate voted on a stripped-down repeal bill. It failed by 51 votes to 49, when three senators of the governing party, John McCain among them, voted against. By then the law had created millions of beneficiaries, and the hospitals, insurers and state governments that depended on it had become its defenders.

The political scientist Eric Patashnik, in Reforms at Risk (2008), studied why some American reforms of the late twentieth century stuck and others collapsed. He found that the durable ones had reshaped the interests around them: they created new groups with a stake in their survival, destroyed or weakened the groups that had benefited from the old arrangement, and built institutions that carried on the reform's work without needing a fresh political victory every year. The reforms that collapsed were those that left the old interests in place and depended on continuing goodwill.

The child tax credit failed every one of those tests. It was temporary by design, so it never had time to create a constituency that knew it would miss it. It had to be renewed by a fresh vote. And it was paid for out of the general budget, in competition with every other claim.

The same logic applies to the organisations of the 99 themselves. In Denmark, Sweden and Finland, union membership has remained far higher than in most rich countries — between half and two-thirds of employees, against an OECD average of around one in six. One important reason, identified by the political scientist Bo Rothstein, is the "Ghent system", named after the Belgian city where it began: unemployment insurance funds are run by or closely linked to the unions, so that joining a union and insuring oneself against unemployment go together. The lock is not a law guaranteeing membership but an institution that makes membership useful. When a Swedish government raised the fees for union-linked unemployment funds sharply in 2007, membership fell by several percentage points within two years, according to the sociologist Anders Kjellberg — a reminder that institutional locks can be loosened too, by anyone who understands how they work.

The great unlocking

The gains and losses since 2008 are best read against an older story: what happened to the gains of the decades after 1945, and which kinds of lock held.

The first book described the great compression of the mid-twentieth century, when the gap between the 1% and the 99 narrowed further and for longer than at any time since the Black Death. Several things held it in place. Top rates of income tax were very high: in the United States the highest rate stood at 91 per cent through the 1950s and at 70 per cent until 1981; in Britain the top rate on earned income reached 83 per cent in the 1970s. Unions were strong. And new public institutions — social insurance, public pensions, national health services, public housing — provided a floor.

Not all of these locks proved equally strong. Tax rates were the weakest. A rate is a number in a law that a single parliament can change in a single budget, and from 1979 onwards one government after another did. By 1988 the top rate of income tax was 28 per cent in the United States and 40 per cent in Britain. Union power proved stronger but was eroded over decades, partly through law: the American Taft–Hartley Act of 1947 allowed states to pass "right-to-work" laws that weakened unions' finances, and by the 2010s more than half the states had done so. (In 2023 Michigan became the first state in more than half a century to repeal one.) The institutions — Social Security, the NHS, public pensions across Europe — proved strongest. They were cut at the edges, but they survived the very governments that most wanted to shrink them.

The pattern is the one Pierson and Patashnik identified. Numbers in a budget are easy to change. Institutions that millions of people use every day are not.

What has not lasted

The years since 2008 offer a long list of gains that were won and then partly or wholly lost, and an equally instructive list of losses that the owners locked in.

After the financial crisis, the United States passed the Dodd–Frank Act of 2010, the most extensive reform of financial regulation since the 1930s. It imposed stricter supervision on banks large enough to threaten the whole system. In 2018 Congress raised the threshold for the strictest supervision from $50 billion of assets to $250 billion. Several banks that the change freed from the tightest oversight, including Silicon Valley Bank, failed or had to be rescued in 2023. The rules had not been repealed; they had been narrowed, quietly, a few years after the crisis that produced them had faded from memory.

In Alaska, as Chapter 1 noted, the state has paid every resident a dividend from its oil fund since 1982, and the fund itself is protected by an amendment to the state constitution approved by voters in 1976. But the constitution protected the fund, not the dividend. For decades the size of the dividend was set by a formula in ordinary law. In 2016, faced with a budget crisis, the governor used his veto to cut the dividend roughly in half, and in 2017 the Alaska Supreme Court upheld his power to do so. Since then the formula has remained on the books but the legislature has set the dividend each year in the budget, usually well below what the formula would pay. The capital was locked in. The payment to citizens was not.

Norway shows how the capital can be protected without a constitutional amendment. Since 1990 the state has put its oil revenues into a fund, now worth well over a trillion dollars, invested abroad and managed at arm's length from the government. Since 2001 a fiscal rule, agreed across the main parties, has limited the amount that governments may spend from it each year to the fund's expected real return — four per cent at first, three per cent since 2017. The rule is not in the constitution, and any parliament could change it. It has held through governments of left and right for more than two decades, because every major party has come to regard it as the condition of the fund's survival, and because voters treat spending beyond it as theft from their children. Chapter 9 will examine the fund as a model of public ownership. Here it matters as a model of locking in by consensus and habit rather than by law alone.

In September 2021, at the same election in which they chose a new state parliament, voters in Berlin approved by 59.1 per cent a referendum calling on the city government to take into public ownership the apartments of every private landlord owning more than three thousand homes in the city — about a quarter of a million flats. The referendum was not binding. The city appointed an expert commission, which reported in 2023 that such a socialisation was compatible with Germany's constitution, even with compensation below market value. The law the voters had asked for was not passed. A clear majority had spoken, and nothing had been locked in, because the vote did not itself create a law, an institution or anyone with the duty to act on it.

The owners, meanwhile, have shown how locking in is done. The tax cuts passed in the United States in 2001 and 2003 were written to expire after ten years; in 2012 most of them were made permanent. The tax cuts of 2017 were written so that the cuts for individuals would expire at the end of 2025; in July 2025 most of them were made permanent, and the amount that a person can leave to heirs free of federal estate tax was raised permanently to $15 million, indexed to inflation — $30 million for a married couple. A temporary measure creates a cliff, and whoever stands to lose at the cliff edge becomes a lobby for permanence. The owners have learned to use the cliff. The 99, in the case of the child tax credit, were pushed off it.

Taxing great fortunes

No gain matters more for this book than the taxation of great wealth, because the first book found that inherited wealth was the single surest way in which the farm reproduced itself across generations. The record since 2008 is mostly one of failure, and the reasons are instructive.

According to the OECD, twelve of its member countries levied an annual tax on individuals' net wealth in 1990. By 2017 only four did: France, Norway, Spain and Switzerland. The next year France replaced its wealth tax with a narrower tax on property. The OECD found that many of the abolished wealth taxes had raised little, had been riddled with exemptions, and had encouraged the wealthy to move their assets or themselves abroad. Inheritance taxes have followed a similar path: Sweden abolished its inheritance tax in 2004, Austria in 2008 and Norway in 2014.

The power of exit was shown again in Norway. In 2022 a new government raised the wealth tax for the richest and increased the tax on dividends. That year at least thirty of Norway's wealthiest people moved to Switzerland, among them the industrialist Kjell Inge Røkke — more, by one newspaper's count, than in the previous thirteen years combined. Economists still debate how much revenue Norway actually lost; some argue that the departures were small relative to the tax base and that the tax continued to raise money. But the episode showed that a single country taxing great fortunes must reckon with the fortunes' ability to leave.

Switzerland, the other surviving case, illustrates a different lesson. Its cantons have taxed net wealth for well over a century, and those taxes raise a larger share of revenue than wealth taxes have anywhere else in the OECD. They are low in rate, apply to a broad base, and have become an accepted part of the Swiss system. But Swiss voters have also shown the limits of what they will accept. In November 2025 they rejected, by 78.3 per cent, an initiative by the youth wing of the Social Democrats that would have imposed a fifty per cent federal tax on inheritances above fifty million francs. During the campaign, several prominent business owners said they would leave the country if it passed.

Efforts to coordinate the taxation of great fortunes internationally have begun, but only in words. In 2024 Brazil, then chairing the G20, commissioned the French economist Gabriel Zucman to design a minimum tax on billionaires. His report estimated that the world's roughly three thousand billionaires paid personal income tax equivalent to about 0.3 per cent of their wealth each year, and that a minimum tax of two per cent of their wealth could raise $200 to $250 billion a year. The G20's finance ministers and then its leaders declared that they would "seek to engage cooperatively to ensure that ultra-high-net-worth individuals are effectively taxed", but committed to no specific measure; the United States and Germany were among the governments opposed. In France, a bill to impose a two per cent minimum tax on fortunes above a hundred million euros passed the National Assembly in February 2025 and was rejected by the Senate in June.

The lesson is consistent. Taxes on great fortunes have been hardest to lock in where the fortunes could leave, where the tax was new and the beneficiaries were diffuse, and where the owners could argue that their departure would cost everyone. They have lasted where, as in Switzerland, the rate was modest, the base broad and the tax old enough to be part of the furniture.

The end of secrecy

There is one large exception to this record of failure, and it shows what can be done when the conditions are right.

In 2007 Bradley Birkenfeld, an American banker working for UBS in Geneva, went to the United States authorities with an account of how the bank had helped thousands of wealthy Americans hide money from the tax authorities. His information — and his own guilty plea to conspiracy, for which he served more than two years in prison — led to one of the most consequential tax cases in history. In 2009 UBS admitted wrongdoing in a deferred prosecution agreement, paid $780 million and agreed to hand over information on thousands of American accounts, breaching the Swiss tradition of banking secrecy that had lasted since the 1930s. The Internal Revenue Service later paid Birkenfeld a whistleblower award of $104 million.

In 2010 the United States passed the Foreign Account Tax Compliance Act, which required foreign banks to report the accounts of American citizens to the American tax authorities or face a thirty per cent withholding tax on payments from the United States. Few banks could afford to be cut off from the world's largest financial market. Other countries then demanded the same treatment, and in 2014 the OECD adopted a Common Reporting Standard under which participating countries' banks report accounts held by foreigners to the tax authorities of the account holders' home countries, automatically and every year. The first exchanges took place in 2017. Today more than a hundred jurisdictions take part, including Switzerland, Luxembourg, Singapore and most of the old havens.

The effect, according to the Global Tax Evasion Report published by the EU Tax Observatory in 2023, has been large. The report estimated that offshore tax evasion by wealthy individuals had fallen by roughly a factor of three in a decade: before the reforms, the great majority of wealth held in offshore accounts went undeclared to the tax authorities; afterwards, most of it was declared. Evasion has not ended — some banks do not comply, and some wealth has moved into assets the system does not cover well, such as property — but it is a lasting change.

Why did this gain last when so many others did not? It met almost every condition that Pierson and Patashnik identified. It was built into an institution — the routine reporting systems of thousands of banks — rather than depending on a fresh political decision every year. It was automatic. It was coordinated across borders, so that the wealth could not simply move to the next haven. And it was enforced by the leverage of the largest market in the world. The end of secrecy closed an exit.

The taxation of companies' profits shows what happens when such conditions are only half met. The EU Tax Observatory estimated that multinational companies shifted about a trillion dollars of profits to tax havens in 2022. In October 2021 some 136 countries agreed to a global minimum tax of fifteen per cent on the profits of large multinationals, and the European Union and many others began to apply it in 2024. In June 2025, however, the G7 agreed that companies headquartered in the United States would be excluded from the main rules that would have enforced the minimum tax on their profits, in return for the United States dropping a proposed retaliatory tax on foreign investors. The world's largest group of multinationals had obtained an exit from an agreement designed to close one.

Breaking up and opening up

Two other kinds of gain since 2008 test the question of durability in a different way: the use of competition law against the largest companies, and the creation of rights over personal data.

The European Union's General Data Protection Regulation, which came into force in 2018, has proved durable. It gave every person in the EU rights to see, correct and delete data held about them, and to object to its use; its enforcement has been slow, but it has produced large fines, including a fine of €1.2 billion against Meta in 2023 over transfers of Europeans' data to the United States. More importantly for this book, it has become the template for data protection laws in many other countries, and, as Chapter 4 showed, it gave drivers in Amsterdam the legal tool to make Uber explain its algorithms. A right held by every individual, enforced by independent regulators and copied across the world, is hard to repeal.

Competition law is less certain. In 2024 and 2025 American courts found that Google had illegally monopolised both internet search and key parts of the technology for selling online advertising. In both cases the judges declined to order the company to sell off the businesses at issue, preferring "behavioural" remedies: bans on certain contracts, requirements to share data or open systems to competitors, and technical monitors to oversee compliance. Google said it would appeal.

The history of antitrust suggests that the difference between structural and behavioural remedies is the difference between a gain that locks itself in and one that must be defended for ever. When the Supreme Court ordered the break-up of Standard Oil in 1911, and when the American government broke up the telephone monopoly AT&T in 1984, the result was a permanent change in who owned what. A rule against a monopoly's conduct, by contrast, lasts only as long as there are regulators willing and able to enforce it, and the history of regulation is full of agencies that were weakened, captured or simply starved of funds once public attention moved elsewhere.

Constitutions and the dead hand

The strongest form of locking in is a constitution, and the evidence on constitutional protection is mixed in a way that matters a great deal for this book's design.

Switzerland offers examples of both kinds. In March 2013 Swiss voters approved by a large majority the so-called "fat cat" initiative, proposed by a businessman named Thomas Minder, which wrote into the constitution a binding vote for shareholders on executive pay and banned golden handshakes. It remains in force. Eight months later, by 65.3 per cent, they rejected the "1:12" initiative of the Social Democrats' youth wing, which would have capped the pay of every company's highest earner at twelve times that of its lowest. The Swiss had been willing to give owners power over their managers' pay. They were not willing to limit it.

Germany's constitution goes further than most. Article 14 of the Basic Law of 1949 guarantees property, but adds: "Property entails obligations. Its use shall also serve the public good." Article 15 permits land, natural resources and means of production to be transferred to public ownership, with compensation, for the purpose of socialisation. It is the clause on which the Berlin referendum relied. It has never been used. A constitutional permission is not the same as a constitutional duty.

Chile offers a warning from the other direction: about trying to lock in too much at once. After mass protests in 2019 over inequality and the cost of living, Chileans voted by a large majority in 2020 to replace the constitution inherited from the Pinochet dictatorship. An elected convention drafted one of the most ambitious constitutions in the world, with extensive social rights, environmental protections and changes to the structure of the state. In September 2022 voters rejected it by 62 per cent. A second, more conservative draft was rejected in 2023 as well. The old constitution remained. A majority had wanted change; it had not wanted everything in one text, decided at one moment, and locked in for good.

There is also a deeper problem, and it bears directly on the design under test. A gain locked into a constitution is protected against the owners. It is also protected against the future. In 1789 Thomas Jefferson wrote to James Madison that "the earth belongs in usufruct to the living": that no generation had the right to bind the next with its debts or its laws, and that every constitution should therefore expire after about nineteen years. Madison replied that a society which remade its fundamental law every generation would be unstable and would invite faction. The philosopher Jon Elster later used the image of Ulysses, who had himself tied to the mast so that he could hear the Sirens without steering onto the rocks, to describe the value of binding one's future self — and the danger that the ropes might hold when the danger was past.

The first book found that the farm's own fences were very often laws that had once been locked in by the owners: the Roman laws of property, the Enclosure Acts, the slave codes, the restrictions on combination that made unions illegal. A society that locked in the Federation's design too tightly would face the fourth test of this book in its sharpest form. If people cannot change it, they have no voice in it; and a design that its members cannot alter is, in the first book's terms, a better-run farm.

What makes a gain last

Gathering the evidence of this chapter, six conditions stand out. Gains since 2008, and in the first book's longer history, have tended to last when they met most of them.

They are universal and felt. The NHS and the Affordable Care Act lasted because millions used them; the child tax credit disappeared before it had become a habit. Benefits for everyone build broader constituencies than benefits for the poor alone.

They are automatic. Bank reporting under the Common Reporting Standard happens every year without a vote. Germany's minimum wage, introduced in 2015 at €8.50 an hour, is reviewed by a standing commission of employers and unions and reached €13.90 in 2026. A gain that must be renewed by a new political victory each year is always one election from loss.

They change structure, not only conduct. Breaking up a monopoly or transferring ownership lasts; rules that tell an owner how to behave last only as long as their enforcers.

They close the exit. Taxes and rules that wealth can escape by moving have been hardest to keep. Those coordinated across borders, or backed by the leverage of a large market, have held.

They have their own defenders and enforcers. Independent regulators, standing commissions, unions and the beneficiaries themselves keep a gain alive when public attention moves on. A referendum that creates no institution, as in Berlin, defends nothing.

They protect the capital, not only the payment. Alaska locked in its fund but not its dividend, and the dividend was cut. What is protected is what lasts.

To these, the problem of the dead hand adds a seventh, which pulls against the others: a gain should be hard for any one faction to reverse, but possible for the people as a whole to revise. The tools for that exist — supermajorities, votes that must be repeated after an election, citizens' assemblies, sunset clauses that bind only the owners' privileges and not the floor.

The tests

This chapter tests the durability of the whole design, and in particular three elements: the floor nobody falls below, the dividend that makes every person a shareholder, and the public ownership of the machines. It also tests the first element, one borderless society, against the evidence on what happens when gains can be escaped across borders.

Precedent. Gains that created constituencies and institutions have lasted for generations: the NHS, Social Security, the Swiss cantonal wealth taxes, the end of bank secrecy. Gains that were temporary, depended on annual renewal, or could be escaped across borders have been reversed: the child tax credit, much of Dodd–Frank's toughest supervision, most of the OECD's wealth taxes, the full Alaska dividend. The owners have locked in their own gains by the same methods.

Evidence. The research of Pierson and Patashnik, the record of automatic information exchange, and the post-2008 experience of financial regulation, wealth taxation and competition law all point to the same conditions for durability: universality, automaticity, structural change, closed exits and institutional defenders.

Capture. The commonest way a gain is lost is not repeal but erosion: a threshold raised, a formula ignored, an agency starved, a carve-out negotiated for the largest firms. Any lock the Federation builds must be protected against the slow capture of the institutions that hold it.

Exit. Two kinds of exit pull in opposite directions. The owners' exit — of wealth across borders — has been the most effective way to escape gains; closing it requires international coordination. The citizens' exit — the right of a future generation to change the rules — must be kept open, or the lock becomes a fence.

The ledger

What survived. The floor, the dividend and public ownership all survive as things that can, on the evidence, be made to last — but only in particular forms. The end of bank secrecy shows that a lasting change in the treatment of great wealth is possible when it is automatic, coordinated and enforced by leverage. The first element of the design, one borderless society, gains support: the most durable gains against the owners were those that crossed borders, and the most common failures were those that wealth could escape by crossing them.

What had to change. The design must lock in its floor and its dividend in a specific way. The dividend must be paid from capital whose ownership by everyone is constitutionally protected, according to a rule that runs automatically and cannot be cut by a single veto or budget; Alaska shows what happens otherwise. The floor must be universal, not means-tested, so that it is defended by everyone who receives it. The public ownership of the machines must be structural — a transfer of ownership, not a set of rules about how private owners should behave — because rules need enforcers for ever and enforcers can be captured. Every lock must have its own defenders: independent institutions and the organised members of Chapter 4. And the design needs a rule for its own revision: hard for any faction to undo, open to change by the people as a whole through slow and deliberate processes.

What remains unknown. Whether international coordination on taxing great wealth will move beyond declarations. Whether structural remedies against the largest technology companies will ever be ordered, and whether they would last if they were. And whether any society can lock in a design firmly enough to withstand the owners, and loosely enough to leave the next generation free — the question Jefferson and Madison argued over, and which no constitution has fully settled.

Bridge: what people believe

The NHS lasted not only because it created a constituency, but because people came to believe in it — in Lawson's word, as something close to a religion. Every lock described in this chapter ultimately depends on belief: on people thinking that the floor is a right rather than a favour, that great fortunes should be taxed, that the machines can belong to everyone. When belief moves, laws follow, in both directions; the welfare state was built on one story about who deserves what, and taken apart in part by another. The first book called the story the deepest fence of all. The next chapter opens Part III by asking what is happening to the story that holds the present farm together — meritocracy — and what a story worth living by would have to contain.