On 15 December 2025, Keir Starmer sat before the House of Commons Liaison Committee and described what happened when he tried to exercise power:

“My experience now as prime minister is of frustration that every time I go to pull a lever there are a whole bunch of regulations, consultations, arm’s-length bodies that mean that the action from pulling the lever to delivery is longer than I think it ought to be.”

Eighteen months after winning a Commons majority of 174, Starmer had discovered that control of Parliament did not necessarily translate into control of the state.

Beyond Starmer’s personal leadership challenges, Britain has become strangely difficult to govern. Governments struggle to build infrastructure, reform public services, reduce immigration or deliver major projects on time. Every proposal seems to encounter another institution, another legal process, another regulator or another veto.

Commentators blame weak leaders, Brexit, Covid or simple incompetence. While these factors matter, they do not answer the deeper constitutional question: how did Britain reach the point where prime ministers can command Parliament yet no longer reliably command the state?

Four decades earlier…

On 11 June 1987, Margaret Thatcher won her third consecutive general election with 42.2% of the vote. Eight years after entering Downing Street, she had already privatised British Telecom, British Gas, British Airways, Jaguar and Rolls-Royce, while beginning the sale of British Steel and many other state-owned industries. Around 1.75 million people had worked in the nationalised industries she inherited in 1979; together they accounted for roughly one-tenth of Britain’s GDP. By the end of the decade, much of that settlement had disappeared.

Inside the Labour Party, the election result prompted a different conversation. If one Prime Minister, elected by 42% of voters, could reshape so much of Britain in less than a decade, perhaps the issue extended beyond Margaret Thatcher herself.

Seventeen months later, after almost a year of drafting, 348 writers, academics and political figures published the declaration that launched Charter 88. It called for a Bill of Rights, devolution, House of Lords reform, freedom of information and a written constitution.

The movement reached the Labour leadership in March 1993. Speaking to Charter 88, John Smith reflected on fourteen years of Conservative government.

“We do have an elective dictatorship. I used to myself believe in the sort of mysteries of the British Constitution. My experience over the last ten, eleven, twelve years has caused me to change my mind quite fundamentally.”

By the time Smith died in May 1994, others had already developed much of the constitutional programme that later became associated with Tony Blair. Donald Dewar had spent decades working towards Scottish devolution. Derry Irvine and a growing circle of Labour lawyers were developing proposals on rights, judicial reform and constitutional restraint. When Blair inherited the leadership of the Labour Party, he also inherited a constitutional project that had been forged throughout the Thatcher years.

The dispersal of power

By 1997, Labour’s constitutional programme had accumulated through nearly two decades of opposition. Blair supplied the parliamentary majority, the political momentum and an appetite for rapid institutional change. During his first term alone, Parliament passed the Bank of England Act, Human Rights Act, Scotland Act, House of Lords Act, Greater London Authority Act and Political Parties, Elections and Referendums Act, and more.

The traditional Westminster constitution concentrated power in one place. Unlike the USA with its three co-equal branches of government, and its checks and balances, in Britain Parliament was sovereign, and the powers of a popular PM with a large majority were essentially absolute.

Blair’s reforms dispersed that power in four directions: upwards to European institutions, downwards to devolved governments, sideways to independent bodies, and inwards into a new layer of constitutional law.

Upwards

Blair entered office after twenty-four years of British membership of the European project. His contribution came through further treaties, wider European competence and an increasingly dense body of law whose authority prevailed over conflicting domestic legislation.

The Amsterdam Treaty, signed in October 1997, expanded EU action across employment, social policy, immigration, asylum and border controls. Britain retained opt-outs in several justice and home affairs areas, while the treaty extended qualified-majority voting, and removed the national veto: Britain could vote against a proposal in the Council and still become legally bound by the result.

European law shaped agriculture, fisheries, environmental standards, competition, state aid, employment rights, public procurement and large parts of trade policy. Regulations applied directly; directives required domestic implementation. Decisions of the European Court of Justice determined how those obligations operated, while the European Communities Act gave European law priority where it conflicted with Acts of Parliament.

Between 1997 and 2009, around 6.8% of UK primary legislation and 14.1% of secondary legislation played some role in implementing European obligations. The government’s own estimate placed the EU origin of legislation carrying a significant economic impact at around 50%. Both measures omit much directly applicable European regulation and vary enormously according to what counts as a law. Their combined message remains clear: ministers worked across broad areas where many of the governing rules had already been settled above the national level.

When David Cameron’s director of strategy, Steve Hilton, arrived in Downing Street he expected the government to implement the programme it had just won an election to deliver. He later recalled the response that repeatedly came back from Whitehall:

“You can’t do that because the EU rules or directives stop you.”

Blair also attached Britain to the eastern enlargement that brought ten new states into the EU in 2004, while opening the British labour market immediately rather than using the transitional restrictions adopted by France and Germany. EU free movement then determined the basic rules governing migration from the new member states.

Across a growing range of subjects, that government inherited laws it could change only through negotiation with other states and European institutions.

Downwards

Blair’s second movement of power travelled towards Edinburgh, Cardiff, Belfast and London.

The Scotland Act 1998 transferred control over health, education, justice, local government, transport, agriculture and large parts of economic development to a new Scottish Parliament. Wales received a more limited settlement, initially based on secondary legislative powers. The Northern Ireland Act implemented the institutional structure of the Good Friday Agreement. London gained an elected mayor and assembly with responsibilities spanning transport, policing strategy, planning and economic development.

Blair paired this spending autonomy with remarkably little revenue autonomy. The original Scottish Parliament received one major fiscal lever: the ability to vary the basic rate of income tax by up to three percentage points. Holyrood never used it. The block grant from Westminster continued to fund almost the whole devolved budget, allowing Scottish ministers to decide how money was spent while the UK government remained responsible for raising most of it.

Later governments expanded Scottish tax powers considerably. By 2025-26, devolved funding sources were forecast to provide 49% of the Scottish Government’s day-to-day budget. That figure reflects reforms made principally after Blair, including wider control over income tax and the devolution of transaction and landfill taxes. At the start of the system, the corresponding share stood at roughly 11%.

Germany demonstrates the distance between Blair’s settlement and mature fiscal federalism. Its Länder receive 42.5% of personal income tax, 50% of corporation tax and, in 2022, 50.5% of VAT through constitutionally structured revenue-sharing arrangements.

Under Blair’s model, Holyrood could announce more generous services and attribute fiscal pressure to Westminster. Westminster could increase the block grant and attribute poor outcomes to Holyrood. Voters now had a blurred answer to the most basic democratic question: who paid for it?

That design fits devolution’s political origins. Blair regarded it as a means of accommodating Scottish national identity, preserving the Union and completing the Northern Irish peace settlement. The settlement delivered autonomy over services while retaining the Treasury as the financial centre of the state.

Power moved closer geographically. Accountability fragmented institutionally.

Sideways

The Bank of England received operational control over interest rates within days of Labour taking office. Ofcom took charge of broadcasting and telecommunications regulation. Natural England became the statutory guardian of the natural environment. The Judicial Appointments Commission assumed a central role in selecting judges. Hundreds of executive agencies, public corporations, commissions and non-departmental public bodies administered policy through statutory duties that ministers could influence, fund or amend, while often lacking the power to direct individual decisions.

Together, they created a quango state in which ministers remained answerable for outcomes produced by institutions operating under their own legal duties.

Natural England supplies a particularly clear example. European habitats law, retained domestic regulations and court decisions required new housing in affected catchments to demonstrate “nutrient neutrality”. Natural England issued advice to local planning authorities identifying where the requirement applied. Councils then withheld permissions until developers could prove that a project would add no net nutrients to already polluted waterways.

By 2023, the Home Builders Federation estimated that the policy had held up 160,000 homes and reduced annual delivery by around 41,000. The government accepted that roughly 140,000 proposed homes would otherwise have been subject to the rules and announced legislation to remove housing development from the requirement. The House of Lords defeated the change. Natural England’s interpretation therefore continued to shape housing supply across dozens of local authorities, despite a government committed to increasing construction.

Ofcom exercises a similar form of delegated authority across communications. Its decisions determine broadcasting licences, impartiality rules, competition requirements, spectrum access and, more recently, the systems online platforms must operate. A minister may set broad policy or bring new legislation. Ofcom applies its statutory duties to individual companies and cases without receiving day-to-day instructions from the Secretary of State.

This structure dispersed blame as effectively as power. During the Somerset floods of 2014, ministers, the Environment Agency and Defra passed responsibility among one another while communities tried to establish who had chosen the underlying flood-management priorities. An Institute for Government review used the episode to illustrate the central problem of arm’s-length government: public bodies possess operational autonomy, ministers retain political responsibility, and the boundary between the two becomes clearest only after failure.

Different classifications have produced totals ranging from fewer than 300 formal arm’s-length bodies to more than 400 public bodies and regulators. Their combined expenditure includes enormous programmes such as NHS funding, making headline cost figures misleading. Their constitutional significance lies in the number of decisions placed behind statutory walls: important enough for government to answer for, independent enough for ministers to disclaim direct control.

Within

The final transfer took place inside the law itself.

A small group of Acts can change the conditions under which later laws are written, interpreted and administered.

The Human Rights Act’s provisions operate across the statute book. Section 3 requires courts to interpret all legislation, past and future, compatibly with Convention rights wherever possible. Section 4 allows senior courts to declare primary legislation incompatible. Section 10 permits ministers to use a remedial order to amend legislation following a declaration of incompatibility or an adverse Strasbourg judgment. Section 19 requires the minister introducing every government Bill to state whether its provisions comply with the Convention.

Cabinet Office guidance requires departmental lawyers to prepare a legal issues memorandum before a Bill is introduced. That memorandum must examine ECHR compatibility, alongside the Bill’s exposure to legal challenge. Parliament’s Joint Committee on Human Rights examines every government Bill and can seek written explanations, expert evidence and changes where it identifies a significant risk.

A policy proposal therefore enters Whitehall carrying a second question alongside cost, practicality and public support: will a court regard it as a proportionate interference with a Convention right?

The system reaches beyond legislation. Every public authority - government departments, councils, police forces, prisons, regulators and much of the NHS - must exercise its powers compatibly with Convention rights. A broad statutory discretion can therefore become a narrower operational power once officials account for the HRA and subsequent judgments.

Courts also began recognising a category of “constitutional statutes”. In the 2002 Thoburn judgment, Lord Justice Laws included the Human Rights Act, Scotland Act and European Communities Act among laws carrying a special constitutional status. His formulation distinguished them from ordinary Acts: an ordinary statute could be displaced implicitly by a later inconsistent statute, while a constitutional statute required express words or an unmistakably specific intention. Erskine May now records that principle in its account of parliamentary law.

The HRA also connected domestic policy to an external and developing body of jurisprudence. British courts were required to take Strasbourg judgments into account, while adverse decisions generated political and legal pressure for domestic compliance.

The HRA established a legislative pattern that Brown expanded on: the Climate Change Act 2008 and Equality Act 2010 had the same roots throughout the system, creating an ever thicker web of legal obligations through which every policy now has to pass before it reaches parliament, and again upon implementation.

Other Blair reforms reinforced the same inward movement. For example, the Constitutional Reform Act placed judicial independence under statutory protection, reconfigured the office of Lord Chancellor, created the Judicial Appointments Commission and provided for a Supreme Court physically and institutionally separate from Parliament.

The traditional constitution relied heavily on political restraint exercised within Parliament. Blair’s settlement increasingly expressed restraint through statutes, legal duties and institutions able to interpret them.

Each reform carried a separate justification. Their accumulated effect left Westminster responsible for the country while reducing the range of decisions that a parliamentary majority could make quickly, directly and alone.

Fear, Faith and Control

Perhaps the primary motive for Blair’s constitutional carnage came from Labour’s experience of Thatcher: a fear of concentrated power in the hands of a PM commanding the Commons. Blair inherited that conclusion from John Smith, Charter 88 and the constitutional reformers around his party. His own account of devolution shows that dispersing power was often instrumental rather than a deep commitment to local democracy. Blair said,

“I was never a passionate devolutionist. It is a dangerous game to play. You can never be sure where nationalist sentiment ends and separatist sentiment begins.”

He accepted a Scottish Parliament because he believed it would preserve the Union; he accepted the wider constitutional programme because Labour had spent eighteen years learning to fear an unrestricted Westminster government.

Blair added a second instinct of his own: confidence that specialists could remove important decisions from political distortion. “What matters is what works” became New Labour’s governing shorthand, treating many political disputes as practical problems with discoverable answers. This ran through Bank independence, regulators, quangos and judicial appointments. Elected politicians would set broad objectives; experts would apply evidence, professional standards and supposedly neutral judgement. Every transfer also insulated the decision from voters, because ministerial authority forms the mechanism through which electoral choices normally reach the state.

A third instinct appeared when popular opinion threatened a project Blair considered correct. In April 2004, after months of resistance, he promised a referendum on the proposed European Constitution and challenged its opponents: “Let the battle be joined.” Labour renewed that commitment in its 2005 manifesto. French and Dutch voters then rejected the Constitution, while British polls pointed heavily towards defeat. Most of its institutional substance returned through the Lisbon Treaty, negotiated at Blair’s final European summit and ratified under Gordon Brown without the promised vote, on the grounds that it constituted a different document. Blair trusted referendums that legitimised devolution and London government; when a referendum threatened European integration, he fell back on weasel words to claim a legal distinction to replace popular consent. His constitution therefore rested on two connected suspicions: that elected governments might exercise too much power, and that voters might use direct power to choose the wrong answer.

What Blair got right

For all his constitutional changes, two stand clear as long-term successes.

First was Bank of England independence. On 6 May 1997, four days after the election, Gordon Brown transferred responsibility for setting interest rates from the Chancellor to the Monetary Policy Committee. Governments retained control of the inflation target, while specialists determined the rate required to meet it. The arrangement insulated monetary policy from pre-election manipulation without disguising who controlled the broader economic settlement. Later mistakes by the Bank provide an argument for stronger scrutiny, rather than returning monthly interest-rate decisions to the Treasury.

Northern Ireland supplied the strongest case for dispersing political power. The Belfast Agreement of 10 April 1998 created power-sharing institutions, recognised that Northern Ireland’s constitutional position rested upon consent, and established structures linking Northern Ireland, the Republic and the two national governments. Devolution here served a concrete purpose: communities with incompatible national allegiances could govern together without forcing either to surrender its identity. The settlement has repeatedly faltered, yet it helped bring thirty years of sustained conflict to an end and remains the foundation of Northern Ireland’s constitutional order.

What Blair got wrong

The list of Blair’s partial and outright constitutional failures runs much longer than his successes. His reforms weakened the cleanest line of authority in British democracy: voters elected a sovereign Parliament, Parliament sustained a government, and that government possessed the power to carry out its programme.

Blair promised a referendum on the proposed European Constitution in 2004, then repeated that commitment in Labour’s 2005 manifesto. After French and Dutch voters rejected the Constitution, most of its institutional provisions returned through the Lisbon Treaty. Blair negotiated the political settlement at his final European summit, and Gordon Brown’s government ratified it without a referendum by insisting that Lisbon constituted a different document. The legal distinction offered exactly the sort of manoeuvre that encouraged voters to believe the political class would evade their verdict whenever it expected the wrong answer.

Blair compounded that breach through the handling of eastern enlargement. Britain granted immediate labour-market access to citizens of the new member states in 2004, while France and Germany used transitional controls. The government expected annual migration from the accession countries of between 5,000 and 13,000; the actual inflow quickly reached many times that level.

These decisions helped produce the referendum Blair had tried to avoid. Eastern European immigration became one of the most visible consequences of EU membership, while the abandonment of the promised vote supplied evidence that reform through ordinary elections had become impossible. Brexit carried many causes, and Blair’s choices supplied two of the most combustible ingredients: rapid migration that the government had badly underestimated, and a constitutional promise withdrawn when defeat appeared likely.

Devolution produced a different failure of design. Scotland, Wales and Northern Ireland received extensive authority over spending and public services, while Westminster continued to raise most of the money. A House of Lords committee later summarised the original settlements plainly: the devolved administrations had substantial powers to spend and “little or no power to raise” revenue. Later legislation transferred more taxation to Scotland, although Blair’s founding structure separated political generosity from fiscal responsibility.

The Human Rights Act began from a defensible ambition and grew into a permanent constitutional filter applied across government. Official guidance describes Convention analysis as “an integral part of the policy-making process”, beginning during policy development rather than at the end of drafting. Each new policy must pass through legal memoranda, compatibility analysis, anticipated case law and the prospect of judicial review. Policies emerge narrower, slower and more procedurally elaborate, while courts acquire growing influence over questions involving immigration, policing, welfare, employment and economic regulation.

The creation of the Supreme Court reinforced that cultural shift. The Law Lords had operated successfully within the House of Lords through one of the characteristically British arrangements that looks absurd in a diagram and works in practice. Their physical and institutional separation encouraged the language of three co-equal constitutional branches without supplying the codified constitution that normally defines and limits judicial power. Parliament remains sovereign in law. Successive governments increasingly behave as though an adverse judgment settles the political question.

The quango state completed the diffusion. Public authority must still lead back to an elected minister with the power to intervene and the duty to defend the outcome. Under Blair’s model, statutory independence frequently severed those two elements. Natural England could help hold up tens of thousands of homes through nutrient-neutrality rules; regulators could shape entire markets; agencies could establish priorities with major political consequences. A minister can be removed by voters. An arm’s-length body survives the election and explains that its hands are tied by the legislation under which it operates.

Blair’s constitutional settlement arose from a deliberate preference for restraint over execution. His generation had watched Thatcher use the full power of Westminster and resolved to prevent a future government from doing the same. Blair added his own confidence in courts, specialists and European institutions, then evaded direct public consent when he expected it to obstruct his preferred outcome. Britain emerged with a Prime Minister responsible for almost everything and sovereign over steadily less.

Blair’s safeguards constrain governments carrying out unpopular transformations and governments carrying out popular ones alike. The constitution has no mechanism for distinguishing between the two. It simply places more institutions between an election and its consequences.

Starmer’s levers connect to nothing because Blair’s settlement was designed to disconnect them.

Labour has more MPs than any British party has had since the parliament of 2001. It has a majority big enough to lose by-elections, suspend rebels, endure scandals, offend its base, bore the country and still carry on governing. Somewhere in those 403 people, there should be five or six obvious future prime ministers, and at least one figure who makes donors, journalists and civil servants think: yes, fine, that one could run the thing.

Instead, the conversation keeps drifting back to Andy Burnham.

This is strange enough on its own. Burnham was an MP for sixteen years. He sat in Cabinet. He ran for the Labour leadership in 2010 and came fourth. He ran again in 2015 and lost to Jeremy Corbyn by 40 points. Since then he has been Mayor of Greater Manchester, where his reputation has somehow inflated from regional executive to lost statesman of the Labour movement.

His great proof of administrative genius is buses.

To be fair, Greater Manchester’s bus reform is not nothing. Local control over routes, fares and franchising is useful, and Burnham deserves some credit for making it politically salient. But useful local government should not be confused with national leadership. Britain is a nuclear-armed G7 state with a stagnant economy, a swollen welfare bill, low productivity, low trust, and a planning system designed by people who hate civilisation. Running the country requires rather more than taking the Bee Network to Bury.

There is nothing inherently wrong with reaching outside Parliament in a crisis. Other countries have done it with genuine external heavyweights. Canada reached for Mark Carney, a former governor of both the Bank of Canada and the Bank of England. Italy reached for Mario Draghi, a former president of the European Central Bank.

Labour’s external heavyweight is Andy Burnham: a former cabinet minister, twice-leadership loser and mayoral bus man.

In a healthier party, Burnham would be one option among many. His record in Cabinet would be weighed. His achievements in Greater Manchester would be assessed in proportion.

Burnham’s rise to become the sole plausible leader is an indictment of everyone else. This achievement is only because the field has become obviously thin. A party only reaches for its bus replacement leader when the normal route has failed.

Heavy is the head that wears the crown

Coronations have a nasty habit of looking clever until voters get involved.

Parties like them because they feel disciplined. No messy contest, no internal argument, no awkward questions about the candidate’s record, judgement or weaknesses. Everyone smiles, the factions pretend to be united, and the press is told that the serious people have behaved seriously.

Then the public gets a vote.

Gordon Brown became Labour leader without a real contest and lost in 2010. Rishi Sunak became Conservative leader without a real contest and lost in 2024. Kamala Harris inherited the Democratic nomination without a normal competitive primary and lost later that year. Rather than eliminating weakness, coronations merely postpone its discovery.

The Burnham conversation already has the faint smell of inevitability about it. If he is the answer, he should have to survive an argument. The fact that parts of Labour seem to want the conclusion without the contest is not a sign of confidence. It is another sign that the bench is thin.

It seems that the Labour Party lack the ability to look forward 12 months, and see the almost certainty that Burnham will be in the same position in the country and in the polls as Starmer was a month ago. Once again, everyone will be regretting that they didn’t truly test Burnham’s vision for the country when they had a chance.

The Substitute Bench

The obvious defence of Burnham is that the alternatives are worse. Unfortunately for Labour, this is also the prosecution’s case.

Take Wes Streeting. His pre-MP career ran through student politics, charities, public-sector consulting and local government. This is a very modern Labour CV: polished, political, media-ready, fluent in the language of reform, but utterly devoid of any evidence of executive command, or understanding of economic growth.

Angela Rayner perhaps has a more grounded claim. Care work is real work: hard, human, underpaid and socially necessary. It gives her a connection to ordinary public services that many politicians lack. But it fails almost every test relevant to national leadership. It does not show that she has run a large institution, managed a serious budget, made strategic trade-offs, delivered complex reform, or been accountable for outcomes at scale. Her UNISON experience is even weaker as proof of statecraft. Trade union politics may teach grievance-handling and internal faction management, but as preparation for running the country it is thin gruel.

Then there is Ed Miliband, which is where the panic becomes visible. Miliband has already led Labour into a general election and lost. He is policy-literate and experienced, but he is not an answer to the leadership problem. That a man who already failed the national test can return a decade later as a plausible leader or chancellor is a further damning indictment of Labour’s woes.

Shabana Mahmood is the most revealing case because she should be one of Labour’s strongest options. She has a proper professional background as a barrister, holds a major office of state, and is pursuing immigration reduction in a country where the public plainly wants lower immigration. In a rational party, that combination would make her formidable. In Labour, it makes her suspect. The party has somehow found a senior figure with professional ballast, executive office and a position closer to public opinion than activist opinion, and turned that into a liability because she is seen as too right wing.

Al Carns is the strongest version of the external-CV argument. He has the kind of serious non-political experience Labour otherwise lacks: a decorated military career, command experience, and time advising Defence Secretaries. He cannot be dismissed as another creature of the politics-adjacent pipeline. But his case also shows the limit of raw material. Once inside government, even impressive people can be processed into the same machine. If the military CV candidate cannot visibly force a serious armed-forces settlement, outside experience is necessary but insufficient. The machine can digest almost anything.

Rubbish in, rubbish out

The thinness of Labour’s leadership field is fundamentally a problem of recruitment.

According to a Sunday Times analysis, at least 224 of the 257 new Labour MPs elected in July 2024 came from charities, communications or lobbying agencies, or had previously been political employees, ie just shy of 90%.

This is the statistic that explains the Burnham boomlet. A party whose new MPs come so heavily from charities, lobbying, comms and political employment should not be surprised when its leadership field looks like politics talking to itself. Those worlds can produce campaigners, message-carriers, factional operators and fluent performers. But apparently, they cannot produce a leader.

That is the missing quality. Too many Labour MPs have worked in institutions orbiting politics: organisations whose incentives are access, advocacy, pressure, narrative, patronage and internal positioning. It does not usually prove that someone can run a large organisation, or has built a business, managed complex operations, commanded a bureaucracy, taken responsibility for failure, or been held accountable by markets, courts, patients, customers or hostile reality.

It also helps explain Labour’s strange detachment from growth. A party drawn so heavily from the politics-adjacent world will naturally talk more fluently about redistribution, rights, services, access and fairness, than about building, selling, hiring, investing and expanding.

The Lost Ballast

Old (ie New) Labour was full of the usual political mix: mediocrities, cowards, obsessives, time-servers, capable ministers and the occasional serious figure. The difference is that many of them had ballast.

Jack Straw and Blair were barristers. Margaret Beckett worked in metallurgy. Tessa Jowell was a psychiatric social worker. Alan Johnson was a postman. Alistair Darling and Douglas Alexander were solicitors. David Blunkett and Gordon Brown were lecturers. Stephen Timms spent 15 years in telecoms.

These were ordinary careers as much as grand ones. That is part of the point. They gave politicians identities formed outside Westminster. They had worked in courts, universities, hospitals, councils, unions, companies and public services before becoming creatures of the whips’ office. They had been shaped by institutions with purposes beyond briefing journalists, lobbying ministers, managing the news cycle or climbing the party ladder.

These experiences do not automatically make someone a statesman. They do give them some formation before politics gets to them.

Plenty of modern MPs are bright, polished and industrious. The missing quality is depth of formation. Older politicians more often arrived in Parliament with a life already partly built elsewhere. Modern politicians too often arrive fluent in politics before they have acquired much obvious authority outside it.

Take a look in the mirror

The Conservatives had the same broader recruitment pool. Around 2005, their front bench included people with serious professional and commercial backgrounds. David Davis spent 15-plus years at Tate & Lyle. Theresa May worked at the Bank of England and in financial services. Michael Howard and Dominic Grieve were barristers. David Lidington worked for BP and Rio Tinto-Zinc. Liam Fox was a GP.

The old Conservative Party had plenty of mediocrities, bores and credentialled disappointments. Credentials can be wasted. Still, the raw material was different. These figures had worked in companies, banks, courts, surgeries and industry before Westminster fully digested them.

The modern Tory comparison cuts two ways. On paper, today’s Conservatives retain more private-sector ballast than Labour. Kemi Badenoch worked in software, banking and digital media. Mel Stride founded and ran a business for around 20 years. Robert Jenrick (not a Tory now, I know!) was a corporate lawyer and worked at Christie’s. Claire Coutinho worked at Merrill Lynch, helped start a business, worked at KPMG, and then moved through the think-tank and special-adviser route.

That is a stronger set of commercial CVs than Labour can offer near the top of its leadership conversation. Yet after the 2024 disaster, the Tory field still looked thin. More private-sector experience helped, and the party still failed to produce a golden generation waiting to govern.

Labour remains the cleanest case because it has the numbers and still looks short of leaders. The wider British political class has the same disease in milder form: narrower recruitment, thinner formation, more professional politics, less obvious authority.

The Machine Selects for Obedience

The cause is fairly simple. Parties reward the qualities party machines want, then act surprised when those qualities fail to add up to leadership.

Candidate selection has become risk management. Headquarters wants candidates who are loyal, polished, ideologically legible, unlikely to embarrass the party, willing to repeat the line, and safe enough to survive the campaign. Still, the safest candidate for the machine is often the least interesting candidate for the country.

The Institute for Government makes the basic point: MPs become the pool from which ministers, cabinet members and prime ministers are drawn. A serious party would therefore treat selection as an exercise in national capacity. Instead, too often it becomes an exercise in internal control. Find the people who will behave. Filter out the people who might cause trouble. Reward the people who understand the machine.

Labour’s 2024 selections produced reports of top-down choices, compressed timetables and local members being left with little say were more than procedural grumbles. They showed candidate selection becoming a bottleneck controlled by headquarters. The people who passed through were the people most acceptable to the machine.

Dominic Cummings argued years ago that Parliament needed people very different from the usual political class: people with quantitative ability, problem-solving skills and experience managing complex organisations. That is exactly the missing group. Westminster has plenty of people who can talk about systems. It has far fewer who have run difficult systems before asking for power over the state.

Simon Hart’s account from the Conservative side points the same way. Candidate identification, selection, training and discipline shape the quality of government. A bad recruitment system produces bad MPs, weak ministers and thin leadership contests. By the time a party is choosing a leader, the damage was done years earlier: in the approved list, the selection meeting, the parachuted candidacy and the safe seat.

That is why the Burnham problem goes beyond Burnham. The route into Parliament produces the alternatives. A machine that selects for obedience produces obedient people, not leaders.

Conclusion

Burnham is the symptom, not the disease.

Labour has 403 MPs, a huge majority, a crowded front bench, and a new intake large enough to reshape the party for a generation. Yet the leadership conversation still circles back to a man the party has already rejected twice.

That should embarrass Labour more than Burnham. He is doing what ambitious politicians do. The stranger fact is that the party has made his pitch plausible.

A serious governing party should have several people who look ready to take command. Labour has few people who have shown authority at scale, independently of the political machine, while being accountable for real-world outcomes.

Until our political parties, their whips, and their MP selections move away from demanding low-risk candidates with little real world experience, we are doomed to repeat our ten-year-and-counting conveyor belt of short-term, weak PMs.

Keir Starmer has been on the ropes for weeks. Labour MPs are restless, the polls are ugly, and Westminster has started doing what Westminster does best: pretending the next man will solve all the problems created by the last man.

So Labour now looks as if it may drift from Starmerism to Manchesterism.

Starmerism claims to offer stability, seriousness, fiscal discipline and competent administration. Manchesterism is the Andy Burnham alternative: public control of buses, visible mayoral activism, local pride, housebuilding promises, and the idea that the state should be more muscular, more rooted, and less embarrassed about intervening.

There are real differences between the two. Starmer is a cautious lawyer managing decline. Burnham is a mayor who has spent years learning that people like buses that turn up.

But the central problem remains the same: Britain is not growing fast enough.

In The Great Stagnation, I discussed how US GDP per capita grew by around 1.2% annually from 2008 to 2024, while the UK managed less than half of that, at around 0.48%. Over sixteen years, that small difference compounds to a huge one.

Americans got richer. Brits got excuses.

Since 2008, Britain has spent a lot of time asking how to stop banks failing again. That was understandable. But a financial system has more than one job. It has to avoid blowing up the economy. It also has to help finance the economy.

Britain may have remembered the first part and neglected the second.

How we got here

After the dot-com crash and 9/11, America cut interest rates and tried to keep growth moving. Housing became the obvious outlet.

American governments had spent years encouraging wider home ownership, including for poorer households and historically excluded communities. It helped expand the social and political permission structure around mortgage credit. The dangerous part came when weak loans could be mass-produced, bundled, rated, sold, insured and borrowed against.

Subprime (ie bad) mortgages rose from 8% of US mortgage originations in 2003 to 20% in 2005. Among subprime borrowers, the share using ARMs rose from around 60% in 2001 to 76% in 2004.

Thousands of mortgages were put into a pool for investors. The top tranches were paid first, so rating agencies could label them AAA even when the underlying mortgages were much weaker.

By the end of 2008, more than 90% of CDO tranches had been downgraded. More than 80% of AAA CDO bonds were eventually downgraded to junk. The system had built an industrial process for turning fragile credit into apparently pristine paper.

Then there were the extreme leverage levels: by November 2007, Bear Stearns had reached nearly 38:1 leverage. The major US investment banks were around 30:1. Fannie Mae and Freddie Mac, including loans owned and guaranteed, had combined leverage of around 75:1 by the end of 2007.

In June 2007, national house prices were already down 4.5%, and around 16% of subprime adjustable-rate mortgages were delinquent. In July 2007, two Bear Stearns hedge funds heavily exposed to mortgage securities collapsed. Fannie Mae and Freddie Mac were taken into conservatorship in September 2008. Lehman Brothers filed for bankruptcy on 15 September 2008. AIG was rescued the next day.

Britain had built its own fragile banking system.

Northern Rock was the first warning. It had grown aggressively and relied heavily on wholesale funding rather than ordinary deposits. In June 1998, its assets were £17.4bn. By June 2007, they were £113.5bn. When wholesale funding froze, the model cracked. In September 2007, savers queued outside branches in the first major British bank run for around 150 years. Northern Rock was taken into public ownership in February 2008.

Then came the bigger banks.

The government injected £45.5bn of equity into RBS. It put £20.3bn into Lloyds Banking Group. Across the crisis interventions, the UK state provided £137bn of public money in loans and capital to stabilise the financial system.

The wider economy paid for it. UK GDP fell by more than 6% between the first quarter of 2008 and the second quarter of 2009. It took five years for the economy to regain its pre-recession size. Unemployment rose from 5.2% in 2008 to 8.4% in 2011.

The pre-2008 system had failed. It had too much leverage, too little capital, fragile short-term funding, bad mortgage risk, rating-agency failure, opaque derivatives and regulators who did not see the whole machine until it was already breaking.

So the question is not whether Britain needed financial reform. It did.

The question is whether it chose the right reforms, stopped at the right point, and remembered that a financial system has to do more than survive. It also has to finance growth.

The sensible reforms

Before 2008, banks could run with tiny cushions of real equity and huge piles of borrowed money. Bear Stearns reached nearly 38:1 leverage by November 2007. The major US investment banks were around 30:1. Fannie Mae and Freddie Mac, including loans owned and guaranteed, were around 75:1 by the end of 2007. At those levels, an asset does not need to collapse. A small fall is enough.

Basel III raised the floor. Banks had to hold more common equity, more liquid assets, and more capital against risky exposures. The ordinary minimum Common Equity Tier 1 requirement became 4.5%, plus a 2.5% conservation buffer, before extra buffers for larger or riskier banks. By the mid-2020s, major UK banks had CET1 ratios around 15%, while large US banks were closer to 12% to 13%.

Stress testing also made sense. Before the crash, regulators looked too much at individual firms and too little at the system. A bank could look fine by itself while half the sector was exposed to the same property assumptions, the same short-term funding markets, and the same ratings models.

The US moved first. In 2009, the Federal Reserve stress-tested 19 major bank holding companies and found that 10 needed extra capital, with a combined shortfall of about $75bn. That forced banks to raise money and helped restore trust. Britain later built its own Bank of England stress tests. The 2022/23 UK test used a scenario with GDP down 5%, unemployment at 8.5%, house prices down 31%, and Bank Rate at 6%. Banks were expected to survive that without immediately becoming a taxpayer problem.

Resolution reform addressed the other obvious failure. In 2008, governments could either let a major bank fail and risk chaos, or rescue it and prove that large banks had a state guarantee. That was the “too big to fail” problem. Shareholders kept the upside. Taxpayers got summoned after the fire had already started.

Britain and America both tried to make large banks fail-able. Banks had to prepare living wills. Regulators gained stronger powers to restructure failing firms. Bail-in rules meant some creditors could absorb losses before taxpayers. The UK built its MREL regime, requiring large banks to hold layers of debt and equity that could be written down or converted in a crisis. The US used Dodd-Frank, living wills, orderly liquidation powers, and loss-absorbing capacity rules for the biggest banking groups.

These reforms attacked real weaknesses: too little capital, too much leverage, fragile funding, weak system-wide supervision, and no credible way to let large banks fail safely.

The fork in the path

Britain and America both made banks safer. Britain and Europe then went further.

Some of this was not direct 2008 banking reform. MiFID II, Solvency II and pension regulation belong to the wider post-crisis culture of financial caution. But that is part of the story. After 2008, risk became something to separate, document, capitalise, stress-test and avoid.

British ring-fencing applied to banks with more than £35bn of core deposits and material investment banking activity. Retail deposits, payments and basic lending were placed inside a protected bank. Trading, underwriting and investment banking sat outside it. If the riskier parts of the group failed, the ordinary bank was meant to keep working.

America used the Volcker Rule instead. It restricted proprietary trading and limited banks’ involvement with certain funds. It did not recreate Glass-Steagall. The largest US banks remained universal banks, combining retail banking, corporate lending, investment banking, underwriting, trading and advisory work inside large financial groups.

MiFID II did something similar to the equity market. From 2018, investment managers had to separate payments for trading from payments for research. The aim was transparency. Fund managers should not hide research costs inside trading commissions.

The side effect was weaker coverage. Large companies still attracted analysts. Small and mid-cap companies became easier to ignore. Research budgets as a share of assets under management reportedly fell by more than 50% after MiFID II.

Insurers and pension funds moved in the same direction. Solvency II-style rules pushed insurers towards capital discipline and caution. Pension schemes moved heavily towards liability matching. Trustees and advisers focused on reducing volatility, matching promises and avoiding regret.

Britain now has vast pools of long-term savings and very little domestic risk appetite. UK pension funds once held more than half their assets in UK equities. That figure has fallen to 4.4%. Corporate defined benefit schemes are even lower, at around 1.4%.

The call from inside the house

The most useful evidence comes from the Bank of England itself: in ‘Staff Working Paper No. 892, Side effects of separating retail and investment banking: evidence from the UK’, Matthieu Chavaz and David Elliott studied what ring-fencing did to banks before it came into force in 2019.

Before ring-fencing, a universal bank could use cheap retail deposits across the group. After ring-fencing, those deposits had to sit inside the protected retail bank. That made some activities cheaper to fund and others more expensive.

Banks responded by shifting towards domestic mortgage lending. They cut mortgage rates, gained market share, and expanded retail lending. Chavaz and Elliott estimate that ring-fencing explains around 10% of the fall in mortgage spreads between 2013 and 2019, making them cheaper.

The other side moved in the opposite direction. Banks reduced their provision of credit lines and underwriting services to large corporates, i.e. the reform made banks better placed to lend against houses and less active in parts of corporate finance.

On the corporate side, Chavaz and Elliott found that when banks lost access to cheap deposit funding, the size of syndicated loans they offered fell: an 11% fall in deposit funding was associated with a 7% fall in loan size, with larger effects for credit lines and ordinary corporate loans. Simply put, ring-fencing helped make mortgages cheaper, but made some banks less willing or able to support business finance.

This isn’t a complete explanation of British stagnation - it would be absurd to blame weak growth mainly on ring-fencing when Britain also has NIMBYism, expensive energy, slow infrastructure, weak public investment, tax complexity and poor state capacity. But it is a useful piece of the puzzle.

Britain already had a problem with property-heavy growth. Too much capital flowed into land, houses and existing assets. Too little flowed into new firms, productivity, technology, infrastructure and scale-up finance. Ring-fencing appears to have nudged the banking system further in the same direction.

Protecting ordinary deposits after 2008 was a reasonable aim. But the trade-off is now visible. Britain made retail banking safer and cheaper. It may also have made the financial system even more biased towards mortgages over productive risk.

Moderate fixes

Five reforms could help keep what is working, while opening up finance for growth:

1. Loosen ring-fencing where resolution now does the same job.

Ring-fencing currently applies to banks with more than £35bn of core deposits. That threshold should rise, and the separation rules should be loosened where banks already have strong capital, liquidity, stress tests, bail-in debt and credible resolution plans. Ordinary deposits and payments still need protection. But the UK does not need two separate systems doing the same job while cutting retail banking off from useful corporate finance.

2. Stop making property lending the easy option.

Capital rules should be reviewed against a simple test: do they make it easier to lend against existing houses than to finance productive business investment? A diversified pool of loans to real firms should not automatically be treated as more suspect than another pile of mortgage lending. The goal here is fewer hidden incentives pushing credit into property.

3. Rebuild SME and mid-market securitisation.

Britain should create a standardised market for SME and mid-market loan pools. Not the opaque pre-2008 version, where weak loans were chopped up, blessed and sold as magic paper. The new version should be dull by design: standard contracts, public loan-level reporting, simple structures, and originators keeping at least 10% first-loss exposure. That would let pension funds and insurers buy diversified productive-credit assets without pretending they can assess every individual company loan themselves.

4. Undo the worst effects of MiFID II for small and mid-cap equities.

Research unbundling was meant to make costs clearer. It also made research easier to cut. Large companies still get analyst coverage. Smaller listed firms become invisible. The UK should make it easier for brokers, exchanges and fund managers to fund research on small and mid-cap companies, with clear disclosure rather than rigid separation.

5. Let pensions and insurers take more productive risk when the return justifies it.

UK pension funds now hold only around 4.4% of their assets in UK equities. Corporate defined benefit schemes are even lower, at around 1.4%. That is not a healthy sign for a country that claims to want more domestic investment. Pension and insurance rules should make it easier to hold infrastructure debt, growth equity, venture funds, productive credit and scale-up finance where the risk and return are credible. The goal is to stop treating all long-term risk as a regulatory embarrassment.

A step farther

One more radical option would be a National Growth Credit Auction.

The government could set a target each year: for example, £20bn of long-term finance for productive UK firms, infrastructure suppliers, advanced manufacturing, energy projects and scale-ups. Banks, insurers, pension funds and private-credit funds would then bid to provide that finance.

The bid would not be “give us money”. It would be “this is the smallest tax or regulatory adjustment we would need to provide this amount of long-term credit”.

One bank might ask for lower capital charges on a tightly defined pool of mid-market loans. One insurer might ask for a specific Solvency II adjustment for infrastructure debt. One pension fund might ask for a tax change on long-term productive-credit vehicles. The Treasury would choose the cheapest credible bids.

The private institution would take the first losses. The eligible assets would be tightly defined. The terms would be public. The performance would be reviewed each year. Firms that gamed the scheme would be excluded from future auctions.

This would be very different from the usual “buy British” nonsense. The real problem is that too little capital reaches useful firms on workable terms.

A British ISA says: please buy British assets because flag.

A National Growth Credit Auction says: show us the cheapest way to make productive lending happen.

That is the right question. Do not force capital to chase weak returns. Make financial institutions compete to provide useful risk at the lowest public cost.

At its peak in 1989, ozone depletion appeared in about 50 New York Times stories in a year. Then, slowly, the story faded. It stopped being a regular feature of public life. It became one of those half-remembered anxieties from the end of the 1900s, filed somewhere between acid rain and quicksand.

Problems often vanish from the news because people get bored, governments lose interest, or journalists find a newer disaster. But the ozone layer faded for a better reason: the world had started to solve it.

Many debates between science and policy never reach a clean ending. Nuclear power is still argued over after decades; nutrition advice seems to change every few years.

This article is about the cleaner cases. The moments where science identified a problem, policy acted on it, and the outcome improved.

1. The ozone layer

In 1985, three scientists from the British Antarctic Survey published a paper in Nature showing large springtime losses of ozone above Antarctica. Ozone values over Antarctica had started falling around 1970 and kept falling into the 1990s. NASA satellite data soon confirmed that it was a regional-scale hole in the ozone layer.

Ozone is a molecule made of three oxygen atoms. Most of it sits in the stratosphere, more than 10 kilometres above the surface, absorbing much of the Sun’s harmful ultraviolet radiation.

The damage came from chemicals that had once seemed almost ideal. Chlorofluorocarbons, or CFCs, were stable, non-flammable and useful. They appeared in spray cans, refrigeration, air conditioning, and industrial cleaning. Their stability was the problem: they did not break down easily near the ground. Instead, they drifted upward into the stratosphere, where ultraviolet light broke them apart, releasing chlorine atoms, which destroy ozone.

By the mid-1980s, scientists had shown that ozone depletion was real, man-made and likely to worsen if emissions continued. The worst losses were over Antarctica, but thinning was also observed over the Arctic and mid-latitudes. The concern was that more UV at ground level meant more skin cancer, more cataracts, possible immune effects, and damage to agriculture and marine life.

The Montreal Protocol, agreed in 1987, controlled the production of CFCs and other ozone-depleting substances. Later amendments strengthened the controls as the science became clearer and substitutes became available.

UNEP’s 2022 assessment says nearly 99% of banned ozone-depleting substances have now been phased out. WMO says the ozone layer is expected to return to 1980 values by around 2040 for most of the world, 2045 over the Arctic, and 2066 over Antarctica, assuming current policies remain in place.

One study estimated that the Montreal Protocol will prevent 443m cases of skin cancer and 63m cataract cases in the US alone for people born between 1890 and 2100, compared with a world with no controls on ozone-depleting substances. And that is just one country.

Today, the ozone layer is not fully healed. The Antarctic hole still appears each spring. Some chemicals linger for decades. Recovery depends on continued compliance. But the battle for our policy response is over and won.

2. Goitre

For much of history, goitre looked like a disease of geography. It clustered in inland and mountainous regions where iodine was scarce in the soil and food supply. The connection took time to prove. In 1813, the Swiss physician Jean-François Coindet argued that seaweed helped treat goitre because it contained iodine. In the 1830s, Jean-Baptiste Boussingault proposed iodine-rich salt as prevention.

Iodine is a trace element used to make thyroid hormones. These hormones help regulate growth, metabolism and brain development. The body cannot make iodine, so it has to come from diet. Coastal populations often got more through fish, seaweed and other marine foods. Inland populations could get much less, especially where iodine had been washed out of the soil.

The thyroid sits in the neck. When iodine is too low, it struggles to produce enough hormone. The body pushes the gland harder, and the thyroid can enlarge. That swelling is goitre. The visible neck swelling was the obvious sign, but iodine deficiency could also damage pregnancy, infant development and learning.

In the early 1900s, the northern US had a recognised “goitre belt” around the Great Lakes, Appalachians and the north-west. Some US Public Health Service surveys in the 1920s found goitre rates of 70% to 100% among schoolchildren in parts of Michigan, Minnesota and Wisconsin. During the First World War draft, 31% of military candidates from one Michigan region were reportedly judged unfit for service because their goitres were too large for them to button a uniform.

In 1917, David Marine and Oliver Kimball gave iodine to schoolgirls in Akron, Ohio, and showed that it sharply reduced new cases of goitre. Michigan then became the first U.S. state to sell iodised salt commercially, starting on 1 May 1924.

By 1935, goitre incidence in Michigan had fallen by 74%-90%, with the largest reductions among children who had consistently used iodised salt.

The global version became universal salt iodisation. In 1980, the first global WHO estimate suggested that 20% to 60% of the world’s population was iodine deficient. By 2020, UNICEF estimated that 89% of the global population used salt with some iodine, though quality and coverage still vary. Nearly 1 billion people still lack access to iodised salt.

3. Leaded petrol

Lead poisoning had been recognised for centuries before anyone put lead into petrol. In 1921, Thomas Midgley Jr. and colleagues at General Motors found that tetraethyl lead stopped engine knock, allowing cars to run more smoothly and efficiently. It was cheap, effective, and commercially attractive. It was also poisonous.

By 1924, workers producing tetraethyl lead at Standard Oil’s Bayway refinery in New Jersey began hallucinating, convulsing and collapsing. Five died. The building became known as the “loony gas building”.

Tetraethyl lead made engines less likely to knock, which allowed higher compression and better performance. After combustion, lead entered air, dust, soil and bodies. Children were especially vulnerable because lead damages the developing brain. It can lower IQ, impair attention and increase behavioural problems. There is no useful amount of lead in a child’s blood.

The early fights focused on workers and acute poisoning. The later case against leaded petrol came from environmental measurement and child health. Herbert Needleman showed that children with higher lead exposure performed worse at school, helping shift attention from dramatic poisonings to lower-level population-wide damage.

In 1973, the US EPA issued rules to reduce lead in petrol under the Clean Air Act. The amount of lead used in US petrol fell by 99.8% from 1976 to 1990. Blood lead levels fell with it. The median blood lead level among American children aged 1 to 5 was 0.6 µg/dL in 2020, down 96% from 1980.

The final country to end sales of leaded petrol for road vehicles was Algeria in 2021. UNEP estimated that the global ban prevents more than 1.2m premature deaths each year and saves around $2.5tn in economic costs.

Lead remains in old paint, pipes, soil, industrial sites and aviation fuel. The damage already done by leaded petrol also lingers in contaminated dust and land. But one of the largest mass poisonings of the 1900s has been stopped at source.

4. Acid rain

Acid rain was first named in the 1800s by Robert Angus Smith, a Scottish chemist working in Manchester, after he noticed that rain was more acidic in polluted industrial areas. For a long time, the idea remained mostly local: smoke, coal, dirty air, damaged buildings.

In the 1950s, Scandinavian scientists began documenting acid damage to lakes and streams. In 1963, scientists in New Hampshire collected rainwater with a pH of 3.7, far more acidic than the expected background level of around 5.6. Hydrogen ion concentration changes tenfold with each whole pH unit.

Acid rain includes snow, fog, cloud water, particles and gases that deposit acids onto land and water. Sulphur dioxide and nitrogen oxides react in the atmosphere and return as sulphuric and nitric acids.

Acidified lakes and streams became harder for fish and other aquatic life to survive in. Soils lost nutrients and released aluminium, which could damage roots and water systems. Acid deposition stressed some forests; limestone and marble buildings also suffered as acidity dissolved their surfaces.

In the US, the main policy response came through the 1990 Clean Air Act Amendments. Title IV created the Acid Rain Program, aimed at reducing sulphur dioxide and nitrogen oxides from electric power generation. The sulphur dioxide programme used cap-and-trade. Power plants needed allowances for their emissions, and the total cap fell over time.

Annual sulphur dioxide emissions from US power plants fell by 94% from 1990 to 2019. Wet sulphate deposition, a common indicator of acid rain, fell by more than 70% between 1991 and 2022.

Europe moved through its own mix of national rules and international agreements, including protocols under the Convention on Long-Range Transboundary Air Pollution. Sulphur dioxide emissions fell sharply across many countries as coal plants fitted scrubbers, switched fuels, or closed.

5. Smallpox

Smallpox caused fever, pain and a deep pustular rash. Many survivors were left scarred or blind. The virus spread from person to person and had no animal reservoir.

Older forms of variolation had existed for centuries, using material from smallpox itself to produce a controlled infection. They reduced risk, but could still spread the disease. In 1796, Edward Jenner showed that exposure to cowpox could protect against smallpox. showing immunity could be induced without using smallpox itself.

Smallpox remained a global killer long after vaccination was discovered. By the mid-1900s, richer countries had largely eliminated it, but the disease still persisted across parts of Africa and Asia. The last stage required a system that could find cases quickly and stop each chain of transmission.

In 1967, the WHO launched the Intensified Smallpox Eradication Programme. Mass vaccination mattered, but surveillance and containment became central: find cases, isolate patients, trace contacts and vaccinate the people around them.

Ring vaccination used the biology of the disease against it. Smallpox spread only between people, symptoms were visible, and infectious cases could be traced. Once health workers found a case, they could build a wall of immunity around it.

The WHO says the intensified programme eliminated smallpox from Latin America within four years, then from Asia four years after that. India recorded its last case in 1975. The last naturally acquired case was recorded in Somalia in October 1977. In May 1980, the World Health Assembly declared the world free of smallpox.

Samples of the virus remain in high-security laboratories, and governments still plan for accidental or deliberate release. But natural smallpox is gone. A disease that killed, blinded and scarred people for centuries was removed from human circulation through vaccination, surveillance and the decision to finish the job.

6. Trans fats

Trans fats began as a food technology success. In the early 1900s, manufacturers learned to turn liquid vegetable oils into solid or semi-solid fats through partial hydrogenation. These fats were cheap, stable, long-lasting and useful in margarine, shortening, baked goods, frying oils and processed foods. They helped food last longer and gave manufacturers a reliable alternative to butter and animal fats.

Fats are made from fatty acids. In ordinary unsaturated fats, the shape of the molecule usually has a bend in it. Partial hydrogenation changes some of those bonds and creates trans fats, which are straighter and more solid.

The debate over butter, seed oils and saturated fat is messy. Industrial trans fats were different: they had no nutritional purpose, clear cardiovascular harms, and technically feasible substitutes.

Industrial trans fats raise LDL cholesterol, lower HDL cholesterol, and increase the risk of coronary heart disease. They had no nutritional need and could be replaced. WHO says trans fat intake causes more than 278,000 deaths each year worldwide.

One European Heart Network summary found every 2% of energy intake from trans fats was associated with about a 25% higher risk of heart attack or death from heart disease.

In 2003, Denmark limited industrial trans fats to 2% of fats and oils in food, which effectively removed partially hydrogenated oils from the food supply. Consumers did not need to read every label or understand lipid chemistry. Manufacturers had to reformulate.

The US required trans fat labelling from 2006, then the FDA ruled in 2015 that partially hydrogenated oils were no longer generally recognised as safe. For most uses, manufacturers could no longer add them to foods after June 2018, with final compliance extended to January 2021.

In 2018, WHO launched its REPLACE package to support worldwide elimination of industrial trans fats. The core answer is settled: remove industrial trans fats from the food supply, rather than asking millions of people to dodge them one packet at a time.

7. Rinderpest

Rinderpest was a viral disease of cattle and buffalo, with fever, mouth lesions, diarrhoea and very high mortality in susceptible herds. In societies built around cattle, rinderpest could destroy food, wealth, transport, farming power and social status in one sweep.

In the late 1800s, rinderpest entered Africa through infected cattle in the Horn of Africa and then spread south and west. The 1888-1897 panzootic killed over 90% of cattle in some regions. Ethiopia may have lost around a third of its human population from the resulting famine.

The virus spread through close contact between infected and susceptible animals. Herds moved for grazing, trade, war, migration and slaughter. Once the disease entered a region with little immunity, it could burn through livestock at extraordinary speed.

Earlier vaccines helped in parts of Africa and Asia, but the major breakthrough came from Walter Plowright’s work in Kenya in the 1950s. His tissue culture rinderpest vaccine was cheap, effective, produced long-lasting immunity and could be made at scale.

Vaccination campaigns reduced rinderpest, but conflict, weak veterinary systems and cross-border animal movement allowed it to survive. The final push combined vaccination, surveillance, laboratory confirmation and international coordination. Rinderpest had useful biological weaknesses: it had no long-term carrier state, animals either died or recovered with immunity, and the vaccine worked well.

The last confirmed outbreak was in Kenya in 2001. In 2011, FAO and the World Organisation for Animal Health declared rinderpest eradicated. It became the first animal disease eradicated worldwide, and only the second disease of any kind after smallpox.

8. Itai-itai disease

Itai-itai disease emerged in the Jinzu River basin in Toyama Prefecture, Japan. The name came from the cries of patients: itai, itai - “it hurts, it hurts”. The disease caused severe pain in the spine and joints, brittle bones, fractures and kidney damage. It affected mainly women in rice-farming communities who had lived for years with cadmium pollution in their food and water.

The pollution came from the Kamioka mines upstream. Mining and smelting released cadmium into the Jinzu River. Irrigation water carried the metal into rice paddies, and rice absorbed it from the contaminated soil and water. Cadmium built up in the body, damaged the kidneys, and contributed to osteomalacia, a softening of the bones.

The first cases were noticed in the early 1900s, but the cause took decades to prove. Dr Noboru Hagino reported the disease in 1955 and later linked it to heavy metals in the Jinzu River basin. In May 1968, Japan’s Ministry of Health and Welfare officially announced that Itai-itai disease was caused by chronic cadmium poisoning from the Kamioka mine.

Residents filed suit against Mitsui Mining & Smelting in 1968. The main issue was causation: whether cadmium discharged from the Kamioka mine had caused the disease. The residents won at the first trial in 1971, then again at the second trial in 1972.

Japan built much of its modern pollution-control system in the late 1960s and early 1970s, under pressure from cases like Minamata disease, Yokkaichi asthma and Itai-itai disease. Contaminated paddy fields were identified, restored or taken out of ordinary use. Monitoring continued at the mine and in the affected basin.

Toyama Prefecture still describes annual on-site inspections at Kamioka mine, carried out by affected residents with experts, and continued company countermeasures to prevent recurrence. The disease was local, painful and slow to prove, but the policy answer became clear: identify the pollutant, prove the pathway, stop the discharge, clean the soil, and make the polluter pay.

Conclusion

These cases worked because the problem could be made concrete.

That did not make the politics easy: industry resisted; governments moved slowly; the science took time; victims had to fight. Some policies created new problems. But the chain of action was unusually clear: identify the cause, prove the pathway, remove or control the source, and measure whether the problem shrank.

Most public policy is messier than that. Housing, crime, growth, migration, education and healthcare are not single-cause problems. They have trade-offs, winners, losers, feedback loops and institutions that fail in several directions at once. Science can inform them, but it rarely supplies a clean switch.

The stranger cases are the ones where the facts are relatively clean and the argument still never ends. Nuclear power is the obvious example. It is one of the lowest-carbon sources of electricity, yet public understanding still lags. In YouGov polling published in April 2026, only 68% of men and 31% of women said nuclear produces little to no carbon emissions.

Evidence is necessary, but it is not always enough. These eight successes had science, but they also had unusually practical levers, often requiring private sector change, rather than billions in public sector investment. Ban the chemical. Fortify the salt. Vaccinate the herd. Remove the additive. Cap the emissions. Stop the discharge.

The problems that remain are often the ones where the lever is harder to pull, the benefits are slower to arrive, or the public has never been given a clear account of the facts.

Science can find the cause. Policy can act on it. The rare thing is getting both to happen before the argument eats the solution.

“The UK has a far flatter income distribution than the Communist Soviet Union.
The UK take home minimum wage for working a full time job (40-hours) is now £22,555.

At £100k salary, the take home is £68,558.
That is a net income ratio of 3.04:1

We are now at the point where the wage compression and taxes in the UK means that the difference between minimum wage and a top 5% salary is a net income difference of only ~3x.

In the USSR using the same comparison, this figure never fell below 5:1
It's actually even worse in reality because the person earning £100k in the UK often has student loans.
Britain is nominally capitalist but functionally communist. China is nominally communist but functionally capitalist.”

The next day, a GB News interview with the editorial director of the IEA, largely confirmed the story to be true.

The UK vs The Reds

Naturally, I wanted to investigate, and ran into a few hiccups.

Firstly, while the tweet does correctly point out that a £100k salary is roughly a top 5% salary, the USSR ratio of top 5% vs minimum wage is quite hard to reconstruct from public data. The USSR mainly published in deciles, so for my calculations, I’ve instead used a comparison of the minimum wage vs a top 10% earner, and for historical analysis, used a bottom 10% pre-the introduction of the minimum wage under Blair.

According to historians from Cambridge, this top 10% vs bottom 10% ( or “p10:p90”) ratio held fairly stable between 1970 and 1990 in the USSR, at a low of 3.0x and a high of 3.3x.

In 2025 in the UK, a pre-tax salary of £83.5k gets you into the top decile, which post-tax equates to £58.9k. As mentioned in the original tweet, we’re assuming this hypothetical high-earner isn’t a young City of London finance bro, paying a further 9% tax towards his student loans. At the other end, our minimum wage worker makes £25.4k, or £21.8k after tax, implying a p10:p90 multiple of 2.7x.

So, for 2025, it does appear that the UK’s wages are flatter than they were under the USSR.

Two key caveats:

• On the top side, there is a huge difference between a top 10% salary in the UK of £83.5k, and the top 1%, starting at £201k, so this doesn’t tell us about the skew at the very top end in modern Britain. That said, Stalin seemed to have a pretty comfy lifestyle.

• On the bottom side, we are looking only at someone working full time on minimum wage. For people complaining that benefits are too generous in the UK, this data completely excludes that segment of society.

But obviously the situation in 2025 represents the UK after 14 years of communist Conservative government - was the p10:p90 ratio always this bad?

Using 5-yearly segmented data, it appears that the answer it yes:

A reminder that this data picks up the bottom 10% income in 1995, but then a full time worker on minimum wage in 2000, explaining the apparent jump under early Blair, and would be slightly flatter using the older metric.

It’s notable that even once that metric settles in, the compression of wages flattens considerably during the 2010-2024 Conservative period.

Britain vs the world

I was curious whether this was a uniquely British form of secret communism, or whether other countries were also crypto-Stalinists. Selecting our cousins over the water, the French, and plucky Singapore, we can see that there is a reasonable variation among nations:

While I couldn’t source reliable data from Singapore pre-1990 or the USA pre-1980, the picture above is pretty clear: the UK is worse than communism, France is on a par, and the USA and Singapore enjoy small amounts of freedom.

One might now start to wonder whether this also reflects a country’s ability to grow, and so I’ve checked whether there is a correlation between annual GDP per capital % growth over the 1970-2025 period and the above multiple:

With a correlation of 0.89 (from a small sample), it does appear that there is indeed a fairly strong correlation between a less-flat income distribution and higher economic growth.

So, assuming we want more in the UK, the question is, do we increase our multiple by changing the top 10% or the bottom 10%?

Unless we can encourage the private sector to pay higher pre-tax wages to the top 10%, a brave politician running on this issue must decide what is the most palatable cause to run on: lower taxes for the rich, or cutting the minimum wage?

The rest is communism.

For this piece, I’ll define stagnation as a multi-year period in which real GDP per capita grows by less than 1% annually. By that standard, the UK has been stuck in an extraordinary slump since 2008 - one so long and so slow that it has been mistaken for normality.

Quick definitions for readers not comfortable with economic jargon: “real” growth means growth adjusted for inflation; GDP is the total output of the economy; GDP per capita is that output divided by the population, which makes it a better rough guide to whether a country is getting richer per person.

Two decades of flatlining

The story most people know is simple enough: in 2008, the financial crisis hit. Growth weakened. Then, just as things were beginning to recover, Covid hit and the economy took another blow. On the usual GDP numbers, that story is true as far as it goes. Across the G7, average annual GDP growth from 2008 to 2024 was markedly lower than in any of the earlier periods shown here. The post-2008 era was plainly weaker than the decades before it.

For the UK, 1.7% annual growth from 2008 to 2024 is poor by recent historical standards. It is well below the 2.8% seen in 2000-07, the 2.4% of the 1990s, and the 2.8% of the 1980s. But on its own, it still looks more like a disappointment than a national emergency.

This is because headline GDP still softens the truth. It gives the impression of an economy that kept moving, just more slowly than before. That is why this era has so often been described as disappointing rather than historic. The usual GDP numbers capture the slowdown. They do not yet capture the stagnation.

Let’s look at the per capita figures

Once population growth is taken into account, the picture changes sharply. Across the G7, average annual GDP per capita growth from 2008 to 2024 was again lower than in any of the earlier periods shown here. But this time the scale of the decline is much clearer

For the UK, average annual GDP per capita growth fell to just 0.48%. Italy managed only 0.12%.

Only the USA stayed above 1%.

This is the real break. On a headline GDP basis, the post-2008 years look like a long period of weakness. On a per capita basis, they look like something much more serious: a sustained failure to make the average person meaningfully richer. That is why this period deserves to be called a stagnation.

Comparing GDP and GDP per capita directly for the post-2008 period reveals how much headline growth has flattered to deceive:

Japan’s weak headline GDP growth masks the fact that GDP per capita grew much faster, so individuals there did better than the aggregate numbers suggest.

In the USA and Canada, the two measures are roughly in-step; in Germany and France, headline GDP still flatters exaggerates per-person growth.

In the UK, the 1.7% annual GDP growth rate falls to under 0.5% on a per-person basis - completely inadequate for a country that expects rising living standards.

And Italy is even worse. There, almost all of the apparent growth in the aggregate figures disappears once population is taken into account.

The worst stagnation in British history?

The length of this period is what makes it so unusual. Since 1945, the UK has had bad years and bad runs, but not a 17-year stretch in which average real GDP per capita growth remained below 1%. That is what makes the post-2008 period different. It is not just weak, it is weak for an exceptionally long time. The post-war era was built around the assumption that living standards would keep rising at a decent pace. Since 2008, that assumption has broken down.

There are some possible historical comparisons, but none quite fits. One is the interwar period. Britain returned to growth after the early 1920s, but it was slow, uneven and accompanied by high unemployment and a wider sense of economic drift. The Productivity Institute notes that UK growth in the 1920s averaged only 1.4% a year, and that the period was marked more by “doldrums” than by any strong recovery. That is probably the closest modern analogue. But even that comparison is imperfect. Interwar Britain contained sharper shocks, more obvious social distress, and a more visibly abnormal economic environment than the one we have lived through since 2008. Even then, that 1.4% average is almost triple our current stagnation, and for a shorter period.

The other possible comparison is the Long Depression that began in 1873. But that too is only a partial match. The late 1800s brought deflation, financial instability and intense anxiety, yet historians still argue over whether it was really a depression in the modern sense at all. Some measures of output and even real income per head continued to rise. In other words, it felt grim, but it was not the same kind of long, per capita stagnation shown in the charts above.

The historical data gets shakier as you go further back, especially before 1900. But my search suggests that this is the longest clear British slump of its kind in at least the modern era, and quite possibly in the last 300 years, or longer. That should be a central fact of British politics. Politicians talk about productivity, investment, living standards and fiscal pressure as separate problems. They rarely describe the underlying condition plainly: Britain has endured an unparalleled long period of near-stagnation, and it is still not being addressed as such.

Why stagnation has been tolerated

As well as the unusual length of this crisis, it is markedly different for its subtle nature.

A country with 15% inflation knows it has a problem. A country with 10% unemployment knows it has a problem. Both create obvious pain, obvious headlines and obvious political pressure. Stagnation does not work like that. It arrives slowly. It spreads its damage across wages, tax revenues, public services, investment and personal expectations. It feels less like an emergency than like a general thinning-out. That is one reason Britain has tolerated it for so long. A crisis that comes as forgone growth is easier to ignore than one that comes as visible collapse.

In 2008, UK real GDP per capita stood at about $44,500 (using 2010 dollars for a stable marker). By 2024 it had risen to roughly $48,000. That is an increase of just 7.9% across sixteen years. If Britain had matched the US rate of real GDP per capita growth over the same period, it would have reached about $54,300 instead - a gap is roughly $6,300 per person. In other words, British output per head would be around 13% higher today if the UK had simply grown at the same per-person pace as the US since 2008.

And this points to the political illusion at the centre of the whole period. Britain reached $48,000 of GDP per head in 2024, and almost nobody treats that as an emergency. But imagine a different path. Imagine Britain had reached roughly $54,000 by 2020, then spent four years shrinking back to the same $48,000 endpoint. That would be treated as a national economic disaster. Ministers would be under intense pressure. The media would call it a crisis. There would be no shortage of demands for a new direction. The fact that we arrived at the same destination through years of weak growth rather than visible decline does not make the damage smaller. It only makes it less legible.

That is the real danger of stagnation. It does not look like a crisis, so it is not treated like one. It produces no single crash, no clean villain and no obvious moment of reckoning. Instead it creates a country that is permanently more strained than it should be, with weaker wage growth and lower ambition. Another five years of this would not feel dramatic in the moment. That is exactly why it would be so damaging.

Conclusion

The point of this essay is simply to make clear just how bad this problem has been, and how poorly it has been recognised. This is not just a British story. It is a wider G7 story, with the USA the main exception.

Across much of the rich world, the post-2008 period has been marked by extraordinarily weak growth in output per person, yet very few politicians speak about it with anything like the clarity it deserves. A crisis this large should have produced a new economic agenda. Instead it has mostly produced patchwork measures, modest reforms and evasive language.

Perhaps that is why so much hope is now being loaded onto AI. It is one of the few remaining ideas that promises growth without forcing governments to admit how poor the underlying record has been. If AI delivers a genuine productivity boom, it may change the picture completely. But if it does not, the emptiness of the current response will become much harder to hide. After nearly two decades of stagnation, much of the rich world is still governed by people who speak about low growth obliquely, timidly, or not at all.

“The cost of a given level of intelligence is falling by a factor of ten every year,” Sam Altman, CEO OpenAI, February 2025

“AI could wipe out half of all entry-level white-collar jobs,” Dario Amodei, CEO Anthropic, May 2025

In 2024, global AI spending was approximately $235bn, according to IDC. By 2026, Gartner expects worldwide AI spending to reach $2.52tn.

The question is no longer whether AI will change the economy, but how to think clearly about what comes next. We need to strip the economy back to first principles. Before we talk about growth, jobs or productivity, we need to ask: what are our basic institutions actually for?

Economics gives one answer. A firm combines labour, capital and raw materials to produce goods or services for sale. A philosopher might use a different word - telos, meaning purpose. The telos of a biscuit company is to make biscuits.

But the self-stated purpose of an institution is often not its most socially useful one.

A school exists to impart knowledge and basic skills.
A university exists to develop expertise.
A firm exists to produce goods or services for sale.

But each also has what I will here call a shadow telos: a secondary function that is rarely stated openly, but which society ultimately depends on. The shadow telos of a school is to look after children while parents work, and to socialise them into ordinary life. The shadow telos of a university is to certify and filter young adults for the labour market.

The detach

Let’s use a thought experiment.

Tomorrow, Sam Altman announces a breakthrough - ChatGPT Education Oneshot:

All children from the age of 4 can sit in front of a laptop for a single hour and look at a flashing screen, and they will instantly learn all the information they would otherwise have taken to the age of 18 to learn in school.

The telos of the school system has, in one sense, been totally solved by AI.

However, it soon becomes obvious that providing the telos of the school (education) without providing the shadow telos of the school (childcare and socialisation) falls very short of the mark, and we still need schools, even without any academic learning. A government that responded by firing all the teachers, selling the buildings and declaring victory would trigger economic chaos overnight.

Let’s return to our firm. The biscuit firm’s telos is to make biscuits. In doing so, it hires sales staff, accountants, delivery drivers and so on. By 2032, the CEO of Biscuitworld realises ChatGPT 7.0 can do 95% of each role, and so the sales team of 50 drops to a sales team of 3, and the other teams follow. The telos of the biscuit firm remains unchanged, and perhaps even more biscuits are made.

But now we must consider the shadow telos of a firm in a capitalist economy.

Before the industrial revolution, almost the entire planet lived in poverty, but for a few at the top. But then everything changed, when the world went through multiple revolutions simultaneously: the industrial revolution of course, but also through it revolutions in finance and business. The industrial revolution created huge new wealth that could now be spread to the masses, but it was the financial and business revolutions that were the mechanism by which far more of that wealth reached ordinary people, rather than a nightmare steampunk feudalism.

In a capitalist economy, this is the hidden role, the shadow telos, of every firm: to take wealth as it is created and distribute purchasing power to employees and investors.

Of course, political choices can shape the details of how it is divided up (taxation, minimum wages and so on). But AI could lead to a wholesale separation of the telos and shadow telos - we can keep making biscuits, while distributing the income of the firm to smaller and smaller numbers of people. The economy can only function if the wealth is sufficiently spread around so individuals can create demand for other firms to meet.

The fault lines

The argument so far rests on a premise: that AI will allow firms to produce roughly the same output with fewer workers, especially fewer junior white-collar workers. That premise is plausible, but it is not yet proven. This section is an attempt to map the main routes by which the detach could happen.

Firstly, by fewer people doing the same work. Here the early signs are already visible. The Institute of Student Employers found that UK graduate hiring fell by 8% year-on-year in its 2025 survey. The same survey found 140 applications per graduate vacancy, more than triple the number from two decades earlier.

There are also direct signs of thinner graduate pipelines in sectors that sit right in AI’s line of fire. According to reporting on the UK graduate market in 2025, Deloitte reduced its graduate scheme by 18% and EY cut graduate recruitment by 11%. PwC UK, meanwhile, received 60,000 applications for about 2,000 entry-level roles in 2025, even as its graduate intake edged down and the firm spoke openly about AI reshaping the traditional consultancy pyramid.

Second, the thinning of firms. In many knowledge businesses, the model is pyramidal. A small number of partners sit at the top, supported by a broad base of juniors doing drafting, research, slide-making, and all the rest. AI strikes first at exactly those tasks. Even if the firm still needs senior judgment, it may need far fewer people underneath. The result will likely be smaller graduate intakes, flatter teams, and a profession that offers far fewer ways in.

Third is margin compression in knowledge work. If AI sharply reduces the labour needed to produce routine legal drafts, tax computations, software modules, or marketing copy, then it lowers the cost base for both incumbents and challengers. In more competitive markets, cheaper inputs invite undercutting, driving down prices and, therefore, firms have less money to hire workers. Per Reuters, 40% of legal tasks could be automated, threatening the traditional Big Law billing model. The same basic logic applies in accounting, consulting, recruitment, design and parts of software.

The fourth fault line is more subtle. Large parts of the modern economy exist to cope with knowledge inefficiency. We employ people to search, summarise, draft, compare, format, review and forward information from one desk to another. If intelligence becomes radically cheaper, a surprising number of sectors may discover that they were merely managing friction. Remove enough of that friction, and you can erase entire layers of work.

Since the launch of ChatGPT in 2022, Adzuna and Indeed’s data both suggest graduate roles advertised in the UK were down 33% on the previous year and at their lowest level in seven years.

Finally, we have timing. Even if the optimistic story is ultimately true, and AI creates new industries, products and categories of work - destruction may still arrive faster than creation. You can lose 50,000 junior coding, analyst or support roles much faster than new businesses can start to hire them. The rebound may come, but if it comes slowly, that is still a major economic and political problem.

No individual mechanism has to be total to matter. Even partial movement in all of them at once would seriously affect the 68% of UK workers who have a white collar job, and would be enough to put strain on a wage-based capitalist economy.

Can the market solve this by itself?

At this point, the obvious objection is Jevons’s paradox.

Named after the 1800s economist William Stanley Jevons, the idea shows when something becomes cheaper to use, people often end up using more of it, not less.

Take driving - when cars became more fuel-efficient, the cost of driving a mile fell. You might expect total spend on fuel to drop with the price drop. In reality, people often drive more when driving becomes cheaper, and so a price drop can lead to a total spend rise. A major Yale review of the evidence found that these rebound effects are real, with many short-run estimates in the 5% to 25% range, and a broader range of roughly 5% to 40%. In plain English: efficiency gains often lower costs, but part of the saving is taken back through greater use.

If intelligence becomes much cheaper, the economy may simply demand far more of it.

Imagine that by 2032 one software engineer, armed with powerful AI tools, can do the work that once required 50. At first glance, that looks catastrophic for software jobs. But cheaper software also changes what the rest of the economy can afford. A chain of a dozen restaurants that currently muddles through with generic booking systems, manual rota planning and separate finance tools might suddenly find it worthwhile to hire one engineer to build a custom system around its exact needs. The same logic could apply to care homes, dental chains, and hundreds of other mid-sized businesses that currently live with bad software because bespoke systems are too expensive.

That is what Jevons would suggest for AI. Intelligence gets cheaper. So the economy may demand more of it.

Six paths to 2032

There are really only three economic questions here.

First, does AI actually work?

Second, if it does work, does labour demand recover?

Third, if labour demand does not recover fast enough, can the state build some other mechanism for distributing income and preserving social stability?

From here, I can see six broad paths to 2032, and my guess at percentage of likelihood.

1. Flop - 10%
   AI is not transformative enough to remake the economy. Firms use it, productivity rises a bit, some jobs change, but the basic structure of work remains intact. This would look more like the arrival of better software than the arrival of a new economic age.

2. Heaven - 5%
   AI and robotics become so powerful that humans simply no longer need to work. Machines do almost all economically useful labour. Output explodes, scarcity shrinks, and everyone enjoys a permanently higher standard of living without needing wages to survive.

3. Hell - 5%
   AI goes catastrophically wrong. Not just a bad labour-market transition, but mass civilisational damage or extinction. Eliezer Yudkowsky’s phrase is the clearest version of this fear: “If anyone builds it, everyone dies.” That is the world in which misaligned AI leads to millions of deaths.

4. Everything works out great - 20%
   AI destroys some jobs, but Jevons’s paradox and wider market adjustment do their work. Lower costs create new demand quickly enough. New industries emerge. New roles absorb the displaced. By the end, most people still work, but they do better jobs in a richer economy.

5. Long transition crunch - 30%
   The long-run story may still be broadly positive, but the timing is ugly. Jobs are destroyed faster than new ones are created. For 5 to 20 years, millions of middle-class people fall into unemployment. This is the scenario most like a white-collar version of deindustrialisation or even, at the sharper end, a depression-style labour glut.

6. Structural post-work drop - 30%
   Labour demand falls and stays lower. The economy still produces plenty, perhaps more than ever, but it simply does not need as many humans to work. That could mean unemployment rising from 5% to 15% and staying there, or 50%, or 90%.

The most likely outcomes are not the cleanest ones. That is why the key task for governments is not to bet everything on either doom or rebound. It is to prepare for the cases 5 and 6, which by my estimate represent (very roughly) a 60% chance: the years in which AI works, but society has not yet worked out what to do about it.

The role of government

A modern state governs a social order. Social orders hold together when enough people feel that they have a place in the system and a share in its prosperity. Firms have long helped provide that settlement. As they produced goods and services, they also spread purchasing power through wages. AI may preserve the first function while weakening the second.

At that point, the state inherits the gap.

Its task must be to preserve demand, social peace and a sense of participation while the old wage mechanism weakens. That task could involve tax reform, wage subsidies, transfers, broader capital ownership, public employment, universal basic income, or some mixture that nobody has yet designed well.

I see very few politicians, civil servants or commentators treating this as the central issue. Plenty of people talk about whether AI works. Plenty talk about whether it beats China, lifts GDP, or threatens particular professions. Far fewer ask: what happens to a capitalist society once one of its main distribution mechanisms begins to fade?

That should be the central question.

The years ahead may still bring a smooth adjustment. They may also bring a long crunch, or a structural fall in labour demand. In each of those worlds, the central duty of government looks similar. It has to keep the economy stable. It has to keep prosperity widely spread. It has to ensure that abundance still reaches the public in a form people can live on.

What if government gets this wrong?

History offers several warnings here.

Feudalism was what a society looked like when it never really solved the distribution problem at all. Wealth and power sat at the top. Most people lived near subsistence. An AI economy that concentrates output and ownership in a few hands could recreate that same failure in modern form.

Communism tried to solve the distribution problem directly. The price was coercion, bureaucracy and stagnation. Its lesson: distribution by brute force creates its own pathologies very quickly.

The Great Depression showed what happens when labour supply vastly exceeds labour demand for years on end. In 1933, US unemployment reached roughly 25%, or about 14m people. By 1932, one in four workers was unemployed. People sat idle, got poorer, lost faith, and radicalised. A society with millions of surplus white-collar workers would face the same basic danger, even if the cause this time came from technology rather than finance.

Deindustrialisation offers a more recent warning. In theory, displaced workers move to where the new jobs are. In practice, many never fully recover. Formal retraining rates for displaced workers were often very low, and even among Trade Adjustment Assistance participants, training participation fell from 31% in 1995 to 18% in 1999; of those leaving the programme, only 56% earned at least 80% of their previous wage. In the hardest-hit China shock areas, 78.2% of the rise in transfer payments came through Social Security and disability-linked payments - in other words, through labour-force exit rather than clean re-entry into equally good jobs.

Then there is the simplest warning of all: exit. If governments try to patch the distribution problem too late, too crudely, or too heavily in the form of huge tax increases on the few who retain jobs and wealth, capital and talent will look for the door.

The longer governments wait, the narrower their options may become.

Conclusion

The AI debate is still far too fuzzy.

Capitalism has worked as well as it has because production and distribution were tied together. Firms made things, and in making them, they spread purchasing power through society. If AI begins to separate those two functions, then the central political problem of the next decade will become very simple: how does a society remain stable once production stays high, but wage distribution weakens?

I do not claim to have solved that problem here. I am making a narrower claim: governments, economists and commentators should treat this distribution issue as their main problem, and they should start doing so now.

Because a solution could be needed as soon as 2027 or 2028.

If AI works as advertised, our futures will be decided by who gets paid, by what mechanism, and whether the system still feels worth belonging to.

And then, as far as I can remember, Indonesia vanished.

I have read international news almost every day for most of my life, and yet Indonesia occupies only the faintest corner of my memory. Before starting this piece, the country brought to mind Bali, orangutans, and a hazy association with palm oil. For a nation of such scale, that feels extraordinary.

Indonesia is the 4th most populous country on earth. More than 280m people live across its thousands of islands, scattered between the Indian and Pacific Oceans. A country like that ought to arrive in the mind already furnished with images, stories and reference points. Somehow, at least in Britain, it barely does.

That oddness grows sharper when set beside other giant states. Russia fills the headlines through war, power and history; Brazil via the Amazon, football, Carnival and dramatic politics; Pakistan appears through conflict, religion and its charged relationship with India. Indonesia, by contrast, is almost invisible.

So this article began in curiosity, and in a certain embarrassment as well. I realised that one of the largest countries in the world existed in my mind as little more than a blank patch on the map. What follows is a journey into the history, geography, politics, people and ecology of a country that deserves far more attention than it usually receives.

History

Modern humans were present in the wider Indonesian region by around 40,000 years ago, and the islands later received major waves of Austronesian-speaking settlers whose maritime cultures spread through much of the archipelago. By the 2nd millennium BCE, these seafaring communities were carrying crops, pottery, boat-building traditions and languages across island Southeast Asia.

When Europeans first arrived, they encountered a rich and busy trading zone. For centuries, Indonesian ports had sat within commercial networks stretching to India, China and the Middle East. Hindu and Buddhist kingdoms had risen and fallen; Islam had spread widely through trade and conversion, especially from the 1200s onward. The Portuguese came first in pursuit of the spice trade, though the Dutch soon proved more formidable. Through the Dutch East India Company, founded in 1602, and later through the Dutch state itself, the Netherlands gradually turned commercial footholds into imperial control. In many places, rule began with monopolies, forts and alliances before hardening into direct conquest. By the late 1800s and early 1900s, the Dutch had largely welded the islands of the future republic into a single colonial possession - the Dutch East Indies.

Colonial rule came crashing down during the Second World War. Japan swept aside Dutch authority in 1942, and when Japan surrendered in August 1945, Indonesian nationalists moved quickly. Sukarno and Mohammad Hatta declared independence on 17 August. The Dutch sought to return, and four years of war, diplomacy and international pressure followed. By 1949, the Netherlands had transferred sovereignty, and Indonesia emerged as an independent state built from the shell of the old colony.

The early republic lurched through unstable parliamentary politics before Sukarno concentrated power under his system of “Guided Democracy”. Then came the vast rupture of 1965-66, when an attempted coup, army reprisals and mass killing opened the way for General Suharto’s New Order regime. Suharto ruled for more than thirty years, combining authoritarian control with economic development and tighter state direction. The Asian financial crisis shattered that system in 1998 and set off the Reformasi era, which brought democratisation, greater press freedom and a much looser political order. Modern Indonesia still carries the imprint of all these phases at once - anti-colonial nationalism, New Order centralisation, and the energetic, messy democracy that followed.

A nation of Islands

Geography explains a great deal about Indonesia. This is the world’s largest archipelagic state, with more than 17,000 islands spread across roughly 5,000 km - about the distance from London to Tehran. The main islands are Sumatra, Java, Kalimantan, Sulawesi, Papua, Bali and the chain of the Lesser Sundas beyond it. Population is heavily concentrated in the west, above all on Java, which alone holds roughly half of the country’s people, and with Sumatra, the two have c.80% of the population. Much of it also sits on the Pacific Ring of Fire, which helps explain the dense chain of volcanoes running through the archipelago and the rich soils that have supported so many people.

The natural world is one of the richest anywhere on earth. Indonesia holds tropical rainforests, mangroves, mountain forests, peatlands, coral reefs and seagrass meadows, while its seas form part of the Coral Triangle, the global centre of marine biodiversity. Orangutans survive on Sumatra and Borneo, Komodo dragons on a handful of eastern islands, and Papua is famous for birds-of-paradise. Sumatra still supports one of the most extraordinary mixes of large wildlife anywhere in Asia: tigers, rhinos, elephants and orangutans in the same broad ecosystem.

The plant life deserves a brief mention too. Indonesia’s forests contain thousands of species, including Rafflesia arnoldii, the world’s largest individual flower, and the titan arum, another giant bloom famous for its smell. All of this natural wealth lives under heavy pressure from logging, plantation agriculture, mining and wider environmental damage.

The people

With roughly 284 million people, Indonesia is the fourth most populous country on earth, and around 59% of its population now lives in urban areas.

Indonesia has hundreds of ethnic groups and several hundred languages, with the Javanese forming the largest single group at c. 40% of the population. Sundanese come next, followed by a long tail of smaller peoples spread across the islands. In many countries, that sort of diversity would pull hard against unity. In Indonesia, the great adhesive has been the language of Bahasa Indonesia. It was based on Malay, promoted through the nationalist movement, and turned into a shared public language across an archipelago whose people often speak a local language at home as well. That linguistic achievement gave Indonesia a national voice broad enough to bind together a country that could easily have fractured into regional blocs.

Around 87% of Indonesians are Muslim, which makes it the largest Muslim-majority country in the world. Yet Islam in Indonesia sits alongside large Christian minorities in some regions, a Hindu-majority Bali, and smaller Buddhist, Confucian and local belief communities. This gives the country a distinctive religious profile: deeply Islamic, though varied in practice and region, and shaped as much by local history and island-level difference as by any single national mould.

In living standards, Indonesia sits in that broad middle ground that so much of the modern world occupies. That means modern skylines, mass motorbike ownership, shopping malls, smartphones and an enormous consumer market, though also overcrowding, weak infrastructure in many places, and large gaps between richer and poorer regions. Jakarta and the western core of the country feel very different from parts of eastern Indonesia. A rising middle class exists, though so does a great deal of precarity.

Indonesia’s own statistical agency put the national poverty rate at 8.57% in September 2024, while the World Bank’s updated international lines produced much higher shares under broader lower-middle-income and upper-middle-income thresholds. Indonesia has made major progress over recent decades, and millions have entered a more comfortable life, while a very large share of the population still lives close to the edge by rich-world standards.

The Economy

In 2024 the Indonesian economy grew by 5.03%, and its nominal GDP reached about $1.4tn - smaller than Mexico’s, though far larger than Thailand’s, and enough to make it the biggest economy in Southeast Asia. GDP per head came in at about $4,925, which puts it in the same broad range as countries such as Algeria rather than the rich world.

Services made up about 43.8% of GDP in 2024, industry 39.3%, manufacturing alone 19.0%, and agriculture 12.6%. Agriculture still matters hugely as a source of work, with the World Bank showing roughly a quarter of total employment in the sector, even though its share of output is much smaller. This is a country with malls, factories, mines, plantations and rice fields all sitting inside the same economic picture.

Indonesia is a major exporter of palm oil, coal, nickel, tin and natural gas. Palm oil shipments alone reached $2.37bn in October 2024, up 25.35% on the year, while the country has become strategically important in nickel thanks to battery supply chains and the electric-vehicle industry. That gives Indonesia a role in several of the biggest global arguments of the moment - energy, industrial policy, green technology and environmental damage.

China is the main destination for non-oil-and-gas exports at $5.66bn that month, with the United States and India next behind. On the import side, China was also dominant at $6.43bn. Among product categories, animal or vegetable fats and oils - mainly palm oil - are the biggest export categories, while electrical machinery and equipment recorded the biggest increase among major import categories. Other trade data for 2024 also points to refined petroleum as Indonesia’s single largest import, which fits a country that exports major resources while still running an oil-and-gas trade deficit.

Tourism matters too, though it is less central to the economy than Bali’s image might suggest to outsiders. Indonesia recorded 1.28 million international visitor arrivals in September 2024, up 19.53% on the year, with Malaysians, Australians and Chinese visitors leading the list. Domestic tourism is even bigger: BPS recorded 757.96 million domestic tourism trips from January to September 2024.

Politics

Indonesia is a presidential republic, so the president serves as both head of state and head of government, appoints ministers, and sits at the centre of executive power. Parliament matters as well, though the system’s main public face is the presidency, especially during elections.

Indonesia is a multi-party democracy, with several nationally important parties rather than two giant blocs. In the 2024 legislative election, the biggest parties by vote share included PDI-P, Golkar, Gerindra, PKB, NasDem, PKS and the Democratic Party. Coalitions matter a great deal, which means presidential politics often revolves around alliances, bargaining and personality as much as party labels.

The current president, Prabowo Subianto, took office in October 2024 after years as one of the country’s most recognisable and controversial political figures. He is a former special forces commander and former defence minister, long associated with a harder, more nationalist style of politics. His path to the presidency was unusual. He lost the presidential race twice, challenged one result, then joined Joko Widodo’s cabinet, softened his public image, and returned as the winning candidate in 2024.

The 2024 election itself looked straightforward on the surface. Prabowo won in a single round with 58.59% of the vote, which spared him a run-off against Anies Baswedan and Ganjar Pranowo. Yet the campaign carried a larger political story behind it. Joko Widodo, known universally as Jokowi, had reached his term limit, though his influence loomed over the race, and Prabowo benefited heavily from his association with the outgoing president. The sense of continuity deepened further because Prabowo ran with Gibran Rakabuming Raka, Jokowi’s eldest son. So the election had the feel of both succession and continuity, with a touch of dynastic manoeuvre thrown in.

As for the issues, three seem to have mattered most. The first was the economy in the broadest sense: jobs, prices, growth, food security and the hope of steadily rising living standards. Candidate pledges and campaign coverage leaned heavily on growth, welfare, school meals, inflation and access to opportunity. The second was corruption, a permanent theme in Indonesian politics and one that still carries weight with voters. The third was continuity versus democratic quality. Plenty of Indonesians liked Jokowi’s development record and wanted more of it, while critics worried about institutional weakening, nepotism and the growing shadow of dynastic politics.

On the way up?

There you have it - the invisible giant of Indonesia. According to Goldman Sachs, per the chart above, by 2050 Indonesia is to jump from the 17th largest economy in the world to the 4th largest.

At some point in that transition, we should really start paying it a little more attention.

It was only weeks after the election. The markets were watching, the government was new, and the message was urgent. In that speech, Reeves claimed she had uncovered a £22bn “black hole” in the public finances.

Three months later, in her October Budget, roughly £40bn of tax rises were announced: employers’ National Insurance increased; income tax thresholds remained frozen; inheritance tax reliefs were reduced. The fiscal rules were projected to be met in year five.

In November 2025, the tightening continued. The Office for Budget Responsibility assessed that policy measures would raise around £26bn a year by 2029-30. Headroom remained narrow. The five-year target still held.

Then the scrutiny began. Reports questioned how the £22bn figure had been constructed and how closely the political framing matched the watchdog’s underlying assessment.

But beneath these arguments sat a simple fiction.

Every headline number in that debate - the £22bn gap, the £40bn package, the £26bn yield - depended on five-year productivity and revenue forecasts that have been based on wildly optimistic projections that were wrong year after year after year.

Stop and look at the above chart. Pause and actually look at it.

The black line is reality. Output per hour worked in the UK since the financial crisis.

The blue lines are forecasts made at different points over the last fifteen years. Each one projects productivity forward from that moment.

From 2010 onwards, the projections assume that productivity will begin to grow again at something close to its pre-2008 trend. In the early years, that was a reasonable assumption. The financial crisis was recent. Many economists expected a cyclical rebound once the shock passed.

But look at what happens next.

By 2014, the rebound had not materialised.
By 2015, it still had not.
By 2017, more than a dozen forecasts had already assumed recovery that failed to appear.

And yet the central projection kept drawing the same shape. The line bends upward again. Next year will be different.

We are now eighteen years on from the crisis. The chart contains roughly twenty-five successive revisions. After the first ten wrong ones, they start to edge the level down slightly. Yet, each one still assumes that the inflection point is close.

The false optimism persists.

Forecasting error is inevitable, but persistent directional error is different. When the same mistake is made repeatedly, the question is no longer whether growth will recover, but why the baseline keeps assuming that it will.

What evidence, exactly, justifies the expectation that productivity will suddenly return to trend in the near future? What structural change has occurred that was absent in the previous ten projections?

Nothing. Just wildly wrong predictions, again and again.

And what happens if the line does not turn?

Productivity drives tax receipts and the debt ratio. The £22bn “black hole”, the £40bn tax package, and the projected £26bn a year by 2029-30 all rest on one assumption: productivity growth accelerates over the next five years.

If it does not, the economy is smaller than forecast. Tax receipts fall short. The projected £26bn no longer closes the gap.

More tax rises will follow.

Population Forecasts

The United Nations publishes long-term population projections for every country on earth. They come in three variants - high, medium, low - driven mainly by alternative assumptions about fertility. A rate of 2.1 is needed for a stable population. Small differences in total fertility rates compound, over decades, into enormous differences in population size.

Take Thailand.

In the UN’s 2022 World Population Prospects, Thailand’s fertility rate was projected to settle at around 1.3 births per woman under the medium variant, and roughly 1.1 under the low.

In January 2026, Thailand recorded approximately 39,000 births - down from around 42,000 in January 2025. A fall of roughly 7% in a single year.

The current fertility rate sits close to 1.0.

That is below the UN’s low variant. The one that was supposed to represent the pessimistic case.

In the chart above, the red line is reality. The blue lines are successive UN projections for annual births in South Korea, Colombia, and Chile.

The pattern is identical to the productivity chart. Each forecast assumes the decline will slow. Each one is drawn from a moment of modest optimism about where fertility is heading. And each one, in turn, is overtaken by a fall steeper than the model expected.

South Korea is the scariest case. The UN’s projections have been revised downward repeatedly. The actual birth count has fallen through the floor of every one of them.

This is persistent directional error. The same mistake, made the same way, for the same reason: next year will be different. The inflection point is close. The line will bend.

It has not bent.

South Korea’s population is currently projected to fall from 52 million today to around 22 million by 2100 under the medium variant. The low variant - the UN’s own pessimistic case - puts it at around 16 million. South Korea’s total fertility rate in 2025 sits at around 0.75. If that rate holds, each generation is roughly 37% the size of the one before it. Two generations from now, the implied population is around 7 million: well below even the pessimistic projection, and the rate has been falling.

Nigeria offers a different illustration of the same problem. In 2012, the UN projected Nigeria’s population would reach 914 million by 2100. By 2024, that figure had been revised down to 477 million, as fertility fell faster than the models anticipated.

Just like that, nearly 440 million people vanish from the model. Oops.

Bad forecasts everywhere…

The pattern is not confined to government forecasters.

The above chart shows the private sector making the same error. JP Morgan’s forward rates market - some of the most liquid, most scrutinised pricing in the world - has repeatedly anticipated Federal Reserve rate rises that did not arrive, across the 1999 cycle, the 2004 cycle, and beyond. The blue lines fan upward. Rates stay flat or fall. The market reprices. And then, in the next cycle, it does it again. The mechanism generating the error is different from a government spending model, but the shape of the mistake is identical.

The Department for Transport has been doing the same thing with roads since 1989.

Every forecast projects car traffic rising sharply. Every actual outcome falls below every forecast variant. Across more than two decades of successive revisions, the gap between projected and actual traffic growth has been persistent and directional. The methodology has been updated. The models have been refined. The line keeps pointing up. Reality does not follow.

Migration forecasting has produced the same pattern.

Before EU enlargement in 2004, the Home Office forecast that net annual migration from the eight new accession countries - Poland, the Czech Republic, the Baltic states - would run at between 5,000 and 13,000. The actual figure was around 50,000 a year, a miss of roughly four times the top of the range. The error was large enough that when Romania and Bulgaria joined the EU in 2007, the government chose to publish no forecast at all for what would happen when their labour market restrictions eventually lifted. The lesson drawn was to stop predicting, or at least to stop doing so publicly.

Conclusion

The argument here is not that forecasters are incompetent, or that prediction is futile. It is narrower than that.

When errors are random, they cancel out over time. When they are directional - when the same institutions, using the same methods, miss in the same direction, across decades, across domains, across countries - something structural is producing them. Incentives? The difficulty of publishing a forecast that says things will get worse? The institutional momentum of a baseline that has always assumed recovery?

The question worth sitting with is why the errors, across such different institutions and such different domains, keep pointing the same way.

So when you next hear a politician stating what the economy, public debt, immigration - take your pick - will look like under their amazing new plan, it’s incumbent on us all to remember the forecasters’ track record, and consider the incentives they have for a consistent bias.

Almost exactly a year later, on 26 January 2026, a coalition of US special operations, naval forces and air assets executed a precision operation in Venezuela, culminating in the capture of President Nicolás Maduro. American officials described it as a narrow “decapitation” mission: targeted, swift, and with minimal reported US casualties, followed by a controlled hand-off to Venezuelan opposition elements. Although formal figures remain opaque, the operation’s rapid execution has been widely described in Washington as a strategic success that avoided prolonged engagement. A success that perhaps is at the back of Trump’s mind, suggesting interventions are easier and better than he thought.

This week, the US has dramatically increased its deployed military assets in and around the Middle East amid escalating tensions with Iran. Two aircraft carriers – the USS Gerald R. Ford and USS Abraham Lincoln – are reported in regional waters, alongside multiple destroyers, fighter jets (including F-35s and F-22s) and over 150 aircraft repositioned to bases in Europe and the Middle East. Diplomacy in Geneva continues, but Tehran has been given a tight ultimatum on its nuclear activities even as evacuation of embassy personnel underscores the heightened risk of confrontation.

For much of my generation, the phrase “interventionism” conjures Iraq and Afghanistan: long, costly wars that became quagmires without clear end states. What is less understood are the cases where limited, defined interventions have achieved their stated objectives with precision and relative strategic clarity.

This article examines five (well, six really) of the most illustrative examples of post-1945 interventionism to ask a simple question: what parameters make an intervention successful?

Sierra Leone, 2000

Sierra Leone’s war began in 1991, but by 2000 it had entered a far more dangerous phase. Roughly 50,000 people had already been killed over the decade, and Freetown had endured a brutal assault in 1999 in which thousands were murdered and mutilated. A fragile peace agreement collapsed, and in May 2000 the Revolutionary United Front advanced again while UN peacekeepers were captured and disarmed in large numbers. The UN mission hesitated, under-equipped and politically constrained. The immediate fear was state collapse and a second urban massacre that could have pushed deaths sharply higher.

At that point, the UK government under Tony Blair authorised Operation Palliser. It was initially framed as an evacuation of British nationals, but the calculation was clear: if Freetown fell, the war would reset. Around 1,000 British troops deployed rapidly, securing Lungi airport and key infrastructure. While the UN mission struggled to regroup, Britain acted decisively, signalling that it would not allow the capital to be overrun. This was a former colonial power choosing to intervene quickly while others deliberated.

British forces established control of the airport within days, providing a secure logistical backbone. When the RUF tested British resolve by kidnapping soldiers later that year, UK Special Forces mounted a precise rescue operation. Simultaneously, British officers began restructuring and training the Sierra Leone Army. The psychological shock of a disciplined, technologically superior force changed the battlefield dynamic. As rebel advances stalled, momentum shifted.

The UK embedded advisers, reinforced the UN mission once it stabilised, and supported disarmament and security-sector reform. By 2002, the war was formally declared over. Disarmament programmes dismantled militia structures, elections followed, and Sierra Leone has avoided a return to civil war since. The intervention was small, time-limited, and applied at the moment when chronic conflict threatened to tip back into mass killing.

East Timor, 1999

East Timor’s crisis in 1999 followed a rapid political shift rather than a long civil war. After decades of Indonesian control, a UN-backed referendum in August 1999 delivered a decisive vote for independence. Within days, pro-Indonesian militias, launched a coordinated campaign of arson, forced displacement and killing. Around 1,000 to 1,500 people were killed in a matter of weeks, and c. 250,000 were forced across the border into West Timor. Much of the capital, Dili, was destroyed. The pace of violence and displacement suggested that, without intervention, fatalities would rise sharply as militias consolidated control.

International pressure mounted on Jakarta, financial markets reacted, and Indonesia signalled willingness to accept an international stabilisation force. The UN authorised a multinational mission. Australia led the deployment, assembling INTERFET with contributions from regional partners. The objective was clearly defined: restore order, secure key infrastructure, enable humanitarian access, and create conditions for a UN transitional administration.

The deployment began with 2,500 troops landing in Dili in the first wave, rising to around 11,000 within weeks under Australian command. Forces secured the capital, airport and major population centres in short order. Militia groups withdrew across the border or dispersed rather than confront a well-equipped multinational presence. Within weeks, the large-scale campaign of arson and killing came to an end. By early 2000, the majority of refugees had begun returning under international supervision, and humanitarian supply chains were operating at scale.

INTERFET transitioned authority to a UN administration that governed temporarily while institutions were built. In 2002, East Timor became formally independent. The country has faced political and economic challenges since, yet it has remained sovereign and free of a return to mass violence.

Yugoslavia part 1: Bosnia, 1995

Bosnia’s war began in 1992 following the collapse of Yugoslavia and Bosnia’s declaration of independence. What followed was a multi-sided conflict marked by siege warfare and systematic ethnic cleansing. By mid-1995, an estimated 100,000 people had been killed. Sarajevo had been under siege for more than three years. In July 1995, over 8,000 Bosniak men and boys were murdered at Srebrenica despite its designation as a UN “safe area”. UN peacekeeping forces were present but constrained by mandate and rules of engagement. The combination of rising civilian casualties, the humiliation of safe-area failure, and the risk of further territorial consolidation by Bosnian Serb forces shifted the strategic calculation.

In August 1995, after a second massacre at Sarajevo’s Markale market, NATO launched Operation Deliberate Force. The campaign involved sustained air strikes against Bosnian Serb military positions, command infrastructure and ammunition depots. Around 400 aircraft conducted more than 3,000 sorties over roughly two weeks. The objective was to compel a cessation of offensive operations and force parties back to negotiations. There was no ground invasion, and no desire for regime change in Belgrade.

The air campaign altered battlefield momentum quickly: the Bosnian government and Croatian forces advanced on the ground while Serb positions degraded under sustained strikes. Within weeks, a ceasefire framework took shape. By November 1995, negotiations in Dayton, Ohio produced a peace agreement that preserved Bosnia as a single state divided into two entities. NATO then deployed a 60,000-strong Implementation Force to stabilise the settlement.

The war ended in December 1995. Bosnia remains politically complex and institutionally cumbersome, yet it has not returned to large-scale armed conflict. The intervention combined delayed escalation with concentrated force, diplomatic coordination and a defined settlement. It ended a war that had produced six-figure fatalities and repeated mass atrocities.

Yugoslavia part 2: Kosovo, 1999

The Dayton settlement ended the Bosnian war but left the wider Yugoslav question unresolved. Slobodan Milošević remained in power in Belgrade, and tensions shifted south to Kosovo, a province of Serbia with an overwhelming Albanian majority. By 1998, clashes between Serbian security forces and the Kosovo Liberation Army had escalated into a campaign of village burnings and population displacement. By early 1999, several thousand people had been killed and hundreds of thousands displaced internally. Western governments assessed that the pattern seen earlier in Bosnia risked repeating itself on a larger scale if left unchecked.

Talks at Rambouillet sought autonomy arrangements for Kosovo within Serbia. When negotiations collapsed, NATO authorised an air campaign without a UN Security Council mandate, calculating that a Russian veto would block formal approval. The objective was to compel Serbian forces to withdraw from Kosovo and accept an international security presence.

Operation Allied Force began in March 1999. Over 78 days, NATO aircraft conducted roughly 38,000 sorties, targeting Serbian military infrastructure, command facilities and air defences. During the opening phase, Serbian forces accelerated expulsions, pushing approximately 800,000 Kosovar Albanians across borders into Albania and North Macedonia. As the campaign continued and Serbia’s military capacity degraded, pressure intensified on Belgrade’s leadership. In June 1999, Milošević agreed to withdraw forces from Kosovo.

NATO deployed around 50,000 troops under the Kosovo Force, establishing a secure environment while the UN administered the territory. Large-scale organised violence ceased. Refugees returned in substantial numbers over the following months. Kosovo later declared independence in 2008, recognised by many Western states, though not universally. The intervention carried legal controversy and geopolitical cost, yet it achieved its defined military objective: Serbian forces left Kosovo, and the immediate campaign of mass displacement ended.

Iraq, 1991

In August 1990, Saddam Hussein’s Iraq invaded and annexed Kuwait. Iraq emerged from the Iran–Iraq War burdened by debt and economic strain, and Baghdad accused Kuwait of depressing oil prices and siphoning crude from shared fields. Within days, the Kuwaiti ruling family displaced, and hundreds of civilians were killed during the occupation and thousands detained. Iraqi forces positioned themselves along the Saudi border, placing a large share of global oil supply under Iraqi control. The strategic concern extended beyond Kuwait: annexation risked becoming permanent and further expansion remained possible.

The US assembled a broad coalition under UN Security Council authorisation, with over 30 countries contributing forces. By January 1991, approximately 700,000 coalition troops were deployed in the Gulf region, including around 540,000 from the US. The objective was defined and narrow: expel Iraqi forces from Kuwait and restore its recognised government. The mandate focused on reversing annexation rather than reshaping Iraq’s internal order.

Operation Desert Storm began with a sustained air campaign on 17 January 1991. Over five weeks, coalition aircraft flew more than 100,000 sorties, degrading Iraqi command structures, air defences and logistics. A ground offensive commenced on 24 February. Within 100 hours, Iraqi forces in Kuwait collapsed or retreated. Coalition military fatalities numbered under 400, with US deaths at 294. Estimates of Iraqi military deaths range from 20,000 to 35,000. Kuwait was liberated by the end of February.

Coalition forces halted at the Kuwaiti border and a ceasefire formalised the end of active hostilities. Kuwaiti sovereignty was restored, oil exports resumed, and the territorial status quo ante re-established. The campaign applied overwhelming force to a limited objective and concluded once that objective had been achieved.

Cambodia, 1978

The previous cases were Western-led coalitions acting beyond their borders. Cambodia presents a different example: a non-Western regional power intervening next door. By late 1978, the Khmer Rouge regime had ruled Cambodia for nearly four years. During that period, an estimated 1.5 to 2 million people – roughly a quarter of the population – died through execution, starvation and forced labour. The regime also launched repeated cross-border raids into southern Vietnam, massacring civilians in frontier villages. With the killing apparatus still active and the Cambodian state hollowed out, mortality projections under continued rule remained catastrophic.

In December 1978, Vietnam launched a full-scale invasion. Within two weeks, Vietnamese forces entered Phnom Penh. The Khmer Rouge regime collapsed in January 1979 and its leadership fled toward the Thai border. The objective centred on removing a hostile regime responsible for cross-border attacks and mass killing. Vietnam installed a new government, the People’s Republic of Kampuchea, staffed partly by Cambodian defectors.

The fall of Phnom Penh ended the organised system of extermination. Execution centres closed and forced collectivisation structures unravelled. Civil war persisted along the border regions through the 1980s, with continued fighting that cost tens of thousands of lives. Yet the mortality rate never returned to the scale of 1975–78. Even a fraction of the earlier annual death rate sustained over several more years would likely have added hundreds of thousands to the toll.

After Vietnam’s intervention, China launched a punitive invasion of northern Vietnam in 1979. Western governments continued to recognise the Khmer Rouge’s UN seat for years as part of broader Cold War alignment. The operation reshaped regional politics and prolonged diplomatic isolation for Hanoi. Its humanitarian effect, however, was immediate and decisive: the regime responsible for one of the 1900s’ worst episodes of mass killing ceased to govern.

Conclusion

Across these cases, common threads recur. First, the objective was specific and finite – expel an army, secure a capital, compel withdrawal, remove a regime engaged in mass killing. Second, force was applied decisively and then stopped once that objective was met. Third, intervening states avoided assuming responsibility for redesigning entire political systems. The settlement was imperfect in every case, but the mission parameters were clear enough that success could be recognised when it arrived.

Applied to Iran in 2026, that discipline would matter. A strike designed to degrade defined nuclear facilities or specific military assets is conceptually different from a campaign aimed at regime change or societal transformation. Trump could pursue tightly scoped coercive action tied to measurable conditions – for example, inspections, withdrawal from certain activities, or verifiable dismantlement. He should resist escalation toward occupation, open-ended air campaigns, or objectives framed in terms of remaking Iran’s internal order. The historical record suggests that limited goals paired with overwhelming force can succeed; expansive goals without a defined end state tend to expand beyond control.

Part two of that essay presented 25 fresh ideas of my own, such as legalising selling your kidney to the NHS, a National Obesity Service to pay people to lose weight, and a trial option of caning criminals, like Singapore.

A few weeks later, I found myself thinking of more and more ideas I should have said, until I had a second set of 25 fresh ideas, such as letting billionaires fund medical trials and abolishing the penny.

It then took me a few months to get to ideas 51 to 75, such as declaring war on the Caymans and paying children to get better exam results.

Ten months later and I’ve finally made it to 100 fresh policy ideas.

As a reminder, I’ve (mostly) stuck to the same rules as before: no boring tinkering of 1% tax increases, no niche things that affect 20 people, and no blowing wild holes in the budget, spending over £1bn to bribe the voters.

With that said, let’s dive in to something sensible.

A - Environment

1. Release Wild Bears into Scotland

Scotland currently spends £40–50m per year managing wild deer. Wolves and bears were exterminated by the 1700s. With no apex predators, red deer populations expanded rapidly and were never brought back under natural control. Today Scotland has around 2 million deer, far above ecological carrying capacity. Across the Highlands, native woodland fails to regenerate because saplings are eaten before maturity. Peatlands are damaged, reducing carbon sequestration. Riverbanks are stripped of vegetation, increasing erosion and downstream flood risk. The response is large-scale culling, and so each year 100,000–200,000 deer are shot.

The proposed policy is the reintroduction of apex predators. A phased programme would prioritise wolves, supported by lynx and brown bear populations in remote Highland zones. Predators reduce deer numbers, but their larger effect is behavioural: deer avoid open valleys and river corridors and move more frequently, reducing concentrated grazing pressure. This allows woodland, scrub, and riverbanks to recover without continuous human intervention.

When wolves were reintroduced to Yellowstone National Park in 1995, elk numbers fell modestly, but elk stopped grazing continuously in open valleys and along rivers. Vegetation recovered and riverbanks stabilised, reducing erosion and altering river shape. Beavers recolonised regenerated waterways, creating wetlands that supported fish, birds, and amphibians. Scavengers benefited from carrion left by wolf kills. Finally, tourism expanded sharply: wolf-related tourism alone now generates tens of millions of dollars annually for surrounding communities.

2. Legalise Kei Trucks

Anyone who has been on holiday to Japan will have noticed that cities, villages, farms, and building sites are full of tiny, boxy pickup trucks quietly carrying tools, plants, ladders, or rubble. These are kei trucks: lightweight utility vehicles under 700 kg, with engines capped at 660 cc. In the UK, vehicles like this are effectively banned by regulation. Instead, tradespeople, councils, and small firms are pushed towards 2–3 tonne diesel vans, oversized for short urban trips. This leads to higher fuel use, higher particulate emissions, more road wear, worse congestion, and vehicles built for motorway freight doing five-mile city jobs.

Kei trucks are built for exactly these tasks. They are cheap, simple, easy to repair, and ubiquitous across Japan’s construction, agriculture, and municipal services. Legalising them would allow councils and small firms to replace heavy vans with vehicles matched to real load requirements. Lower vehicle mass cuts energy use, tyre particulates, and brake dust. Smaller footprints reduce parking pressure and congestion. Electrification is also easier: batteries scale with weight, making small electric utility vehicles cheaper and less resource-intensive than large electric vans.

The environmental gains come from deregulation, rather than costly subsidies. Japan’s cities show that dense urban logistics do not require large vehicles. Kei trucks operate safely at urban speeds, coexist with pedestrians, and impose far less damage on roads. The UK already allows oversized SUVs for private use. Allowing genuinely small work vehicles would cut emissions, improve air quality, and reduce urban clutter by letting the right-sized tool exist.

B - Increase British TFR

3. Introduce Paid Grandparents’ Leave

The UK’s total fertility rate (TFR) – the average number of children per woman – is now ~1.4, well below the 2.1 needed for population stability. Survey data shows a persistent gap between desired and achieved family size of around 0.5 children per woman, driven less by attitudes than by cost, time pressure, and lack of support during the first years after birth.

The policy would introduce paid grandparents’ leave, tightly capped and deliberately modest. One grandparent would be eligible to take up to 2 days per week of paid leave for the first 6–12 months after birth. This targets the highest-stress childcare window without creating a full substitute for formal provision. Because the leave is part-time and time-limited, costs scale slowly. Even at statutory pay rates, this would be far cheaper than expanding nursery capacity or extending full parental leave, while directly easing return-to-work pressure for parents.

European studies find that access to reliable grandparental childcare increases the probability of a second birth by ~5–10%, particularly among women in full-time work. Sweden allows parental leave to be shared across caregivers, while Japan has trialled formal grandparent childcare support in response to a TFR below 1.3. The UK already relies heavily on grandparents. A small, cheap, targeted policy would reduce friction at the margin where fertility decisions are actually made.

4. The Baby Lottery

Small, predictable cash payments for having children have been tried repeatedly and have mostly failed. Hungary expanded family benefits heavily, fertility rose briefly, then fell back. The pattern is consistent: guaranteed money largely subsidises people who were already planning to have children. It does little to change decisions at the margin.

Human psychology massively overvalues small probabilities of large rewards. That is why lotteries work.

I propose we award £100,000 to the parents of every 500th child, provided at least one parent is a UK citizen. In 2023, the UK had roughly 600,000 live births, implying around 1,200 winners per year. The total cost would be £120m annually. If this increased the TFR by even 0.1, the fiscal return through a larger future workforce would dwarf the cost.

The social effect matters as much as the expected value. With 1,200 winners per year, almost every town would see a winner annually. People would know someone who won. Local news would cover it. Friends, colleagues, and relatives would talk about it. That creates salience in a way abstract policy never does. Most parents would never win, but many would feel that they might.

C - Health

5. Address Birth Trauma

Birth trauma affects a large share of women giving birth in the UK. Around 30–35% describe their birth as traumatic. Around 4–5% develop childbirth-related PTSD, equivalent to roughly 30,000 women per year. Rates rise sharply following emergency caesarean, instrumental delivery, severe tearing, unmanaged pain, or perceived neglect during labour. This places childbirth among the largest sources of new trauma cases each year.

Around 10% of mothers experience postnatal depression. Surveys report that roughly 20% of women with birth injury or trauma experience reduced capacity to work in the months following birth. The estimated lifetime cost of perinatal mental illness is ~£8bn per annual birth cohort, with around 70% of that burden falling on children through later developmental, behavioural, and educational impacts. Maternity-related clinical negligence accounts for approximately £1.3bn per year in NHS compensation payments.

This policy would introduce structured post-birth debriefs at 2 weeks and 6 weeks, using short trauma screening tools and direct referral pathways. Provide rapid access to trauma-focused CBT and EMDR for those screening positive. Expand continuity-of-carer midwifery models, where women see the same team across pregnancy and birth. Sweden and Norway run continuity models at much higher coverage than the UK and report childbirth-related PTSD rates closer to 1–2%. Several Australian states use structured debrief and trauma screening, showing lower PTSD symptoms at 6–12 months postpartum. These systems correlate with fewer chronic mental-health cases, higher maternal labour retention, and fewer severe incidents.

6. Heart Disease – A Double Kick

Heart disease remains the UK’s leading cause of death. It kills around 160,000 people per year, compared with roughly 150,000 from cancer. It also drives a large share of chronic illness, disability, and healthcare use. Cardiovascular disease costs the UK economy an estimated £19–£30bn annually, combining NHS spending, lost productivity, and informal care. Direct NHS costs alone are around £9–10bn per year, largely driven by preventable heart attacks, strokes, and long-term complications. Despite this, public attention and research funding skew heavily towards cancer, even though heart disease is more predictable and more cheaply preventable.

The first intervention is pharmacological: statins reduce major cardiovascular events by 20–30% in older populations. Age is the dominant risk factor, yet uptake remains patchy. The policy is to make statins free and routine for everyone over 65, with opt-out rather than opt-in prescribing. This mirrors practice in parts of the US, where preventive statin use is far higher. Large trials show serious adverse effects are rare, while the number needed to treat to prevent one heart attack or stroke falls sharply with age. At population scale, the cost of statins is low and the avoided cost of hospital admissions, surgery, and long-term care is large.

The second intervention targets diet: excess sodium intake raises blood pressure and cardiovascular risk. Average salt intake in the UK remains around 8.4g per day, well above the 5g guideline. The policy is to require food manufacturers to replace a proportion of sodium chloride with potassium-based salt substitutes in processed foods. Randomised trials show this reduces stroke risk by 10–14% and major cardiovascular events by ~13%, with minimal taste impact. Countries including China, Finland, and Peru have implemented salt-substitution or reformulation programmes at scale and recorded sustained blood-pressure reductions. Combined with statins, this delivers a compounding reduction in heart attacks, strokes, and NHS demand at low recurring cost.

7. Improve Air Quality in Public Buildings

Poor indoor air quality increases respiratory illness, cardiovascular disease, and cognitive impairment. Fine particulate matter (PM2.5) raises heart and lung mortality even at low concentrations. Elevated indoor CO₂ reduces concentration and decision-making speed. In schools, classrooms above 1,500 ppm CO₂ show lower attention and weaker test performance. Large studies find improved ventilation raises test scores by 3–8% and reduces absenteeism by 5–10%. In hospitals and care settings, poor ventilation increases airborne infection and staff sickness.

Set higher mandatory indoor air standards for public buildings covering ventilation rates, CO₂ limits (eg below 1,000 ppm), and particulate filtration. Require continuous CO₂ monitoring in schools, offices, and hospitals. Upgrade ventilation and install HEPA filtration in high-occupancy spaces. Portable HEPA units cut indoor PM2.5 by 40–60%. Ventilation upgrades typically cost c. £20 per m² as a one-off, with modest running costs when demand-controlled systems are used.

Finland, Germany, and Japan enforce high ventilation standards in public buildings. Finland recorded lower respiratory illness and reduced teacher sick leave after implementation. Economic modelling finds benefit–cost ratios above 5:1, driven by fewer sick days and higher productivity. A 1% reduction in absenteeism across schools and public-sector workplaces would save hundreds of millions annually.

8. Put Lithium in the Water

Mental illness and suicide impose large, persistent costs. Around 6,000 people die by suicide each year in the UK. Depression and mood disorders drive NHS spending, welfare use, and lost productivity. Lithium has been used for decades in psychiatry to stabilise mood and reduce suicide risk. Less discussed is trace lithium in drinking water. At concentrations thousands of times lower than clinical doses, naturally occurring lithium shows population-level associations with better mental-health outcomes, without sedation or dependence.

Studies from Japan, Austria, Texas, Greece, and Denmark compare regions with different natural lithium levels in water. Areas with higher concentrations consistently show 10–20% lower suicide rates, even after adjusting for income, unemployment, and healthcare access. A large Danish cohort study also found lower dementia incidence at modestly higher exposure.

Adding trace lithium at these levels would cost pennies per person per year, using existing water-treatment infrastructure. Even a 10% reduction in suicides would save hundreds of lives annually and reduce emergency care, long-term mental-health treatment, and productivity losses. The individual effect is small, but the population effect is large. Few interventions offer such low delivery cost with broad mental-health gains.

9. Triple P Parenting

Child behaviour problems drive large downstream costs. Around 5–10% of children meet criteria for conduct disorder, with much higher rates of school exclusion, later criminal justice contact, and adult mental illness. Parenting quality is one of the strongest predictors. Poor early behaviour multiplies costs across education, health, and policing. In the UK, children with persistent conduct problems cost the state £70,000–£100,000 more each by adulthood than their peers, largely through special educational needs, mental-health services, and later offending.

Triple P (Positive Parenting Program) is a structured, evidence-based parenting system delivered at different intensities, from light-touch advice to targeted support. It focuses on consistency, predictable consequences, and emotional regulation. Randomised trials show 20–35% reductions in child behaviour problems, improved parental mental health, and lower rates of abuse and neglect. Effects appear across income groups and persist for years when delivered properly. The programme scales well because most families need only low-intensity support.

Triple P has been rolled out nationally in Australia, New Zealand, parts of the US, and Ireland. Large evaluations show reductions in child maltreatment rates of 10–25% and sustained drops in behaviour-related school and health referrals. Cost–benefit analyses consistently find returns of £4–£7 for every £1 spent, driven by avoided special education, social care, and criminal justice costs. Delivered early, this is one of the cheapest ways to reduce long-term state failure.

10. Adopt the Swedish Malpractice Model

Clinical negligence has become a balance-sheet problem for the NHS. Outstanding liabilities now sit at around £80–90bn, larger than recent annual NHS budget increases and still rising. Annual cash pay-outs are £2–3bn, with maternity cases alone accounting for roughly £1.3bn per year. These figures reflect long-tail liabilities rather than day-to-day care, and they grow automatically even when clinical quality improves. The system locks the NHS into decades of future payments, crowding out investment in staff, equipment, and prevention.

A large share of this spend never reaches patients. Legal fees, expert witnesses, court costs, and administration consume around 30–40% of total negligence spending. Cases take years to resolve, compensation is uneven, and incentives skew towards defensive medicine. The system is adversarial, slow, and expensive by design.

Sweden uses a no-fault malpractice model. Patients receive compensation when harm occurs, without proving negligence, through an insurance-based scheme. Legal costs are minimal, claims resolve faster, and compensation is more predictable. Administrative costs are a small fraction of UK levels, and total spending on medical injury compensation is far lower per capita. Adopting this model would cut legal overhead sharply, stabilise NHS liabilities, and redirect billions from lawyers to patients and prevention.

11. Nursing Reform

The UK has a chronic nursing shortage. Vacancies sit at around 40,000 posts, with worse gaps in acute, mental-health, and community care. Short staffing raises mortality, increases medical error, and lengthens hospital stays. NHS trusts spend £3–4bn per year on agency and bank staff, often paying 2–3× standard hourly rates. Reliance on temporary staff reduces continuity of care and accelerates burnout among permanent nurses, feeding a self-reinforcing shortage cycle.

Nursing remains structured around a single academic route, poorly suited to mid-career entrants. Reform should open multiple entry paths. Expand paid, employer-led training routes similar to Germany and Switzerland, where trainees earn while qualifying. Create accelerated conversion programmes for healthcare assistants, paramedics, and science graduates. Remove unnecessary placement bottlenecks and fund supervised practice directly within hospitals. These changes shorten training time, widen the applicant pool, and reduce drop-out.

Retention matters as much as recruitment. Establish nurse hubs modelled on Denmark, combining scheduling support, childcare coordination, and career development in one place. Introduce discounted key-worker housing near major hospitals in all large cities, targeting early-career nurses who face the highest living costs. Countries using these models show higher retention and lower agency spend. Every 10% reduction in agency reliance saves hundreds of millions annually. Nursing reform pays for itself through staffing stability, safer care, and lower long-run costs.

12. Local Clinical Triage Centres

We’ve all had the nightmare of needing to book the GP: the phones open at 8:00am, by 8:01 there’s a 15-person queue and by 8:04 there’s none more to be booked.

My brother is a GP, and his surgery actually solved this problem. Their simple solution was to have one rotating doctor permanently answer the phone. As a clinician answers first, it resolves many calls immediately, and decides who actually needs an appointment. That cuts repeat calls, removes reception bottlenecks, and stops minor problems becoming urgent through delay.

Other countries show how much demand can be filtered safely before it reaches a GP. In the Netherlands, nurse-led telephone triage resolves around 50–60% of patient contacts without a GP appointment, with emergency referral rates falling by 15–30% after adoption. Studies show no increase in adverse outcomes. In Norway, nurse triage in out-of-hours services reduces GP consultations by around 20% and emergency department attendance by 10–15%, while maintaining high patient safety. Both systems generate reliable data on call volumes, symptoms, outcomes, and escalation rates, allowing continuous improvement rather than blind rationing.

My proposed solution is to merge these models locally. Create clinical triage centres covering seven or eight surgeries, staffed by around 10 triage nurses supported by two doctors. All first-line calls go to nurses, and then 10–20% of calls escalate immediately to a doctor when the presentation is unclear, risky, or atypical. Doctors focus on complexity, nurses handle volume, and patients get faster decisions. This ends the 8am scramble, reduces A&E leakage caused by access failure and frees GP time for face-to-face care.

D - Housing

13. Legalise Sash Windows and Air Conditioning

I’ve already written at length about the tyranny of the UK building code, but if I could change just one item in it, it would be the window regulations. England has effectively banned traditional sash windows and made mechanical cooling regulatorily and culturally hostile, even as summers get hotter. This matters because heat kills. In the UK, heatwaves now cause roughly 2,000–3,000 excess deaths in severe summers, concentrated among the elderly and those in poorly ventilated housing. Across Europe, the 2022 heatwave caused around 60,000 excess deaths. Heat-related mortality is rising faster than cold-related deaths are falling.

Current rules make adaptation harder than necessary. Building regulations prioritise airtightness and winter heat retention, and new-build standards often require passive cooling measures before mechanical cooling can be considered. In London, the mayor’s planning policies have explicitly discouraged the inclusion of air conditioning in housing designs because of concerns about energy usage and a preference for ventilation and passive approaches. Local planning permission is often triggered for standalone cooling units, especially in conservation areas or for older buildings, discouraging installs.

We must reverse these defaults. Legalise sash windows, high-opening designs, and effective shading as of right. Treat air conditioning and efficient air-to-air heat pumps as health adaptations and permit their installation by default, particularly in bedrooms, care homes, and flats. Modern inverter AC units make cooling vastly cheaper than emergency healthcare and lost productivity in heatwaves. Countries such as Japan, much of the southern United States, and parts of the Mediterranean treat AC as standard safety infrastructure. Bringing British regulation into line with this reality would cut heat deaths, improve sleep and productivity, and make housing fit for the climate Britain now faces.

14. Listed Building Reform

England has over 370,000 listed building entries. Around 9,000 are Grade I, 21,000 are Grade II* , and about 339,000 are Grade II. Listing places legal controls on what owners can change, both externally and internally. Any alteration that affects the building’s “special interest” requires listed building consent from the local authority. This applies equally whether the building is a cathedral or a modest house.

Scotland operates a similar framework using Categories A, B, and C, covering 46,794 listed buildings. In both systems, listing often applies to the entire structure rather than specific features. That means owners need permission for minor, everyday changes: replacing windows, adding vents, upgrading heating, installing secondary glazing, or removing unsympathetic later additions. Local councils act as gatekeepers, interpreting heritage impact on a case-by-case basis, often conservatively.

This structure creates friction for ordinary people. Applications take months, require specialist consultants, and are frequently blocked or delayed for marginal reasons. One estimate suggests councils spend around 4,000 working days per year handling minor retrofit applications, with only around a third decided within the 8-week target despite most being approved. We should narrow that scope. Keep tight controls for Grade I and Grade II* buildings and Scotland’s Category A. For the rest, introduce a clear permitted-works list covering routine upgrades that preserve appearance while improving liveability. This would protect genuinely important heritage while letting people maintain and adapt their own homes.

E - Education

15. University Admin Purge

UK universities have shifted staff composition away from teaching and research. In 2003–04, there were roughly 1.7 non-academic staff per academic. By 2021–22, that ratio had risen to around 2.5 to 1. Over the same period, student numbers rose modestly, students per academic increased, and students per administrator fell sharply. This expansion was driven by internal growth in compliance, reporting, student services, marketing, HR, safeguarding, and managerial layers.

The outcome has been worse conditions for both staff and students. Non-academic roles now account for over half of total university payroll, while academic pay has fallen sharply. Between 2009 and 2022, average real-terms academic pay dropped by 15–20%, even as total staffing costs rose. At the same time, student satisfaction failed to improve. National Student Survey results have hovered in the low-80% range for over a decade, with declines in categories tied most closely to academic staff, particularly teaching quality and assessment feedback. Fees rose, admin grew, and the core service stagnated.

Administrative growth also reshaped university culture: risk-aversion expanded rapidly: safeguarding, compliance, and reporting systems multiplied, often without clear evidence of benefit. Demand for formal counselling and support services has risen three- to four-fold over roughly the same period, partly reflecting the displacement of informal academic support by bureaucratic systems. My fix is to set target ratios for non-academic to academic staff and require institutions above them to shrink back over time. Merge administrative functions across institutions and sunset compliance roles unless they demonstrate measurable improvements in teaching, research, or completion. Universities exist to teach and research. Their staffing should reflect that again.

16. Retrain 90% of Primary School Classroom Assistants as Tutors

Primary schools in England employ large numbers of classroom assistants. There are over 180,000 teaching assistants in primary schools alone. These roles are intended to support classroom teachers, but large evaluations have found weak effects when assistants are used mainly for general classroom support or supervision. At the same time, a strong evidence base shows that structured small-group tutoring delivers substantial gains in reading and maths. Randomised trials, including UK pilots, consistently find that small-group tutoring produces the equivalent of two to four additional months of learning per year.

The policy is to retrain 90% of classroom assistants as small-group tutors working with groups of two or three pupils throughout the week. Their role would be tightly defined: tutors would deliver short, frequent sessions focused on core skills: phonics, reading fluency, basic writing, number sense, and arithmetic. They would use standardised, scripted materials aligned to the school curriculum, with clear progression and frequent checks for understanding. This does not require specialist teachers. The content is narrow, repetitive, and diagnostic rather than creative. Teachers retain responsibility for whole-class teaching and planning; tutors focus on targeted catch-up and consolidation.

Retraining existing assistants uses an already-funded workforce. With around 180,000 assistants, even after exclusions and attrition roughly 150,000 tutors could operate nationally. If each tutor works with three pupils at a time for four sessions per week, millions of tutoring sessions could be delivered annually. The learning gains from small-group tutoring are large by education standards - the effect is bigger than reducing class sizes and most ed-tech interventions. Few school-based interventions outside early phonics match this impact. At scale, this would materially reduce attainment gaps, cut later remediation costs, and raise basic literacy and numeracy without expanding headcount.

17. An Ultra Grammar School Network

The original grammar school system selected around 25% of pupils at age 11. That structure placed three quarters of children outside the academic track. Secondary moderns became an early path to failure, expectations fell, and the social signal dominated educational outcomes. When one child in four is labelled “top”, the remaining majority absorbs a clear inverse message. Political support collapsed and the system was replaced by comprehensive schooling.

A new grammar network should operate at a much smaller scale. Selection should be limited to around 2% of students nationally. At that level, missing out reads as normal. These schools would be few, explicit, and elite, with fast-paced curricula, advanced maths and science, heavy reading loads, and teachers trained to push the very top end. Entry would be based on sustained academic performance across several years. The curriculum could be open source, with explicit places reserved each year for late developers to join as the cohort grows.

Separating the top few percent produces large academic gains. Research on accelerated and high-ability grouping shows that students taught together at an appropriate pace finish secondary school close to a full academic year ahead of equally able peers in mixed-ability settings. Even simpler grouping without acceleration delivers gains of four to five additional months. These advantages compound. Students reach degree-level material earlier, take harder subjects at school, and arrive at university prepared for genuinely advanced work. Breakthrough innovation and scientific leadership come from a very small share of the population. Developing that share raises the national ceiling and feeds growth across the whole economy.

18. Mandatory Minimum Quotas of Injured Children

Over the past thirty years, childhood in the UK has become dramatically safer in physical terms. Unintentional injury rates among children have fallen by roughly 50% since the early 1990s. Playground equipment has been softened, climbing trees discouraged, contact play restricted, and adult supervision expanded. Over the same period, mental health outcomes have moved sharply in the opposite direction. The share of children aged 6–16 with a probable mental disorder rose from around 10% in 2004 to nearly 20% by the early 2020s. Hospital admissions for self-harm among teenagers roughly doubled between 2000 and the late 2010s. Jonathan Haidt’s argument is that these trends are linked: remove everyday physical risk, and children fail to learn how to assess danger, regulate fear, and recover from mistakes.

Behavioural studies consistently find that children who engage in unsupervised or risky play show better emotional regulation, lower anxiety, and stronger social skills. Countries and communities with higher levels of independent outdoor play report lower childhood anxiety and fewer stress-related disorders. Meanwhile, the near-elimination of minor injuries has coincided with a surge in institutional responses to perceived risk: bans on games, constant adult monitoring, and a school culture that treats all physical danger as failure. Scrapes and bruises have declined; panic attacks and school refusal have risen. The risk profile has shifted from visible, low-cost harm to invisible, long-term damage.

Schools would be expected to record a minimum rate of minor injuries per pupil per year, limited to trivial incidents requiring basic first aid. Schools falling far below the expected range would trigger inspection for excessive risk aversion rather than praise. Climbing frames, rough play, contact sports, and unscripted outdoor activity would return as normal parts of childhood. A grazed knee at eight beats an anxiety disorder at sixteen. Children who fall early, learn faster, and grow up steadier cost the system less and cope with the world better. Where schools fail to improve, we draft in the bears from policy 1.

19. BBC 2.0

The BBC faces a long, visible decline. Linear TV audiences have fallen sharply over the past decade, younger viewers have largely disengaged, and trust has been eroded by repeated rows over political bias, overpaid presenters, and internal scandals. Spending has drifted away from education, culture, and serious factual output towards entertainment formats and star-led programming. Multiple reforms have been attempted. Editorial guidelines have been tightened and rewritten. Senior management has been reshuffled. None of this has reversed audience decline or restored broad legitimacy.

The core problem is structural - a single institution tries to be everything at once, while fighting internal culture wars and defending legacy talent contracts. BBC 2.0 would be built in parallel as a clean-slate public broadcaster with a new charter, new staff, and a narrow remit. Political and current-affairs output would be explicitly pluralist by design. One third of the political budget would fund left-leaning programming, one third centrist, and one third right-leaning. Each audience would know there are programmes made for them, without pretending to a single neutral voice. Five regional production hubs would anchor commissioning outside London, spreading jobs and talent across the country.

Spending priorities would be hard-coded: entertainment would be capped at 15% of production spend. There would still be room for mass-appeal formats and talent-development shows, but the centre of gravity would shift back to factual, educational, science, arts, and high-end historical drama. BBC 2.0 would be smaller than the current BBC and explicitly temporary in ambition: a pipeline for writers, producers, journalists, and presenters to move on to the private sector, rather than a place to warehouse stars on seven-figure salaries. Over time, funding would be transferred gradually from the existing BBC to BBC 2.0 as audiences and trust followed. This approach reforms public broadcasting without trying to fix an institution that has lost the ability to fix itself.

F - Economy

20. Mandatory HR Externalisation

The expansion of HR in the UK is largely a post-2008 phenomenon. Since the financial crisis, HR roles have grown far faster than almost every other occupational group. There are now around 470,000 people working in HR in the UK, an increase of roughly 65–70% since 2010, according to ONS-based estimates. Over the same period, total employment grew by around 10–15%, and growth in core private-sector functions such as operations, sales, and production was far weaker. HR now accounts for nearly 1.5% of the UK workforce; among advanced economies, only the Netherlands has a larger proportion of its workforce in HR roles.

This growth has predictable effects on how organisations operate. HR is oriented toward risk avoidance, and once embedded it expands its remit. Hiring slows as procedures multiply. Discipline becomes legalistic. Training becomes mandatory rather than useful. Each incident produces a policy; each policy produces compliance work; compliance work justifies further expansion. Unlike operational functions, HR rarely contracts, because its internal success metric is the reduction of risk rather than the delivery of output. Over time, speed, judgement, and managerial authority are crowded out by process.

There is a clear precedent for fixing this. Internal audit functions once expanded in a similar way, producing box-ticking, conflicts of interest, and bloated teams. The solution was mandatory externalisation. Audits were separated from management, priced as discrete services, and exposed to competition. Quality improved and internal sprawl stopped. The same model should now apply to HR. Any organisation with more than five internal HR staff would be required to externalise the function. Providers would unbundle and sell services individually: recruitment support, disciplinary advice, compliance guidance, training. Firms would buy what they need rather than carry permanent overhead. External providers would compete on judgement, speed, and proportionality. The aim is to streamline organisations, restore managerial discretion, and raise tolerance for productive risk.

21. Kill Off Subsidy Miners

Across the UK economy, a growing class of firms exists primarily to extract subsidies rather than produce value. These “subsidy miners” cluster around planning gain, renewable incentives, government procurement, regional development grants, and complex tax reliefs. In energy alone, UK renewable support schemes have involved tens of billions of pounds over the past two decades. Planning gain mechanisms such as Section 106 and the Community Infrastructure Levy regularly add £30,000–£70,000 per home in high-demand areas, creating entire businesses focused on navigating, arbitraging, and monetising the system rather than building efficiently.

Annual consultancy spending by the UK government runs at £20–25bn, much of it tied to managing compliance and bids for public programmes. In farming, over £3bn per year flows through subsidy schemes, with a large share captured by landowners and intermediaries instead of increasing productivity. In housing, energy, and transport infrastructure, firms invest more in grant eligibility, legal positioning, and lobbying than in cost reduction or innovation. Talent follows the money: engineers become bid writers; builders become form-fillers. This dynamic increases cost inflation, slows delivery, and dulls competition.

The policy response has two parts:

• First, compress, cap, and sunset subsidies wherever possible. Any subsidy running longer than five years must be reauthorised with published cost-per-outcome metrics. Schemes that reward activity rather than measurable outcomes should be scrapped or redesigned.

• Second, introduce a rent-seeking clampdown regime. Firms that derive more than 30% of revenue from public programmes over a rolling five-year period must be designated “subsidy-dependent” and face special rules, including reduced eligibility for new grants. Similar approaches exist in Nordic countries, where government-procurement officials screen for subsidy dependence and can disqualify bids from firms that rely excessively on public programmes. The aim is to force firms back toward making money by building, selling, and improving things, rather than mastering paperwork.

22. Yozma UK

The UK is very good at basic research and very bad at turning it into companies. We produce around 7% of the world’s academic research but capture only 2–3% of global venture capital value. Public money funds discoveries, then the commercial upside is often realised elsewhere. Israel and Singapore faced similar problems and fixed them directly, using the state as a catalyst rather than a permanent owner.

Israel’s solution was Yozma, launched in 1993 with just $100m of public capital. The government co-invested alongside private venture funds and offered them an option to buy out the state’s stake at a discount if investments succeeded. Within a decade, Israel’s venture capital market grew more than ten-fold. Today Israel has the highest number of start-ups per capita in the world and spends over 5% of GDP on R&D, with technology exports forming a core part of the economy. The state made money, private capital followed, and the programme wound itself down.

Singapore’s equivalent is A*STAR, a network of applied research institutes closely linked to industry. It employs around 5,000 researchers and focuses on turning lab work into manufacturable products. A*STAR has spun out hundreds of companies and helped anchor major semiconductor, biotech, and advanced manufacturing firms in Singapore. The model is explicit: fund applied science, embed researchers with firms, and measure success by commercial uptake rather than papers published.

The policy is to build a UK version with a narrow remit. A single national applied-research and venture platform would co-invest alongside private capital, focus on scale-up rather than discovery, and take equity stakes rather than issue grants. Funding would be capped, time-limited, and recyclable. Success would be measured in firms created, exports grown, and private capital crowded in. We could keep the upside of British ideas in Britain, and turn world-class research into world-class companies rather than footnotes in someone else’s success story.

23. A public R&D Grand Solutions Lab

The UK already funds research, but it does so in a fragmented and cautious way. Programmes tend to favour broad, visible themes where private investment is already active and political credit arrives quickly. The real value of public R&D lies elsewhere: problems with clear social benefit, long timelines, high uncertainty, and weak incentives for private capital. These are precisely the areas where markets fail and where state-backed ambition can unlock breakthroughs rather than marginal gains.

A small number of “grand solution” programmes would be funded at meaningful scale and given long time horizons. Each programme would unite academic researchers, engineers, and industry around a concrete target rather than a research field. Initial candidates could include:

• Carbon-negative steel and cement
  Steel and cement together produce around 14% of global CO₂ emissions, much of it baked into the chemistry itself. Clean alternatives currently cost 40–70% more than conventional production, making private scale-up unattractive. A UK programme could aim to build a pilot plant producing 1 million tonnes a year at cost parity within ten years, roughly 2% of UK output, proving decarbonisation without offshoring.

• Antibiotic resistance
  Antibiotic resistance already causes 1.3 million deaths a year, with forecasts of 10 million by 2050. No new antibiotic class has reached patients since 1987. Development costs exceed £1bn, while successful drugs often earn under £30m a year, leaving the market broken. A focused public programme could target new mechanisms, share screening infrastructure, and restart progress where private capital has withdrawn.

• Ultra-deep geothermal drilling
  Geothermal power delivers 90%+ capacity factors, yet supplies under 1% of global electricity. Most accessible heat lies below 10km, where temperatures destroy conventional tools. Experimental wells cost £10–20m each, stalling private investment. A public programme could turn this from an extraction problem into an engineering one, unlocking constant, zero-carbon baseload energy.

• Next-generation desalination
  Two-thirds of the global population already face severe water stress for part of the year, rising to three-quarters by 2050. Desalination supplies water to 300 million people but uses 3–4 kWh per cubic metre and produces large volumes of brine waste. A UK programme could focus on advanced membranes and low-grade heat, strengthening both exports and domestic water security.

The lab’s role would be to bridge the gap between research and deployment. Funding would be large, time-limited, and tolerant of failure. Success would be measured by working pilots, cost parity, and real-world uptake rather than papers published. Methods would be left to researchers; goals would be explicit. Most projects would fail. A few would change entire industries. Progress comes from giving experts permission, backing ambition, and accepting failure as the price of discovery.

G - Government

24. The Department for Ungovernment

As I wrote about recently, the modern British state has a ratchet problem. New rules, bodies, and processes are easy to create and almost impossible to remove. Over time, this accumulates into a dense web of permissions, reporting requirements, and precautionary checks that slow everything down. No individual regulation looks unreasonable in isolation. The problem is their combined weight. Britain now struggles to build housing, infrastructure, power generation, or public services at speed, even when there is political agreement and funding in place.

The Department for Ungovernment would exist for one purpose: systematic removal. Its job would be to identify rules, approvals, and processes whose original justification no longer applies, whose costs outweigh their benefits, or whose effects duplicate other controls. Each year it would publish a rolling list of repeals, simplifications, and mergers across departments, with ministers required to actively defend any rule they wish to keep. Silence would mean deletion. The burden of proof would flip.

This would change incentives across the state. Civil servants would know that new rules face future scrutiny rather than permanent survival. Departments would be forced to prioritise outcomes over process. The goal is not chaos or deregulation for its own sake, but restoration of state capacity. Governments that can only add layers eventually lose the ability to act. A department dedicated to subtraction makes everything else possible again.

25. The Department for McKinsey & Co

The UK state spends an extraordinary amount on external consultants. Annual public-sector spending on management consultancy now sits at around £20–25bn, depending on the year, with large spikes during crises that never fully unwind. Once embedded, consulting contracts tend to renew, expand, and fragment across departments, creating dependency rather than capability.

The policy response is to internalise the skillset. A new “Department for McKinsey & Co” would be created inside the Civil Service as a permanent internal consultancy. It would recruit heavily from top-tier consulting firms, targeting managers and engagement leads rather than partners. The offer is explicit: around a 25% pay cut in exchange for roughly a 50% reduction in working hours, predictable schedules, and public-service prestige. This is an attractive trade for a large pool of experienced consultants approaching burnout. The department would be organised along standard consulting lines: finance and turnaround, digital transformation, operations, procurement, HR and workforce reform.

Crucially, the department would have first right of refusal on any major outsourcing or consultancy spend across government. External consultants could still be used, but only where the internal team formally declined or lacked capacity. Over time, institutional knowledge would accumulate instead of walking out the door every six months. If the department displaced even 20–25% of current consultancy spending, it would save £4–6bn per year by the end of a single parliament, while improving continuity and accountability. Governments outsource because they lack capacity. Build the capacity, and the money stops leaking.

Conclusion

There you have it, 100 policy ideas, fresh to the UK political conversation. This was done by one nerd in his spare time in under a year. At the next election, none of us should be settling for the complete lack of thinking from our political parties.

The list above was not meant to win your support for every one, but to push you to half-adopt a couple with your own twist, and push policy ideation forward.

In 2025, at Coachella Valley Music and Arts Festival in California, tens of thousands of people chanted against Margaret Thatcher, led by performers born in the 1990s. The crowd skewed even younger. Thatcher had finished her 11-year premiership before any of them were alive.

A year earlier, her statue in Grantham was defaced with paint.

This is not especially unusual. Thatcher remains one of the most hated figures in British political culture, decades after leaving office. For many younger people, she exists as a fixed symbol: hard-right, cruel, neoliberal, evil.

But how many of the people chanting could describe her actual economic policies in any detail? How many could explain why a British electorate, exhausted and angry, put her into power in the first place? How many know what the country looked like in the decade before 1979 – the inflation, the strikes, the power cuts, the emergency measures, the repeated sense that government no longer worked?

Thatcher was a reaction to a specific economic and institutional breakdown that is now largely forgotten. To understand why Britain changed in the 1980s, you first have to understand how broken it felt in the 1970s.

The post-war boom

For roughly two decades after 1945, Britain experienced steady improvements in living standards: real wages rose; mass unemployment disappeared; home ownership expanded. Consumer goods that had once been luxuries became normal. By the late 1960s, most households were materially better off than their parents had been.

In this world, the state played a large role in managing demand. Key industries were nationalised, trade unions were integrated into wage bargaining, and governments of both parties prioritised full employment, even at the cost of higher inflation. For much of the 1950s and early 1960s, this worked well enough.

But beneath the surface, Britain’s performance was already lagging. Productivity growth was weak compared with France and West Germany. Investment rates were low. Manufacturing struggled to modernise. Exports underperformed.

By the mid-1960s, Britain repeatedly ran into balance-of-payments problems. Governments responded with familiar tools: spending cuts, credit restrictions, wage restraint, then a return to expansion once the immediate crisis passed.

The 1967 devaluation of sterling was a turning point. It was intended to restore competitiveness, but it also signalled that Britain’s economic model was no longer stable.

By 1969, Britain was not poor, but it was brittle. Living standards were still high by historical standards, and employment remained strong. Yet growth depended on continual intervention, wage restraint was already politically fraught, and confidence in economic management was eroding. The post-war settlement had delivered gains, but it was increasingly reliant on improvisation to hold together.

The labour market

By the end of the 1960s, trade unions sat at the centre of British economic life. Union density was c. 58% of the workforce, and in sectors such as coal, steel, rail, docks, power generation, and car manufacturing it was far higher. In many large firms, almost every worker belonged to a union.

Pay bargaining was commonly conducted at industry or national level. Agreements covered wages, hours, job classifications, and working practices across entire sectors. Individual firms had limited room to diverge.

Closed shops were widespread: in many workplaces, union membership was a mandatory condition of employment, either on hiring or after a short qualifying period. Losing union backing could mean dismissal even when employers wished to retain a worker. By the mid-1970s, an estimated 5–6m workers were employed in closed-shop arrangements.

Working practices accumulated over time and proved difficult to remove:

• In British Leyland plants, tasks were tightly demarcated, with separate grades responsible for narrowly defined actions. Workers could refuse to move a tool a few feet or perform a task assigned to another grade. In some car factories, disputes arose over who was permitted to change a lightbulb or operate a particular machine.

• In the docks, the system of registered dock labour restricted who could unload ships and under what conditions. Disputes over manning levels and job allocation regularly delayed cargo, with knock-on effects across supply chains.

• In printing, long-standing compositors’ rules limited the introduction of new technology well into the 1970s, contributing to repeated newspaper shutdowns.

Industrial action extended beyond single workplaces. Sympathy strikes and secondary action allowed workers to stop work in support of disputes elsewhere. Lorry drivers might refuse deliveries to a struck plant. Power station workers might reduce output during disputes involving miners.

Governments managed the economy through negotiation with this system. When inflation accelerated and real wages came under pressure, the system became harder to manage, as settlements in one sector fed quickly into demands elsewhere

The wage–price spiral

By the late 1960s, Britain’s inflation problem was already developing, but the institutional context is easy to miss today. For much of the post-war period, currencies were not freely traded; exchange rates were set politically.

Defending a chosen exchange rate required constant intervention. When Britain imported more than it exported, pressure built on sterling. To hold the rate, governments raised interest rates, restricted credit, and cut domestic spending.

The system failed visibly in 1967, when the government devalued the pound by 14%. The decision instantly raised the price of imports. Britain relied heavily on imported food, fuel, and industrial inputs. Devaluation increased the cost of oil, fertiliser, animal feed, machinery, and components in a single step.

In 1971, the Bretton Woods international finance system collapsed, ending fixed exchange rates between major currencies. Sterling moved into a more uncertain managed float, with confidence swings feeding directly into prices.

In 1973, the situation deteriorated sharply. Following the Yom Kippur War, several oil-producing states in the Middle East cut production and restricted exports to Western countries. The global oil price roughly quadrupled within a year. Britain, still heavily dependent on imported oil, faced an immediate rise in energy costs.

Transport costs rose across road, rail, and shipping. Industrial users faced sudden increases in input prices.

By 1974, inflation exceeded 20%.

Wage bargaining struggled to keep pace. Pay deals were typically agreed once a year. Inflation eroded real wages between settlements. A worker receiving a 10% rise at the start of the year could see prices rise faster than pay before the next negotiation.

Large settlements in one sector set expectations elsewhere. In 1974, miners secured a 35% pay increase, quickly followed by similar demands in other industries. Employers raised prices to fund higher wages. Retail prices adjusted upward.

By the mid-1970s, inflation had become a dominant feature of daily life. Pay packets, energy bills, and prices adjusted rapidly, often within the same year. Economic planning relied on short-term fixes rather than stable expectations.

In 1970, a pint of beer was 10p and an average car c. £550. By 1979, the beer was up to 37p and the car to £2,400.

For the equivalent, imagine London beer prices reaching £27 a pint by 2034.

Fighting inflation

By the early 1970s, both major parties were committed to controlling pay directly. Governments attempted to slow inflation by limiting wage increases across the economy, intervening in disputes, and approving or rejecting individual settlements. The broad approach remained consistent as administrations changed.

These policies reached deep into ordinary employment relationships. Pay limits applied regardless of sector or circumstance. That meant that if an individual employee, working for a private company with no union involvement, agreed with their employer to a pay rise above the permitted ceiling, that agreement could be unlawful. The state reserved the right to block it.

Between 1961 and 1974, Britain operated under some form of prices and incomes policy for around half the period. In 1972, the Conservative government imposed a legally binding system. Pay rises were capped at fixed amounts rather than percentages.

A substantial administrative apparatus supported this. The Prices and Incomes Board examined wage claims, reviewed pricing decisions, and issued rulings across large parts of the economy.

By 1973, inflation was running at over 9% and rising quickly. Pay controls lagged behind prices. Governments responded by refining formulas, tightening thresholds, and returning repeatedly to negotiation.

State involvement expanded while control diminished.

The three-day week

In early 1974, disputes in the coal industry triggered a national energy emergency. The National Union of Mineworkers had pressed for large pay rises; an overtime ban reduced coal output sharply. Coal still produced the bulk of Britain’s electricity. When stocks ran low, generating capacity fell and power supplies became uncertain. To conserve energy, the government introduced the three-day week on 1 January 1974, limiting commercial use of electricity to three consecutive days each week.

The measure was literal: businesses could only use electricity on their assigned days, and working hours were capped. Essential services such as hospitals, supermarkets, and newspaper printing presses were exempt, but most factories and offices had to plan around power allocations. Television broadcasters were required to end transmissions early to save energy.

One local newspaper reported that areas such as Lowestoft and Diss saw thousands of workers temporarily idle because of the three-day week; in one district more than 2,000 people were put out of work or temporarily stopped as a direct result of the electricity rules.

Some businesses attempted 12-hour shifts on allotted days to squeeze more production into limited hours, while others simply closed for two days at a time.

The toll on government negotiators was visible. Amid endless talks with unions, senior officials became exhausted and erratic; a widely reported episode during this period saw the Cabinet Secretary, Sir William Armstrong, so overwhelmed by the pressure of negotiations that he was found lying on the floor of Number 10 Downing Street and subsequently taken for medical treatment. On another occasion he started stripping naked during a meeting.

The crisis carried straight into politics. In February 1974, with the three-day week still in force and negotiations stalled, Edward Heath called a general election under the question “Who governs Britain?”. The answer was unclear. The Conservatives won the largest share of the vote, Labour won the most seats, and neither secured a majority. Heath attempted to remain in office, then resigned days later.

The change of government did not resolve the underlying problems. Industrial disputes continued through the spring and summer. Inflation remained high. Pay negotiations restarted under a new administration with limited credibility and little room to manoeuvre. By autumn, political authority remained fragile and parliamentary arithmetic unstable. In October 1974, Britain went back to the polls for a second time in the same year.

Two general elections in one year was highly unusual. The previous occurrence had been in 1910. In 1974, it reflected a system struggling to govern amid economic pressure, industrial conflict, and exhausted institutions.

North Sea oil and the IMF crisis

North Sea oil is often treated as a lost inheritance, or as something later handed away by Conservatives. The chronology matters. Britain discovered oil in the late 1960s, but large-scale production arrived slowly. During the worst years of inflation and industrial conflict, oil revenues were limited and arrived too late to stabilise the economy.

Production began to rise meaningfully from the mid-1970s. Output increased from roughly 13m tonnes in 1976 to around 80m tonnes by 1980. In the early years, Britain still imported much of its energy and faced persistent balance-of-payments pressure.

The ownership framework also cuts against later narratives. It was a Labour government, under James Callaghan, that decided against full nationalisation. The British National Oil Corporation took stakes and oversight roles, but production and profits largely flowed through private companies. The basic structure that allowed oil revenues to accrue to private firms was in place before 1979.

By 1976, inflation had recently exceeded 20%, sterling was falling sharply, and foreign exchange reserves were draining. International confidence weakened. In September 1976, the government applied to the International Monetary Fund for a loan of around $3.9 billion, the largest the IMF had ever approved at that point.

The episode carried a strong sense of national embarrassment. Contemporary reporting described Britain going “cap in hand” to an external lender, with policy conditions negotiated abroad rather than set domestically. The requirement to agree spending cuts and fiscal restraint reinforced the impression that economic control had slipped away.

The Winter of Discontent

By the winter of 1978–79, the strain built up over the decade finally spilled into daily life. Inflation had eased from its mid-1970s peak, but pay restraint remained in place. Public-sector workers were asked to accept limits that lagged prices after years of erosion. Many refused.

Strikes spread quickly. Lorry drivers walked out, disrupting food distribution. Petrol stations ran dry in parts of the country.

Local councils were hit hard. Refuse collectors struck across large parts of the country. Rubbish accumulated in city centres and residential streets. In Leicester Square, piles of bin bags stacked several feet high became a national image of the winter. Newspapers ran photographs of foxes and rats tearing open refuse in the cold.

Gravediggers also walked out. In Liverpool, the dead were stored temporarily in a disused factory after burials stopped. In Tameside, a council used a furniture warehouse to hold coffins.

Hospitals remained open, but ancillary staff struck. Cleaners, porters, and ambulance drivers took action in some areas. Emergency cover existed, but it depended on ad hoc local arrangements rather than central coordination.

The government struggled to impose coherence. Agreements reached in one sector triggered immediate demands elsewhere. In January 1979, as disruption mounted, James Callaghan returned from an overseas summit to face headlines suggesting he was out of touch with events at home, a perception that proved hard to shake.

By early 1979, the public no longer trusted negotiated restraint to hold. The decade ended with exhaustion rather than resolution.

Conclusion

This account has focused on the economic mechanics of the 1970s: inflation, pay control, energy shortages, and external constraint. Even on those terms alone, the decade was defined by fragility, improvisation, and a shrinking margin for error.

It also leaves much untouched. Alongside the economic crisis ran a parallel political farce: unstable leadership, exhausted institutions, and a governing culture that often veered into the surreal. The later years of Harold Wilson’s career, including the influence of Marcia Williams, sit firmly in that category, as do the internal dramas of a state struggling to function under pressure.

But for those wondering why the UK would vote for a Conservative government bent on crushing the unions, the above should give a window into a world that seems almost entirely alien to anything we can imagine today.

For readers who want a fuller sense of the atmosphere and political detail around the 1970s, I’d strongly recommend the four-part series on 1974 by The Rest Is History, which captures the instability, personalities, and sheer oddness of the period far better than any single article could.

On 20 January 2025, the day Donald Trump was sworn in for his second term, he signed Executive Order 14158 establishing the Department of Government Efficiency, or DOGE.

The basic premise was hard to dispute: in fiscal year 2024 the US ran a $1.8 trillion deficit, and debt held by the public stood at $28.3 trillion. If you are already borrowing at that scale, “stop wasting taxpayers’ money” is hardly a fringe belief.

DOGE quickly paired that principle with examples of waste that sounded like parody, like a $33k USDA contract for “Guatemala youth conservation corps development consultants”, or a $144k USDA contract for a “Mexico sustainable landscapes consultant”.

It was not long before the problems started: despite the name, DOGE was not a cabinet department created by Congress, with a Senate-confirmed secretary, and budgetary authority that survives a news cycle. It was created by executive order, and could be reshaped or extinguished just as easily by another order.

Next was credibility. DOGE’s public face was Elon Musk. That made it louder, but also made it political, personal, and brittle. Even on its own terms, DOGE’s accounting quickly became disputed. DOGE claimed $55bn of savings early on, while a Reuters analysis of the partial data it had released suggested contract cuts were really worth c. $8.5bn.

The third problem was basic governing competence. Reporting at the time described key DOGE staff as extremely young and with little to no government experience. One widely-cited report identified six key individuals in influential roles aged 19 to 24.

Predictably, the project did not settle into durable machinery. Musk formally stepped down on 28 May 2025. By November 2025, Reuters reported that DOGE “doesn’t exist”, having been disbanded well before its scheduled end, with former staff moved into other roles and functions absorbed elsewhere.

So DOGE leaves a useful lesson: yes, government wastes money, but the real failure was design. A serious efficiency programme cannot be a celebrity-led, executive-order contraption that hunts for viral contracts and calls it a statecraft revolution. If you want long-term gains, you need something institutional, cabinet-level, and boring in the right way. Today’s essay is an attempt to steelman DOGE, to show what it could have been.

How deep the rot goes

A true attempt to roll back unnecessary parts of the state must not just look at unnecessary spending, but unnecessary laws, and the process that creates them.

To take an example: in the UK the national building regulations were introduced in 1965 and ran to eight pages. Today, the statutory regulations sit atop a mass of Approved Documents and guidance running to well over a thousand pages. Nothing dramatic happened, the scope simply expanded, year after year.

The YIMBY movement has spent years attacking the 1947 Town and Country Planning Act, but the more interesting point is how long it took for anyone to care. It took seventy years of accumulated damage before the system became politically loud enough to question. The obvious implication is that there are many other regimes on the books right now doing harm, but still stuck in the “too small / boring to notice” phase.

These examples point to two simple problems.

Firstly, all bureaucracies exist to self-perpetuate. No government department in history has ever said “we’ve solved the issue - you can fire us all now”.

Politicians make this worse by outsourcing democratically granted power to quangos (Quasi-Autonomous Non-Governmental Organisations). They do it because it is convenient. It pushes responsibility away from ministers and makes failure easier to blame on “independent experts”. Combine that instinct with self-perpetuation and you create unelected bodies whose remit steadily widens, while democratic control steadily thins.

Take Ofcom: it was established in 2003, merging five existing regulators, with a remit focused on broadcasting standards, telecommunications competition, spectrum allocation etc. Over time that remit widened. In 2020, Ofcom took on regulation of video-sharing platforms under new media rules, and after the Online Safety Act 2023 it was given broad powers over online content, including duties on platforms to assess and mitigate illegal content. What began as a technical regulator now shapes how platforms monitor and moderate everyone’s speech, effectively acting as a legal backstop for policing “harmful” and offensive expression.

The second principle is repeal failure. We repeal a small number of laws only when they cause such a stink that they reach the desk of a leader of one of the main parties. The structure of government means it is nobody’s job to prune outdated or overgrown law before it becomes a national blockage.

It’s simply not credible to believe that any government is taking the issue of overregulation, mission creep or conflicting laws seriously if not one person has it in their formal remit, and everyone hopes someone else will get to it.

The knock-on effect

Creeping laws and regulations do not just slow growth and empower NIMBYism. They also stop government being able to function. The system fills with counteracting duties, veto points and arms-length bodies until ministers discover that the levers of power do not connect cleanly to outcomes:

• In 2010, then conservative PM David Cameron warned that there is “no one lever” government could simply pull to make things happen.

• On 15 December 2025, at a Liaison Committee session, Starmer said: “every time I go to pull a lever there are a whole bunch of regulations, consultations, arm’s-length bodies that mean the action from pulling the lever to delivery is longer than I think it ought to be”.

This is the natural result of too many counteracting laws and stakeholders jamming up the system. Prime ministers hope one change can help. But no one is slimming down the full structure, so the number of parts between the prime minister’s hand and the end of the lever keeps growing.

One civil service insider told me that every decision made is taken under constant fear of being ruled illegal, for countermanding “the Human Rights Act 1998, the Equality Act 2010, the Climate Change Act 2008, the Environment Act 2021, the Habitats Regulations 1994, the Children Act 1989, the Care Act 2014, the Planning and Compulsory Purchase Act 2004, the Localism Act 2011, or the Public Services (Social Value) Act 2012”, to name but a few. No one in government can act without an army of lawyers signing off every step.

The forever expansion of the state bureaucracy shows up in tax too. In 2009 the UK tax code was reported at 11,520 pages. More recently, commentators commonly cite a figure north of 23,000 pages. By contrast, countries with simpler tax frameworks such as New Zealand (at roughly 3,000 pages) are often praised for codes that are substantially easier to work with, both in statutory text and overall regulatory burden.

Whatever the exact count, the direction is the point. Complexity is not a bug. It is what you get when the system is built to add, and not built to subtract.

Above, the actual full set of tax books we use in accounting for our mess of a system

Who has tried this?

Several governments have attempted, at least temporarily, to impose discipline on the growth of the state. The common feature of the more serious efforts is that they focused on systematic deletion, not marginal efficiency gains.

In New Zealand, between 1986 and 1996, the number of core public service departments fell sharply, large areas of commercial activity were removed from ministerial control, and departments were forced to publish clear statements of purpose tied to a narrow set of outputs. Public employment fell by roughly 20% over the period, and entire functions were either abolished or transferred out of the state. Crucially, this was accompanied by a legal and administrative shift that required departments to justify why an activity belonged in government at all, not just how efficiently it was delivered. When that discipline weakened in later decades, scope began to re-expand.

A more explicit deletion exercise occurred in Canada during the 1994–1997 Programme Review under Jean Chrétien. Every federal department was assessed against six tests, including whether its activities still served a legitimate public purpose and whether they needed to be delivered by the federal government at all. Federal programme spending fell by around 20% in real terms, dozens of programmes were eliminated outright, and several departments were merged or closed. The review worked precisely because it forced ministers to make binary choices. It failed in the long run because the mechanism itself was dissolved once the crisis passed.

A third, narrower example is Sweden, which since the mid-1990s has operated strict multi-year expenditure ceilings covering central government. These ceilings are binding, politically salient, and force explicit trade-offs between priorities. New commitments must displace old ones. This did not repeal large volumes of law, but it materially slowed the automatic expansion of state activity by making “add without delete” fiscally impossible. When political consensus around the ceilings weakened, their constraining effect weakened with it.

Where governments imposed formal deletion rules, scope shrank. Where restraint relied on political will alone, it decayed. None of these systems permanently solved state overreach. But all demonstrate the same point: overreach can be resisted, but only if subtraction is made a standing institutional function rather than a one-off reform.

That is the missing piece DOGE never addressed.

The Department for Ungovernment

If an actual, grown-up version of DOGE is to work, it has to be a permanent, serious department, not a taskforce or a review unit. The Secretary for Ungovernment would need a seat at Cabinet. Frankly, (sorry Mum and Dad!) it would be a more important role than the Secretary of State for Northern Ireland.

Every quango should be reviewed. Each one would be tested against its original statutory remit and shrunk back where it has expanded beyond it. Where democratically granted power has been pushed into unelected and unaccountable bodies, that power should be returned to Parliament. Shutting a quango down should also be an option.

Starting with building regulations, and then moving outward, every regulation should be scrutinised line by line and asked a basic question: is this something that genuinely requires the force of law, or is it simply something we like? Guidance that has quietly hardened into de facto law should either be formalised properly or stripped back.

Every law more than, say, twenty years old should be considered for repeal and replacement. The goal would be tighter scope, clearer intent, and plain English, rather than the kind of bizarre legalese that does not believe in punctuation.

A specialist tax team should be tasked with simplifying the tax code to bring it into line with comparable countries, rather than remaining ten times the length for no defensible reason.

The broad “everything” laws listed by my civil service insider should be scaled back so they cannot be used as all-purpose vetoes on decisions to which they are only tangentially related.

....I was very tempted just now to include that the Department for Ungovernment should get involved in procurement too, cutting back on poor and wasteful defence and NHS contracts.

But that would be me falling for my own trap: every good idea expanding its scope till its a mess. The Department for Ungovernment would fail the moment it tried to fix everything.

The above remit is enough.

It’s 1 May 2029. Against all odds, I’ve won the UK general election. An upstart YIMBY party surged once Prime Minister Angela Rayner finally pushed through proportional representation, and Farage’s 2028 tiramisu-based scandal did the rest.

I appoint you Chancellor.

Your first day is brutal. No briefings. No scene-setting. Just back-to-back meetings with ministers who all want more money and all insist they are non-negotiable.

First through the door is the Chief Secretary to the Treasury. We have to start with the boring stuff. Debt servicing. Why, they ask, is it suddenly such a large chunk of the budget? Germany pays far less. France too. You nod. There is nothing to cut here without consequences.

Next is the Health Secretary. The NHS budget must go up. Outcomes are irrelevant. Spending is the signal. Anything else would be apostasy against the national religion.

Then Education. A sharp intake of breath. Did you know, they say, that the UK spends a smaller share of government money on education than almost all its peers? Germany. France. Even the US. What kind of future-facing country does that make us look like? Do you hate children? No? Find more money then!

Finally, the Pensions Minister, just before lunch. This one should be easy. Years of the triple lock. Political deference. A grey vote that has been carefully fed. Except there is a problem. Somehow, despite everything, the UK still spends less of its public budget on pensions than Germany or France.

You break for lunch slightly dazed.

Debt cannot be cut. Health must rise. Education wants more. Pensions are already politically sacred and still somehow underweight. Defence is waiting in the wings. Everyone is right. The numbers do not add up.

You pull the numbers for the UK, the United States, Germany, France, China, and Singapore. What you discover is not that Britain is uniquely profligate or uniquely austere. It is that every country makes a small number of hard, structural choices early, and then spends decades pretending they are immutable facts of nature.

This is an attempt to map those choices.

The state’s income

The timing is awful. The AI bubble has popped. Credit is tight. Bond markets are jumpy.

We are going to have to run something close to a no-borrowing system. What we tax is what we can spend. If the money doesn’t come in, it doesn’t go out. Any deeper arguments about tax design, incidence, or optimal marginal rates will have to wait for another day. First, we need to know something simpler.

How much of the economy does the state already command?

That means looking at government revenue as a share of GDP – the size of the pie before we argue about how to slice it.

A few things jump out immediately.

Start with the familiar comparison. The UK sits between the US and continental Europe. Larger than the American state, smaller than Germany’s, well below France’s. The UK is not a low-tax country by global standards, but it is also not a maximal European welfare state. It occupies an uncomfortable middle.

America collects a noticeably smaller share of national income than Britain, and far less than France or Germany. That does not make it a small state in absolute terms, but it does shape every downstream choice. There is simply less public money to allocate.

Germany and France sit at the other end. France in particular stands out. Over half of national income flows through the state. Whatever the merits of that model, it gives French chancellors a degree of fiscal room that British ones simply do not have. More money in. More choices available.

Then there is China.

At first glance, the number is baffling. A state that still describes itself as communist collects a smaller share of GDP than the UK, Germany, or France… Are we the communists now?

Your research aide appears, slightly breathless, to explain.

Much of what looks like “state activity” in China happens outside the formal tax-and-spend system. Mandatory savings. State-owned enterprises retaining earnings. The Chinese state is powerful, but it is not fiscally large in the way European states are.

Finally, Singapore. The outlier’s outlier.

A government that takes barely a tenth of GDP looks almost fictional next to European norms, but it is a different architecture entirely. The state mandates saving instead of taxing. Pensions, healthcare, and housing are largely pre-funded by individuals through compulsory schemes. The government does less spending because it forces citizens to do more of it themselves.

Different countries. Different models. Very different amounts of money on the table.

And this is the key constraint your day-one meetings have been ignoring: before a chancellor can choose priorities, they are bound by scale.

Paying for the past

Before a chancellor can fund anything new, there is one bill that must always be paid first. Debt servicing. Interest on yesterday’s decisions. No discretion, no upside, and no way around it.

This is the first constraint on every other choice.

The UK and the US sit at the top of this chart for broadly similar reasons. In the UK, debt rose from roughly 40% of GDP before the 2008 financial crisis to around 70% by the mid-2010s, then jumped again during Covid. In the US, the trajectory was steeper still, with debt climbing from the mid-60s pre-crisis to over 100% of GDP after the pandemic. When interest rates were near zero, this mattered little. Once rates rose, the bill became unavoidable.

Continental Europe tells a different story. Germany entered the financial crisis with lower debt, ran primary surpluses for much of the 2010s, and saw its debt ratio fall before Covid hit. France borrowed more and reduced debt less aggressively, but still benefited from long periods of ultra-cheap financing. The result is that both countries service their debt at a far lower cost relative to the size of their state.

China’s debt expanded rapidly after 2008, but much of it was held domestically and rolled within the state sector, limiting exposure to market rates. Singapore followed a different path entirely, and sits around half a percent. It entered both crises with low net debt and a system designed to pre-fund future liabilities. As a result, rising interest rates barely register in the public finances.

The national religion

If debt servicing is unavoidable, healthcare is untouchable.

Health spending must rise. Outcomes are secondary. Any hint of restraint is treated as heresy. No chancellor survives long by arguing otherwise.

(frontline health spending in dark colours, social care pale on top)

The UK now devotes a larger share of public spending to healthcare and social care than any of the countries in this comparison. This reflects an ageing population, chronic workforce shortages, and a system that funds both frontline care and much of social care through general taxation. Your aide quietly directs you to a brilliant Substack piece showing that, despite this higher spending, almost everyone else still achieves better outcomes.

In the United States, a smaller share of government spending goes to healthcare because so much of the system sits outside the public budget. Total health spending is far higher, but it is fragmented across private insurers, employers, and households.

Germany and France sit in the middle. Both spend a substantial share of public money on healthcare, but their systems are structurally different. Social care plays a larger role, funding is more pre-planned, and pressure is spread across insurance contributions rather than concentrated on annual budgets. The result is high spending, but less political volatility.

China and Singapore again reflect different models entirely. China’s state spending on healthcare remains relatively contained, with large gaps particularly on social care filled by family provision and regional variation. Singapore’s public spending is modest by design, with mandatory savings and co-payments doing much of the work before the state steps in.

Healthcare is not optional. But how it is funded, where it sits in the budget, and how much flexibility it leaves elsewhere varies enormously. For the UK chancellor, it is both the largest live pressure and the hardest to reform.

The future, quietly underfunded

Education is the one line of the budget everyone claims to support and almost everyone quietly squeezes. Unlike healthcare, it does not generate daily headlines. Unlike pensions, it does not have a disciplined voting bloc. Its effects arrive slowly, often after the politicians who made the decisions have moved on.

Despite constant rhetoric about skills, productivity, and “investing in the future”, education in the UK takes a smaller share of public spending here than in every other country in this comparison. According to the Institute for Fiscal Studies, real-terms spending per pupil in England fell by around 9% between 2010 and 2020, the largest cut to school funding in a generation. Yet at the same time, it must be noted that the UK moved up the international PISA rankings, so perhaps a lesson should be taken here.

The United States sits at the opposite extreme. Education absorbs a remarkably large share of public spending, despite the country’s smaller overall state. This reflects how decentralised the system is. States and local governments carry much of the burden, and education is one of the few areas where American voters consistently tolerate higher local taxes. The result is a system that looks underfunded in outcomes, but very heavily prioritised in budgets.

Germany and France cluster together. Neither overspends, neither starves the system. Germany’s relatively modest share masks the importance of vocational training and apprenticeships that sit partly outside traditional education budgets. France’s spending is more centralised and academic, but similarly stable.

China and Singapore again reveal strategic intent. Both devote a large share of public spending to education, despite very different political systems. In China, this reflects a long-running push to expand tertiary education and technical capacity. In Singapore, it is explicit state strategy: human capital is the country’s only natural resource, and the budget treats it accordingly.

All about the bennies

Your aide now brings you to the largest single choice most governments make once debt and healthcare are accounted for: cash transfers. Not services, not investment, but money paid directly to households.

These transfers fall into two very different categories: payments to retirees; and support for people of working age.

They are split below - dark bottom columns for pensions, light top columns for non-pension benefits.

Pensions

Across advanced economies, pensions absorb an extraordinary share of public spending. In France and Germany, roughly one quarter of the entire state budget goes to old-age transfers. In China, it is closer to 30%. These figures are not the result of recent policy experiments. They reflect long-run demographic change interacting with systems designed decades earlier.

In 1970, France had roughly four people of working age for every person over 65. Today it is closer to two to one. Germany shows a similar pattern. When pension systems are funded on a pay-as-you-go basis, this shift translates directly into higher spending unless benefits fall or retirement ages rise. In most countries, neither has happened enough to offset the change.

The UK sits lower than its continental peers, but not because ageing has been avoided. The median age in the UK has risen from 34 in the early 1970s to over 40 today. The difference lies in structure. The UK state pension replaces a smaller share of earnings and relies more heavily on private provision through occupational schemes and auto-enrolment. That keeps public pension spending lower as a share of the budget, even as political pressure pushes benefit levels up at the margin.

The United States follows a similar pattern. Social Security is large in absolute terms, but it operates alongside widespread private retirement saving. Public pension spending therefore takes up a smaller share of the federal and state budget than in countries where the state plays the dominant role in retirement income.

China’s position is more compressed and more abrupt. Pension coverage expanded rapidly as urbanisation accelerated and family-based support weakened. At the same time, fertility collapsed. The result has been a sharp rise in pension spending within a relatively short period, without the gradual institutional adjustment seen in Europe.

Singapore stands apart. Public pension spending is minimal because retirement income is pre-funded through mandatory individual savings. The state enforces contributions but does not promise future pay-outs in the same way. The fiscal burden never appears because the liability never sits on the government’s balance sheet.

Pension spending is already the dominant answer many states give to the question of who public money is meant to support.

While the UK does appear to have this area under control in comparison, you still confirm to your aide that we’re ending the triple lock.

Everyone else

Non-pension benefits generally cover support for people of working age: unemployment, disability, housing, and family payments; but your aide reminds you that many of these benefits do loop back onto pensioners too.

Unlike pensions, this part of the budget is discretionary. It can be cut, and it often is.

The UK sits at the top of this comparison for generosity. Around 12% of public spending goes to non-pension benefits – more than Germany, France, the US, China, or Singapore. That reflects how much risk the British state absorbs during working life, particularly around housing costs and low pay.

Your welfare adviser flags the asymmetry. Between 2010 and 2019, working-age benefit spending fell sharply in real terms, while pension spending continued to rise. Even after that squeeze, this remains the largest non-pension benefits system in the group. Its size comes from breadth. It covers more failures elsewhere.

The US allocates barely 6% of spending here. Working-age risk is left largely to markets and employers. Germany and France sit below the UK, but with a different mix: more family support, less housing assistance, and unemployment insurance tied to previous earnings.

China and Singapore rely a little less on cash transfers. Families, employers, and mandatory savings carry much of the load.

The loudest budget line

Your defence adviser is already waiting.

The war in Ukraine is dragging on. Ammunition stockpiles are thin. Vladimir Putin looks more plastic and more deranged every time he appears on television. Allies are nervous. Defence feels urgent in a way few other budget lines ever do.

The adviser slides a chart across the table.

You’re confused “I thought everyone was pushing up to a 2% commitment.”

“No”, says the adviser. “We push for 2% of GDP, not 2% of government spending, so we look like we're asking for less than we are. We nicked that trick off the '0.7% of GDP for foreign aid' guys.”

Seen through that lens, the chart clicks into place.

The UK devotes around 5% of public spending to defence. Germany, France, and China cluster just below that level. Defence dominates headlines and strategy papers, yet it remains a relatively contained fiscal item. Even sizeable increases here move the budget far less than marginal changes in pensions or healthcare.

The United States sits elsewhere entirely. Defence absorbs over 10% of federal spending. That reflects global commitments, alliance management, and the cost of sustaining a military that underwrites international security far beyond its own borders.

Then there is Singapore.

At roughly 16%, defence looms over the budget. Singapore’s defence spending reflects vulnerability rather than ambition. A small, wealthy city-state with no strategic depth, a history of regional insecurity, and universal national service treats military readiness as an insurance premium.

Defence feels existential because it is tied to survival. In budgetary terms, however, it is rarely the main source of constraint.

Conclusion: what’s actually left to choose

You check the diary. After lunch come energy, transport, justice, policing, local government. All convinced they matter. All arguing over what remains.

But they are not where the budget is stuck.

This exercise has only covered the largest spending blocks, and that is the point. Most public money is already committed before ministers make their case.

Two items dominate the UK’s constraint:

• The UK spends a much larger share of its budget on interest than France, Germany, China or Singapore, and doesn’t have the ‘infinite debt’ abilities of the US hegemon.

• The NHS absorbs a larger share of public spending than peer systems while delivering weaker outcomes.

Together, aligning just these two areas with France and Germany would open up roughly 10% of the budget. That is the difference between constant trade-offs and real choice.

Everything else follows. Pensions are politically fixed. Education has been quietly squeezed. Defence is loud but contained. Non-pension benefits are a little higher than comparators, but not materially.

Budget debates fixate on the wrong fights. The state is constrained by a small number of legacy decisions, not by the number of departments in the room.

You shut the folder. At least now you know where the leverage is.

In September 1992, a twelve-year-old girl stood at a whiteboard in a New Jersey primary school, asked to spell the word potato. Behind her sat a small audience of teachers, aides, journalists, and television crews. She wrote it correctly: P-O-T-A-T-O. At that point, helpfully, the man beside her leaned in. He was the sitting US vice-president, Dan Quayle, on a campaign stop designed to showcase seriousness, competence, and concern for education as the Bush–Quayle ticket fought a tight re-election race against Bill Clinton. He suggested one small amendment; the girl dutifully added it. The board now read P-O-T-A-T-O-E. The cameras rolled. Political folklore was born.

News bulletins looped it, comedians pounced, and Quayle became a national punchline almost overnight. It is still remembered as a textbook example of a campaign gaffe. Yet at the time, almost nothing happened. Polling barely moved, and the election continued to turn on the recession, not a spelling mistake. The potatoe mattered enormously to the media. It mattered far less to voters.

Most famous campaign gaffes fall into the Quayle category: vivid, embarrassing, and essentially inert. What follows is a deliberately shorter list – eight campaign mistakes, drawn from elections across the developed world since the 1970s, where something real did change: polling moved, momentum collapsed, or a damaging narrative became impossible to escape.

At the end, I’ll return briefly to a few gaffes that didn’t matter at all, but are too fun not to mention.

1. Dukakis goes for a ride

It’s summer 1988, and Michael Dukakis is the Democratic nominee for president, running against George H. W. Bush. After the Democratic convention, Dukakis is riding high. In late July he holds a double-digit national polling lead. The problem is that voters are unconvinced he looks like a commander-in-chief.

To fix this, his campaign decides to manufacture an image. On a visit to a military base in September, aides arrange for Dukakis to ride in an M1 Abrams tank, helmet on, smiling from the turret. The intention is a serious leader; what the cameras capture instead is a small, cheerful man bobbing out of a tank like a cartoon character.

The image is replayed endlessly, then weaponised. The Bush campaign folds it into attack ads, using it as visual shorthand for weakness on defence. The press treats it as comedy, but voters treat it as confirmation. Whatever doubts existed about Dukakis’s toughness are no longer abstract - they now have a picture.

Within weeks, the race flips. By October, Bush has opened up a high-single-digit lead, roughly mirroring Dukakis’s earlier advantage. Dukakis never recovers. In November, Bush wins the popular vote by 7.8%. One photograph didn’t lose Dukakis the election on its own, but it locked in a fatal perception at exactly the moment voters were still updating.

2. The train to Frankfurt airport

It is summer 2002, and Edmund Stoiber is the conservative challenger to Chancellor Gerhard Schröder. The race is winnable. Germany’s economy is weak, the government is tired, and the CDU and CSU believe that seriousness and competence should be enough. Stoiber’s reputation fits that theory perfectly.

During a televised appearance meant to demonstrate economic credibility, Stoiber is asked about infrastructure and competitiveness. There’s no public transcript, but one attempted reconstruction is below:

“If you look at the distances, if you look at Frankfurt, you will find that ten minutes you need always easy in Frankfurt to find their gate. If you start from the main railway station, you take the Transrapid in ten minutes to the airport, then start practically at the main station in Munich. This means, of course, that the main station is basically closer to Bavaria, because at the main station many lines from Bavaria run together, and this means that the connection is fundamentally better.”

(clip here for German speakers!)

Within hours, clips circulate with captions asking what he was trying to say. German press coverage treats the moment as comic relief, but also as confirmation of a deeper problem. Stoiber is described as intelligent but unreadable, precise but incapable of speaking plainly to voters.

The campaign tightens, then slips. Schröder’s informal, conversational style begins to poll better against Stoiber’s technocratic delivery. From a pre-gaffe, 3 point lead, on election day, the SPD finishes ahead by just under 2 %, and Schröder remains chancellor. Stoiber loses an election his party had expected to win.

3. Maybot

It is spring 2017, and Theresa May has done something unprecedented in modern Westminster politics. Having taken over as leader of her party after the Brexit vote, she calls a snap general election only 2 years after the last election, on the assumption that Jeremy Corbyn is an easy and unpopular opponent, and that she can turn a small Conservative majority into a landslide. Polling in late April shows the Conservatives comfortably ahead by double digits.

As the campaign unfolds, May’s team lean heavily on tightly scripted slogans, especially “strong and stable leadership”. On 30 April, she sits down for a long television interview with Andrew Marr. After a series of carefully folded answers, Marr, visibly exasperated, tells her that “people can listen to that sort of thing and think it’s a bit robotic”, a criticism that will stick and turn into the mocking nickname Maybot. Her replies are dutiful but circular, repeating the same lines about leadership and growth without offering substantive engagement on policy questions.

Two weeks later, in late May, the campaign is forced into damage control. A controversial social care funding proposal in the Conservative manifesto, quickly branded the dementia tax by opponents, is quietly modified after backlash, with May tersely insisting that nothing has changed despite the shift in posture.

By election day on 8 June, the Conservatives have lost their majority. They go from 330 seats in 2015 to 317, forcing May into a confidence-and-supply deal with Democratic Unionist Party. What began as an attempt to cement dominance with a timid opponent became an unravelling narrative in which scripted solidity looked vapid and evasive to voters still updating their beliefs. May was just lucky that Corbyn was such a weak opponent that she remained PM.

4. Caught in the lie

It is autumn 2011, and Seán Gallagher is the front-runner in Ireland’s presidential election. An independent candidate with a business background, he has run a friendly, non-ideological campaign and benefited from voter exhaustion with party politics. In the final week, opinion polling puts Gallagher clearly ahead, with some surveys showing him around 40%, far in front of the field.

During the final televised debate, Gallagher is challenged over his past relationship with Fianna Fáil. Martin McGuinness alleges that Gallagher personally collected a cheque for the party from a wealthy donor shortly before a fundraising dinner. Gallagher is given the chance to respond, live, in front of the electorate. He denies it outright. He says it did not happen.

Within minutes, the claim begins to unravel: McGuinness says he personally confirmed the call. Journalists pursue it immediately. Gallagher is forced to clarify that he did in fact make the call, but insists it was misunderstood. By the following morning, the story has solidified. What matters is no longer the cheque or the party. It is that the front-runner was caught denying something that turned out to be true.

Gallagher could have used the debate to frame the episode himself. He could have acknowledged the call, minimised its importance, and reinforced his independence. Instead, by choosing denial, he surrendered control of the narrative entirely. The campaign shifts overnight from warmth and competence to credibility and trust.

Gallagher’s support falls by more than 10 % in a matter of days. On election day, he finishes second with 28.5 %, while Michael D. Higgins wins outright. A campaign that had been comfortably on track for victory is undone not by an old connection, but by a single decision to lie when the truth would likely have been survivable.

5. Little Marco glitches

It is February 2016, and Marco Rubio is, briefly, the Republican establishment’s great hope. After a strong third-place finish in Iowa, donors and party figures begin circling. In the days before the New Hampshire primary, polling places Rubio in the high teens and rising, with a plausible path to consolidating the non-Trump vote.

Early in the New Hampshire debate, Rubio delivers a well-rehearsed attack line on Barack Obama’s presidency:

“Let’s dispel once and for all with this fiction that Barack Obama doesn’t know what he’s doing. He knows exactly what he’s doing.”

A few minutes later, under questioning on a different topic, Rubio returns to the same sentence. Word for word. Same cadence. Same phrasing.

Then Chris Christie jumps on it. He explicitly accuses Rubio of relying on memorised soundbites and avoiding real answers. The camera cuts back to Rubio. Given a clean opportunity to respond unscripted, he does something extraordinary: he delivers the exact same line again. Verbatim. For a third time. Christie calls it out, the audience laughs, and the clip goes viral.

On primary day in New Hampshire, Rubio finishes fifth with 11 % of the vote, far below expectations. His momentum collapses, donor enthusiasm fades, and the consolidation that briefly seemed inevitable never happens. Rubio remains in the race for weeks, but the moment when he could plausibly have become the nominee is gone.

6. Australia’s Out of Hand

It is March 2004, and Mark Latham is leading the Australian Labor Party into a federal election against the long-serving prime minister John Howard. Howard has been in office for eight years. The Iraq war is unpopular, interest rates are rising, and the race looks competitive. Early campaign polling has Labor level with or slightly ahead of the Coalition.

At a campaign event in Tasmania, the two leaders meet briefly in front of waiting cameras. Howard extends his hand. Latham grasps his arm instead and pulls him sharply towards himself. Howard stumbles a half step. The interaction lasts only seconds, but it looks aggressive, awkward, and unnecessary. The footage is replayed repeatedly, slowed down, zoomed in, and looped.

Coverage fixes on the physicality rather than the intent. Evening bulletins lead with the clip. Howard’s campaign says nothing. Latham apologises later and describes it as a misunderstanding, but the explanation fails to land. What voters see is not confidence or toughness, but a flash of volatility.

From that point, Labor’s polling begins to soften, particularly among older voters and women. Howard’s argument that Latham is reckless and unsafe stops sounding theoretical. By election day in October, the Coalition wins comfortably. Labor finishes with 38 % of the two-party preferred vote, several points below its earlier position, and Howard increases his majority.

7. It’s my way or the highway

It is 2016, and Matteo Renzi is at the height of his power. Still in his early forties, he presents himself as the moderniser Italy has been waiting for. His Democratic Party has performed strongly in European elections, and through early 2016 polling consistently puts him well ahead of a fragmented opposition.

Renzi decides to push through a package of constitutional reforms aimed at reducing the power of the Senate and speeding up legislation. The substance is technical and poorly understood, but it is not obviously toxic. Instead of running the referendum as a narrow institutional change, Renzi personalises it completely. Speaking publicly, he says: “If I were to lose the referendum I would consider my career over.”

From that moment, the vote stops being about constitutional mechanics and becomes a judgment on Renzi himself. Parties that disagree on almost everything else find a shared incentive to oppose him. A No vote becomes a way to remove the prime minister rather than to reject a specific reform. Renzi attempts to soften the message later, but the framing has already stuck.

In the weeks before the vote, polling tightens sharply. On referendum day in December, the reforms are rejected by 59% to 41%. Renzi resigns the same night.

A proposal that might have passed quietly instead becomes a decisive defeat, not because the reforms were uniquely unpopular, but because Renzi chose to make himself the question. This is a textbook unforced strategic error in which a leader hands voters a lever and then acts surprised when they pull it.

8. Deny everything

It is late 2015, and Mariano Rajoy is heading into Spain’s most unstable election since the transition to democracy. The financial crisis has hollowed out trust in institutions, corruption scandals have damaged the main parties, and new movements are pulling voters away from the traditional system. Rajoy’s People’s Party is still polling first, but its majority is clearly at risk.

During the campaign, Rajoy is pressed publicly about corruption allegations linked to his party. Attempting to deny wrongdoing while avoiding legal exposure, he offers a sentence that immediately collapses under its own weight:

“Everything is false, except some things.”

Clips circulate within minutes. In an election already defined by mistrust, the moment hardens voter judgement. The quote becomes shorthand for evasiveness at precisely the point when voters are deciding whether the old parties still deserve the benefit of the doubt.

On election day, the People’s Party finishes first but loses its absolute majority, falling from 186 seats to 123. Spain enters a prolonged period of parliamentary deadlock and repeat elections. Rajoy survives in office, but the political system fractures.

The best of the rest…

The cases above are the relatively rare moments where something genuinely shifted, where polling moved, momentum broke, or a candidate crossed a line they could not uncross. What follows are my five favourite campaign moments that felt disastrous at the time, dominated coverage, and have survived in political memory ever since. They did not change outcomes, but they are too memorable and too entertaining to leave out.

#5. The EdStone

In April 2015, Ed Miliband launched what was meant to be a visual summary of Labour’s election pledges. The result was a literal stone slab, delivered by crane, engraved with six policy commitments and unveiled in a Kent quarry. The symbolism was unclear, the aesthetics were baffling, and the staging looked faintly apocalyptic. Media coverage was merciless. Commentators joked that Labour had produced its own gravestone. Despite the ridicule, polling barely moved. Labour was already behind, and remained behind.

#4. The Dean Scream

In January 2004, Howard Dean was fighting to regain momentum after a disappointing third-place finish in the Iowa caucus. Speaking to supporters that night, he launched into an upbeat rallying speech listing upcoming states. As the crowd noise dropped out of the broadcast mix, Dean’s excited closing yell sounded shrill and unhinged on television. The clip was replayed endlessly, usually without context. Commentators treated it as evidence of instability. In reality, Dean’s campaign was already sliding before Iowa, and polling showed little immediate movement after the scream itself.

#3. What is a lepo?

In September 2016, Gary Johnson was running as the Libertarian nominee and polling in the high single digits, unusually strong for a third-party candidate. The Syrian civil war was dominating international news, and the city of Aleppo had become a focal point of the conflict. During a television interview, Johnson was asked a straightforward question on Aleppo and replied, “What is a lepo?” Johnson later said he had misunderstood the question, but the moment stuck. However, the electoral effect was limited. His polling dipped slightly, then stabilised.

#2. My hovercraft is full of eels

In the February 1974 UK general election, Jeremy Thorpe embraced novelty as part of a high-visibility Liberal revival. To project modernity and momentum, he toured parts of the south coast aboard a hovercraft, arriving at campaign stops by sea to maximum press attention. Partway through the campaign, the hovercraft broke down and had to be abandoned, leaving Thorpe stranded and photographers delighted. The images quickly turned into the perfect metaphor: a futuristic vehicle, loud and eye-catching, ultimately unable to complete the journey. Despite the symbolism, the Liberals still won 14 seats, their best result for decades.

#1. …oops

In November 2011, Rick Perry was already struggling in the Republican primary after a strong early start. During a televised debate, he attempted to list three federal departments he would abolish as president. He managed two.

I highly recommend watching the video above, as words can’t do it justice.

But if you can’t watch, someone tries to help him out and suggests the EPA. He laughs and says yes, and the moderator asks if that’s right. Perry confirms that no, it’s not that one. The moderator asks if he can name the third department, and he tries again, fails, apologises, and ends with “….oops.”

On 12 December 2025, the United States announced a new strategic initiative called Pax Silica: a coalition of technologically advanced democracies aimed at securing supply chains for silicon, semiconductors, artificial intelligence infrastructure and the critical minerals that underpin all of these industries. The inaugural summit brought together representatives from Japan, South Korea, Singapore, the Netherlands, the United Kingdom, Israel, the United Arab Emirates and Australia.

When discussing with friends, one asked: could a bloc like this actually function as a closed system? Not merely prioritising internal trade, but operating with something close to genuine autarky among its members?

My immediate answer was no - there would be an issue with rare earth elements. But I could not say which materials mattered most, where they were found, how they were extracted, or whether rare earths were even the right place to focus.

So, once again I find myself writing largely to teach myself. It is an attempt to understand how mining actually works, how elemental scarcity differs from political scarcity, and how these constraints shape the real limits of bloc-based trade and strategic autonomy over the next 30 years.

Let’s start at the very start.

In the beginning…

Around 13.8 billion years ago, the universe began with the Big Bang. For the first few hundred million years, it consisted almost entirely of hydrogen and helium, with trace amounts of lithium. Nothing heavier existed in meaningful quantities.

Over the billions of years that followed, stars formed, burned, and died. In their cores, and in supernovae and neutron star collisions, heavier elements were forged. Every naturally occurring element heavier than helium was created in these processes. By the time the solar system formed, the periodic table already existed almost in full.

About 4.5 billion years ago, a rotating cloud of this material collapsed to form the Earth. The elemental mix present then is, for practical purposes, the mix we still have today. Only negligible quantities of new elements have been created since, through radioactive decay or rare natural nuclear reactions. Mining operates on a fixed stock of atoms that has existed since the planet first cooled.

That fixed inventory was not spread evenly. Geological processes concentrated some elements into a small number of locations and left the rest dispersed beyond practical use. Ancient continental cores, known as cratons, preserved certain mineral systems. Plate tectonics, volcanism, and hydrothermal activity concentrated others. These processes operated under specific conditions and over finite windows of geological time.

Take samarium. Like most rare earth elements, it exists in trace amounts across much of the Earth’s crust. Economically meaningful concentrations formed only where rare alkaline and carbonatite magmas appeared early in Earth’s history. As a result, large samarium-bearing deposits are found in only a handful of places, including Bayan Obo in Inner Mongolia, Mount Weld in Western Australia, and parts of southern Brazil. Regions without this history did not acquire usable concentrations later.

This constraint applies broadly. Almost every element exists almost everywhere at trace levels. Extraction becomes viable only when geology has already done enough of the work. Below certain concentrations, recovery is energetically, financially, or environmentally unviable. By the time politics enters the picture, the underlying map of feasible supply has long since been drawn.

Even where suitable deposits exist, extraction only proceeds if three conditions are met. The ore grade must be high enough to justify the energy required to move and process it. The surrounding infrastructure must make transport and refining feasible. And the legal and environmental costs must remain below the expected value of the output.

New mines take years to permit, finance, and build, even in favourable jurisdictions. Price spikes can accelerate marginal projects, but they do not create new geology. The result is a supply system that responds slowly, unevenly, and with long lags.

Common people

Before turning to rare earth elements, it is worth starting with three metals that are neither rare nor exotic, but which dominate modern industrial life. Copper, lithium, and aluminium account for a large share of global mining activity, whether measured by tonnage moved, capital invested, or downstream economic value. They are common, bulky, and essential. If supply fails here, nothing further up the value chain works.

Copper

Copper sits at the centre of the modern economy. It is essential for electrical wiring, motors, transformers, power grids, data centres, and almost every form of electrification. There is no scalable substitute with comparable conductivity, durability, and cost.

By value, copper typically represents around 15–20% of global non-fuel mining output, placing it among the most economically important mined materials. Demand growth is tightly linked to grid expansion, electric vehicles, renewable energy, and AI infrastructure. It is also, uniquely, one of the few strategic materials that occasionally appears in policy discussions and police blotters, having been rediscovered more than once in the form of missing cable along local railway lines.

Production is geographically concentrated. Chile and Peru together supply roughly 40% of global output, with additional production from the Democratic Republic of the Congo, China, and the United States.

Most copper comes from very large deposits formed around ancient volcanic systems. These deposits are not narrow veins running through rock, but vast bodies of mineralised stone where copper is spread thinly throughout. The Andes contain some of the world’s largest and best-preserved examples of this geology, which is why Chile and Peru dominate global production. They are mined using enormous open pits, with huge volumes of rock crushed and processed to extract relatively small amounts of metal. Modern copper mining works through scale rather than richness, which is why falling ore grades translate directly into higher costs and longer lead times.

Lithium

Lithium has become strategically important through its role in rechargeable batteries. Electric vehicles, grid storage, consumer electronics, and military systems all rely on lithium-ion chemistries in some form.

Although lithium only accounts for roughly 0.5% of global mining output by value, it sits at a critical bottleneck. The main constraint is timing: known lithium resources are plentiful, yet converting them into battery-grade material requires long permitting processes, water access, specialised processing plants, and chemical conversion steps that scale slowly. Demand accelerates faster than any of these stages can respond.

Production comes primarily from two very different sources. Brine extraction dominates in Chile, Argentina, and Bolivia, while hard-rock mining is led by Australia.

Brine operations pump mineral-rich groundwater into large evaporation ponds, concentrating lithium salts over months or years before further processing. Hard-rock mining involves conventional open-pit extraction followed by chemical conversion. In both cases, permitting, water use, and processing capacity limit how quickly supply can respond to rising demand.

Aluminium

Aluminium is often overlooked precisely because it is so common. It is used in construction, vehicles, packaging, aircraft, power transmission, and industrial machinery. By tonnage, aluminium production ranks among the largest material flows in the global economy, accounting for roughly 8–10% of all non-fuel minerals extracted worldwide, even though it represents a much smaller share of mining value.

The raw ore, bauxite, is widely distributed, with major deposits in Australia, Guinea, Brazil, and China.

Bauxite is first refined into alumina, then smelted into aluminium using electrolysis cells that consume enormous amounts of electricity. Competitive production requires decades of cheap, stable power, tying aluminium output to hydropower, coal, or long-term state-backed energy systems. Control over energy and processing capacity therefore matters more than control over ore.

Rare Earth Elements

Rare earth elements are a small group of chemically similar metals that sit together on the periodic table. They are not especially rare in the Earth’s crust, but they are difficult to separate, because they occur mixed together and behave almost identically during processing, often requiring dozens to hundreds of chemical extraction stages to isolate individual elements. Modern economies rely on them quietly and extensively, particularly for permanent magnets used in electric motors, wind turbines, precision guidance systems, sensors, and advanced electronics.

What makes rare earths distinctive is not their volume, but their leverage. By weight, they account for well under 0.1% of global mineral extraction, yet small amounts can determine whether high-performance systems work at all. Compared to the bulk metals that underpin industrial life, rare earths operate at the margin, but that margin increasingly defines what advanced economies can and cannot build.

Neodymium and Praseodymium

Neodymium is a core input for modern permanent magnets. It enables compact, high-power electric motors used in electric vehicles, wind turbines, industrial machinery, and precision robotics. Removing it forces motors to become larger and heavier, reducing efficiency and power density.

Praseodymium plays a closely related role, contributing to magnetic strength and thermal stability. Like neodymium, it lacks a viable substitute in high-performance applications.

Neodymium and praseodymium occur together in the same minerals, in roughly fixed ratios, and pass through the same processing chains. They are therefore traded as a single material, usually referred to as NdPr. NdPr magnets sit at the centre of modern electrification: around 85–90% of all high-performance permanent magnets rely on NdPr chemistry, with electric motors accounting for the majority of demand. Small changes in magnet quality translate directly into vehicle range, turbine output, and system efficiency.

Usable NdPr resources exist in several countries, but supply is tightly concentrated. China accounts for roughly 70% of global NdPr mining and around 85–90% of NdPr separation, refining, and magnet production, giving it effective control over the downstream supply chain. Australia supplies around 10–15% of global NdPr ore, largely through a single major producer, while the United States and parts of Africa contribute smaller volumes with limited processing capacity.

Extraction does not produce discrete neodymium or praseodymium streams. Ore contains many rare earths mixed together, requiring long chemical separation chains with repeated solvent extraction stages. These processes are capital-intensive, generate significant waste, and scale slowly.

Dysprosium

Dysprosium is used in very small quantities, but it solves a specific failure mode. When NdPr magnets heat up, their magnetic strength drops. Dysprosium is added to prevent that loss, allowing motors to operate at higher temperatures without degrading.

This matters most where space, weight, and cooling are constrained. Electric vehicle motors, military systems, aerospace components, and compact industrial drives rely on dysprosium-doped magnets to maintain performance under load. Without it, systems still function, but they become bulkier, heavier, or less reliable.

Dysprosium is far rarer in usable concentrations than neodymium or praseodymium. It represents less than 1% of total rare earth output by volume, yet its absence can determine whether a motor design is viable. Supply is highly constrained, as China produces around 80% of global dysprosium, with much of the remainder coming from Myanmar, often through informal or environmentally damaging mining operations. Other countries possess trace resources, but few have economically viable deposits or the processing capacity required to separate dysprosium at scale.

As with NdPr, extraction does not yield dysprosium directly. It is recovered as part of mixed rare earth streams and isolated late in the separation process, after many chemical stages. This makes dysprosium supply especially sensitive to disruptions in processing capacity rather than mining alone.

Cerium and Lanthanum

Cerium is the most abundant rare earth element in commercial use. It is used primarily in catalytic converters, oil refining catalysts, glass polishing powders, and emissions control systems.

Lanthanum plays a closely related role. It is used in refining catalysts, hydrogen storage alloys, optical glass, and nickel metal hydride batteries. While it attracts far less attention than magnet rare earths, many industrial processes depend on it quietly.

Cerium and lanthanum occur together in the same minerals and are usually produced as a pair. Together they account for roughly 60–70% of total rare earth output by weight, making them the dominant rare earths by bulk rather than by value. Their challenge is not finding supply, but making it economic.

China dominates global cerium and lanthanum production, accounting for around 60% of worldwide output. Outside China, the largest contributions come from Africa. South Africa and Madagascar together account for roughly 10–15% of global production, largely from monazite sands and carbonatite-associated deposits. Additional resources exist in Namibia and Tanzania, often produced as by-products of mineral sands or phosphate mining.

Unlike cobalt, African cerium and lanthanum production is largely industrial rather than artisanal, with fewer labour-abuse concerns but persistent challenges around regulation, processing capacity, and radioactive waste management.

Extraction is mechanically straightforward but commercially awkward. Cerium and lanthanum are easier to separate than heavier rare earths, but prices are low and volatile. Processing costs, waste handling, and radioactive by-products often dominate project economics. Many deposits that are geologically viable remain undeveloped because margins are thin.

Strategic implications

The material constraints described above define the space in which strategy operates. Access to geology, processing capacity, energy, and time determines what outcomes are feasible. This makes full autarky an unrealistic objective, even for friendly blocs. No plausible coalition controls all the inputs required to sustain a modern industrial economy from extraction through to finished systems.

Political attention tends to focus on mining. Mines are visible, local, and easy to frame as strategic wins. Outcomes are shaped further downstream. Refining, separation, and stable energy supply determine whether raw materials can be turned into usable inputs. Aluminium output tracks electricity availability as closely as bauxite; rare earth supply depends on chemical separation long after ore is extracted.

The materials discussed so far describe only a fragment of the broader picture. Modern supply chains depend on hundreds of inputs, each with its own failure modes. One important omission is cobalt. It is used in lithium-ion batteries to improve energy density, thermal stability, and lifespan, particularly in electric vehicles and grid storage. Around 70% of global cobalt mining takes place in the Democratic Republic of the Congo, and estimates suggest 10–20% of Congolese cobalt output comes from artisanal mining, with tens of thousands of children involved in related activities. This problem is specific to certain materials and mining methods, but it illustrates how ethical, political, and supply risks can concentrate in ways that have little to do with chemistry.

Over the next thirty years, strategies that work will share a small number of characteristics. They will prioritise processing and energy alongside extraction. They will accept some dependence and manage it deliberately rather than pretending it can be eliminated. They will build redundancy where chemistry allows it and stockpile selectively where it does not.

On 2 September 2025, the US military carried out airstrikes on small vessels in the Caribbean it said were linked to Venezuelan narcotics trafficking, marking a shift from sanctions enforcement to the use of lethal force. That escalation deepened on 10 December, when US authorities seized a sanctioned oil tanker operating out of Venezuela, followed by additional US naval deployments to the region to enforce oil export restrictions. Venezuelan authorities responded by escorting tankers with naval vessels, while Washington has insisted these actions do not amount to a formal blockade.

While this confrontation could easily fizzle out, it is also possible that Venezuela becomes one of the defining geopolitical flashpoints of 2026. Either way, the country is likely to feature heavily in headlines in the months ahead. This piece is therefore a background guide to how Venezuela reached its current position, and how its economy and society function day to day.

The national origin

Venezuela is a country roughly 4x the size of the United Kingdom, located on the northern edge of South America, with a long Caribbean coastline. Its geography is varied: dense jungle in the south, wide plains known as the llanos across the centre, mountains in the west, and a relatively narrow but strategically important coastal strip. Human habitation dates back at least 8,000 years, with indigenous groups living by hunting, fishing, and small-scale agriculture long before European contact.

Europeans first arrived in 1498, when Christopher Columbus sighted the Venezuelan coast on his third voyage. Early Spanish settlement focused less on land conquest than extraction, particularly pearl fishing along the islands and shallow coastal waters, which relied heavily on forced indigenous labour. The territory remained a marginal part of the Spanish Empire, lacking the large mineral deposits that drew sustained imperial attention elsewhere. Venezuela declared independence from Spain in 1811, followed by a long and destructive war that ended in 1821, driven by local elites and military leaders seeking autonomy from a distant colonial state rather than mass popular mobilisation.

At the start of the 1900s, Venezuela was relatively poor by Latin American standards. Its economy was based on agriculture and cattle ranching, infrastructure was limited, and the state was weak and fragmented after decades of civil conflict. Unlike Argentina or Chile, it had not yet found a clear export engine.

Black Gold

Oil was discovered in commercial quantities in the early 1900s, with the first major strike at Zumaque I in 1914. By the 1920s, foreign firms, mainly British and American, were extracting crude at scale, and by the 1930s Venezuela had become one of the world’s largest oil exporters. Oil rapidly displaced cattle, coffee, and cocoa as the dominant source of income, exports, and state revenue.

This transformation created a classic rentier state. The Venezuelan government did not need to tax its population heavily to function. Instead, it distributed oil rents through public employment, subsidies, and infrastructure spending. That arrangement brought real gains. Living standards rose, cities expanded, and by the mid-20th century Venezuela was one of the richest countries in Latin America on a per capita basis, attracting migrants from across the region and southern Europe. Caracas in particular developed the institutions and lifestyle of a middle-income country rather than a post-colonial backwater.

As the state was funded by oil rather than taxation, accountability mechanisms remained weak. Political competition focused on controlling the distribution of oil revenue, not on improving productivity or building a diversified economy. Manufacturing remained limited, agriculture stagnated, and imports filled the gap. When oil prices were high, this fragility was invisible. When prices fell, it became immediately painful.

Oil nationalisation in 1976 formalised this model rather than changing it. The state-owned oil company, PDVSA, was initially well run and technically competent, staffed by trained engineers and managers who operated with a degree of autonomy from day-to-day politics. For several decades, this worked tolerably well. Venezuela enjoyed long periods of macroeconomic stability and rising consumption, even as the underlying economy became more dependent on a single commodity.

By the late 1990s, however, the limits of the system were clear. Oil revenue was volatile, inequality remained high despite decades of spending, and the non-oil economy was weak. The state was large but brittle, rich on paper and poor in resilience. When a political movement emerged promising to reclaim oil wealth for the people, it was building on a structure that already existed rather than inventing something new.

Chávez

Hugo Chávez was a rather unconventional political outsider: in 1992, as a mid-ranking army officer, he led a failed military coup against the elected government. He was imprisoned, then pardoned two years later, and entered politics with his coup attempt having elevated him to national prominence. This trajectory mattered: from the outset, Chávez viewed electoral politics as one route to power rather than its sole source, and his later willingness to weaken institutions followed a pattern established before he ever won office.

Chávez took office in 1999 at a time of widespread frustration with Venezuela’s political class. For many ordinary Venezuelans, especially in poorer urban areas, the early years of his presidency brought visible changes. Public spending increased, new social programmes were rolled out in health, education, and food distribution, and access to basic services improved in neighbourhoods that had long been neglected. For a large part of the population, daily life became more secure through the 2000s, particularly while oil prices were high.

These gains should not be read as the result of European-style social democratic policy. Many programmes were poorly administered, highly politicised, and vulnerable to corruption. Distribution was often discretionary rather than universal, with benefits expanding sharply ahead of elections and contracting afterward. Competence varied widely, waste was common, and loyalty increasingly mattered more than performance. For recipients, this still represented real improvement over what came before, but it was not a stable or rules-based welfare system, and was a rot that would later cause a sharp collapse.

The opposition during this period was real but uneven. It was strongest among the middle and upper classes, business owners, parts of the professional class, and residents of wealthier urban areas, especially in eastern Caracas. Many in these groups experienced Chávez’s rule as disruptive and threatening, particularly as property rights weakened and state control expanded. The opposition also included parts of organised labour and civil society, but it struggled to unite around a coherent alternative or leadership.

Political conflict escalated in the early 2000s, most sharply during the failed coup attempt in 2002 and the subsequent oil strike in 2002–03. After this point, the balance of power shifted decisively. The government moved to marginalise opponents institutionally: opposition media faced increasing pressure, senior civil servants and managers were dismissed, and loyalty became a prerequisite for advancement in the public sector. Elections continued, but the playing field became steadily less even.

For most ordinary people, however, this period was not experienced as a collapse. Food was available, wages were paid, and consumption rose. Subsidised imports kept prices down, and the state absorbed economic shocks through spending. Emigration existed but was limited, largely involving professionals, business owners, and politically active opponents. The idea of leaving the country was present, but not yet dominant.

By the time of Chávez’s death in 2013, Venezuela was politically polarised but not yet socially disastrous. Daily life functioned, institutions still operated, and the economy had not yet contracted. The mass exodus that would later define the country’s crisis had not begun in earnest.

Maduro

Nicolás Maduro assumed the presidency in 2013 following Hugo Chávez’s death, after narrowly winning a special election. A former bus driver and union organiser, Maduro lacked Chávez’s personal authority and public connection. His claim to power rested primarily on continuity and loyalty to Chávez’s legacy rather than on an independent political base.

The early years of Maduro’s presidency were not marked by immediate collapse. For a short period, the existing system continued to function. Oil revenues were still sufficient to fund subsidies and imports, and the government attempted to maintain price controls and currency pegs inherited from the Chávez era. The decisive break came after 2014, when global oil prices fell sharply. With little fiscal buffer and a highly distorted economy, the state struggled to adapt. Shortages emerged, inflation accelerated, and basic administrative capacity began to erode.

The economic contraction that followed was extraordinary in scale. Between 2013 and 2021, Venezuela’s real GDP fell by an estimated 70–75%, one of the largest peacetime economic collapses ever recorded. Output per person fell back to levels last seen in the mid-20th century. This was a systemic collapse, in which production, imports, public services, and state capacity all contracted simultaneously.

As conditions worsened, Maduro’s approach shifted. Economic mismanagement was compounded by increasing political repression. Opposition victories, most notably the parliamentary elections in 2015, were neutralised through court rulings and executive action. Independent institutions were hollowed out further, and security forces played a growing role in maintaining control. Unlike under Chávez, legitimacy increasingly depended on coercion rather than consent.

For the opposition, space narrowed rapidly. Political leaders were barred from office, imprisoned, or pushed into exile. Protests were met with force, and legal routes to power were systematically blocked. Media outlets faced closures or were brought under indirect state control. While opposition sentiment remained widespread, particularly in urban areas, it was fragmented and constrained by fear, fatigue, and limited organisational capacity.

It was during this period that the mass exodus began. From around 2015 onward, emigration accelerated, first among professionals and skilled workers, then across all income groups. By 2023, an estimated 7 to 8m people, roughly a quarter of the population, had left Venezuela, making it one of the largest peacetime displacement crises in modern history.

The composition and routes of this migration shifted over time. In the early years, those leaving were disproportionately middle-class and affluent, travelling by air to the United States, Spain, or other parts of Latin America. As conditions deteriorated, migration became poorer, slower, and more regional. Large numbers crossed on foot into Colombia and Brazil, while others travelled onward to Peru, Ecuador, and Chile. In the late 2010s and early 2020s, increasing numbers began moving north through Central America, with Venezuelans becoming one of the fastest-growing groups encountered at the US southern border.

This outflow acted as both a safety valve and a structural loss. Emigration reduced immediate political pressure on the government while stripping the country of skilled labour, working-age adults, and informal support networks. Families fragmented, remittances became a central income source, and departure became a normalised life strategy rather than an exceptional one.

Daily Life

For most people in Venezuela, daily life now operates on a parallel system. The formal state still exists, but it is not the primary organiser of ordinary activity. Survival depends on access to dollars, informal networks, and constant adjustment.

Salaries are often still paid in bolívars, but prices are quoted in US dollars. Formal wages are extremely low in dollar terms, commonly equivalent to $20–$50 per month. Most households rely on a mix of informal work, remittances from relatives abroad, or dollar-denominated side income. Cash dollars circulate widely and are more trusted than the banking system.

Food is generally available, but affordability and quality vary sharply. Staples such as rice, pasta, corn flour for arepas, eggs, and chicken can usually be found. Beef, dairy, cooking oil, and imported packaged foods are expensive or intermittently scarce. Diets are carb-heavy and protein-light. People substitute constantly, changing meals based on what is affordable that week rather than preference.

Infrastructure is unreliable. Electricity cuts are common outside wealthier districts, and water supply is intermittent in many areas, sometimes arriving only a few days per week. Storing water in tanks and containers is routine. These shortages shape daily schedules and household routines rather than appearing as emergencies.

Public hospitals lack staff, equipment, and basic supplies. Doctors are poorly paid and many have emigrated. Private clinics and pharmacies exist, but treatment depends on the ability to pay in dollars. Medicines are often available, but at prices that place them out of reach for many.

Violent crime is lower than at its peak, but trust remains low. People limit travel by time and neighbourhood, avoid unnecessary movement after dark, and rely on informal transport. Outside major cities, fuel shortages and transport breakdowns constrain mobility further.

Venezuela 2025

By 2025, Venezuela is no longer in freefall. The economy has stabilised at a low level after the quiet abandonment of price controls and the spread of de facto dollarisation. Imports have partially returned, inequality has widened, and urban markets function unevenly. Politically, Nicolás Maduro remains in power through institutional control, military loyalty, and external support. The regime is weaker than a decade ago, but also more durable.

What has changed is how Venezuela intersects with the outside world. Sanctions have pushed oil exports into opaque channels involving intermediaries, ship-to-ship transfers, and buyers willing to operate at the margins. At the same time, Venezuela has become a key transit state for cocaine moving out of Colombia, exploiting weak border control, poorly monitored ports, and maritime routes that overlap with sanctions evasion. These activities are not the state’s formal policy, but parts of the security apparatus have been implicated as sources of revenue in a sanctioned economy.

By the mid-2020s, oil smuggling, drug trafficking, and sanctions enforcement were operating in the same physical space. For the US, this collapsed what had previously been treated as separate problems into a single operational theatre in the Caribbean. The first use of force in 2025 came through strikes on vessels described as drug-linked, which provided a legal and political basis for expanded naval presence. Once in place, enforcement against oil shipments followed.

This helps explain Trump’s renewed focus. During his first term, Venezuela’s collapse was largely inward-facing. Migration to the US was lower, oil still flowed through conventional markets, and drug trafficking was handled primarily as a law-enforcement issue. In 2025, the spillovers are harder to ignore. Migration has domestic salience, sanctions evasion has become more physical, and maritime routes link drugs and oil into a single enforcement problem.

The result is a shift from rhetorical pressure to direct control of movement. This is less about sudden interest in Venezuela and more about a situation that has become operationally entangled with US security priorities in a way it was not before.

This substack, like many in the YIMBY world, has spent plenty of time describing the negative economic consequences of NIMBYism in the UK and elsewhere, and not just to the housing market. You’ve probably heard before our many woes, like the fact that the UK builds homes that are 25–30% smaller than new French ones, takes five to seven years to approve a major housing scheme that Tokyo clears in roughly twelve months, and pays construction costs 30–60% higher than Germany.

What we talk about far less is what happens when a country actually delivers the kind of housing boom we keep arguing for. Who has done it well, and what changed beyond rents falling? What happens to mortgage holders who fall into negative equity? And how do we distinguish a healthy supply-led boom from the credit-fuelled disasters that produced Spain’s ghost stock or China’s empty districts?

This piece walks through these questions and sets out why a serious YIMBY programme avoids the risks behind the world’s worst housing bubbles.

Short-term effects (years 0–5)

The short-term effects of a genuine housing boom show quickly: rents cool, price growth slows, vacancies stay low, and households absorb new stock almost immediately. These effects appear within two to five years in every city that has ever built at scale in response to real demand.

Tokyo added roughly 110,000–140,000 homes a year through the 2010s, and central-ward rents fell by about 5–10% between 2012 and 2017. Houston’s mid-2000s boom coincided with a 20% population increase, yet median rents stayed flat. Montreal built roughly 40,000 units a year after 2016 and recorded the slowest rent growth of any major Canadian city. New construction reducing nearby rents is one of the most robust findings in urban economics. Most recently, Austin built tens of thousands of new apartments between 2021 and 2023 and is now seeing some of the sharpest rent falls in the United States. Median asking rents dropped by about 9% year-on-year in 2024, with one-bedroom rents down roughly 17% from their 2021 peak.

Tokyo’s construction surge kept price inflation in central wards at roughly half the rate seen in other global capitals. Montreal’s price growth diverged sharply from Toronto and Vancouver once output rose, with annual increases several percentage points lower. Seoul’s mid-2000s New Town projects released more than 100,000 units, pulling down deposits and sale prices in affected districts relative to their neighbours. Decades of evidence show that cities with high supply elasticity experience far slower house-price inflation during demand shocks.

Tokyo’s new housing wasn’t empty for long: vacancy in the city fell from around 9% to about 6% through the 2010s while completions stayed high, which means people moved in as soon as homes were built. Houston shows the same pattern: neighbourhoods that added tens of thousands of units absorbed them immediately because population growth and construction moved together. Seoul’s New Town districts in the mid-2000s released more than 100,000 units, and occupancy rose straight away.

Construction activity also delivers short-term economic gains without destabilising prices. Tokyo’s housing cycle lifted construction employment by 5–10% in the early 2010s while rent pressure fell. Seoul’s mid-2000s building programmes produced similar jumps in local construction jobs but kept deposit levels down in the same period. Montreal’s post-2016 boom pushed construction investment up by roughly 20% over five years while recording the slowest rent growth of any major Canadian city. Building more raised economic activity and eased housing stress at the same time.

Texas shows the same effects at state scale. Its population rose by ~40% since 2000, and the housing stock expanded by almost exactly the same percentage. In each early boom cycle, rent and price growth stayed mild because new homes were absorbed as fast as they appeared. A housing boom does not need to be confined to one metro. A large region can build quickly, fill quickly, and avoid the short-term price spikes seen in constrained markets.

Austin, Texas

Medium-term effects (years 5 to 15)

When cities build at scale for a decade, the main effect is higher long-run economic growth. Housing stops acting as a brake on the labour market. People move more. Firms hire more easily. Commuting costs fall. Investment becomes smoother.

From 2005 to 2020, Tokyo added more than 1m homes, and average rents in central wards stayed flat or edged down while the population passed 14m. Vienna built roughly 5,000 to 7,000 social homes a year through the 2000s and 2010s, and rent levels remained among the most stable in Europe.

Elastic supply also produces milder downturns, which supports growth by reducing risk. Tokyo’s price index fell by roughly 10% after 2008 and then settled, compared with drops of 20 to 30% in the UK and more than 40% in parts of the US. Smaller cycles mean lower financial stress and more consistent investment.

Mobility is another key long-run driver. Tokyo pairs high vacancy with high completions, which produces short search times and high movement: more than 10% of households move each year. Seoul’s expansions opened entire new districts for young families and cut overcrowding from 12% to around 8% between the mid-2000s and mid-2010s. Vienna’s continuous building keeps waiting times short: roughly 40% of social-housing applicants receive a flat within a year, and turnover stays high by European standards. A thicker housing market raises productivity by shortening job searches and reducing mismatch.

Continuous building also renews the capital stock. Tokyo’s mid-2010s construction wave lifted construction employment by roughly 5 to 10% while replacing older, inefficient buildings. Seoul linked its housing expansions to major transport upgrades that improved commuting for several million people.

Finally, and most importantly, lower housing costs raise real incomes and support stronger GDP growth. Tokyo kept rent-to-income ratios near 20 to 25% through the 2000s and 2010s while real wages grew by roughly 10% and Tokyo’s prefectural GDP rose by 20 to 25% over the same period. Vienna held rent shares close to 20% for more than a decade, and Austria’s real GDP per capita grew by roughly 15%, helped by stable housing costs freeing up disposable income.

Negative equity and how to handle it

For young people wanting on the housing ladder, falling house prices are great, but not so great if you get a mortgage at the peak.

Negative equity happens when a home falls in value below the mortgage still owed. You buy for £300k with a £300k mortgage. Prices fall, the home is now worth £250k, yet the contract stays fixed at £300k. If payments slip, the bank can repossess, sell at £250k, and still pursue you for the difference.

A key harm comes from immobility. Selling requires cash you do not have. Moving to another city is blocked. Downsizing is blocked. Refinancing is blocked. Overpaying for a house is a financial mistake. Negative equity removes the ability to move at all.

Roughly half of Irish mortgaged households fell into negative equity after 2008. Workers who lost jobs in construction or finance could not move to Dublin or abroad because selling required cash they could not raise. Mobility collapsed, arrears climbed, and unemployment stayed elevated for years.

In Nevada, more than 60% of borrowers were underwater in 2010. Many who wanted to relocate to Phoenix, Denver, or Texas were unable to sell without absorbing a large loss. Job matching stalled and unemployment stayed well above the national average for a long time. Housing mispricing turned a regional crash into a labour-market problem.

So how have different countries dealt with negative equity traps before? To take our example above, household buys a home for £300k with a £300k mortgage. Prices fall and the home is now worth £250k. Selling clears £250k but leaves £50k still owed. The question for policy is what happens to that £50k.

In non-recourse US states, the £50k ends at the point of repossession. The lender sells the home for £250k and writes off the remaining £50k. The borrower walks away with no long-term liability. The advantage is speed: households regain mobility immediately, and labour markets reallocate people quickly. The cost is pushed onto lenders, and mortgage credit tends to be priced accordingly.

Ireland warehoused the loss instead. A £250k active loan continues as normal, and the remaining £50k is shifted into a separate low-interest or zero-interest loan. Monthly payments fall, arrears stabilise, and the household remains in place. The long-term issue stays unresolved. Selling still requires finding that £50k, so labour mobility remains limited even though day-to-day strain is lower. It contains the crisis but does not clear it.

Iceland eliminated the shortfall outright. The lender cuts the loan balance to the new value of the home. The £50k is removed from the borrower entirely and recognised immediately as a loss by the bank. This was a political choice: Iceland prioritised resetting household balance sheets so the wider economy could recover: households could move, refinance, or downsize without carrying historic debt. Banks took the hit upfront, and the state ultimately supported the system to prevent collapse. The result was faster labour-market adjustment and a quicker recovery.

The UK stretched the problem out. The £300k balance stayed £300k, but lenders allowed interest-only periods or much longer mortgage terms. Monthly costs fell, repossessions slowed, and the acute phase eased. The structural issue did not change. A sale at £250k still leaves £50k to be paid. Many households stayed put for years because the underlying numbers gave them no other option. Mobility remained weak, and regional job adjustment stayed slow.

These models show that states have multiples ways to prevent negative equity from turning into an economic disaster. All of them were deployed during a sudden, violent price collapse in 2008, when shortfalls were large and arrived almost overnight. A housing boom, by contrast, produces smaller pockets of negative equity spread over longer periods, and usually alongside rising wages. The numbers are less extreme, the adjustment slower, and the policy tools easier to deploy.

Good booms, bad booms

All the cases in the first two sections were positive, supply-driven booms. Tokyo, Vienna, Seoul, Houston, and Austin all built enough homes to keep rents stable or falling while their economies grew. That is not the only kind of boom that exists.

A good boom is one where supply expands fast enough to hold rents steady or push them down while the economy grows.

Bad booms look different. Supply lags badly. Prices rise because credit expands, not because new homes exist, and so rents rise even during the boom. China’s 2015–2023 cycle produced more than 50m empty or semi-empty units because investors bought off-plan but did not occupy. Spain’s 2003–2008 cycle saw prices rise by 150% while completions fell by 40% from their peak. Dubai’s mid-2000s boom had price growth above 20% a year despite vacancy levels that made little sense for real residents. GDP often looks strong in the short run because construction investment surges, then collapses when credit tightens.

Bad booms usually emerge from the same causes. Credit expands far faster than real supply. China grew mortgage credit by more than 15% a year through the late 2010s while household formation slowed. Spain offered tax relief on mortgage interest through the 2000s, pushing thousands of buyers into second homes. Ireland’s banks in 2005–2007 issued investor mortgages at loan-to-value ratios above 90%, even while completions slowed. Planning systems often block real building but allow speculative pre-sales of units not intended for occupation.

The warning signs show up early: new homes relative to new households drop well below healthy levels: Spain fell to 0.6; China’s big cities to 0.5–0.7; Ireland to 0.7 before the crash. Rents rise even as prices surge, which is the clearest scarcity signal. Vacancy collapses in real neighbourhoods even as empty investor stock builds elsewhere. Investor mortgages take a large share of lending: more than 40% in Ireland by 2006, more than 30% in Spain by 2007, above 35% in China’s Tier-1 cities before the slowdown. Price-to-income ratios rise every year while completions barely move. Approval times lengthen and construction bottlenecks persist.

There’s plenty of examples of good governments who paid attention to the above signs, and intervened early enough to avoid a full crash: Singapore tightened loan-to-value ratios in 2010–2013, raised stamp duties on second homes to 7–10%, and imposed strict debt-to-income ceilings. Investor demand fell, yet housing output stayed high at 20–25k units a year. Prices cooled rather than collapsed. and rents flattened.

New Zealand introduced tougher investor lending rules in 2021–2022 and removed interest deductibility for investment properties. Price inflation stopped almost immediately even though completions were still climbing above 45k a year. Rents stayed roughly flat, which showed the underlying scarcity was being addressed.

Canada’s BC and Ontario imposed foreign-buyer taxes of 15–20% and added vacancy taxes. Investor activity fell sharply, but construction continued. Vancouver’s price growth slowed from 15% a year to near zero by 2018, while rental vacancy rose from 0.6% to 1.3%, small in absolute terms but a meaningful shift in a tight market.

Conclusion

Bad booms come from fake demand. Credit expands, investors buy multiple units, planning blocks real building, and prices rise faster than incomes. Rents climb because scarcity never changes. When credit tightens, the whole cycle reverses and the labour market takes the hit.

Good booms come from fixing supply constraints. Planning gets easier, construction rises, vacancy increases slightly, and rents fall. Prices stabilise instead of accelerating. The economy absorbs workers more easily, and downturns are smaller because households are not leveraged against rising land values.

Expanding supply creates a boom centred on real households and real output. Expanding credit creates a boom centred on balance sheets and speculation. Only one of these produces lower rents, higher growth, and a labour market that can move.

As we have been treated to Rachel Reeves’ political masterclass over the past week, for some reason, the topic of terrible policies has been on my mind.

Britain hardly has a monopoly on bad ideas: once you start looking, you realise the developed world is littered with policies that raise prices, block growth, and entrench cartels with impressive confidence. Today is a quick run through 10 of the worst of them.

1. Canada’s Internal Trade Barriers

During the Canadian PM debates last year, plenty of viewers blinked when Pierre Poilievre promised “to end the internal tariffs in Canada”. Internal tariffs? The what??? Unsurprisingly, google searches for it soon spiked.

Say you own a craft-beer shop in Ottawa and want to stock a popular Montreal brewery. You can’t simply order a few cases. Alcohol is controlled by provincial monopolies, so the shipment has to move through the Ontario system, be priced under Ontario rules, and comply with Ontario labelling requirements. If you drive it back yourself without permission, you’re technically committing an offence.

Swap the product and the story repeats. A small construction firm in British Columbia can’t simply take on a job in Alberta: different certification rules, different insurance requirements, and often a need to reapply for provincial permits. A trucking company might have to switch regulations mid-route.

This affects normal, day-to-day commerce constantly, and the macro impact is huge. The best estimates suggest these internal barriers reduce Canadian GDP by roughly 3–7%. Canada trades more freely with the United States than with itself.

Other countries used to behave like this, but they dropped it centuries ago. The United States abolished internal tariffs in 1787, Britain did it with the 1707 Union, and Germany cleared its internal borders with the Zollverein in the 1830s. Canada is the only G7 country still running a fragmented market that would have looked normal in the 1700s.

2. The Jones Act

The Jones Act dates back to 1920, when the United States decided that any cargo shipped between two American ports must be carried on a vessel that is US-built, US-owned, US-flagged, and largely US-crewed. It was framed as a national-security measure after the First World War, but it created a protected domestic shipping sector almost entirely insulated from competition.

Shipping from Houston to Boston can cost several times more than shipping from Houston to Rotterdam, and fewer than 100 compliant ships now operate nationwide. A Jones Act tanker can cost 3–5x as much as one built in Japan or South Korea, which prices many domestic routes out of existence. Coastal freight is pushed onto lorries and rail, raising costs and congestion far from the coastline.

Puerto Rico absorbs roughly $1.1–1.5bn a year in extra costs linked directly to the Act. Cars, building materials, food, and fuel all arrive via a needlessly expensive route, even though cheaper foreign vessels pass the island daily. After Hurricane Maria, basic emergency supplies required a federal waiver simply to enter on non-US ships, which delayed support at exactly the wrong moment.

3. The Town and Country Planning Act

The TCPA came in during 1947 and replaced Britain’s old rule-based system with a fully discretionary one. Development rights were nationalised, and nothing could go ahead unless a council granted specific permission. That single shift created a world where predictability vanished and the default expectation became “wait, negotiate, appeal, hope”.

The rules meant everything from a back extension to a housing estate fell under the same discretionary structure, and there was a presumption that you can’t build, unless the council says so. They had no requirement to set any rules, and could deny permission for any arbitrary reason they fancied.

Britain now builds roughly half as many homes per capita as France and about a third as many as the Netherlands. New homes average 76 m², compared with 95–100 m² in much of northern Europe. Planning approvals routinely take a year or more, and large projects can sit in the system for three to five, while Japanese cities approve major schemes in months. Since the mid-1990s, house prices have risen by over 300% while real wages barely moved.

If you want to read my overly-long description of why this policy is terrible, click here.

4. Protectionism run amok

Protectionism survives because it promises simple wins: protect local jobs, stabilise prices, keep foreign competition at bay. In practice it raises costs, shrinks supply, and hardens lobbies that then fight to keep the rules in place. It is one of the few policies where the long-run outcome is the same in every country, yet governments keep it alive because the losses are spread thinly and the gains are concentrated.

Canada’s supply-management system covers dairy, eggs, and poultry through quotas and tariffs that often exceed 200%. Milk in Canada costs 20–40% more than in nearby US states, and quota rights can be worth millions on a single farm.

The US corn-ethanol complex consumes billions of dollars a year in subsidies and mandate support. Around 40% of all American corn now goes into fuel rather than food, with minimal environmental benefit and clear price effects for global grain markets. Farmers in Iowa openly admit that without the mandate they would switch crops, which is why presidential candidates spend every cycle performing loyalty rituals at ethanol plants.

Norway’s dairy tariffs are high enough that the country ran out of butter in 2011 after a small surge in demand. Imports were blocked or priced into oblivion, so shops had empty shelves while foreign suppliers sat a short sail away. The government had to issue emergency quota increases and still struggled to fill the gap.

Japan’s construction state channels public money into rural areas through padded contracts, guaranteed work, and an infrastructure pipeline that keeps going even as populations shrink. Public works spending has hovered around 5–7% of GDP in many years, far above other developed countries. Small towns with a few hundred residents still receive new roads, river embankments, and concrete slopes whose value is far below the price paid.

5. Occupational licensing

Occupational licensing has expanded far beyond anything that improves safety. In the US, about 5% of workers needed a licence in 1970; today it is roughly 25%. Estimates puts the annual cost to American consumers at tens or hundreds of billions in higher prices and lost output. Germany, France, Greece, Australia, and several Canadian provinces show the same pattern: ordinary work pushed into regulated territory for no clear public benefit.

The weakest cases are the low-risk jobs that never needed licensing. More than 1,000 occupations in the US are licensed in at least one state, including hair braiding, interior design, floristry, locksmithing, and massage. Germany’s Meisterpflicht blocks people from opening trades like tiling and plastering without years of formal training. Greece’s old taxi and trucking rules pushed prices up by 30 to 40% and kept new entrants out almost entirely. These systems restrict supply first and foremost.

The sharper failure sits in medicine, where licensing matters but training places are kept artificially scarce. In the UK, the BMA has spent decades resisting expansion and warning that more domestic doctors would “destabilise the profession”. The system now rejects over 20,000 qualified applicants a year for about 9,500 places, then fills the gap with international recruits. In 2022, around 40% of new F1 doctors had trained abroad, and in several recent years international graduates have outnumbered UK graduates entering the NHS. Germany shows the same pattern, with tight caps and steady reliance on Eastern European clinicians. Canada caps residency slots so tightly that hundreds of domestic graduates go unmatched each year while provinces import GPs to fill shortages.

6. Italian notaries

History lovers will know that one of the sparks that lit the American Revolution was the Stamp Act, which forced colonists to pay an official fee for an unnecessary stamp on basic documents. The idea that the state should insert itself into every contract and charge for the privilege looked absurd even in the 1700s.

Modern Italy still does it. Every house sale, company registration, inheritance update, power of attorney, or corporate amendment must go through a notary, and notaries operate as a legally protected cartel. Fees are fixed, numbers are capped, and entry requires a notoriously selective exam with pass rates often below 10%. An Italian notary can earn well over €200k a year, and in some regions more than €400k, largely because the law guarantees their monopoly.

The economic effects show up everywhere. Simple property transfers can cost several thousand euros in mandatory notary fees; corporate registrations proceed more slowly and more expensively than in almost any other EU state; and international investors routinely cite notary costs as a friction point. Italy has built a system where routine paperwork functions as a private tax, collected by a small profession with legally enforced scarcity.

7. The EU’s Common Agricultural Policy

The Common Agricultural Policy began as a way to stabilise post-war food supply and support small farmers, but it became the EU’s largest spending item. It absorbs roughly €55–60bn a year, close to a third of the entire EU budget. Payments are tied mainly to land ownership rather than productivity, environmental impact, or market conditions.

The distribution is heavily skewed, with large landowners capturing most of the money: in many member states, the top 20% of farms receive about 80% of all payments. In the UK’s final years inside the system, some individual estates collected more than €1m a year simply for holding acreage. Several member states still run market-distorting quotas and intervention schemes that keep prices above world levels, long after their original purpose vanished.

Land prices in recipient regions rise because subsidies capitalise straight into the value of acreage. Environmental outcomes stay weak despite the scale of spending, and younger farmers face higher entry costs than they would under a market system. The EU also struggles to open global agricultural trade, since domestic protection must remain in place to justify the subsidies.

8. New Zealand’s unofficial supermarket duopoly

New Zealand’s grocery duopoly came from policy, not preference. Councils limited commercial land through zoning, and competition law allowed supermarkets to attach long exclusivity covenants to nearby sites. These clauses barred future owners from opening a supermarket or leasing to anyone who might compete, often for decades. With the state treating them as ordinary contracts, incumbents built a legally protected perimeter around their stores.

By 2022, these covenants covered hundreds of parcels of land. Some banned food retail within a set radius; others blocked use even when sites sat empty. The Commerce Commission estimated excess profits of about NZ$430m a year, noted that suppliers faced a two-buyer market, and placed New Zealand among the most expensive food markets in the OECD. New entrants couldn’t secure viable sites because most were contractually tied up.

Costco’s arrival in Auckland showed how frozen the market had become. It took a large foreign chain to find land outside the covenant network and push prices down. The government has since moved against these clauses, but the structure remains a clear example of how policy can seal off a market without ever announcing a ban.

9. Price controls

Price controls survive because they promise simple wins: force prices down and claim a victory. Supply always reacts. Firms delay launches, cut investment, or withdraw stock, and the controlled price turns into queues, shortages, or rationing by other means. A small group benefits, and everyone else pays through slower access or higher prices elsewhere.

In pharmaceuticals, Japan and France show how this works in rich, regulated systems. Japan revises prices down every 2 years, often by 20–50%, which pushes companies to treat Japan as a late-launch market. Over the past decade, only about half of new US-approved drugs reached Japan within 3 years, and oncology drugs often arrived 1–2 years later than in the US or Germany. France relies on long pricing negotiations rather than automatic cuts, but the effect is similar. EMA-approved medicines typically take 6–24 months to reach French patients, and France now has lower availability of new cancer drugs than Germany, Switzerland, and the UK. Both systems keep sticker prices low by slowing access rather than reducing real costs.

Rent control shows the same pattern. In San Francisco, a detailed study of the 1990s expansion found that around 15% of affected units exited the rental market through conversion or demolition, raising citywide rents by about 5% for everyone else. Sweden’s system fixes nominal rents well below market levels but produces extraordinary waits: 10 years or more for a central Stockholm flat, and 5–8 years for many suburbs. Berlin’s 2019–21 rent freeze cut listed rents on regulated units but caused new listings in that segment to fall by roughly 30–40%, with sharp increases in unregulated stock. New York’s long-running controls have produced a split market where controlled units are under-occupied and newcomers face some of the highest asking rents in the OECD.

Across both sectors the mechanism is identical: the controlled price becomes a privilege for insiders, and the real price reappears somewhere else - in time, access, quality, or scarcity.

10. US Certificate-of-Need laws

Certificate-of-need laws were introduced in the 1970s to limit “duplication”, but they gave existing providers a formal right to challenge newcomers. If an outpatient surgery centre wants to open, or if a hospital wants a second CT scanner, they must apply to the state. Incumbent hospitals can then file objections claiming local needs are already met, triggering hearings and delays that smaller providers struggle to absorb.

Incentives drive results: independent surgery centres often charge 30–60% less than hospitals for the same procedures, so hospitals use CON rules to keep them out. Dialysis firms trying to enter a region can be blocked by the dominant operator. In many of the 35 states with CON laws, pharmacies, rehab clinics, and even extra MRI machines require approval that incumbents can contest. The law turns competitors into gatekeepers.

Government reviews show that CON states have higher prices, fewer hospital beds per capita, and far fewer independent surgery centres. Border studies find cardiac units 11–15% less available in CON states and nursing homes with fewer beds despite clear demand. Rural hospitals have used CON to block nearby clinics offering cheaper care. The system was sold as cost control, but it largely freezes market entry for the 60% of Americans living in CON states.

Conclusion

Most of these policies share the same structure. The state steps in to “protect” something, locks in a rule that freezes competition, and then spends decades managing the consequences. Prices rise, supply tightens, and incumbents defend the system because it now serves them perfectly.

The striking part is how durable these systems are. Once a policy creates a protected group, reform becomes an uphill fight, because the winners are few and concentrated while the losers are dispersed. The result is a developed world that often behaves like an old guild economy with better branding. Markets are not perfect, but nearly all of the failures in this list come from governments preventing competition, not from competition failing.

When I was at university, my foreign-languages reading list had a habit of drifting into odd places. Our Italian lecturers took us from Dante’s Inferno all the way to modern mafia non-fiction. One book in particular – ZeroZeroZero by Roberto Saviano – had an anecdote I’ve never forgotten.

Investigators in Italy once tracked a container of granite and marble slabs arriving from Latin America. Nothing looked unusual: just heavy stone bound for a construction firm. Buried between the slabs, packed so tightly that even scanners struggled to pick it up, was hundreds of kilos of cocaine.

While most drugs arriving in the UK do not come inside marble, their supply chain is nonetheless a complex system of precisely moving parts. This piece is a walk through a half-dozen of the most popular drugs in the UK, and tracing their path from exotic origins to the toilet stall at your local nightclub.

Cannabis

The earliest clear record of cannabis in Britain comes from the 800s, when Anglo-Saxon medical texts list hænep (hemp) as a remedy and fibre crop. By the 1200s, hemp appears across royal shipyards and monastic estates. None of this is recreational use, but it puts the plant firmly in the country long before Parliament ever cared about it. Modern recreational use only shows up in the 1960s, grows with youth culture, and quietly embeds itself in British life.

Roughly 2–3m people in England and Wales use cannabis each year, making it the most common illegal drug. The market is worth several billion pounds. And unlike most illegal drugs, a large share of the supply chain now sits entirely inside the UK.

Most cannabis used in Britain is grown inside Britain. Police raids give only a partial view, but the scale is clear enough: thousands of grow-houses uncovered each year, almost all in ordinary rental properties. London logged just over a thousand in a five-year stretch. Greater Manchester found more than four hundred in twelve months. These are only the sites police stumble across. The real figure is far higher.

A standard grow-house is a rented house or flat stripped back to the plaster and rebuilt around lights, fans and hydroponics. A single four-bed house can run multiple crop cycles a year and generate six-figure revenue. Multiply that across several thousand properties and you get a domestic market large enough that imports play a secondary role.

The people doing the growing vary by region, but the labour is often exploited: migrants working off smuggling debts, or people kept in houses to tend the crop. Once harvested, the product moves through local networks. There’s no long distribution chain. In most towns, if someone buys cannabis, it has travelled only a few miles.

MDMA / Ecstasy

MDMA is first synthesised in 1912 by Merck during work on haemostatic agents, but it doesn’t enter recreational circulation until the 1970s, when a handful of American therapists and researchers begin using it socially and realise it creates a distinct, reliable euphoria. It spreads into European nightlife in the 1980s and becomes tied to British club culture in the late 1980s and early 1990s, especially the acid house and rave scenes. From then on it sits in the background of nightlife rather than mainstream culture.

Use in England and Wales floats around 0.7–1.0% of adults each year, with higher levels in younger groups. That translates to roughly 300–500k past-year users depending on the survey. Prices stay fairly stable because production is concentrated in a small part of Europe rather than scattered across multiple continents. The UK is simply one end-point of a larger Benelux chemical economy.

MDMA production is concentrated in the Netherlands and Belgium, where a long-standing chemical industry and lax precursor oversight create the conditions for large illegal labs. The inputs are standardised: PMK, safrole derivatives, and other pre-precursors, mostly diverted from China. Chemical suppliers in Europe provide the industrial equipment and solvents. Labs vary from small rural setups to industrial-scale operations capable of producing hundreds of kilos per batch.

Once produced, MDMA leaves the Netherlands in two main forms: powder and pressed tablets. The volumes are modest compared with cannabis or cocaine, so concealment is simple. The most common export route into the UK is postal packages, followed by ferries and haulage into southeast England. Shipments are usually small because the active dose is tiny; a few kilograms of powder is enough to supply a regional market.

Inside the UK, mid-level distributors receive the parcels, break them down, and either press their own tablets or sell the powder directly. Distribution is fragmented: small crews, local dealers, and encrypted-app sellers. There is no long hierarchy. The distance between producer and user is short. In many cases, a pill taken at a British festival has travelled through only two or three hands since it left a Dutch warehouse.

Ketamine

Ketamine was developed in 1962 as an anaesthetic, first used in human surgery in the 1970s, and widely adopted in veterinary practice soon after. Recreational use in Britain appeared in the late 1990s, spread through club scenes in the 2000s, and then settled into a niche but steady presence. It never reached the cultural saturation of MDMA, but it became common enough that most cities developed a small, reliable user base.

Use sits at around 2% of young adults (16–24) each year, with far lower rates among older groups. That works out to roughly 200–300k past-year users. The economics are shaped by the drug’s physical properties: ketamine is light, compact and cheap to produce. A kilogram provides thousands of small doses. That keeps UK supply reliant on frequent, low-volume shipments rather than big consignments.

Most ketamine entering Britain is produced in China and India, where legitimate pharmaceutical plants sit alongside grey-market labs. Ketamine is easier to manufacture than most synthetic drugs. The reactions use common precursors, standard lab equipment and moderate temperatures, which makes production feasible for small operators rather than large industrial plants. The quantities exported illegally are small enough that concealment rarely needs to be elaborate. Shipments are often labelled as veterinary products, research chemicals or industrial powders.

The main import route is the post. Parcels of a few hundred grams move through international mail, private couriers or freight forwarders. Because the active dose is so small, a single parcel can supply an entire local scene for weeks. Larger shipments do arrive through airports and ports, but they’re the exception.

Inside the UK, the distribution layer is thin. A few intermediaries receive the parcels, re-cut the powder and pass it onwards through small networks linked to clubs, nightlife and encrypted-app groups. In many cases, ketamine used in a British flat has travelled from a Chinese exporter to a UK letterbox in under two weeks.

LSD

LSD was first synthesised in 1938 by Albert Hofmann at Sandoz while he was studying ergot derivatives. Its psychedelic effect was discovered accidentally in 1943, and for the next two decades it circulated in psychiatry, research labs and small intellectual circles. Recreational use arrived in Britain in the mid-1960s, attached to counterculture and avant-garde art scenes. Use declined sharply in the 1970s as the cultural wave passed, but it never disappeared.

Modern use is small. Surveys usually show 0.3–0.5% past-year use among adults, rising slightly in university-age groups. That is only a few tens of thousands of regular users. The market is tiny in absolute terms, but it persists because LSD is potent in microgram quantities; whole city-level supplies can be moved in envelopes.

Production is concentrated in Europe. A small number of specialist chemists in countries like Germany, the Netherlands and Czechia synthesised it over the past several decades. The work requires skill, tight temperature control and access to ergotamine derivatives. That limits who can produce it. LSD is not made in small domestic labs. It comes from a very narrow layer of chemists who specialise in this exact set of reactions.

Once synthesised, LSD is distributed as crystal, diluted into liquid, or laid onto blotter paper. These formats are light, easy to hide and cheap to ship. Most LSD entering Britain arrives through the post: letters, parcels, and other low-volume routes. Bulk movement isn’t necessary because a few grams of crystal can produce tens of thousands of doses.

Inside the UK, the chain is minimal. LSD moves through university networks, small festival-linked circles and private groups. There is rarely a hierarchy because the market is too small to support one.

Cocaine

Cocaine’s first recorded use goes back to pre-Columbian Andean societies, where coca leaves were chewed for endurance and altitude tolerance. The active compound was isolated in 1859, when Albert Niemann extracted cocaine hydrochloride in a German laboratory. Its medical use took off in the 1880s, when surgeons adopted it as a local anaesthetic and Sigmund Freud promoted it as a treatment for fatigue and depression. Recreational use followed through chewing-gum, lozenges and cocaine-infused wines such as Vin Mariani. By the early 1900s, cocaine was circulating in Britain and the US as an ingredient in patent medicines. The modern global market emerged only in the 1970s–1980s, when coca cultivation scaled up in the Andes and smuggling routes expanded. Britain’s consumer market grew steadily from the 1990s onward as prices fell and purity stabilised.

About 2.6% of adults in England and Wales report past-year use. That is around 900k people. Powder cocaine sits in the middle of the UK drug economy: not as widespread as cannabis, not as niche as ketamine or LSD. The UK pays among the highest prices in Europe because it sits at the end of the supply chain and relies on entry through northern European ports rather than direct bulk shipments.

Most cocaine used in Britain begins in Colombia, Peru or Bolivia, where coca is grown on small farms. Roughly 200,000–250,000 households depend on the crop across the region. The surroundings are often violent. Homicide rates in Colombian coca zones such as Cauca and Nariño can reach 50–70 per 100,000. For comparison, Chile, which sits outside the coca economy, records about 4 per 100,000. Ecuador, now a major transit corridor, has seen its national rate climb to around 46 per 100,000, one of the fastest increases in South America.

Leaves are processed into coca paste and then into cocaine base in small rural labs. These are usually sheds or outbuildings with a few workers, not the industrial complexes of cartel mythology. Across the Andes, the coca–cocaine economy employs several hundred thousand people in farming, processing and logistics. Once refined, the cocaine base moves to mid-level operators who convert it into cocaine hydrochloride for export. At this stage the chain is worth $10–14bn a year, but the global retail market reaches $150–180bn, most of it captured outside the Andes.

From here, the chain diverges, but the modern pattern is consistent. Large volumes leave South America through Brazil, Ecuador, Colombia’s Pacific coast and Venezuela. Almost all significant shipments for Europe now travel by container ship, hidden in legitimate cargo. The main entry points are Antwerp, Rotterdam and Hamburg. Britain is a second-hop market. Cocaine rarely enters directly into UK ports in bulk because northern European ports are larger, busier and easier to exploit.

Once in Europe, the drug moves into the hands of wholesale brokers based around Belgium, the Netherlands and the western Balkans. Albanian-speaking networks have become prominent over the past decade because they operate consistently at both the mid- and upper-tier levels. These groups move multi-kilo lots across borders using a mix of hauliers, couriers and storage sites in the Netherlands.

The UK layer sits below that. Kilo quantities arrive from Dutch and Belgian hubs into southeast England, often via ferry or through mixed freight. From there, the market fractures into regional distributors and local crews. Cutting agents are added at various stages. By the time cocaine reaches a town-level dealer, it has passed through several hands but travelled mainly within Europe, not across the Atlantic. The “international” part ends at Rotterdam.

Heroin

The earliest recorded use of the opium poppy goes back thousands of years, where raw opium was taken as a pressed resin for pain relief and sedation across the eastern Mediterranean and Persia. The modern drug only appeared much later. Morphine was isolated in 1805, and heroin (diacetylmorphine) was first synthesised in 1874. It entered mass circulation in 1898 when Bayer sold it in syrups, tablets and injectable ampoules as a cough remedy and a supposed safer alternative to morphine. Restrictions followed in the early 1900s once dependence became obvious.

Large-scale illicit production developed after the 1950s, when opium from Turkey, Iran and Afghanistan began moving into small rural labs that refined it into morphine base and then into heroin No.3 or No.4. These labs were simple operations using basic equipment, which allowed production to expand quickly.

Britain saw little heroin use until the late 1970s, when cheap Afghan-Pakistani heroin arrived through London and the Midlands. Use rose rapidly in the 1980s, especially in deindustrialised towns. By the mid-1990s Britain had one of the largest heroin-using populations in Europe. New use has since fallen, but a large long-term cohort remains.

Economically, heroin has a small user base but the highest harm. Past-year use sits around 0.1–0.2% of adults, yet opioids cause roughly 1,200–1,300 deaths a year in England and Wales. Violence inside the UK market is limited, but upstream it is far higher. In parts of Afghanistan’s Helmand and Nangarhar provinces, homicide rates have reached 30–50 per 100,000, compared with about 2–3 per 100,000 in stable neighbours like Jordan or Uzbekistan.

Most heroin used in Britain originates in Afghanistan, which supplied over 80% of global illicit heroin until 2023. Farmers collect raw opium, which is boiled into morphine base and then refined into heroin in small labs. Tens of thousands of people are involved in these early stages.

From Afghanistan the drug travels along the Balkan route through Iran, Turkey and the Balkans into western Europe. Turkish and Kurdish groups dominate the mid-points. By the time heroin reaches Turkey it has multiplied in price many times over.

Britain enters at the final step. Heroin stored in the Netherlands, Belgium or Germany is moved into the UK through lorries, ferries and parcel freight, usually in kilogram quantities. Inside Britain the chain is stable: regional wholesalers, local groups, street-level sellers. The geography still mirrors the towns that saw the 1980s and 1990s epidemic.

If there is a common thread, it is how ordinary the logistics are. A few plants, a few labs, some shipping containers, some postal sorting centres and a handful of local groups. The drama sits mostly in people’s imaginations. The reality is a set of supply chains that would look unremarkable if the products were anything else.