Some will. That has never been the interesting part of the question.
The interesting part is how many, which ones, for how long, and how much of the tax base goes with them. Those are four separate questions with four separate answers, and British public debate has spent a decade collapsing them into one word.
We can answer them, roughly, because the United Kingdom ran the experiment. In 2017 the government removed a substantial tax privilege from a group of very wealthy residents and then watched, through HMRC's own records, what they did. The results have been analysed twice, independently, by academic researchers and by HMRC itself. They broadly agree.
They do not tell us what a wealth tax would do. Nothing tells us that yet. But they tell us a great deal about how this argument is usually conducted.
Why this objection wins arguments it hasn't earned
"The rich will leave" has a structural advantage over almost any reply. It is intuitive, it feels like worldly realism rather than ideology, and in conversation it cannot be falsified. Nobody has the counterfactual to hand. The person raising it is not required to say how many, or on what evidence, and rarely does.
It also arrives pre-supplied with numbers. In June 2025 a report from Henley & Partners, an advisory firm that sells residence and citizenship services, projected a net outflow of 16,500 millionaires from Britain that year. The figure was repeated everywhere. It was a forecast rather than a count, it defined a millionaire as anyone with a million dollars in liquid investable assets, and it inferred people's locations partly from LinkedIn and business directories rather than from tax records. There was no control group and no attempt to establish what had caused any individual move.
None of that stopped it becoming the number most people can quote. Tax Policy Associates published a forensic critique the following month arguing the underlying figures were not empirically derived at all. We are not in a position to verify that allegation and do not repeat it as our own. We do not need to. A commercial forecast, on a population a hundred times larger than the one under discussion, with no causal identification, is simply not evidence about the effect of a tax.
Meanwhile the actual administrative evidence, which does exist, is quoted almost never. That asymmetry is the reason this article is long.
Three different things called leaving
Start with the confusion underneath most of this argument.
Saying you will leave is a statement about how you feel, usually made to a journalist or a survey. Surveys of wealthy individuals reliably find that a substantial minority "would consider" relocating in response to higher taxes. This is real information about sentiment and no information whatever about behaviour. People say a great many things they do not do, particularly when asked about a tax they dislike by someone who will print the answer.
Planning to leave is a step further and still not a move. It shows up in advisory-firm enquiry volumes, which rose sharply after the 2024 announcement. Enquiries are a leading indicator of something, though not necessarily of departure, and the firms reporting them have an interest in the story.
Actually ceasing to be UK tax-resident is the only one of the three that appears in tax data, and it is the only one with fiscal consequences.
Even that third category needs splitting, because the administrative record measures tax residence rather than physical severance. Someone who spends most of the year abroad, keeps a house in London, retains directorships, and continues to draw UK income has left in the sense that matters to HMRC's residence test. Whether they have left in any other sense is a different question. The research bears this out: among those who did leave in response to the 2017 reform, more than half were still reporting UK income three years later.
So the accurate phrase is "ceased to be UK tax-resident". The gap between that and "fled Britain" is roughly the width of the entire public argument.
One more distinction, and it is the one that cuts against us. Departures and revenue losses are not proportional. Wealth is distributed so unevenly that a handful of people can represent a large share of any tax base built on it. A statement that one per cent of affected taxpayers left tells you nothing about what happened to receipts. If those leavers sat at the very top, the fiscal effect could be many times their headcount share. We will come back to this, because it is where the strongest version of the objection lives.
What the evidence shows
The British experiment
Until 2017, non-domiciled residents could pay UK tax only on income they brought into the country. The reform removed that treatment from anyone resident for fifteen of the previous twenty years. For those affected, the share of their income they could keep fell by 18.9 per cent. That is a large tax shock delivered to a group with unusually good options.
Arun Advani, David Burgherr and Andy Summers analysed the result using HMRC administrative records, comparing people just over the fifteen-year threshold with otherwise similar people just under it. Their finding: 4.9 per cent of the affected group left over five years.
The dynamics matter as much as the total. Annual departures among this group already ran at roughly five per cent before the reform, because internationally mobile people move. After the reform the rate rose by about six percentage points, then fell back to where it had been. What the reform changed was the timing of departures that were coming anyway. HMRC's later work confirms no delayed second wave.
HMRC ran its own evaluation, published in October 2024, and reached a compatible answer by a different route. Unaffected non-doms were four to five per cent more likely to leave over the period, which is the background churn. Affected taxpayers were ten to twelve per cent more likely to leave. The reform is responsible for the gap.
Then the part that rarely gets quoted. Among those who stayed, UK-reported income rose by an average of £600,000 three years on, and UK income tax paid rose by an average of £190,000. Across the group HMRC tracked, additional tax collected exceeded £700 million in each of the first two years and passed £1 billion by the third. HMRC's own conclusion is unambiguous: the extra tax from those who remained more than compensated for the revenue lost from those who left.
That is one reform, on one unusually mobile population, and it taxed foreign income rather than accumulated wealth. It is not a forecast for anything else. What it establishes is narrower and still useful: a large tax increase on very wealthy people produced real emigration, at a scale far below what was predicted, and raised substantial net revenue anyway.
The reform nobody can yet judge
There is an obvious question here, and the answer is unsatisfying. The non-dom regime was abolished outright in April 2025 and replaced with a residence-based system. Surely we know what happened?
We do not, and we will not for a while. Tax residence is established through Self Assessment returns, and returns covering the year to April 2026 will not be filed and processed until 2027. Until then the available material falls into three categories, none of which answers the question. There are forecasts produced for fiscal scoring, which are assumptions rather than observations. There are counts from advisory firms, which measure enquiries and client activity among a self-selecting group. And there is early payroll data, which captures salaried employees and therefore misses precisely the people whose wealth sits in offshore structures rather than a payslip.
Both sides of this argument have spent the past year citing whichever of these supports them. Neither is entitled to. A departure figure for the 2025 abolition does not currently exist in any form that would survive scrutiny, and anyone quoting one is quoting a projection or a proxy.
We will report the administrative evidence when it arrives, including if it cuts against us. The temptation to declare victory on an early proxy will be considerable, and doing so would be as wrong for us as it is for anyone else.
Where wealth taxes themselves have been studied
For evidence on wealth taxes specifically, the strongest source is a study of Scandinavian administrative data by Katrine Jakobsen, Henrik Kleven, Jonas Kolsrud, Camille Landais and Mathilde Muñoz, forthcoming in the American Economic Review. Their estimate is that raising the top wealth tax rate by one percentage point reduces the stock of wealthy taxpayers by about two per cent. The wider economic effects they measure are small: employment down 0.02 per cent, investment down 0.07 per cent, value added down 0.10 per cent.
Two per cent per percentage point is a real cost and a manageable one. It is also drawn from small, densely connected economies where the nearest alternative jurisdiction is a short drive away.
Switzerland is worth a paragraph purely to disarm it, because a figure from there circulates widely in this debate. Brülhart, Gruber, Krapf and Schmidheiny found that cutting the wealth tax by one percentage point raised reported taxable wealth by at least 43 per cent over six years. The number is often repeated as though 43 per cent of rich people move. The authors attribute only about a quarter of the effect to taxpayer mobility, and that mobility is between Swiss cantons, which requires changing neither country nor currency nor language. Most of the response is reporting and valuation behaviour. It is powerful evidence that wealth taxes provoke avoidance, and it says almost nothing about who emigrates.
Norway, which is the hard case
Norway raised its wealth tax at the top in 2022, and the response was sharp. Christine Blandhol's analysis finds that out-migration among affected households rose from 0.2 per cent to more than two per cent in the year of the reform. That is a tenfold increase, and it is not the sort of number that can be waved away by pointing at averages.
Forty per cent of those leaving were firm owners. Their companies subsequently recorded revenues about 12.6 per cent lower than comparable firms whose owners stayed. Blandhol estimates a long-run reduction in aggregate output of 1.3 per cent, though that particular extrapolation is disputed in the literature and we treat it as contested rather than established.
Stated at its strongest, and it deserves to be: a recurrent wealth tax can produce a sudden order-of-magnitude jump in departures at the very top, and those departures take real businesses with them. Averages across the wealthy population hide this, because the response is sharpest at the very top, which is also where most of the taxable wealth sits.
Three features of the Norwegian case limit what it can tell us about Britain. The 2022 package raised dividend tax at the same time, so the departure response cannot be attributed to the wealth tax alone. Norway's wealth tax begins at around 1.7 million kroner, a threshold roughly eighty times lower than the one proposed here, which means it reaches ordinary business owners whose wealth sits locked inside illiquid companies and who may have to pull cash out of a working business to pay it. And Norway has since tightened its exit tax, which changes the calculation for anyone contemplating the same move today.
What we do not know
The most important limitation is the simplest. No study estimates the effect of a two per cent annual tax on wealth above £10 million, because no such tax has existed anywhere in a form close enough to measure. Every figure above describes a different instrument, a different threshold, or a different country. Anyone offering a confident number for this proposal, in either direction, is extrapolating.
The 2017 evidence has a specific weakness. Non-doms had already migrated once, by definition, and hold foreign ties, foreign family and foreign assets. They are more mobile than a UK-born business owner in Leeds with the same net worth. The 4.9 per cent figure is best read as an upper bound for the most mobile segment rather than an average for the wealthy.
There is a limitation running the other way. Income and accumulated wealth are different bases. A recurrent charge on the stock reaches people who have little income against which to pay it, and may provoke a response that differs in kind as well as degree, so the 2017 result sets no bound on a wealth tax in either direction.
The evidence on headcount is also much stronger than the evidence on the tax base. We can say with reasonable confidence how many people left in these cases. We cannot say what share of taxable wealth left with them, and Norway is the reason that matters. If departures skew towards the top of a Pareto-distributed base, a small percentage of people can carry a large percentage of the revenue.
And there is a channel that produces no departures at all: people who would have come and now do not. A tax can shrink the wealthy population by deterring arrivals while every existing resident stays. That effect is real, is often larger than outward migration, and is close to invisible in the data. It should never be folded into an emigration figure, and we do not have a UK estimate for it.
Departure is a design variable
Here is where the public argument usually stops, and where it should start.
The objection assumes that if wealthy people can leave, the policy fails. That only follows if departure is costless and unconditional. It is neither, and the states that levy these taxes have spent decades building instruments that alter the calculation. Britain would be designing in 2026 with the benefit of watching all of them.
An exit charge taxes accrued gains at the point residence ends, treating assets as sold on the day before departure. Canada does this, as does Australia, and Germany applies a version to substantial corporate shareholdings. It removes the option of accumulating a gain under UK protection and realising it elsewhere tax-free. The most interesting argument for it comes from an unexpected direction. Blandhol's Norwegian study, the same paper that documents the sharpest departure response on record, goes on to model an exit charge on the market value of the firm, and finds it curbs tax flight, particularly among more productive entrepreneurs, while raising aggregate output.
That is a modelled result rather than an observed one, and the distinction matters: the finding describes what the author's framework predicts under its own assumptions. Nobody has run the controlled experiment. But it is notable that the analysis most often cited against wealth taxation contains, in its own second half, a case for the instrument that answers it.
The objection to it is serious. A founder holding £100 million of private company shares has no cash corresponding to the gain, and a charge falling due on departure can force a sale or a valuation fight. Deferral arrangements exist, as in Canada, but they shift the burden onto HMRC as a long-term creditor with an enforcement problem. Any UK version would have to solve illiquidity, and no jurisdiction has solved it elegantly.
A trailing liability keeps someone within the tax net for a defined period after they go. This is less exotic than it sounds, because Britain already does it. The temporary non-residence rules tax gains made during an absence of five years or less, charging them in the year the person resumes UK residence, though they generally reach only assets held before departure. The inheritance tax changes that took effect in April 2025 keep a former long-term resident in scope for between three and ten years, scaling with how long they lived here; ten years requires twenty years of prior residence, and three is the ordinary case. Germany applies an extended liability for ten years to former residents who move to low-tax jurisdictions.
Its weakness is that a tail shifts the margin rather than removing it. People may leave earlier to start the clock, and prospective arrivals may decline to start it at all. It also runs into the UK's double taxation treaties. Where a treaty allocates taxing rights over a person or a gain to the destination state, the domestic claim can give way, and whether that happens depends on the treaty in question.
Source-based charges tax UK assets regardless of who owns them or where they live. This is the most treaty-secure of the three, and again Britain already does it: non-residents pay capital gains tax on UK land, and on companies deriving at least 75 per cent of their value from UK land where the seller holds at least a quarter of the company. Whether that treatment could extend beyond land to substantial holdings in UK operating businesses is an open question, and it is the one a UK proposal would have to answer. The cost is that it discourages exactly the inward investment the country wants, and it invites ownership restructuring through foreign holding companies. It is much cleaner for land, which cannot be moved, than for corporate equity, which can be reorganised.
None of these is free. Each buys a reduction in one behaviour at the price of distortion somewhere else, and any serious proposal has to say which trade it is making instead of pretending the instruments are costless. All three are already law in one country or another.
The principle
A policy is not defeated by an objection existing. It is tested by whether its design has an answer.
This distinction is worth defending, because the alternative is a standard no tax has ever met. Every tax provokes avoidance. Income tax provokes income-shifting; corporation tax provokes profit-shifting; capital gains tax provokes deferral. We do not conclude from this that the taxes should not exist. We conclude that they need anti-avoidance rules, and then we argue about the rules.
The wealth tax debate is conducted differently, because "they will leave" is treated as a natural law rather than a behavioural response to a specific set of incentives that a legislature sets. Once it is stated as a design question, the argument becomes tractable. How large is the response likely to be? Which instruments reduce it? What does each cost? Those questions have partial answers, which is what policy is usually built on.
We are not claiming the design problem is solved. The evidence on exit charges is thin, the treaty position needs specialist review, and the illiquidity problem is unresolved. What we are claiming is narrower: the objection describes a constraint that policy can act on. Conceding otherwise hands the argument over before it has been made.
Where to go next
The exit charge carries more of this argument than any other instrument, and it deserves more than the paragraphs it gets here. Our next piece looks at how one would actually be designed for the UK, what Canada and Norway have learned from running theirs, and whether the liquidity objection can be answered or only mitigated.
Correction — 11 September 2026. This piece cites a claim whose mandatory companions were added to the evidence register after publication. The qualification below is part of the claim and belongs with it.
The 4.9 per cent departure figure is a central estimate, not the finding. The authors' own sensitivity check puts the stock effect between 3.0 and 12.3 per cent, depending on which counterfactual emigration rate is assumed for the treatment group (KB-202). They also removed the two tax years between the reform's announcement in July 2015 and its commencement in April 2017 from their estimation sample, because departures had already begun before the tax changed (KB-198, KB-210) — so the estimate measures the response to the rate rising, not to learning it would rise. HMRC's own evaluation of the same reform, on the same administrative records, found deemed-domiciled taxpayers 5 to 8 percentage points more likely to leave than the control group, and recorded that more than 9,800 of them remained in the UK and that their number has since increased (KB-209).
The millionaire-count figures in circulation are modelled, not counted. That applies to the Adam Smith Institute's figure as much as to the forecast discussed above: its 442,000 sterling millionaires, "down by 7% since 2024", is derived by fitting a Pareto tail distribution with a single parameter. Nobody is counted (KB-257).
Sources and evidence status
All figures traced to primary sources. Advani, Burgherr & Summers, Taxation and Migration by the Super-Rich, using HMRC administrative data (working paper). HMRC, Evaluation of the change to UK Deemed domicile policy 2017, 30 October 2024. Jakobsen, Kleven, Kolsrud, Landais & Muñoz, Taxing Top Wealth, American Economic Review. Blandhol, Curbing Tax Flight?, Princeton (working paper). Brülhart, Gruber, Krapf & Schmidheiny, Behavioral Responses to Wealth Taxes, AEJ: Economic Policy, 2022.
Statements of law in the design section are traced to the relevant tax authority or statute, retrieved 14 August 2026. Canada Revenue Agency — Dispositions of property for emigrants of Canada. Australian Taxation Office — How changing residency affects CGT. Skatteetaten — Exit tax. HMRC helpsheet HS278 — Temporary non-residents and Capital Gains Tax, updated 6 April 2026. HMRC — Inheritance Tax if you're a long-term UK resident, in force from 6 April 2025. HMRC helpsheet HS307 — Non-resident Capital Gains on direct and indirect disposals of interest in UK land and property. Aussensteuergesetz — sections 2 and 6.
The United Kingdom has no general exit charge on ceasing residence and no open consultation on one. HMRC's Capital Gains Manual CG13400 is explicit that no legislation deems cessation of UK residence a disposal for all categories of person.
Contested: Blandhol's 1.3 per cent long-run output estimate is disputed in an unrefereed preprint and is presented here as contested. The allegation that Henley & Partners' figures are not empirically derived is reported as an allegation by Tax Policy Associates, not adopted.
No figure in this article is an estimate for a 2% annual UK tax on wealth above £10 million. No such estimate exists.
0 comments