Key takeaways
- A £10,000 share purchase costs £100 in Irish stamp duty, £50 in the UK, £40 in France and nothing in the United States, on each authority's published rate.
- France lifted its transaction tax to 0.4% on 1 April 2025, and Italy doubled its share rate to 0.4% on 1 January 2026, halved to 0.2% on regulated markets.
- At the UK's 0.5% rate, replacing a portfolio once a year costs 9.54% of it over 20 years and 13.96% over 30, before any other charge.
- UK stamp taxes on shares raised £4,320 million in 2024 to 2025, up 35% from £3,200 million the year before, on a rate that hasn't changed.
- The ECB measured a 10% fall in French trading volume after the 2012 tax; an AMF study puts the fall at 20% with no liquidity effect.
What stamp duty on shares costs across eight markets, on a £10,000 purchase
You're about to buy a share, and you want to know what the state takes before your money reaches the market. The answer has little to do with where you live. It has almost everything to do with where the company was incorporated.
On rates published by each tax authority and in force in August 2026, a £10,000 purchase attracts £100 in Ireland, £50 in the UK, £40 in France, £20 in Italy, £20 in Spain, £15 in Switzerland, £10 in Hong Kong and nothing in the United States. The chart above ranks them. Those eight figures are the whole headline, and the rest of this piece is about the two ways the headline misleads.
The first is that the base matters more than the rate. Every one of these taxes carves out something large, and the carve-outs move the effective cost further than the percentages do. The second is that a rate you pay once is trivia. A rate you pay every year is a drag, and drag compounds.
The tax follows the issuer's nationality, not the investor's
In the UK, stamp duty on shares runs on the place of issuance. GOV.UK puts the charge plainly: when you buy shares electronically "you usually pay a tax or duty of 0.5% on the transaction". What decides whether that charge bites isn't your residence. It's the company's.
HMRC's Stamp Taxes on Shares Manual sets out the exclusion at STSM041080. Securities issued by companies incorporated outside the UK fall outside SDRT under FA86/S99(4), unless they sit on a register kept in the UK or are paired with shares in a UK company. The AMF's 2017 review of transaction taxes describes the same principle from the outside: "it is the nationality of the company that issues the shares that geographically determines the tax's scope, rather than the nationality of the counterparties or intermediaries that carry out the transaction."
That's why a London-based investor buying a US-incorporated technology company pays no UK stamp duty at all, while an investor in Frankfurt buying a UK-incorporated company pays the full 0.5%. It's also why the tax is difficult to escape by moving your account. Spain and France build their taxes the same way. France taxes a security only where it is "issued by a company whose registered office is located in France and whose market capitalisation exceeds €1 billion on 1st December of the year before taxation", and Spain applies the same €1,000 million test to Spanish issuers.
Four of these schedules moved in the last 18 months, and three of them went up
Transaction taxes have a reputation for sitting still. They haven't.
France raised its rate first. Article 98 of the finance law of 14 February 2025 "porte à 0,4 % le taux" of the tax on acquisitions of capital securities, applying to acquisitions from 1 April 2025. On a £10,000 purchase of a qualifying French share, that moved the charge from £30 to £40.
Italy went further. The Agenzia delle Entrate now sets the rate on transfers of shares in Italian-resident companies at 0.4%, noting that the rate "è stata dello 0,2% per le operazioni e i trasferimenti effettuati fino al 31 dicembre 2025". The rate is halved for transfers arising from trades concluded on regulated markets or multilateral trading facilities, so an ordinary exchange purchase is charged 0.2%. High-frequency trading on the Italian market moved from 0.02% to 0.04% on the same date.
The UK moved the other way, and only for a narrow group. From 27 November 2025, HMRC's UK Listing Relief exempts companies newly listed on a UK regulated market from the 0.5% Stamp Duty Reserve Tax charge on agreements to transfer securities, for three years from the date of listing.
The United States changed too, though the number is small enough to disappear into a spread. The SEC's order of 27 February 2026 set the Section 31 fee at "$20.60 per $1,000,000 effective on April 4, 2026", up from $0.00 per million before that date. It's charged on covered sales, so a US purchase carries no transaction tax and a £10,000 sale carries about 21 pence.
What replacing a portfolio every year does to the arithmetic
The rate on one trade is the least interesting number in this piece. What matters is how often you cross the tax. Here's what a full replacement of the portfolio each year removes, cumulatively, in each market. A replacement means selling one holding and buying another, so the round-trip cost is the sell-side rate plus the buy-side rate.
| Market | Round trip | 10 years | 20 years | 30 years |
|---|---|---|---|---|
| Ireland | 1.0% | 9.56% | 18.21% | 26.03% |
| United Kingdom | 0.5% | 4.89% | 9.54% | 13.96% |
| France | 0.4% | 3.93% | 7.70% | 11.33% |
| Italy | 0.2% | 1.98% | 3.92% | 5.83% |
| Hong Kong | 0.2% | 1.98% | 3.92% | 5.83% |
| United States | 0.00206% | 0.02% | 0.04% | 0.06% |
These are cumulative haircuts, computed from the published rates above, and they hold whatever the market does. The tax is a share of the amount traded, so it takes the same slice of a portfolio that tripled as of one that halved. That's the reason this table doesn't need a return assumption to work.
Quarter the turnover and you quarter the annual drag. Replacing 25% of a UK portfolio each year costs 2.47% over 20 years rather than 9.54%. Replacing it every second year costs less again. The variable you control isn't the rate, and it isn't the market. It's how many times you cross the toll.
Set that against a UK portfolio's other running costs. The cost of investing in the UK comes to 0.49% a year across seven published layers, of which tax is the largest single one. A buy-and-hold investor pays the 0.5% once and then never again. Someone rotating holdings annually pays it as an extra 0.5% a year, which more than doubles their total cost. On identical holdings, stamp duty on shares is a one-off charge for the first investor and a running one for the second.
Hong Kong charges both sides of the trade, and the US charges only the seller
The per-side design is the detail most comparisons drop, and it doubles or halves the real cost.
Hong Kong's rate, effective from 17 November 2023, is "0.1% of the amount of the consideration or of its value on every sold note and every bought note". Both notes, both sides. A £10,000 round trip costs £10 on the way in and £10 on the way out. The headline 0.1% is genuinely 0.2% to anyone who eventually sells.
The UK and Ireland charge the buyer only. Irish Revenue applies "a Stamp Duty rate of 1% on" instruments transferring shares, and is direct about who hands it over: "You must pay Stamp Duty when: you buy existing shares, stocks or marketable securities (shares) and a document (for example, a stock transfer form) or an electronic transfer order, effects the transfer." Ireland is therefore twice the UK's rate on the way in, and identical to it on the way out, which is to say zero.
Switzerland charges through the dealer rather than the trade. The Federal Tax Administration levies the Umsatzabgabe at "1,5 ‰ für inländische Wertpapiere und 3,0 ‰ für ausländische Wertpapiere", on purchases and sales handled by a domestic securities dealer. A Swiss share bought through a Swiss dealer carries 0.15%, and a foreign security bought through that same dealer carries 0.3%. The liability sits with the domestic securities dealer taking part in the trade, as counterparty or as intermediary, so this schedule taxes the intermediary's location rather than the issuer's.
The United States sits at the other end. Its Section 31 fee exists to fund the regulator, not to raise general revenue, and at $20.60 per million it costs a seller of £10,000 of stock roughly 21 pence.
The UK's exemptions are worth more than the UK's rate
A 0.5% headline with three large holes in it is not a 0.5% tax. HMRC's manual at STSM041270 records that section 99(4B) of the Finance Act 1986 "excludes securities admitted to trading on a recognised growth market but not listed on that or any other market from the definition of chargeable securities", in force since 28 April 2014. HMRC's list of recognised growth markets at STSM041330 opens with the Alternative Investment Market, so AIM shares are out of the charge entirely.
Add the non-UK issuer exclusion, which removes foreign shares that aren't on a UK register, and the listing relief of 27 November 2025, which removes new UK listings for three years. What's left is a tax on secondary trading in established UK-incorporated companies.
It still raises real money. HMRC's Annual Stamp Tax Statistics, published 11 March 2026, record Stamp Duty Reserve Tax receipts rising 33%, from £2,295 million to £3,050 million, and Stamp Duty receipts rising 40%, from £905 million to £1,270 million, between 2023 to 2024 and 2024 to 2025. Combined, stamp taxes on liable securities rose 35%, from £3,200 million to £4,320 million, on a rate that didn't move. The rise came from what was traded, not from what was charged.
The same asymmetry runs through the other schedules. Spain's Ley 5/2020 charges "el tipo impositivo del 0,2 por ciento", but only on Spanish issuers whose market capitalisation exceeded €1,000 million on 1 December of the prior year. France's tax has the same threshold. Italy halves its rate for regulated-market trades. A comparison that ranks these markets by the headline rate of stamp duty on shares ranks them wrongly.
Two studies of the same French tax reached different conclusions
France introduced its tax in 2012. Two studies of what happened next reach different conclusions, and the difference is worth refereeing rather than splitting.
ECB Working Paper 2030, by Jean-Edouard Colliard and Peter Hoffmann in February 2017, found that "average trading volume decreased by around 10%, and this development was accompanied by a moderate decline in market quality". The effect was uneven: "the most liquid stocks were largely unaffected, less liquid securities were subject to considerable adverse effects". They also documented "a shift in security holdings from short-term to long-term investors", and noted that the tax raised 198 million EUR from August to December 2012 against an initial projection of 1.1 billion EUR a year.
The AMF's 2017 review by Gunther Capelle-Blancard reads the same literature and concludes that a transaction tax "reduces transaction volumes, by around 20%, without significant effect on stock market liquidity and volatility". It adds that transaction taxes "are neither the catastrophe feared by some, nor the panacea hoped for by others".
These are less contradictory than they look. Both find volumes falling. They differ on whether the fall damages liquidity, and the ECB paper's own heterogeneity result explains why: the aggregate looks flat because large caps absorb the tax easily, while small caps don't. An average across the whole market can hide the part where the effect lives. For someone holding a broad index the ECB's aggregate is the relevant number, and it's small. For someone holding illiquid smaller companies it isn't.
One finding neither disputes is that turnover falls. If you'd like the behavioural half of that story, the evidence on overconfidence in investing covers what high turnover costs before any tax is levied.
What these schedules cannot tell you
Every figure here is a statutory rate as published in August 2026, and statutory rates are the easiest part of a cost to measure. They're also the part most likely to change: four of the eight moved within 18 months, so a table like this has a shelf life measured in quarters.
The cumulative columns are arithmetic, not a forecast. They assume the stated turnover happens every year, that the rate holds, and that no exemption applies. Change any of the three and the number changes. They also isolate the transaction tax and nothing else. Platform charges, spreads, and the tax on your gains and dividends sit on top, and the UK investment tax rates for 2026 to 2027 are a separate arithmetic from this one.
Converting foreign rates into pounds hides a cost of its own. Buying a Hong Kong or Italian share from a UK account means a currency conversion, and currency conversion costs of 0.25% to 0.99% on UK platforms are larger than several of the transaction taxes in the table above.
The market-effects evidence has narrower limits still. Both studies examine one country, one asset class and one policy change. Neither tells you what a tax at a different rate, or on a different base, would do. The ECB paper is a working paper, and it carries the disclaimer that its views are the authors' rather than the ECB's. The AMF review is an argument for the tax, written by an author who says so in the title, and it should be read as one.
What would change the conclusion
If the base changed rather than the rate. The UK's receipts rose 35% in a year without a rate change. A future move on exemptions, whether widening the listing relief or narrowing the non-UK issuer exclusion, would shift the effective cost far more than a few basis points on the headline.
If a large market abolished its tax, or a zero-tax market introduced one. Sweden's 0.5% tax, introduced in 1984 and abolished at the end of 1991, is the standing example that these schedules aren't permanent. The United States has moved from $0.00 to $20.60 per million within a single year, in the opposite direction and on a far smaller scale.
If your turnover assumption is wrong. The whole table rests on how often the portfolio is replaced. Someone who trades once and holds for 30 years pays the UK 0.5% once, which is 0.017% a year, and the ranking above stops mattering to them. Someone rotating annually pays 13.96% of the portfolio over the same period. The rate is set by the tax authority. The frequency isn't, and it's the larger of the two variables by an order of magnitude.
The number worth watching isn't in any of these schedules. It's your own turnover, which is the one input in this arithmetic that nobody else sets.