Every published ruling on the provision has gone the taxpayer’s way. Co-working space counts, a nominee director does not, and an entity with no employees was still excluded.

A regional group holds its Vietnamese operating company through a Singapore entity and sells it for a gain of twelve million Singapore dollars. The Singapore entity has a nominee director, a registered office at its corporate secretary’s address and no staff. The board papers are prepared and the decisions taken in another country. The gain is brought into Singapore, and it is taxable at seventeen per cent, which is about two million dollars.

Now change one thing. The same group on the same deal at the same price pays nothing. That is the outcome if the Singapore entity has two real employees, premises it actually uses, and board decisions genuinely taken in Singapore. That is the whole of section 10L in one comparison, and the difference is not a structure. It is whether the company does anything.

The provision was inserted into Singapore’s Income Tax Act with effect from 1 January 2024. A great deal has been written about it, much of it saying that Singapore now taxes capital gains. That is not what it does. The Government said so in terms during the second reading, stating that the policy objective was not to tax capital gains in Singapore.

What it does is narrower and more specific. Four things have to coincide. The gain must be one that would otherwise fall outside the charge or be exempt. It must arise on a foreign asset. And it must be received in Singapore by an entity of a group with some foreign presence. Where all four hold, the gain is treated as income. There is no new tax and no new rate. The gain is simply taxed as ordinary income at seventeen per cent.

The honest qualification is that a taxpayer who is caught cannot tell the difference. Seventeen per cent on a gain feels exactly like a capital gains tax, whatever the mechanism. Both things are true, and a piece that says only one of them is not much use.

Individuals and standalone companies are outside it entirely

The scope rules exclude more than most summaries acknowledge, and two of the exclusions are substantial. Neither is a concession or a relief that has to be claimed. They are limits on the charge itself, written into the definitions the provision depends on. A person or company outside them never engages with the substance question at all, which is worth establishing before paying anyone to think about substance.

The provision applies only to entities, and the definition of entity expressly excludes individuals. A person selling a foreign asset personally is not within it at all. That is the first thing to establish in any family or founder situation. A great deal of anxious advice has been given to individuals about a provision that cannot reach them. A separate exemption deals with an individual’s gains flowing through a transparent partnership.

It applies only to an entity of a relevant group. A group exists where entities are consolidated line by line, so an associate accounted for under the equity method is outside it. A group is relevant where its entities are not all established in Singapore, or where any entity has a place of business outside Singapore. One foreign branch anywhere is enough to make the group relevant.

A standalone company with no group is therefore outside the provision, and the revenue authority’s own first worked example makes the point. A single Singapore company with an overseas branch is not caught, because the company and its branch are one legal entity. Add a second entity to the picture and the group test is met.

The trigger is the date of disposal rather than the date of receipt. The provision applies to disposals on or after 1 January 2024, and the tax arises in the year the gain is received in Singapore. Receipt includes remittance, applying the amount against a debt of a Singapore trade or business, and using it to buy movable property brought into Singapore.

The substance tests differ for a pure holding entity

The exception that takes a gain out of charge is built into the definition of an excluded entity. It is tested in the period of the disposal rather than the period of receipt, which matters more than it sounds. A group that sells in one year and remits in the next is judged on what the entity looked like in the earlier year. Building substance after the sale and before the money moves does not help.

A pure equity-holding entity must satisfy three conditions. It must submit the returns and accounts required where it is incorporated. Its operations must be managed and performed in Singapore. And it must have adequate human resources and adequate premises in Singapore. There is no expenditure condition for this category.

Any other entity must satisfy two conditions. Its operations must be managed and performed in Singapore, and it must have adequate economic substance there. Adequacy is judged on four matters. They are the number of full-time employees or other persons, their qualifications and experience, the local business expenditure, and whether key business decisions are made in Singapore.

The purity distinction turns on what the entity holds. An entity holding only equity and the incidental assets is pure. A single shareholder loan to a subsidiary can move it into the other category, which changes the test it has to meet. That is a cliff rather than a slope, and it catches groups who did not notice the loan was there.

What the entity has	Treatment
A registered office at the corporate secretary’s address	Not adequate premises
Co-working space or shared premises actually used	Adequate premises
A nominee director	Not a human resource
A director who performs the function	A human resource
No employees at all, in a special purpose vehicle	Can still qualify, tested at the holding entity
Operations outsourced to a Singapore provider	Permitted, on three conditions
One provider serving several group entities	Permitted, if resources match the work
Board decisions taken outside Singapore	Fails the managed and performed test
A prescribed number of staff or floor area	No threshold is prescribed

Outsourcing is expressly permitted for both categories, and that answers a question many groups assume is closed. Three conditions apply. The activities must be carried out in Singapore. The entity must have direct and effective control over the arrangement, with documented monitoring. And dedicated resources must be set aside for it. An arm’s length fee is expected, and one provider may serve several entities where its resources match the work it is doing.

Special purpose vehicles are dealt with separately and sensibly. Three things have to be true of the immediate holding entity. It must have effective control over the vehicle, derive the economic benefits from it, and set the core investment strategies. The substance test is then applied at the level of that holding entity rather than the vehicle. A vehicle with no employees of its own can therefore be an excluded entity.

Every published ruling has gone the taxpayer’s way

Six advance ruling summaries on the provision have been published, between June 2025 and July 2026. All six were decided in the taxpayer’s favour on the substance question. That is the most useful evidence available of where the line actually sits, because there is nothing else. The figures in each are redacted, so the rulings show the shape of an acceptable arrangement rather than its scale.

One ruling concerned a company with a real head office, a Singapore-based chairman with the final say, and board decisions taken in Singapore. It was an excluded entity. Another concerned a special purpose vehicle with no employees at all. It was tested at the level of the holding entity above it, conditional on that entity keeping its incentive. It too was an excluded entity.

The most instructive ruling is the one the taxpayer nearly lost. An incentivised trading company argued that the exclusion for incentivised activities covered its disposal. It lost that argument. The statutory exclusions for incentives are drafted by reference to activities rather than to entities, and the disposal was not part of its qualifying activities. It survived on the substance limb instead.

That is the point for any group holding an incentive. The incentive is not a shield. Eight named incentives are referred to in the exclusions, including the finance and treasury centre, the global trader programme and the maritime sector incentive. The question in each case is whether the disposal formed part of the activity the incentive covers, not whether the company happens to hold one.

Two further points of fact are worth recording. There is no court decision and no Board of Review decision on the provision. And there has been no amendment to it since it was enacted, with the Budget of 12 February 2026 saying nothing about it at all.

Foreign intellectual property is taxed whatever the substance

The one place where substance does not help is intellectual property. A group that does not know this can be caught while doing everything else right. It is also the part of the provision that departs furthest from the stated principle, because the question stops being whether anything real happens in Singapore. For a business holding the wrong category of asset, the answer is fixed before the substance question is reached.

The exclusions that take a gain out of charge apply only to assets that are not intellectual property rights. A gain on a non-qualifying foreign intellectual property right is fully taxed regardless of economic substance. The guide says so directly rather than leaving it to be inferred. Employees, premises and Singapore decision-making make no difference to the outcome.

Qualifying rights are a short list. It covers patents, patent applications, and copyright in software. Trade marks, brands and customer lists are not on it. A business selling a foreign trade mark portfolio is therefore taxed on the whole gain, however many people it employs in Singapore. For a consumer business whose foreign value sits in brands, that is the single most important sentence in the provision.

Where the right does qualify, relief comes through a nexus ratio rather than an exemption. Qualifying expenditure is uplifted by thirty per cent and divided by total expenditure, capped at one hundred per cent. The revenue authority’s own example takes a patent gain of three hundred thousand dollars. Qualifying expenditure is one hundred thousand and non-qualifying expenditure is two hundred thousand. The ratio is 43.3 per cent, and one hundred and seventy thousand dollars is taxed. A three year transitional ratio exists because tracing expenditure cumulatively is genuinely hard.

Losses are allowed, and the symmetry rule is perverse

A loss on a foreign disposal is allowed, but it is ring-fenced in three ways. It can only be set against chargeable foreign disposal gains. It is carried forward rather than set against other income of the year. And it only exists at all if the loss would itself have been within the provision had it been a gain. Each restriction is ordinary enough on its own, and together they produce a result worth noticing.

That produces an outcome worth planning around. An entity with adequate substance is an excluded entity, so its gains are not chargeable. In a loss year that same substance means the loss is not recognised either. Having substance destroys the loss. The revenue authority’s own annex sets this out.

The interaction with the foreign income exemption is also one-directional. The exemptions in section 13 do not apply to a gain caught by section 10L, because the provision overrides the rest of the Act. Disposal gains were never within that exemption in any case, since it covers foreign dividends, branch profits and service income. A dividend paid out of such gains is a separate question and sits outside the provision entirely.

There is no participation exemption and no reorganisation relief. One of the published rulings treated an intra-group share-for-share transfer as a disposal. Nothing about a transaction being internal takes it outside the provision. A group reshuffling its own holdings should establish the position before moving anything. The absence of reorganisation relief is one of the fairer criticisms of the drafting.

The provision is reasonable, and it costs real money to comply with

Section 10L targets genuine double non-taxation, where a gain is untaxed where the asset sits and untaxed where the money lands. It uses the test Singapore’s own policy claims to care about, which is whether anything real happens here. It is narrowly scoped, with individuals, standalone companies and equity-accounted associates all outside it. On the published record the revenue authority has administered it accommodatingly, and it preserves Singapore’s position of not taxing capital gains generally.

One point of background should be stated carefully, because it is often overstated. The provision was introduced in response to European Union guidance on foreign-source income exemption regimes and the international rules on harmful tax practices. Singapore does not appear in either annex of the European Union list, and no verifiable record of a listing or a formal dated commitment was found. The defensible description is pre-emptive compliance with a standard rather than a response to being listed.

The criticisms are real and should be stated at full weight. Adequate is undefined, no threshold is published, and every figure in the published rulings is redacted. Certainty therefore costs professional fees, and that burden falls hardest on small groups, which is regressive. The intellectual property rules punish genuine businesses holding the wrong category of asset. The purity cliff can turn on a single shareholder loan. There is no reorganisation relief. The loss symmetry rule works against the taxpayer in both directions. The relevant group test is crude, in that a single sales office across the causeway puts a company in the same regime as a multinational. And one published question on promissory notes was withdrawn in June 2025 pending further review, leaving deferred consideration unaddressed.

The costs of compliance are not only the substance itself. The most underestimated one is governance discipline. Board meetings have to be held where they are said to be held. Minutes have to record decisions rather than ratify them, and the chain of authority has to be written down. Records must be kept indefinitely under the guide’s own requirements. For intellectual property, expenditure has to be traced cumulatively. And where certainty matters, an advance ruling is the route, with its own fee and preparation cost.

Three claimed costs are not real, and each of them circulates widely. A substantial office is not required, because co-working and shared premises count. Outsourcing is not prohibited, and it is expressly permitted on three stated conditions. Holding companies are not dead either, as the six published rulings demonstrate. A group being told any of those three things should ask where the statement comes from.

What this means for groups, finance directors and advisers

For a group with a Singapore holding company, the work is to establish the position before a disposal rather than after. Substance is tested in the period of the disposal, so arranging it when the sale is agreed is too late. Four items are worth checking specifically. Whether a nominee director is doing the work. Whether the registered office is the only premises. Whether board decisions are genuinely taken in Singapore. And whether a forgotten shareholder loan has moved the entity out of the pure category.

For a finance director, two mechanical points matter more than they look. The group test can be met by a single foreign branch, so a company that considers itself purely domestic may be in scope. The asset test also turns on situs. Singapore immovable property is outside the provision not because of an exclusion, but because it is not a foreign asset. That distinction matters, because an exclusion can carry conditions and a situs point cannot.

For an adviser, a citation trap is worth knowing. The December 2024 amendment to the guide renumbered the paragraphs dealing with special purpose vehicles. Older commentary citing the earlier numbers is now pointing at the test for non-pure entities instead. The current edition is the third, dated 6 June 2025. And the number of entities actually affected by the provision has never been published, so any figure claiming to quantify it is an estimate.

The rule is written as a condition on control rather than as a flat prohibition. The Reserve Bank’s own drafting settles it, and four routes sit outside the restriction altogether.

A founder in Pune runs a profitable engineering services company and wants a Singapore entity. Her investors want one too. She has been told three different things by three advisers. The first said a Singapore company is simply not available to her. The second said it is available and she can put her Indian company underneath it. The third said it depends. All three were working from summaries rather than from the rule, and the rule says something more precise than any of them.

India’s Overseas Investment Rules were notified on 22 August 2022. The provision that governs a resident individual is Schedule III. Its first paragraph permits investment in an operating foreign entity that is not engaged in financial services. The entity must also have no subsidiary or step-down subsidiary where the resident individual has control.

Those last six words are the whole argument. They are in the text, and they mean the restriction is conditional on control rather than absolute. A founder may hold a Singapore company outright, provided that company owns no subsidiary she controls through it. She may also hold a stake without control in a company that does have subsidiaries. What she may not do is combine the two.

A second condition sits in the same sentence and gets far less attention. The investment must be in an operating foreign entity. That term is nowhere defined in the rules or in anything published under them. It is the harder of the two gates for anybody proposing a pure holding company. A company that only holds shares does not obviously carry on business. Almost every discussion of this subject argues about control and passes over the operating condition entirely.

The Reserve Bank’s own drafting settles the control question

Three things establish that the subsidiary condition turns on control, and the third is close to decisive. None of them requires a view about what the policy ought to be. They are the words of the rule, the structure of its proviso, and the way the regulator has drafted its own guidance. Taken together they leave very little room for the absolute reading, though that reading is still in print.

The first is the language itself. The words restricting the subsidiary are qualified by where the resident individual has control. On an absolute reading those words do nothing at all. A court would not readily treat a qualifying phrase in a statutory instrument as surplus.

The second is the proviso. The rules relieve certain categories of acquisition from the restriction. That relief is expressed as applying whether or not the foreign entity has a subsidiary or step-down subsidiary where the resident individual has control. A provision cannot relieve against a condition that does not exist.

The third is the Reserve Bank’s own Master Direction on Overseas Investment. It was issued on 24 July 2024 and updated on 1 April 2026. Its paragraph 22(1) addresses a particular case. A resident individual has invested without control in a foreign entity, and that entity then acquires or sets up a subsidiary. It provides that the individual shall not then acquire control. That sentence makes no sense on an absolute reading. It takes for granted that holding a stake in an entity which has subsidiaries is lawful. It has been in force unamended since the rules were notified.

One caution belongs here. A different paragraph of the same Master Direction lists a foreign entity having a subsidiary as a non-conforming holding, without the control qualifier. It requires disinvestment within six months. An officer at an authorised dealer bank reading that paragraph in isolation could take the stricter line. The conditional reading is the better one, and it is not the only one in circulation.

Control starts at ten per cent and arises by agreement as well as by shares

The definition is wide, and it is why the subsidiary condition catches almost every founder who proposes a holding structure. A founder considering a Singapore parent rarely wants a passive minority stake in it. The whole point of the arrangement is that she directs it. That intention is what brings her inside the definition, usually several times over. No amount of share structuring changes it while she keeps the ability to decide things.

Control arises from the right to appoint a majority of directors, or from control of management or policy. It arises from shareholding and from management rights. It arises from shareholders’ agreements and voting agreements that entitle a person to ten per cent or more of voting rights. It arises in any other manner, directly or indirectly, individually or acting in concert with others.

A founder who holds the majority of a Singapore company clears that bar several times over. So does a founder holding twelve per cent with a board seat and a shareholders’ agreement. The definition is also what makes a subsidiary a subsidiary. An entity is a subsidiary where the foreign entity has control of it.

One question is genuinely disputed. It is whether ten per cent on its own, with no board seat and no agreement, amounts to control. One firm has argued in print that the ten per cent figure should not be read as a standalone test. Its further point is that the Reserve Bank’s directions cannot be treated as exhaustive aids to interpreting rules made by the Central Government. Two others read it broadly, one of them describing the threshold as substantially low. The Reserve Bank itself takes the broad reading in a note to its own Master Direction. Since the bank applies the directions, the broad reading governs in practice whatever its merits.

The second gate is the one that stops a holding company

Schedule III permits a resident individual’s direct investment only in an operating foreign entity. The term is not defined in the rules, in the related regulations, in the directions or in the Master Direction. At least one law firm has publicly identified the gap. Nothing published since has filled it. A founder therefore has to form a view on ordinary meaning, knowing that the bank applying the rule will form its own.

On ordinary meaning, a Singapore company whose only function is to hold shares in an Indian subsidiary is not operating. There is also structural support for treating it as a real restriction. The directions governing investment through the international financial services centre at GIFT City deal with the point expressly. They disapply the restriction on making investment only in an operating foreign entity. A restriction that needed no disapplication would not have been disapplied.

The practical consequence is that the founder-controlled Singapore holding company over an Indian operating business fails twice. It fails the control gate, because the Indian company beneath it is a subsidiary she controls. It fails the operating gate, because a pure holding company does not carry on business. The second failure is the harder one to argue around, and it has had almost no attention.

A trading company is a different matter entirely. Take a Singapore company that signs its own regional customers and employs its own regional staff. It holds the cash those customers pay, and it is plainly operating. Owned outright by the founder and owning nothing in India, it passes all three gates. That is the structure most regional expansions actually need, and a flat prohibition would have wrongly discouraged it.

Configuration	Permitted on the automatic route
Singapore trading company, founder controls, no subsidiary	Yes
Singapore services company alongside the Indian business	Yes
Stake under ten per cent, no control, entity has subsidiaries	Yes
Singapore holding company above the Indian company, founder controls	No, on two independent grounds
Pure holding company with no business, founder controls	No, fails the operating condition
Holding acquired by inheritance, with control and subsidiaries	Yes, expressly without limit
Shares from an employee plan, sweat equity or qualification shares	Yes, the proviso disapplies both conditions
Structure built while non-resident, retained on returning	Yes, the rules do not apply to it
Holding company inside the GIFT City financial centre	Yes, both conditions are disapplied
Lending to or guaranteeing the Singapore company	No, an individual may hold equity only

Four routes sit outside the restriction

The exceptions are not marginal, and two of them are considerably wider than the rule they sit inside. Each one has a different logic. Two reflect the fact that the individual did not choose the structure. One reflects a deliberate policy of routing international business through a designated centre. The fourth follows from the ordinary principle that rules made for residents do not reach back over what a non-resident lawfully did.

Inheritance is the widest. The rules permit acquisition by inheritance without any limit, from a resident or a non-resident. That category also sits inside the proviso. An inherited Singapore company that is controlled and has an Indian subsidiary is therefore lawfully held. It could not have been created, but it can be inherited. A gift is narrower, because gift is not inside the proviso, so all three conditions continue to apply. A gift from a non-resident also engages the foreign contribution rules, which are rarely satisfiable.

Employee share plans, sweat equity and minimum qualification shares are the second route. For them the proviso disapplies both the financial services bar and the subsidiary and control condition. The rules describe the amount as without limit, though any actual remittance still counts against the personal allowance. It requires a genuine employment or directorship connection and a plan applied uniformly across the company. It is not a device for a controlling founder.

The international financial services centre is the third and the most important for anybody who genuinely needs a holding company. The directions disapply both the operating entity condition and the financial services restriction for investment there. They also permit subsidiaries inside the centre, including with control. Outside it, a resident individual may hold a company with subsidiaries only without control. This is the one answer on the automatic route to a founder who wants to control a holding structure.

The fourth is the most valuable and the least known. The Act provides that nothing in these rules applies to an investment made outside India while the person was not resident. Consider a founder who built and held a controlled Singapore holding company with Indian subsidiaries while living abroad. She may continue to hold it on returning to India. The configuration prohibited to a resident is protected where it was created by a non-resident. Fresh subscriptions after the return are not protected, which is where the care is needed.

The two layer relaxation does not rescue an individual

The 2022 rules changed the position on a foreign entity investing back into India. Before them there was no express provision, and the Reserve Bank required specific approval that was in practice not forthcoming. Since 22 August 2022 the rules deal with it directly. They bar a financial commitment in a foreign entity that invests back into India. The bar applies where the result is a structure with more than two layers of subsidiaries. Within two layers it is automatic, with no approval needed.

Most commentary stops there, and that is the commonest error in this area. A founder holding a Singapore company that holds an Indian company is one layer, comfortably inside the two layer limit. But the Indian company is still a subsidiary for the purposes of Schedule III. If the founder controls the Singapore company, the Schedule III condition is breached regardless of how few layers there are.

The two layer relaxation is in substance a relaxation for Indian companies investing abroad. For an individual, Schedule III bites first and bites harder. How the layers are counted is itself openly disputed, with no clarification from the Reserve Bank. At least one firm has questioned whether the relaxation reaches resident individuals at all.

One point of scope is worth stating because it removes an escape route people reach for. Acquiring any amount of unlisted foreign equity is direct investment under these rules. There is no percentage threshold in the first limb of the definition. A Singapore private limited company is unlisted, so a single share is direct investment and all three gates apply. The ten per cent figure is relevant only to listed entities.

The reporting is light and the deadlines are unforgiving

A form is filed through the authorised dealer bank before the first investment is made, and a unique identification number issues. No money may leave India until that number is issued. The Master Direction states expressly that its allotment is not an approval from the Reserve Bank. Evidence of the investment is due within six months or the funds must come back.

An annual performance report falls due at the end of December each year. It is not required where the holding is under ten per cent without control and there is no other financial commitment. For an individual it must be certified by a chartered accountant. The report requires disclosure of any step-down subsidiary acquired or established, so the Schedule III condition is in effect self-reporting.

Late reporting is curable, and the mechanism is published rather than discretionary. A late submission fee runs on a formula, is available for up to three years, and is capped at the amount involved. For a periodical return it is a flat figure. For a transactional return it rises with the amount and the delay. It cures late reporting only, and it does nothing at all for a substantive breach of the conditions.

A substantive breach is a different matter. The Act provides for a penalty of up to three times the sum involved. Where the amount is not quantifiable the figure is two lakh rupees, with a continuing daily charge. Compounding is available. For a contravention of this kind the published schedule gives a fixed amount plus a percentage of the sum involved, subject to an overall cap.

No enforcement has been published on this specific point

The Reserve Bank publishes summaries of its compounding orders, and the term step-down subsidiary does not appear in any contravention description. Orders have been made against individuals under these rules through to September 2026. They concern limits, reporting and transfers rather than the control condition.

That absence cuts both ways and should not be oversold. The rules are young, and compounding is started by the applicant rather than the regulator. An absence of published orders is not evidence of tolerance. There is also no adjudication, tribunal decision or court decision on the point. No official set of questions and answers on the Overseas Investment Rules exists at all.

One false positive is worth flagging for anybody searching the orders. Entries citing rule 5 read with paragraph 1 of Schedule III are made under a different instrument. The Current Account Transactions Rules 2000 have their own Schedule III. They have nothing to do with overseas investment, and a search on the phrase will return them.

In the absence of enforcement, paragraph 22(1) of the Master Direction is the best available evidence of how the Reserve Bank reads the rule. It is nearly as good as a compounding order would be. The paragraph could not have been drafted on any reading but the conditional one.

What this means for founders, returning non-residents and advisers

For a founder building regionally, the structure that works is the one most businesses actually need. A Singapore company that signs its own customers, employs its own people and owns nothing in India passes all three gates. It may be held outright. The capital travels under the personal remittance allowance. That is USD 250,000 for each resident individual in each financial year, so two co-founders hold two allowances. One limit must never be crossed. That company must not acquire the Indian business at any point. From the moment it does, the founder holds a controlled foreign company with an Indian subsidiary beneath it.

For a founder who genuinely needs a holding company, there are two honest answers and neither is an offshore workaround. The international financial services centre permits it on the automatic route, with both conditions disapplied. The investors can instead come into the Indian company directly as inbound investment. That is how most of these situations are actually resolved. A stake below ten per cent without control is a third possibility, and it is a one-way door. A person who went in without control may never acquire control while the entity has a subsidiary.

For a returning non-resident, the most valuable thing in these rules may already be in place. A controlled Singapore holding company with Indian subsidiaries may already exist, built while the founder was abroad. On return it is protected by the statutory carve-out. The documentation establishing when the investment was made, and what the person’s residence was at the time, is therefore worth assembling properly. It is the whole of the protection. Anything subscribed after the return sits outside it.

For an adviser, three points are worth carrying. The bar is conditional on control. A flat statement that a resident Indian cannot own a Singapore company is wrong, and it discourages legitimate business. The operating entity condition is undefined, and it is the stronger objection to a pure holding company. It is almost never raised. A claimed 2025 amendment to these rules also circulates online. It carries no notification number, no date and no operative text, and it is corroborated nowhere. The rules as notified on 22 August 2022 appear never to have been amended.

Vietnam charges nothing on dividends paid to a corporate shareholder abroad. The constraint is the annual audit cycle, and a Singapore holding company does not shorten it.

A precision components manufacturer near Ho Chi Minh City has its first good year. It lost 18 billion dong in 2024 and made 52 billion dong before tax in 2025. Its owner is a Singapore holding company, and the group wants the cash. The finance director has been told that a Singapore parent reduces Vietnamese withholding tax on dividends from ten per cent to five. That advice is wrong in a way that matters, because Vietnam charges nothing on those dividends in the first place.

This is the single most common error in the field, and it is imported from elsewhere. China, Thailand and Indonesia all tax outbound dividends, so a holding company in the right treaty jurisdiction genuinely reduces the rate. Vietnam does not. A dividend paid to a corporate shareholder abroad leaves Vietnam free of withholding tax, whoever the shareholder is and wherever it sits. The treaty figures of five, seven and twelve and a half per cent are ceilings on a tax Vietnam has chosen not to levy.

What Vietnam does impose is a sequence, and it is long. Profit earned in January 2025 reaches the shareholder in June 2026 at the earliest. The constraint is the audit and filing cycle rather than the tax. No structure shortens it, because every step is a statutory precondition rather than an administrative habit. For a group accustomed to sweeping cash monthly, that is the real cost of operating in Vietnam. It is a working capital problem rather than a tax one.

Four of the governing instruments also changed in 2025 and 2026, which means almost all published guidance on this subject is now out of date. The capital account has been renamed. The repatriation conditions have moved into a new decree, with a new condition attached. The transfer pricing decree has been replaced, and the charge on exit has been rebuilt from the ground up.

Vietnam charges nothing on an outbound dividend to a company

The withholding position is better than the one most groups are planning around, and the treaty makes almost none of it better still. That second half is the part advisers get wrong. A treaty sets a ceiling on what a country may charge. Where the domestic rate already sits below the ceiling, the domestic rate applies and the treaty is irrelevant. On two of the four main payment types, Vietnam’s domestic rate is already at or below what the treaty would allow.

A dividend to a corporate shareholder carries nothing. A dividend to an individual carries five per cent. Interest carries five per cent domestically against a treaty ceiling of ten. The treaty is therefore worse than the domestic position, and a Singapore lender gains nothing from it. Royalties carry ten per cent domestically, and here the treaty does help. For patents, designs, formulas, processes and equipment the ceiling is five per cent, which is a genuine halving.

Services paid to a foreign contractor attract a combined charge of five per cent corporate tax and five per cent value added tax. The second of those is generally creditable. The corporate element can be reduced to nothing under the treaty where there is no permanent establishment, which is a claim rather than an automatic position.

The claims procedure itself changed on 1 July 2026, when a new circular replaced the 2013 rules. The burden of proof sits with the taxpayer, substance is weighed over form, and a claim is refused after three years. A Singapore certificate of residence is the precondition for any of it, and Singapore tightened what it takes to get one from 2025.

The money is free but the calendar is not

The sequence is fixed by statute and by the audit requirement, and it runs on an annual cycle with no interim route. There is no mechanism for an interim dividend out of Vietnam in the way a British or Singapore company would declare one. The year has to close, the accounts have to be audited, the tax return has to be filed and the tax paid. Only then does the question of distribution arise at all.

Audited financial statements and a filed corporate tax finalisation are both hard preconditions. Both fall due on the last day of the third month after the year end. An audit is separately compulsory for a foreign-invested enterprise regardless of size. Every tax, late payment charge and penalty must be settled. Since July 2026 a further condition applies, which is that no tax debt may be outstanding or overdue at the moment of remittance.

Accumulated losses block a distribution for the year, and the trap here is a mismatch of two numbers. Corporate tax is computed on taxable income after loss relief, which runs for five years. The distributable amount is retained earnings in the accounts. On the manufacturer described above, tax is charged on 34 billion dong after loss relief. That gives a charge of 6.8 billion dong and an after-tax profit of 45.2 billion. The distributable ceiling is not 45.2 billion. It is retained earnings of 27.2 billion, because the 2024 loss still sits in the accounts.

Notice of the transfer must be given to the managing tax office at least seven working days beforehand, on a prescribed form. There is no approval and no objection period. The claim that the tax office has seven days to object describes a mechanism that does not exist. The notice is a notice.

Taken end to end, profit earned in January 2025 can be distributed in June 2026 and not before. Seventeen months is the figure to plan against, and it assumes everything goes smoothly. A qualified audit opinion, a disputed assessment or an unpaid penalty extends it. So does a late discovery that retained earnings are lower than after-tax profit. The sequence has no mechanism for urgency, and nothing in it responds to the shareholder needing the money.

The capital account was renamed in August 2026

The account through which investment money moves has been known for years by an acronym that no longer exists. A circular issued on 31 July 2026 and effective from 18 August 2026 replaced the 2019 rules in full. It renamed the direct investment capital account as the foreign investment capital account.

Three of the changes are commercially useful. A foreign currency account is no longer mandatory, so the account may be held in dong, in foreign currency or both at a single licensed bank. The account may be opened before the investment registration certificate issues, with limited uses until it is granted. Profits may now be retained in Vietnam and moved from the account to the investor’s payment account to fund another Vietnamese project. That previously required taking the money out and bringing it back.

Two details are worth checking against existing documentation. The ownership threshold that determines whether the account is required is more than fifty per cent, not fifty-one, and it has been since June 2025. Capital contributions, capital transfer payments between local and foreign parties, profit remittance, loan drawdowns and repayments, and returns of capital must all pass through it. Transfers between two foreign parties or two local parties need not.

Capital transfers between a resident and a non-resident generally settle in Vietnamese dong. That places the currency risk on the foreign party for the period between pricing and settlement. It is not a detail that appears in any rate table, and on a large transaction it can exceed the tax. The trading band has been five per cent either side of the reference rate since October 2022.

The exit charge was rebuilt, and it is now payable on a loss

The charge on selling a Vietnamese company changed completely on 15 December 2025. Whether the change is good news depends entirely on whether the seller made money. A charge on the gain and a charge on the proceeds behave in opposite ways at the bottom of the range. Vietnam moved from the first to the second, which helps a profitable seller and penalises an unprofitable one.

A foreign corporate seller now pays two per cent of gross sale proceeds, in place of twenty per cent of the net gain. Take a sale at 1,000 billion dong with a contributed capital of 600 billion. The old charge was 80 billion dong and the new charge is 20 billion. That is a substantial improvement. Now take the same sale where the seller had put in 1,200 billion dong and is therefore losing money. The old charge was nothing and the new charge is still 20 billion dong. The break-even point is a gain of ten per cent of the sale price.

The charge also reaches indirect transfers. Selling the offshore holding company rather than the Vietnamese shares does not escape it, which closes the structure most often proposed for an exit. That was not the position before December 2025, so a plan built on the older rules now produces a charge its authors did not expect. An intra-group restructuring exemption exists, on four cumulative conditions that have to be met together.

Two practical points follow. The tax point may be the date of contract signature rather than receipt of the money. The charge can therefore crystallise before the seller has been paid and converted. The treaty argument that the gain should be taxable only in Singapore is also contestable on several grounds. Those include the principal purpose test and the new burden of proof rules. There is a further question of whether a contribution to a limited liability company’s charter capital is a disposal of shares at all.

Vietnamese corporate tax is where the money actually goes

The headline rate is twenty per cent, and that is where essentially all the tax on a Vietnamese operation is paid. A new corporate income tax law took effect on 1 October 2025 and applies from the 2025 tax period. Implementing decrees and circulars followed through March 2026.

Lower rates of fifteen and seventeen per cent exist for smaller companies, tested on the preceding period’s revenue. Most foreign-invested subsidiaries cannot use them, because the law denies both tiers to a subsidiary or affiliate of a group that does not itself qualify. Incentive rates are genuinely generous where they apply. High technology, software, semiconductors and large projects can obtain ten per cent for fifteen years. A period of full exemption is followed by several years at half rate.

Two changes in the new law bite on common structures. Interest paid to a non-credit institution above the civil code ceiling of twenty per cent a year is wholly non-deductible, which constrains shareholder loans. And industrial parks were removed from the list of incentivised locations for new projects, which a great deal of older site-selection advice still assumes.

A replacement transfer pricing decree took effect on 1 July 2026 and applies from the 2026 period. It raised the revenue threshold for the profitability safe harbour from 200 billion to below 500 billion dong. It also moved the country-by-country reporting threshold onto a euro 750 million group revenue basis. The interest deductibility cap of thirty per cent of earnings before interest, tax, depreciation and amortisation is reported unchanged, with a five year carry-forward.

A tax holiday is where the Singapore company starts costing money

The one case where a Singapore holding company makes a group worse off is the case investors most often want. It follows directly from Singapore’s own conditions rather than from anything Vietnam does. The incentive that attracted the investment is the same thing that breaks the exemption on the way out. Nothing in the Vietnamese rules causes it, and nothing in the Vietnamese rules can fix it.

Singapore exempts a foreign dividend only where three conditions are met. The income must have been subject to tax in the source jurisdiction. That jurisdiction’s highest corporate rate must be at least fifteen per cent. The exemption must also be beneficial to the Singapore recipient. Vietnam’s twenty per cent headline rate clears the rate test comfortably. It still clears it where the subsidiary itself enjoys an incentive rate of ten or fifteen per cent. The test looks at the headline rate rather than at the rate actually applied.

The subject-to-tax condition is the problem. Vietnam levies no dividend withholding tax, so there is no Vietnamese tax on the dividend itself to point to. The condition therefore has to be satisfied by the underlying corporate tax the subsidiary paid. A subsidiary paying twenty per cent satisfies it. A subsidiary sitting inside its four year exemption period has paid no corporate tax at all. The condition therefore fails. The dividend is taxable in Singapore at seventeen per cent, with no credit available.

Singapore also requires more of a holding company than it used to. From 2025 a foreign-owned investment holding company needs management and control in Singapore plus at least one further element. That element is a Singapore-based executive director who is not a nominee, a Singapore-based senior officer, or management by a related Singapore company. Board meetings alone are not sufficient. Without a certificate of residence there is no treaty claim at all, so the one real treaty benefit on royalties depends on maintaining that position.

Cross-border cash pooling does not exist in Vietnam

Groups that run a regional treasury routinely ask how to include Vietnam, and the answer is that they cannot in the normal sense. There is no offshore dong market. Domestic settlement must be in dong. A Vietnamese company lending money offshore is a capital movement rather than a treasury operation.

Three channels move money on a defensible basis, and together they fall well short of a pool. The annual dividend is the largest and the slowest. An inbound shareholder loan is useful in both directions. The interest is deductible against twenty per cent tax while bearing five per cent withholding. Repayment of principal is neither taxable nor a profit remittance. That last point is the real structuring insight here: principal repayments need no audit and no seven day notice.

Management fees and royalties are the third channel. A management fee bears the foreign contractor charge, and the corporate element is far smaller than the corporate tax deduction it generates. A royalty for patents or know-how is the one place the treaty genuinely halves a rate. None of this is pooling. It is a set of separate payments, each with its own documentation, each defensible only to the extent it reflects something real.

Offshore borrowing into Vietnam has its own regime. A loan with a tenor over twelve months must be registered with the central bank, and there is no monetary threshold, only a tenor test. Short-term offshore debt may only refinance offshore debt or pay short-term payables, and it cannot fund working capital or capital expenditure.

What Vietnam gets right, and what it costs

The right to remit profit is statutory rather than licensed, which is not true everywhere in the region. Charging nothing on outbound corporate dividends is worth more than any holding company structure. It compares well against Indonesia at twenty per cent, Thailand at ten and China at ten. The accumulated loss rule is ordinary capital maintenance. Vietnam accepted the International Monetary Fund’s obligations on current account convertibility in 2005. Profit remittance is a current account transaction. The common claim that the dong is non-convertible is therefore too blunt to be useful.

The 2026 reforms also cut real friction. The foreign currency account became optional, and the capital account can be opened before the licence issues. Profits can be retained for reinvestment without a round trip. The transfer pricing safe harbour threshold also rose substantially. Banks now carry published procedural duties they did not have before.

The costs are equally real. The annual cycle traps cash for up to seventeen months with no interim dividend route. Two per cent of gross proceeds is payable on an exit that loses money. Indirect transfers are caught with no credit mechanism. Capital transfers settle in dong, so the foreign party carries the currency risk. Short-term offshore debt cannot fund the things a business actually needs funded. And the sheer pace of instrument change means nothing learned before 2025 can be relied on without checking.

One figure is missing from the public record entirely. No central bank, finance ministry or statistics office series exists for profit actually repatriated from Vietnam. The World Bank’s economic update refers to large profit repatriation from the export sector without giving a number. Several commercial sources publish estimates, and none of them cites an official series.

What this means for investors, treasurers and advisers

For an investor, the planning assumption should be that Vietnam is cheap on tax and slow on cash. Budget for the first distribution to arrive seventeen months after the profit is earned, and fund working capital accordingly. Check whether the subsidiary is inside an incentive period before putting a Singapore company above it. That is the one configuration where the holding company adds a seventeen per cent charge rather than removing one.

For a treasurer, the three channels should be sized and documented separately rather than treated as a pool. The shareholder loan deserves particular attention. Principal repayments sit outside the profit remittance machinery altogether, and they are the only route that does not wait for an audit. The interest rate has to stay inside both the civil code ceiling and the earnings-based deductibility cap. The loan also has to be registered where the tenor exceeds a year.

For an adviser, the instrument list is the work. The repatriation conditions moved into a new decree in July 2026 and acquired a new condition. The capital account was renamed in August 2026. The transfer pricing decree was replaced in July 2026, superseding both of its predecessors. The treaty claims circular was replaced in July 2026. Any memorandum written before the middle of 2026 should be assumed stale until each of those has been checked.

The 2023 change removed a timing escape rather than introducing worldwide taxation. Money earned before 2024, money earned in a non-resident year, and capital itself all stay outside the charge.

A British engineer retires to Chiang Mai in March 2025 and keeps a Singapore brokerage account she opened in 2011. She spends more than 180 days in Thailand in both 2025 and 2026, so she is a Thai tax resident in both years. In 2026 she transfers 2.4 million baht to her Thai bank to live on. Her neighbours have told her that Thailand now taxes her worldwide income and that the money is caught. They are wrong on both counts, and the difference in her case is about 150,000 baht.

Thailand did change its rules, and the change was real. What it did was narrower than almost every account of it. Thailand still taxes foreign income only when it is brought into the country. What disappeared was a timing escape that had stood since 1987. Under it, foreign income remitted in a year later than the year it was earned fell outside the charge entirely.

That escape was not a remittance basis. It was a one-year waiting period, and no comparable system works that way. Removing it brought Thailand into line with how a remittance basis is normally understood. Thailand then declined to go further. A separate draft would have taxed residents on foreign income whether or not it was ever remitted, and that draft is not law either.

This article sets out what the rule now catches and the three large categories it does not. It also deals with an announced relaxation that a great many people are relying on and should not be. The announcement was made by the head of the Revenue Department and reported everywhere. No instrument has followed it. Anyone who has moved money on the strength of that reporting has a liability they do not know about.

Three categories of money stay outside the charge

The useful part of the rule is what it leaves alone, and the Revenue Department has published its own position on each of these. That matters, because the exclusions are not inferences drawn by advisers from the statutory language. They appear in the department’s own worked examples, which is the best evidence available of how the rule will be applied on audit. The examples are also more generous than the drafting strictly required.

Income that arose before 1 January 2024 is not taxable, whenever it is brought in. That is the effect of the second instruction, issued on 20 November 2023, which disapplied the new reading to income derived before that date. Consulting fees earned in 2022 and remitted in 2026 are outside the charge. This is clean grandfathering on a calendar line, and it is more generous than it had to be.

Income that arose in a year when the person was not a Thai tax resident is also not taxable, whenever it is brought in. The Revenue Department’s own worked example covers a Thai national who lived in China from 2007 and returned in 2024. The savings accumulated during those years are untaxed on remittance. Residence is tested in the year the income arises.

Capital is not income, and the Revenue Department says so directly. Only the gain above cost is assessable, computed transaction by transaction. On a bank deposit the principal is untaxed and only the interest is in charge. Unrealised gains are nothing at all. Money that left Thailand and came back is not assessable. Where a remittance is mixed, the taxpayer characterises it on the facts.

That last row deserves emphasis because it is where real liabilities arise. Consider one multi-currency account opened a decade ago, used for salary, dividends and sale proceeds with continuous reinvestment. The holder has a single pool with no natural dividing lines. Reconstructing the composition is possible but tedious. A statement showing the position as at 31 December 2023 is the single most valuable document such a person can hold. The window to obtain one cleanly has largely closed.

The announced relaxation is a draft, and remitting on the strength of it is expensive

In May 2025 the Director-General of the Revenue Department announced that income remitted within two tax years of arising would be exempt. Income brought in later would remain taxable at the normal rates. The announcement was corroborated the following day by a Deputy Director-General. It was widely reported as a change in the law.

It is not law. It requires Cabinet approval and then review by the Council of State, and neither has produced an instrument. The Revenue Department’s own register of newly issued Royal Decrees runs as a complete sequence through October 2026, and none concerns foreign-source income. Its register of ministerial regulations shows nothing relevant either. The department’s news index through 22 October 2026 contains no item on the subject.

Even the form the relaxation would take is disputed. The Revenue Department itself described a ministerial regulation. At least one advisory firm described a Royal Decree. A third said it could be either. That disagreement cannot be resolved, because there is no instrument to read.

Two measures are routinely confused with it, and both are enacted. A Ministry of Finance exchange control notification effective 20 January 2026 exempted receipts below USD 10 million from the repatriation requirement. That is exchange control rather than tax. A Bank of Thailand instruction effective 29 December 2025 imposed stricter documentary checks on foreign currency transactions above USD 200,000. The stated purpose of the second was currency volatility, though the paper trail it creates closely resembles what a tax audit would want.

Anyone who remitted money in 2026 on the strength of the May 2025 announcement is taxable on it. The Revenue Department has issued no public correction to the reporting. Nor has it formally withdrawn the proposal, so the position is not that the relaxation was abandoned. The position is that it was announced, drafted, and never enacted, and a draft creates no exemption. That silence has cost people money.

Residence turns on 180 days, with no apportionment

The test is a count of days in the calendar year, aggregated rather than continuous. At 180 days a person is a Thai tax resident. At 179 they are not. There is no apportionment, no tie-breaker and no partial year. That cliff edge cuts both ways. It makes the test cheap to apply and hard to argue with, which compares favourably with the multi-limbed statutory residence tests used elsewhere. It also means a single extra trip can change a year’s tax position entirely.

Nationality is irrelevant. The Revenue Department states the point explicitly in its own material, without regard to nationality or ethnicity. Two claims circulate widely and both are wrong. The rule does not target Thai nationals returning from abroad, and it does not target foreign retirees. It applies to anyone who crosses the day count.

Companies are not affected at all. The provision sits in the personal income tax part of the Revenue Code and applies to natural persons. A Thai-incorporated company is taxed on worldwide income as it accrues, at 20%, with graduated rates for smaller companies. Remittance is irrelevant to a company in either direction.

The rates themselves are not punitive at ordinary amounts. The first 150,000 baht is exempt, and the top rate of 35% only engages above 5 million baht. Return to the engineer described at the start. The assessable portion of her remittance is 1.4 million baht, which produces about 215,000 baht of tax. That is 15.4% of the assessable slice and 9% of the whole transfer. The arithmetic only looks bad when the layers cannot be separated.

A thirteen week difference in a completion date can be worth more than a house

The sharpest consequence of the rule is not the rate. It is that residence is tested in the year the income arises. That makes the timing of a disposal decisive for anybody planning a move to Thailand. A gain realised while still a non-resident never enters the charge, however long the money sits abroad afterwards and whenever it is eventually brought in. A gain realised after residence begins enters the charge on remittance, which may be years later.

Consider a founder selling a business for a gain of 30 million baht, who intends to settle in Thailand and buy a house there. In the first version the sale completes in August 2025. He spends 70 days in Thailand that year, so 2025 is a non-resident year. He relocates in January 2026 and remits the proceeds. The gain arose in a non-resident year, so the first condition of the charge fails. Later remittance does not revive it. The Thai tax is nil.

In the second version everything is identical except that the sale completes in September 2026, after he has crossed 180 days. He remits in 2027. The gain is now assessable in the year of remittance, and the tax is about 10 million baht. There is no foreign tax credit, because Singapore charged nothing on the gain in the first place.

These are illustrative composites rather than real people, and the allowances are left out, so a real liability would be somewhat lower. The structural point survives the simplification. Ten million baht turns on which side of a day count a completion date falls, and that is a date rather than a judgement. It is the strongest argument there is for doing the work before arrival rather than after.

The treaty credit fails in exactly the case people rely on it for

Thailand and Singapore have a treaty signed on 11 June 2015, in force from 15 February 2016 and effective from 1 January 2017. It provides for Thailand to credit Singapore tax against the Thai tax on the same income, capped at the Thai tax attributable to it.

A credit is worth the lower of the two taxes, which is the point that defeats most expectations here. Where Singapore charged nothing, the credit is nothing. Singapore has no capital gains tax and exempts most distributions received by individuals. A gain realised in Singapore and remitted to Thailand therefore bears Thai tax at up to 35% with no relief at all. The treaty is not failing in that case. It was never designed to relieve tax that one side did not charge.

Two further gaps are worth knowing. There is no unilateral foreign tax credit for an individual receiving income from a country Thailand has no treaty with, so that foreign tax is simply lost. And where tax is paid abroad in one year and the income is remitted three years later, the Revenue Department has published no criteria for matching them. Its assurance that double taxation cannot arise is a statement rather than an analysis.

One route does work, and it is statutory rather than interpretive. A Royal Decree issued in 2022 exempts holders of the Long-Term Resident visa from tax on foreign income brought into Thailand. It covers the wealthy global citizen, wealthy pensioner and work-from-Thailand categories. Because it is a Royal Decree, a departmental instruction cannot override it. The door is narrow: the Board of Investment reports 12,011 approvals across all categories in four years, against 12.57 million personal tax returns filed. One drafting point is unresolved. The exemption is expressed by reference to income derived in the previous tax year. Whether it covers income accumulated over many years is not addressed by any source.

Singapore does not tax an individual on foreign income at all

The contrast is clean for an individual and much less clean for a company. Conflating the two is the commonest error in this area, and it runs in both directions. Advisers quote the company conditions at individuals who are not subject to them. They also quote the individual exemption at companies that cannot use it. The two regimes sit in different subsections and were written at different times for different reasons.

For an individual, Singapore exempts income arising from sources outside Singapore and received in Singapore. For a resident the exemption has depended since 2004 on the Comptroller being satisfied that it is beneficial, which is a formality. There is no subject-to-tax condition and no headline rate test for an individual, because both of those belong to the company regime. One exception is real and catches professional structures. Foreign income received by a resident individual through a Singapore partnership is taxable.

Malaysia brought foreign income into charge from 2022 but exempts resident individuals, on conditions, through to the end of 2036. Hong Kong taxes individuals on a territorial basis and its foreign-source income regime applies only to entities, with natural persons expressly excluded. Of the four, Thailand is the only one that actually taxes an individual’s remitted foreign income today.

For a company the picture reverses in an important way. Singapore’s exemption for foreign dividends, branch profits and service income requires all three statutory conditions to be met. One of them is that the source jurisdiction’s highest corporate rate is at least 15%. That test looks at the headline rate rather than at tax actually suffered, and the Minister may waive the subject-to-tax condition but expressly not the rate test. Thailand’s 20% headline rate clears it comfortably. A Thai company, by contrast, simply reports worldwide income at 20% and claims a credit, with nothing to establish.

The rule is defensible, and the way it was made is not

The substantive change closed a genuine gap. Under the 1987 reading a Thai resident could earn income anywhere and in any amount. Bringing it in on the first day of the following year made it free of tax. Describing that as a remittance basis flattered it. Thailand also declined to abolish the remittance basis altogether, having drafted the power to do it. Money left offshore remains untaxed, and so do unrealised gains. That is more generous than the United States and most of the OECD.

The grandfathering is clean, and the capital and income distinction is recognised in writing by the Revenue Department itself. The residence test is objective and cheap to apply. Several remittance systems impose ordering rules on mixed funds that force income out first. Thailand lets the taxpayer characterise the pool on the facts, which is the more generous approach.

The execution is the problem. A reading that had stood for thirty-six years was overturned by an internal instruction to revenue officers rather than by legislation, with 107 days of notice. The documentation standard is undefined and the burden falls entirely on the taxpayer. One basic question has gone unanswered for nearly three years. Whether spending on a foreign card in Thailand, or drawing on an offshore account from a Thai cash machine, counts as a remittance remains unsettled. The Revenue Department expressly declined to confirm it. Thailand has adopted the common reporting standard, so it has independent data on offshore accounts while the taxpayer has no answer on the basic question.

Two contradictory drafts have also been live at once for more than a year. One would relax the rule and one would abolish the remittance basis entirely. Neither has been enacted, so the policy direction is genuinely unsettled. The May 2025 announcement was read by a great many people as a change that had already happened.

What this means for new arrivals, founders planning an exit, and advisers

For a person moving to Thailand, the work that matters is evidential and it is best done before the move. A statement of every foreign account as at 31 December 2023 establishes the pre-2024 layer. Records of acquisition cost, sale price and gain for each holding establish the capital layer. Without those, a remittance is treated as income in full. On an illustrative 6 million baht transfer, the difference is about 1.5 million baht of tax on identical money.

For a founder planning an exit, the question is the completion date rather than the structure. A gain realised in a year of non-residence is outside the charge permanently, whenever it is remitted. A gain realised after the 180th day is assessable when the money comes in, with no credit for Singapore tax that was never charged. The Long-Term Resident visa is a statutory exemption rather than a concession, and it is worth examining before anybody assumes the ordinary rules apply.

For an adviser, three warnings are worth passing on. The announced relaxation is not law and should not be relied on in any written advice. The Revenue Department’s English-language rate page is out of date, and so is the treaty text it hosts. The operative article is in the 2015 agreement rather than the 1975 one. This subject also attracts an unusual volume of confident, undated and incorrect material online, much of it with no identifiable author. A figure that cannot be traced to the Revenue Department or the gazette should not be used.

The UAE taxes most companies at nine per cent and large groups at fifteen. Its small business exemption now runs to 2029, and almost every published summary still gives the old date.

The United Arab Emirates introduced corporate tax for financial years beginning on or after 1 June 2023. For a company on a calendar year, the first taxed period was therefore the year ended 31 December 2024 rather than 2023. The rate is nil on the first AED 375,000 of taxable income and nine per cent above it. That is a profit threshold rather than a revenue one.

Two things about the regime are routinely misstated, and both move the answer. The first is that the UAE is tax free, which it has not been for more than two years. The second is that free zone companies are exempt, which is not what the law says either. Free zone treatment is a conditional nil rate on defined categories of income, and it can be lost for five tax periods at a time.

There is also a change that most published summaries have not caught up with. Small Business Relief gives full relief where revenue does not exceed AED 3 million. Its expiry was extended from 31 December 2026 to 31 December 2029 by a ministerial decision issued on 29 July 2026. A great deal of advice written before August 2026 assumes the relief is about to lapse. It is not.

This article sets out what the UAE regime actually charges and where it stops being nine per cent. It then compares the two jurisdictions on each point that matters. Where the UAE is better, that is said plainly, because a comparison written any other way is not worth reading.

Nine per cent is the headline, and the effective rate is lower

The nil band below AED 375,000 pulls the effective rate well under the headline. It does so across the range that most owner-managed businesses occupy. A comparison of headline rates therefore overstates the gap at the small end and understates it at the large end. Both comparisons are set out below, because the headline figure is what gets quoted and the effective figure is what gets paid. Neither is the whole answer on its own.

The headline comparison puts the UAE at nine per cent against Singapore at seventeen. Hong Kong charges 8.25 per cent on the first HKD 2 million of profit and 16.5 per cent above it. Only one company in a connected group may claim the lower tier. Ireland charges 12.5 per cent on trading income and twenty-five on non-trading income. The United Kingdom charges twenty-five per cent above GBP 250,000.

On the UAE rules, a company with AED 1 million of profit pays an effective 5.63 per cent. At AED 3 million that is 7.88 per cent, at AED 10 million 8.66 per cent, and at AED 50 million 8.93 per cent. On the Singapore rules for the 2026 year of assessment, a company with SGD 1 million of profit pays an effective 15.26 per cent before any rebate. At SGD 10 million that is 16.83 per cent.

Singapore’s reliefs behave differently from the UAE’s, and the difference is worth understanding. Partial tax exemption exempts seventy-five per cent of the first SGD 10,000 of chargeable income and half of the next SGD 190,000. The maximum exempt amount is SGD 102,500, which saves SGD 17,425 of tax. That is a fixed cash amount, and it does not grow as profits grow. The start-up exemption is more generous at SGD 21,250, and it lasts for three consecutive years of assessment only.

The start-up exemption also carries conditions that exclude the structures most often proposed. The company must be incorporated in Singapore and tax resident for the year of assessment. It must have no more than twenty shareholders holding directly, all of them individuals, or one individual holding at least a tenth. Investment holding companies and property developers are excluded outright. A one-year rebate also applies for the 2026 year of assessment, and one-year measures are best left out of a structural comparison.

Above a certain group size both charge fifteen per cent

The comparison changes entirely at the top end, and this is the point to establish before any structuring work begins. Both jurisdictions have adopted the international minimum tax, and both have set it at the same rate and the same threshold. Above that threshold the question of which has the lower corporate rate stops having an answer. A great deal of advice still treats the nine per cent as though it applied at every size, and it does not.

The UAE enacted a domestic minimum top-up tax by a cabinet decision issued on 31 December 2024 and effective from 1 January 2025. It charges fifteen per cent on groups with consolidated revenue of EUR 750 million or more in two of the four preceding years. The nine per cent and the free zone nil rate are both computed first, and the liability is then topped up to fifteen.

Singapore’s equivalent arrived at the same time and on the same terms. The Multinational Enterprise (Minimum Tax) Act 2024 introduced both an inclusion rule and a domestic top-up tax at a fifteen per cent minimum effective rate. The threshold is EUR 750 million, and it applies to financial years beginning on or after 1 January 2025. Same rate, same threshold, same currency, same start date.

The consequence is that the UAE’s rate advantage disappears completely above that size. A group above EUR 750 million pays fifteen per cent in either place. Both regimes allow a substance-based carve-out of five per cent of payroll and five per cent of tangible assets, and both have a de minimis election. For a group of that scale the corporate rate is no longer a reason to choose between the two.

Free zone treatment is a conditional nil rate, not an exemption

A qualifying free zone person pays nil on qualifying income and nine per cent on everything else. Four conditions have to be met. The company must maintain adequate substance in the UAE, derive qualifying income, not have elected into the ordinary regime, and comply with the transfer pricing rules. Compliance with transfer pricing is itself a condition of the nil rate, which is unusual and easy to overlook.

The qualifying activities list was replaced in 2025 and the replacement was backdated. A ministerial decision issued on 28 August 2025 repealed the 2023 list and applied retroactively to 1 June 2023. The current list covers manufacturing, processing and trading in qualifying commodities. It covers holding shares and securities for investment, ship operation and reinsurance. It covers fund management, wealth and investment management, and headquarter and treasury services to related parties. It also covers aircraft financing and leasing, distribution from a designated zone, and logistics.

The excluded activities matter as much. Transactions with natural persons are excluded, other than ship operation, fund management, wealth management and aircraft financing. Banking is excluded. Insurance is excluded, with reinsurance and headquarter services preserved. General finance and leasing is excluded, with the specific carve-outs preserved. Ownership of immovable property other than commercial property inside a free zone is excluded.

The de minimis rule has two tests and the binding one is often the second. Non-qualifying revenue must not exceed five per cent of total revenue or AED 5 million, whichever is lower. Above AED 100 million of total revenue the absolute cap governs. Breach of the de minimis rule removes the nil rate from the start of that period and for the following four tax periods. A single bad year therefore costs five.

Substance requirements for free zone entities have tightened rather than relaxed. The UAE cancelled its separate economic substance reporting regime for financial years ending after 31 December 2022. The Ministry of Finance announced that on 14 October 2024. Substance then moved into the corporate tax law and became more demanding. Core income-generating activities must be performed in the zone, with adequate assets, adequate full-time qualified employees and adequate operating expenditure for each activity. Outsourcing is permitted only within the zone and with adequate supervision.

Two newer requirements add real cost. Every qualifying free zone person needs audited financial statements regardless of size, for tax periods from 1 January 2025. A second requirement starts with tax periods beginning 1 January 2026. A free zone distributor relying on the designated zone activity must file an independent auditor’s agreed-upon-procedures report, using a prescribed sampling method. It falls due within thirty days of the return deadline. Failure to file it means the conditions are not met.

Outbound payments are where the UAE wins clearly

The strongest point in the UAE’s favour is the one least often made, and it has nothing to do with the corporate rate. It concerns what happens when money leaves the country. A corporate rate is paid once on profit, while a withholding tax is paid on every payment that crosses the border. For a structure that pays regular service, management or interest fees out, the withholding position is the larger annual number. This is the comparison that decides where a regional services company should sit.

The UAE charges nil withholding tax on a non-resident’s income from the country where it is not attributable to a permanent establishment. The nil rate is statutory rather than administrative, and there is no return and no registration. One qualification is worth noting for the long term. The provision allows the Cabinet to specify another rate by decision, so this is a withholding regime set at nil rather than the absence of one. No such decision has been identified as at October 2026.

Singapore’s position is quite different, and the figure that bites is not the one usually quoted. Technical assistance, service fees and management fees paid to a non-resident are charged at the prevailing corporate rate of seventeen per cent on a gross basis. Interest and loan-related payments are charged at fifteen per cent, rent for movable property at fifteen, and royalties at ten. Dividends carry no withholding under the one-tier system. For a regional services or holding structure, that seventeen per cent on service and management fees is what actually costs money.

There is a further asymmetry that runs against the UAE, and it matters to anybody placing a UAE subsidiary under a Singapore parent. Singapore’s foreign-sourced income exemption requires the foreign headline tax rate to be at least fifteen per cent. The UAE’s headline rate is nine. UAE-sourced dividends, branch profits and service income received in Singapore therefore fail that test, and a Singapore parent cannot use the exemption on them.

The treaty networks are counted on different bases

The UAE’s Ministry of Finance states that 137 double taxation agreements have been concluded, on a page last updated on 18 October 2026. It does not publish the split between signed and in force. Singapore reports 96 comprehensive agreements signed and 93 in force as at 31 March 2026, in its revenue authority’s annual report. Those two counts are on different bases, and printing them side by side without saying so is misleading.

What can be said is that the UAE’s network reaches considerably further into Africa, Central Asia, the former Soviet states and the Middle East. Singapore’s network is deeper and more tested across Asia. For the four markets that matter most to a Southeast Asian group, there is no gap at all. Both the UAE and Singapore have agreements with India, Indonesia, Vietnam and China, and any claim otherwise is wrong.

One real asymmetry runs the other way. The Germany and UAE agreement expired on 31 December 2021 and has not been replaced. The German federal trade agency stated in April 2024 that there were no current efforts to conclude a new one. German unilateral relief does not help either, because the UAE does not tax the income. Singapore’s agreement with Germany remains in force.

The compliance load is real in both places

The UAE requires a master file and a local file where a company’s own revenue reaches AED 200 million. The same applies where consolidated group revenue reaches AED 3.15 billion. Documentation is due within thirty days of a request. The return itself requires separate disclosure of related party transactions above AED 40 million in aggregate and AED 4 million by category. Connected person disclosure starts at AED 500,000. Those return thresholds are frequently confused with the documentation thresholds, and they are much lower.

Returns are due nine months after the end of the tax period. Registration deadlines ran on a licence-month ladder through 2024 for companies existing before March 2024, and run three months from incorporation for companies formed after it. Penalties are specific. Failure to register costs AED 10,000. A late return costs AED 500 a month for twelve months and AED 1,000 a month after that. Late payment carries fourteen per cent a year charged monthly. Failure to keep records costs AED 10,000, doubling on repetition within two years.

Singapore’s load is differently shaped. Transfer pricing documentation is required above a revenue threshold, with separate per-transaction thresholds. The threshold for goods and loans rose to SGD 2 million from the 2026 year of assessment. A five per cent surcharge applies to any transfer pricing adjustment even where no additional tax results. Audit is required only where a company exceeds two of three size thresholds. A small Singapore company may therefore be audit exempt, where every qualifying free zone person must be audited.

Point of comparison	Which is better	The honest qualification
Rate for owner-managed business	UAE, by roughly half	Each in its own currency
Small business relief	UAE, a full exemption to 2029	Revenue cliff edge, losses forfeited
Outbound service and management fees	UAE, nil against seventeen per cent	Cabinet may set a rate by decision
Founder taxation	UAE, no personal income or capital gains tax	Often decisive, and rarely the headline
Treaty breadth	UAE, 137 against 96	Different bases, no Germany agreement
Groups above EUR 750 million	Neither, both fifteen per cent	Same rate, threshold and start date
Domestic dividends	Neither, both effectively nil	Different machinery, same answer
Regime stability	Singapore	Five UAE instruments replaced since 2023
Audit for a small company	Singapore, exemption available	Every free zone person is audited
Bringing profits to a parent	Singapore, except from the UAE	Nine per cent fails the fifteen per cent test

Where the UAE is genuinely better

Five advantages survive scrutiny. The nil withholding on outbound payments is the strongest and the least disputable. Seventeen per cent on gross service and management fees is a cost a regional structure pays every year. The rate for an owner-managed business is roughly half Singapore’s across the middle of the range. Small Business Relief is a full exemption rather than a reduction. It covers AED 3 million of revenue with no profit cap, and it now runs to 2029.

There is no separate economic substance filing regime any more, which removes a recurring compliance exercise from the calendar. The substance itself is still required, and the corporate tax law now sets a higher bar than the old regulations did. What has gone is the annual notification and report rather than the obligation. For a founder, there is also no federal personal income tax and no individual capital gains tax. That is frequently more decisive than anything about the corporate rate.

A genuine nil rate with a statutory fifty year horizon also exists for a defined set of activities, and Singapore has no as-of-right equivalent. Manufacturing, processing, commodity trading, headquarter and treasury services, reinsurance, fund and wealth management, aircraft financing, logistics and designated zone distribution all qualify. For a business that actually does one of those things in a free zone and can carry the substance, that is a real and durable position.

Where Singapore holds the advantage

The honest case for Singapore here is not the tax rate, which is higher. It is predictability. Five instruments have been replaced in three years. The UAE has replaced its qualifying income decision, and repealed and retroactively replaced its qualifying activities decision. It has repealed and replaced its audit decision, and cancelled its economic substance filing regime. It has added a top-up tax and a distribution audit requirement, and extended Small Business Relief by three years. A structure built on the 2023 rules has had to be reconsidered more than once.

The five period cliff is a second point, and it is the one that should be modelled rather than noted. A single breach of the de minimis rule costs the nil rate for that year and the four that follow. Above AED 100 million of revenue the binding test is the absolute AED 5 million cap rather than the five per cent proportion. A growing company therefore tightens the constraint on itself as it grows. That is a severe consequence for what may be an administrative failure. There is no equivalent in Singapore.

The third is what can be done with the profit afterwards. A UAE subsidiary’s profits cannot be brought into a Singapore parent under the foreign-sourced income exemption. The nine per cent headline rate fails the fifteen per cent test. For a group that wants to pool regional profit in Singapore, that is a structural problem rather than a detail.

Both jurisdictions have also converged on the same answer to the same question. Capital gains are exempt in both subject to substance, under the participation exemption in the UAE and under section 10L of the Income Tax Act in Singapore. Neither offers a version of itself without substance. Anybody selling an empty company in either place is selling something the law has already addressed.

What this means for founders, groups and advisers

For a founder choosing where to put an operating business, the decision should turn on where the work happens and where the money goes afterwards. Where the business performs a qualifying activity in a free zone with real people and real premises, the UAE nil rate is durable. It should be taken seriously. If the business will pay significant service, management or interest fees out of the country, the UAE’s nil withholding is worth more than the corporate rate difference. If profits need to pool in Singapore, the nine per cent headline rate creates a problem that structuring does not solve.

For a group approaching EUR 750 million of consolidated revenue, the corporate rate comparison should be set aside. Both jurisdictions charge fifteen per cent above that threshold from financial years beginning on or after 1 January 2025. Planning should concentrate on the substance-based carve-out, and on where real payroll and tangible assets actually sit. That is the only part of the calculation still moving.

For an adviser, two practical points are worth carrying into the next conversation. Any UAE analysis written before August 2026 should be rechecked. The Small Business Relief expiry and the qualifying activities list have both changed, and the activities change was retroactive. And the effective rate figures in this article are calculations from the published rules rather than published figures, computed in each jurisdiction’s own currency. They compare rate structures and not equal amounts of money, and they should be presented to a client that way.

Since April 2021 a foreign national on a work visa has their foreign assets outside Japanese inheritance tax, however long they have lived there. Taking permanent residence forfeits that.

Japan’s inheritance tax reaches a top rate of fifty-five per cent, which is the highest in the developed world. Most of what is written about escaping it is wrong, and some of it is expensively wrong. The single most useful fact is also the least known. A foreign national living in Japan on a work visa has had their foreign assets outside Japanese inheritance and gift tax since 1 April 2021. Length of stay makes no difference to that.

That relief is not a loophole. It is a deliberate policy choice, stated as such, to make Japan workable for skilled foreign professionals. Before April 2021 the same person was taxed on worldwide assets once they had been in Japan for more than ten of the previous fifteen years. The condition was repealed on the side of the person who dies or gives. It remains in place on the side of the person who receives.

The relief attaches to the visa category and not to the person. Japan’s immigration law divides residence statuses into two appended tables, and the relief covers table one only. Every work-related status sits in table one, including engineer, specialist in humanities and international services, intra-company transferee, business manager, highly skilled professional, professor and student. Permanent residence, spouse of a Japanese national, and long-term resident sit in table two, and none of them has ever qualified.

That makes permanent residence an expensive thing to apply for. Consider a professional who has lived in Japan for a decade on a work visa and then takes permanent residence. That step moves their worldwide assets into the Japanese inheritance tax base. It starts the clock on the exit tax at the same time. Almost nobody is told either thing before they file the application.

The tax is charged on the recipient, on a notional share

Japanese inheritance tax works differently from the British and American systems, and the difference changes the arithmetic. It is a tax on the person who receives rather than on the estate. The rate bands are applied to each heir’s hypothetical statutory share of the estate after the basic deduction. They are not applied to the whole estate, nor to what any heir actually receives.

The basic deduction is thirty million yen plus six million yen for each statutory heir. Heirs who renounce are still counted, and adopted children are capped at one where there are biological children and two where there are none. A family of a spouse and two children therefore has a deduction of forty-eight million yen before any tax arises.

That deduction is the real complaint about Japanese inheritance tax, and it is a domestic complaint rather than a cross-border one. It was cut in the 2013 reform. Before that reform the deduction stood at fifty million yen plus ten million yen for each heir. The effect of the cut is visible in the published filing statistics, and it is larger than anything the cross-border rules have done.

In 2014 about 4.4 per cent of Japanese deaths produced an inheritance tax liability. In 2015, the first year under the reduced deduction, that doubled to about eight per cent. It has risen in almost every year since, and 2024 was the first year above ten per cent, at 10.4 per cent of 1,605,378 deaths. The taxable value assessed in 2024 was 23.4 trillion yen and the tax 3.2 trillion yen.

Against other systems, Japan is at the top and not by a small margin. South Korea reaches fifty per cent above three billion won. The United States reaches forty per cent, with an exclusion of USD 15 million for deaths in 2026. The United Kingdom reaches forty per cent above a nil rate band of GBP 325,000. A further residence band of GBP 175,000 tapers away above GBP 2 million. Singapore abolished estate duty entirely for deaths on or after 15 February 2008.

Foreign assets leave the net only when both parties have been gone ten years

The rule that defeats most planning is a double test, and it is routinely described as a single one. Take two Japanese nationals. Foreign assets fall out of the Japanese inheritance tax base only when both of them have been without Japanese domicile for more than ten years. That applies to the person who dies and to the person who receives.

The ten year lookback itself is often quoted as five years. It was five until the 2017 reform, which extended it to ten with effect from 1 April 2017. Stale summaries are still in wide circulation. The practical meaning is that one generation leaving is not enough. The next generation has to move as well, and stay moved for over a decade, with the death falling after both clocks have run.

A second rule is also commonly misstated. One measure once allowed a donor who had left Japan to give foreign assets free of Japanese gift tax after a short period. It was repealed on 1 April 2021. The advice to leave Japan, wait two years and then give is no longer correct, and following it now produces a Japanese gift tax charge.

Gift tax is harsher than inheritance tax at the same amount, which is why giving things away before leaving usually backfires. The annual exemption is 1.1 million yen. The top rate of fifty-five per cent is reached at thirty million yen on the general table. Where the gift comes from a parent or grandparent, it is reached at forty-five million yen. The scope rules mirror the inheritance tax rules, with the same table one definition and the same clocks.

One further change is worth knowing because the transition is keyed to the date of death rather than the date of the gift. Gifts made within three years of death are added back for deaths up to 31 December 2026. For deaths between 2027 and 2030, gifts from 1 January 2024 onwards are added back. The full seven year add-back applies only to deaths from 1 January 2031.

The exit tax charges unrealised gains at 15.315 per cent

Japan taxes unrealised gains on departure, and the measure has been in force for departures on or after 1 July 2015. It applies where a person holds at least one hundred million yen of assets in scope. They must also have had a domicile or residence in Japan for more than five years in aggregate within the preceding ten.

The threshold is aggregated across the whole holding, including loss positions and tax-exempt government bonds. Unsettled derivative and margin positions enter at their deemed settlement gain or loss. The assets in scope are securities, which covers listed and unlisted shares, investment trusts and bonds, together with silent partnership interests and unsettled trades. The rate is 15.315 per cent, being fifteen per cent with the reconstruction surtax, and local inhabitant tax does not apply.

Here the visa rule helps again. Periods spent on a table one status do not count towards the five years. A person on a work visa is therefore outside the exit tax however long they have been in Japan. A person who has taken permanent residence is inside it. Transitional rules mean the earliest such a person could have crossed the five year line was 1 July 2020.

Deferral is available for five years and can be extended to ten. It requires a tax agent appointed before departure, a claim made on the return, security equal to the deferred tax and interest, and an annual continuation notice. The charge is cancelled in full if the person returns within the period still holding the asset. Deferral is therefore real relief, and it is conditional on paperwork that has to be in place before anybody boards a plane.

The exit tax also has two limbs that bite while the owner is still in Japan. A gift of in-scope assets to a non-resident triggers it. So does an inheritance passing to a non-resident. A family that moves one child abroad and leaves the parents in Japan has created an exit tax event on the parents’ death, not avoided one.

A Singapore holding company makes the position worse

Shares in a Singapore company are foreign-situs property under Japanese law, because shares are situated where the issuing company has its head office. That fact leads a great many people to the wrong conclusion, because situs only matters to a person taxed on Japanese assets alone.

Where Japanese inheritance tax reaches a person’s worldwide assets, a holding company removes nothing. The shares themselves are an asset. A portfolio worth twenty million dollars moved into a Singapore company becomes shares in a Singapore company worth about twenty million dollars. It is taxable at that value. Interposing an entity changes the description of the asset. It changes neither the value nor the taxpayer’s status, and those are the only two things that matter.

The move	Effect on Japanese inheritance tax	Effect on the exit tax
Singapore bank account	None	None
Singapore holding company	None, and adds annual Japanese income tax	None
Singapore trust	None, and may be deemed a gift on creation	None
Singapore address while living in Japan	None	None
Holding a table one work visa	Foreign assets out of scope since 2021	Period does not count towards five years
Taking permanent residence	Worldwide assets into scope	Starts the five year clock
Both parties without Japanese domicile ten years	Foreign assets out of scope	Departure itself is the charging event

The holding company also creates a new annual cost. Japan’s controlled foreign company rules reach resident individuals and not only companies. A ten per cent holder is within charge where Japanese ownership exceeds half the company. A pure holding company fails the business content test outright, so it is treated as a paper company. The exemption threshold for a paper company is an effective rate of twenty-seven per cent, and the general threshold is twenty per cent. Singapore’s headline rate of seventeen per cent is below both.

The result is the worst of both outcomes. The owner pays Japanese income tax on the company’s income every year, and the shares remain fully inside the inheritance tax base. Disclosure obligations follow as well. A Japanese resident holding more than fifty million yen of foreign assets at the end of the year must file a foreign asset report. It falls due by the end of June.

Trusts do not solve it either, and the legislation addresses them directly. A person who becomes a beneficiary without adequate consideration is deemed to have acquired the trust rights from the settlor by gift or bequest. The provision also deems them to have acquired the assets and liabilities in the trust. A separate provision deems the trustee to acquire where there is no beneficiary and a prospective beneficiary is a relative of the settlor. That one is aimed squarely at discretionary structures.

Changing residence is the only lever, and it carries its own bill

Residence is the only thing that changes the scope rules, and it changes them on statutory terms rather than on paperwork. Japanese law treats domicile as a question of fact, turning on where the centre of a person’s life actually is. Deregistering an address does not move it. A Singapore address maintained while the family, the work and the home remain in Japan changes nothing.

The arithmetic also needs doing in the right order. The exit tax is certain, immediate and charged at 15.315 per cent on unrealised gains. The inheritance tax is uncertain and more than ten years away. For a holding with large accrued gains, the exit charge can exceed the eventual inheritance tax. The basic deduction and the spouse’s relief are what close the gap. That calculation should be done before anybody commits to moving.

Two conclusions follow for different readers. A table one visa holder already has the relief on the side of the person who dies, and does not need to leave Japan to get it. They should establish whether the intended recipient is independently taxable on worldwide assets, because that is where the exposure actually sits. A Japanese national with Japanese children has no cross-border planning available at all, and should be told so plainly rather than sold a structure.

Japan’s own direction of travel is domestic

The annual tax reform acts are the place to look for what is coming, and the two most recent ones say something useful by omission. The 2026 reform was enacted on 31 March 2026. It contains nothing on the exit tax, nothing on inheritance or gift tax scope, and nothing on foreign trusts or holding companies. The 2025 reform contained nothing on those subjects either.

What the 2026 reform did address was domestic valuation. Rental real estate acquired or built within five years before death will be valued at normal transaction value rather than at the lower assessed value. That applies to events on or after 1 January 2027. Fractional real estate products are treated the same way. A minimum tax on high incomes was also tightened, with the deduction halved and the rate raised from the 2027 tax year.

The pattern is consistent across both years. Japan is closing domestic valuation gaps and tightening the income side. It has left the cross-border scope rules alone since 2021. A planner should therefore assume that the scope rules are stable and that the domestic valuation rules are not. The real estate valuation change is the one that will alter numbers already in existing plans, and it takes effect in just over a year.

Singapore taxes less here, and it is not nothing

Singapore abolished estate duty for deaths on or after 15 February 2008. The abolishing Act was passed that year, and the original Act of 1929 remains on the books for earlier deaths. There is no inheritance tax, no gift tax and no estate duty in Singapore today. That position has held for eighteen years, which is worth saying because it is the kind of exemption other jurisdictions have introduced and then withdrawn.

What Singapore does tax should be said, so that nobody reads this as a claim that Singapore taxes nothing. Individual income tax reaches twenty-four per cent above SGD 1 million. Corporate income tax, goods and services tax at nine per cent, property tax, stamp duty and withholding tax all apply. Stamp duty on residential property includes an additional buyer’s duty, and there is a separate conveyance duty on the transfer of property-holding entities.

One Singapore rule now works against the substanceless holding company specifically. Section 10L of the Income Tax Act applies to disposals of foreign assets on or after 1 January 2024. It taxes gains received in Singapore by an entity in a relevant group where that entity lacks adequate economic substance in Singapore. The test is applied at entity level, and a pure equity-holding entity meets it only with Singapore management, people and premises. An empty Singapore company is therefore worse in Singapore terms than it was before 2024, quite apart from what Japan does to it.

What this means for professionals, families and advisers

For a foreign professional in Japan, the position should be established before any structure is discussed, because the answer is often better than expected. A table one visa holder’s foreign assets are outside Japanese inheritance and gift tax on the side of the person who dies. Their time in Japan also does not count towards the exit tax. The question to settle is the recipient’s status, and the question to think hard about is whether permanent residence is worth what it costs.

For a Japanese family considering a move, the two clocks are the whole of the planning. Both the person who dies and the person who receives need more than ten years without Japanese domicile. The death has to fall after both periods have run. That is a decision about where a family lives for a generation rather than a tax structure. The exit tax should be quantified first, because it is usually the largest number in the plan.

For an adviser, the defensible work is determining and documenting status under the four statutory definitions and planning around the clocks. The indefensible work is interposing an entity, because the controlled foreign company rules make a Singapore holding company a negative-value transaction for a Japanese resident. A structure that adds annual tax and removes nothing from the base is not a plan. A client who has been sold one should be told what it is costing.

The Supreme Court struck down the emergency tariffs in February 2026. What replaced them leaves barely two and a half points between Southeast Asian exporters.

The tariff regime that most trade planning still assumes no longer exists. On 20 February 2026 the Supreme Court held that the International Emergency Economic Powers Act does not authorise the President to impose tariffs. The vote was six to three. The decision came in the consolidated appeals heard as Learning Resources against Trump, with the Chief Justice writing for the majority. Every tariff imposed under that Act was terminated by executive order five days later. Collection stopped for goods entering on or after 24 February 2026.

About USD 166 billion had been collected under those orders. Roughly USD 126 billion of it had been certified as refunded with interest by the start of October 2026. The Court itself did not order refunds, and the refund process came out of separate litigation in the Court of International Trade. A class covering importers of record who paid between February 2025 and February 2026 was certified on 8 October 2026, and the government’s appeal is pending.

What replaced the emergency tariffs is narrower and differently built. The country layer now rests on a forced labour determination made under Section 301 of the Trade Act. It was published on 28 July 2026 and took effect from 24 July 2026. It sets twelve and a half per cent for most Asian economies and ten per cent for Indonesia and Malaysia. The practical consequence is a change of kind rather than degree. Where a country once carried a rate, a product now does.

The spread between Southeast Asian exporters has collapsed

Under the emergency orders the gap between Asian exporters ran from ten per cent to fifty per cent. An exporter could change its landed cost materially by changing where goods were made. That was the arithmetic behind the relocation of assembly work out of China across 2025. The gap is now two and a half points. The same decision, taken on the same grounds today, would be worth a fraction of what it was worth eighteen months ago.

Singapore, Vietnam, Thailand, China and Hong Kong all sit at twelve and a half per cent. Indonesia and Malaysia sit at ten. For a business that moved production to Vietnam on tariff grounds between 2025 and early 2026, the arithmetic that justified the move has largely disappeared. The remaining large differentials are attached to goods rather than to places.

Those product duties are where the money now is. Steel, aluminium and copper carry fifty and twenty-five per cent under Section 232 proclamations effective from 6 April 2026. Those rates now apply to the full customs value. Patented pharmaceuticals carry up to one hundred per cent. Automobiles and furniture carry twenty-five per cent, lumber ten. A narrow semiconductor duty of twenty-five per cent took effect on 15 January 2026. China’s separate Section 301 stack from the 2024 four-year review survives at rates between seven and a half and one hundred per cent by product.

Two different numbers circulate as the American tariff rate, and mixing them is the most common error in the field. The collections measure divides duty actually received by the value of goods imported. It stood at 2.4 per cent in 2024 and rose to 7.7 per cent across 2025, the highest since 1947. In August 2026 it sat at 7.1 per cent. The statutory average, which weights published rates against trade, was about eleven per cent in August 2026. The gap between the two is exclusions, preferences and goods that do not arrive.

Singapore faces a duty the United States did not impose over trade imbalance

Singapore’s position is unusual and worth stating precisely. The United States runs a surplus with Singapore, and has done every year from 2019 to 2025 except 2020. In 2025 that surplus was USD 3.4 billion on goods and USD 29.5 billion on services. There is no deficit grievance to answer. The twelve and a half per cent comes from a forced labour finding, not from an imbalance.

The free trade agreement does not solve it either. The agreement reduces the most favoured nation component of the duty, and Section 301 duties sit outside that. Singapore has no country-specific exclusion list. The Ministry of Trade and Industry stated on 24 July 2026 that about one third of Singapore’s domestic exports to the United States are affected.

A larger exposure is still forming. On 11 March 2026 the United States initiated a Section 301 investigation into excess capacity covering sixteen economies, Singapore among them. The named sectors are chemicals, semiconductors, electronics and machinery. Any duty arising from it could stack on the existing twelve and a half per cent. For an exporter in those sectors, that pending investigation matters more than the duty already in force.

The import data also contradicts a pitch that was common through 2025. American imports from Vietnam rose from USD 66.4 billion in 2019 to USD 193.9 billion in 2025. Imports from Thailand rose from USD 33.4 billion to USD 91.3 billion. Imports from Singapore rose from USD 26.4 billion to USD 38.2 billion, which is below both 2023 and 2024. Singapore is not absorbing diverted Chinese volume the way its neighbours are.

Origin is decided by substantial transformation, and routing is not origin

The rule that matters most to anybody considering a Singapore entity is the origin rule, and it is unforgiving. Non-preferential origin turns on substantial transformation. Work done in a country must give the goods a new name, character or use. Simple assembly does not qualify, and the regulations treat an operation involving five or fewer foreign parts as simple assembly. Preferential rules under a free trade agreement do not decide whether a Section 301 duty applies.

The forty per cent transshipment duty that featured in so much 2025 commentary came from an emergency order of 31 July 2025. It applied in place of the emergency rate, and that rate has now been struck down and terminated. No transshipment duty appears in the Section 301 forced labour notice or in any Section 232 proclamation. Its current status is therefore unclear rather than settled, and no exporter should plan on either answer.

The exposure is far larger than the duty, and this is the figure that should end the conversation. On a USD 10 million shipment the duty at twelve and a half per cent is USD 1.25 million. The negligence ceiling is twice the duty avoided. Gross negligence is four times. Fraud is the full domestic value of the merchandise, which is USD 10 million. A false origin declaration risks eight times the amount it was meant to save, and that is before any criminal exposure.

A middleman sale can lower the value, and it does not touch origin

There is one legitimate structure in this area, and it is narrow. The first sale rule allows the dutiable value to be the price in an earlier sale in the chain rather than the price to the American importer. A Singapore trading company may buy from an Asian manufacturer and sell to an American buyer. In the right circumstances the manufacturer’s price becomes the customs value.

The conditions are strict and the documentation burden is real. There must be a bona fide sale at each step. The goods must be clearly destined for the United States at the time the middleman contracts. Each sale must be at arm’s length. The supporting documentation requirements were set out by customs in a decision published on 2 January 1997 and have not become easier since.

What this does is reduce the base on which duty is calculated. It does not change the rate, and it does not change the origin. Take a product carrying twelve and a half per cent. A first sale structure that moves the customs value from USD 10 million to USD 8 million saves USD 250,000 of duty. That is a real saving, obtainable lawfully, and it is an order of magnitude smaller than what the tariff-avoidance pitch implies.

The profit side is a separate question entirely, and it is governed by transfer pricing rather than by customs. Singapore requires transfer pricing documentation above a revenue threshold. It also applies a five per cent surcharge on any adjustment, even where no additional tax results. The per-transaction documentation threshold for goods and loans rose to SGD 2 million from the 2026 year of assessment. A margin left in a Singapore company that does not perform the functions earning it will be adjusted.

The remittance tax is one per cent, and it applies to cash

A separate measure is widely misdescribed, and the correct version is much narrower. A one per cent excise applies to certain outbound money transfers under a provision added by the budget reconciliation act signed on 4 July 2025. It took effect for transfers after 31 December 2025. Proposed regulations were published on 13 April 2026, and final regulations have not been confirmed as published.

The scope is where the misreporting sits. The tax applies to cash, money orders, cashier’s cheques and traveller’s cheques. It does not apply to transfers from a bank account. It does not apply to debit or credit cards, regardless of where the card was issued. It does not apply to cheques, automated clearing house transfers or prepaid cards. Transfers of fifteen dollars or less are excluded.

Two further points matter for a business. Citizenship and immigration status are irrelevant to the charge, despite the versions reported during the bill’s passage. The sender must be a consumer transferring money for personal, family or household purposes, which places commercial transfers outside the measure altogether. The rates of five per cent and three and a half per cent that circulated widely were House versions. The enacted law carries neither of them. It also dropped the citizenship test that accompanied one.

A separate proposal would raise the rate to twenty-five per cent and extend it to all non-citizens. It was introduced on 21 May 2026 and has not been enacted. Section 899, the retaliatory tax measure that caused considerable alarm in mid-2025, was never enacted either. It was removed from the Senate bill before the vote of 29 June 2025, following an understanding reached with the Group of Seven. The act signed on 4 July 2025 does not contain it.

Where a Singapore entity does not help at all

The tariff follows the goods. A Singapore company does not change where something was made, and a customs officer assessing origin does not look at the certificate of incorporation. For goods manufactured in China, the Chinese Section 301 duties apply wherever the invoice is raised. The new twelve and a half per cent sits on top of them. For steel, aluminium, copper, pharmaceuticals, automobiles, furniture and lumber, the Section 232 duties are product-based and indifferent to country.

The free trade agreement does not shield Singapore-origin goods from the new duty. On the Ministry’s own figures, one third of Singapore’s domestic exports to the United States are affected. Singapore is also one of the sixteen economies in the excess capacity investigation, so its exposure may grow rather than shrink. None of this is improved by a holding structure.

What a Singapore company does for a business selling into the United States is commercial rather than tariff-related. It provides a contracting entity with an established legal system, multi-currency banking, treaty access across Asia and a workforce for regional functions. Where the company performs real functions, the profit attributable to those functions can properly sit there. That case stands on its own merits, and it does not need a tariff claim attached to it.

What this means for exporters, importers and advisers

For an exporter, the first task is to classify by product rather than by country. The country layer is now ten or twelve and a half per cent across the region. The decisive question is whether the goods fall under a Section 232 proclamation or a China-specific list. A business that relocated production for tariff reasons in 2025 should rerun the calculation. A two and a half point spread may no longer justify the cost of the move.

For an importer of record that paid emergency tariffs between February 2025 and February 2026, there is money to recover and a process already running. About three quarters of what was collected had been certified as refunded by early October 2026. A separate claims process for finally liquidated entries opened on 6 October 2026, and a class action covering that period was certified on 8 October 2026. Any business that paid those duties should establish where its entries sit rather than wait.

For an adviser, the tariff landscape should be checked against the issuing documents rather than against commentary. A large number of automatically generated tariff tracker sites still publish rate tables that were made obsolete in February 2026, and they carry no dates. The status of the transshipment duty is genuinely unresolved, and the honest answer to a client asking about it is that nobody can state it with confidence. What can be stated with confidence is the penalty arithmetic, and it is the part of the conversation worth having first.

Natural resource exporters must place all their proceeds in a state bank for twelve months. The rule restricts the money rather than freezing it, and it has changed four times in nineteen months.

An exporter of Indonesian coal, palm oil, nickel or frozen shrimp must now bring the whole of its foreign currency proceeds into the Indonesian banking system. Those proceeds must then stay there for a year. The obligation runs from the export declaration. The money has to land by the end of the third month after the month of that declaration. Since June 2026 it has to sit at a state-owned foreign exchange bank rather than any licensed one.

The common description of this is that Indonesia has locked up exporters’ dollars. That is not quite what the rule does, and the difference is worth money. Five categories of spending reduce the amount an exporter has to leave on deposit. The funds can also be pledged as security for rupiah credit. Half the proceeds can be converted to rupiah each month. A twelve month placement carries a nil final tax rate, where a one month placement would be taxed at ten per cent.

What is genuinely difficult about the rule is not its severity. It is the pace of change. The rule was rewritten in February 2025, in February 2026, in June 2026 and again in September 2026. A treasurer cannot plan a year of working capital against a regulation that moves twice a year. This article sets out the position as at October 2026, what is permitted inside it, and the much larger change that arrived alongside it.

The rule covers four sectors and starts at USD 250,000 per declaration

The retention requirement applies to export proceeds from mining, plantations, forestry and fisheries. It also covers goods processed from those inputs. Scope is set by tariff line rather than by description. An exporter therefore has to check its own tariff codes rather than reason from the name of its sector. The current list runs to more than a thousand lines across mining, plantations and forestry alone.

The threshold is the detail most often reported wrongly. The obligation bites at USD 250,000 per export customs declaration. It does not apply per exporter or per year. A trader sending many small consignments can therefore fall outside the rule entirely. A single large shipment brings the full requirement. The voluntary scheme that once allowed smaller exporters to opt in was removed in March 2025.

Oil and gas exporters are not exempt, and the claim that they are is wrong. They retain thirty per cent for three months rather than everything for twelve. The relevant tariff lines are listed in the same decree. From September 2026 a group of qualifying mining exporters moved to the same thirty per cent for three months. That change is dealt with further below.

Five uses reduce what has to be left on deposit

The regulation sets out what the retained proceeds may be spent on. Amounts spent that way reduce the placement obligation rather than breaching it. The permitted uses are government and tax obligations, dividends paid in foreign currency, and procurement of materials, capital goods and services in foreign currency. The fourth is repayment of foreign currency loans. The fifth is conversion to rupiah, up to half the export value each month.

The procurement limb is wider than most summaries allow. Bank Indonesia has confirmed in its published guidance that this spending is not confined to goods unavailable domestically. The loan repayment limb covers working capital borrowing and not only capital goods finance. An exporter with foreign currency input costs and foreign currency debt may therefore find that very little is actually left idle.

One published breakdown shows how exporters behaved under the rule. USD 22.9 billion entered special accounts in March and April 2025. Exporters applied USD 14.4 billion of that to permitted uses and retained USD 7.6 billion in foreign currency. Of the amount applied, USD 12 billion was converted into rupiah. Only USD 194 million reached foreign currency term deposits at the central bank.

Proceeds can secure a rupiah loan but not a dollar one

The collateral rules are specific, and the distinction matters commercially. Retained proceeds may be pledged as security for rupiah credit. The regulator has stated that pledging does not affect the bank’s gearing calculation. They may not be pledged for a foreign currency loan. The facility is capped at the amount pledged, a written declaration is required, and a default does not extinguish the placement obligation.

Foreign exchange swaps are permitted on the buy side only. The swap must be done at the bank holding the placement. The proceeds must return to the originating foreign currency account. The tenor must fall inside the remaining placement period. Sell-side swaps and linked forwards are not available. The placement itself can sit in a special account, a term deposit at a state bank, central bank instruments, or government securities bought in the primary market.

This is the point exporters have pressed hardest in public. In May 2026 the chairman of the advisory council at the Indonesian mining professionals association put the complaint plainly. The money is there, he said, but the business has to borrow it back at a higher rate than the deposit pays. That is a cost of capital argument rather than a liquidity one. It is also the honest version of the grievance.

Export services have been suspended, and the enforcement record is public

There is a single sanction, and it is a serious one for a trading business. The Minister of Finance may suspend an exporter’s access to export services, which customs then executes. There is no fine on the exporter. The suspension does not remove the underlying obligation. Bank Indonesia supervises the inflow and the placement, and the financial services authority supervises the accounts themselves.

The enforcement has happened. The cumulative number of exporters blocked rose from 23 in March 2024 to 176 by January 2025. Some of those were unblocked after remediation. All of that enforcement relates to the earlier thirty per cent requirement rather than the current one. No count has been published since January 2025, which means the current rate of enforcement is unknown rather than low.

Sixty four mining exporters returned to thirty per cent in September 2026

The government has given ground in stages, and the most recent concession is narrow but real. From September 2026 a defined group of mining exporters may retain thirty per cent for three months. They may do so at any of fifteen designated banks, five of them state-owned and ten of them not. The criteria require an Indonesian company in mining with a shareholder of at least ten per cent from a partner country.

The partner countries are the United States, China, Hong Kong, Australia and Canada. Of 537 registered mining exporters, 64 qualify, which is about twelve per cent. An eligible exporter has five working days to opt out in writing, and silence counts as acceptance. The choice is irreversible and applies to everything the company exports. The list is reissued every month.

Whether the rule has achieved its stated aim is not settled by the reserve figures. Reserves stood at USD 156.5 billion at the end of December 2025. They stood at USD 146.3 billion at the end of September 2026. They are therefore lower after the rule than before it. Bank Indonesia’s own releases do not mention the retention scheme at all. If the objective is domestic dollar liquidity and rupiah demand, which is what officials actually say, the evidence is better.

The rupiah leg cannot leave the country at all

Three restrictions compound, and the second is the least discussed. The proceeds are held onshore at a state bank for a year. The rupiah created by conversion cannot then be transferred abroad. Indonesia maintains a prohibition on moving rupiah out of the country, alongside bans on offshore rupiah forwards and on rupiah credit to non-residents. Conversion is therefore a one-way movement into the domestic economy.

The daily and monthly rupiah transfer limits quoted in many summaries come from a 2016 regulation. That regulation was revoked in July 2022, and those figures should not be cited as current law. The current thresholds are not published. Outbound foreign currency transfers must carry the value, the purpose and the parties. They must also state the financial relationship between the parties, with supporting documents required above a threshold that is also unpublished.

Offshore borrowing has its own regime. An annual loan plan falls due in the middle of March each year, and monthly reporting on the fifteenth. A credit rating report is required. Non-revolving drawdowns must pass through an Indonesian foreign exchange bank. A drawdown outside one attracts a sanction of a quarter of a per cent of the amount, capped in rupiah. Late reporting is charged daily, and persistent failure is notified to the parent company and the creditor.

A state company becomes the only exporter of coal and palm oil

For a trading business the larger change is not the retention rule at all. From June 2026 exports of strategic natural resources are routed exclusively through a state export entity. That entity acts as sole exporter and can set the price. The first commodities named are coal, palm oil and ferro alloy. All palm oil derivative exports must go through it after 31 December 2026, with full implementation from 1 January 2027.

That changes the question. The retention rule asks what an exporter may do with its proceeds. The single gate asks whether a private trader can be the exporter of record at all. Any structuring work for Indonesian natural resource trade done in late 2026 that ignores the second question is answering the wrong one.

Singapore has had no exchange control since 1978

Singapore has no retention requirement and no surrender requirement. There is no repatriation percentage, no holding period, no designated bank and no conversion cap. That is not a recent liberalisation. A Monetary Authority of Singapore notice issued on 25 May 1978 exempted all persons from the obligations imposed under the Exchange Control Act. It took effect on 1 June 1978. It removed exchange control formalities for all payments and capital transfers, and abolished the statistical returns that went with them.

What Singapore does require is corporate rather than monetary. A company holds an annual general meeting and files an annual return within seven months of its financial year end. Late filing carries a penalty. An audit is required where the company exceeds two of three thresholds for revenue, assets and headcount. Tax is assessed at seventeen per cent. An estimate of chargeable income is due within three months of year end, and the return by the end of November.

A trading company receiving its own export proceeds does not need a payment services licence in Singapore. The licensing requirement attaches to the business of providing a transfer service to other people. A company moving its own money provides a service to nobody. One residual currency measure remains in place, and it concerns Singapore dollar lending to non-resident financial institutions. It is addressed to banks rather than to exporters.

The rule has a rationale, and predictability is the fair criticism

Indonesia exported USD 282.91 billion of goods in 2025. It held USD 146.3 billion of reserves at the end of September 2026. A country exporting at that scale while holding that level of reserves has a real structural question. Five months of import cover is also thinner than it looks. The misinvoicing concern is also substantive rather than rhetorical. A review of about 6,500 declarations worth more than USD 14 billion produced a preliminary estimate of around USD 5 billion a year in understated value. The government was explicit that this is not additional money already earned abroad.

Structures in which an Indonesian exporter sells cheaply to a related offshore trader are a real leakage. A Singapore firm should say so rather than pretend otherwise. The rule is also not confiscatory in design. The exporter keeps the money and earns a return on it. A twelve month placement pays nil final tax, against up to twenty per cent on ordinary investment income. The proceeds may be spent on five categories, pledged for rupiah credit without a gearing consequence, and used in buy-side swaps. The government has visibly responded to representations at each revision.

Two criticisms survive that. The first is predictability rather than severity, and it is the stronger point. The rule changed four times in nineteen months. The eligible bank universe was narrowed and then partly reopened. The qualifying mining list republishes every month with a five day opt-out. Working capital cannot be planned against that.

The second concerns the state bank limb specifically. A dollar held at a foreign-owned bank incorporated in Indonesia is inside the Indonesian financial system exactly as much as a dollar at a state bank. That limb has no evident monetary justification. The September 2026 concession reopened ten non-state banks for 64 exporters, which rather concedes the point.

What this means for exporters, treasurers and advisers

For an Indonesian exporter, the first task is mechanical, and nobody can do it from a summary. The tariff codes of every line actually exported determine whether the rule applies. The USD 250,000 test then applies declaration by declaration. The second task is to map foreign currency outflows against the five permitted uses. Every dollar of genuine foreign currency procurement, dividend or loan repayment reduces the placement obligation rather than breaching it.

For a treasurer running a regional group, the planning assumption should be that the Indonesian limb is slow and the rest of the group is not. Proceeds of an Indonesian export are reached wherever the group sits. The useful work is therefore sequencing rather than relocation. A Singapore company genuinely supplying inputs, genuinely lending, or genuinely receiving dividends is a permitted counterparty under the regulation itself. That is a better footing than a structure the rule was written to catch.

For an adviser, two warnings are worth passing on. The published scans of the 2026 amendments are poor quality, with percentages mangled by character recognition. Any figure quoted in a contract should therefore come from the gazette rather than a downloaded copy. The guidance on moving retained proceeds abroad for permitted procurement is also internally inconsistent. Indonesian counsel should settle that question before anybody commits in writing.

The IMF put the dollar at 56.70% of world reserves in mid-2026. Central banks bought a record 289 tonnes of gold that quarter, into a falling price.

A particular story has been circulating for two years. The dollar is being abandoned, central banks are piling into gold, and money is moving east. Each part of it has a grain of truth and each part is overstated in the retelling, usually by people selling something.

The underlying data is published quarterly by the International Monetary Fund and annually by the World Gold Council, and it is worth reading directly. What it shows is a slow drift away from the dollar over a decade. Most of the quarterly movement is explained by exchange rates rather than by decisions. Gold buying has fallen back from its 2022 to 2024 peak.

None of that makes the question irrelevant to a business. It changes what the honest answer is. The case for holding company money across more than one currency does not rest on predicting the dollar’s decline. It rests on where the money is earned and where it has to be spent, and it is stronger for not pretending otherwise.

The dollar held 56.70% of world reserves in the second quarter of 2026

The latest figure was released on 30 September 2026 and covers the second quarter of that year. The dollar accounted for 56.70% of global official foreign exchange reserves, down from 57.18% in the first quarter. Total world reserves stood at about US$13.2 trillion, up on the quarter before.

The euro was second at 20.60%, having gained about half a percentage point in the quarter. The yen was third at 4.96% and the renminbi fourth among the currencies the IMF published levels for, at 2.11%. The fifth-largest reserve currency on the most recent full data is the Canadian dollar rather than the renminbi, which is a detail most summaries skip.

The renminbi point is worth pausing on, because it runs against the headline narrative. Its share is small and it has fallen since early 2022. Whatever is happening to the dollar, the money is not going into the Chinese currency in any quantity that shows up in reserves.

The decade-long drift is real, and smaller than it looks

Taking a longer view removes most of the noise. The dollar accounted for 65.8% of disclosed reserves at the end of 2015 and 57.8% at the end of 2024. That is a fall of eight percentage points over nine years, or a little under one point a year.

A fall in share is not a fall in holdings. Total reserves grew over the same period, so central banks can hold more dollars than before while the dollar’s percentage declines. Anyone presenting the share as evidence that dollars are being sold has skipped a step.

There is also a measurement break that matters. Until December 2025 the IMF reported shares of allocated reserves, meaning only those countries that disclosed their composition. Since then it has imputed the composition of non-reporters so the data covers all of the world’s reserves. The restatement moved the numbers by about eight tenths of a point on its own. A single line running from 2015 to 2026 mixes two different denominators.

Most quarterly moves are the exchange rate, not a decision

This is the point that gets lost most often, and the IMF has made it explicitly. In the second quarter of 2025 the dollar’s share fell by almost a point and a half. At constant exchange rates it would have been essentially flat. Currency movements explained 92% of the decline, and the IMF concluded that the shift reflected valuation effects rather than active reallocation by central banks.

The mechanism is simple enough. When the dollar weakens against other reserve currencies, the dollar value of euro and yen holdings rises. The dollar’s share then falls without anyone having traded anything. That accounts for much of what gets reported as a retreat from the dollar.

The second quarter of 2026 is the exception, and it is the reason the latest release is interesting. Exchange rates were range-bound within about two per cent, so valuation effects were limited. The IMF attributes that quarter’s movement mainly to other price changes and to portfolio reallocation. It is the first recent quarter where the shift looks like a decision rather than an artefact.

Central banks bought less gold in 2025 than in each of the three years before

The gold half of the story is the one most frequently inflated. Central banks bought 863 tonnes on a net basis in calendar 2025, according to the World Gold Council’s figures published on 29 January 2026. That is a large number by historical standards and a smaller one than 2022, 2023 or 2024.

The 2026 picture is mixed and has already been revised once. The first quarter was initially reported at 244 tonnes and cut to 57 tonnes by erratum in July 2026, after a large volume was reclassified. The second quarter was 289 tonnes, a record for any second quarter and up 62% on the year before. The first half came to 345 tonnes, the weakest first half since 2022.

The price context cannot be left out. Gold traded above US$5,500 an ounce in January 2026 and was around US$4,190 in early October 2026. It is roughly a quarter below its peak and up only a few per cent over twelve months. The genuinely interesting fact is that central banks bought a record second-quarter volume into a falling price, which says more about intent than a rising price would.

No major central bank has published a policy of holding fewer dollars

This is the claim that fails most comprehensively on the evidence. A search for an official statement of intent to reduce dollar holdings produces nothing from any major central bank. The European Central Bank said as much in January 2026, describing the dollar as likely to remain by far the largest international currency.

What does exist is a gold target rather than a dollar target. Poland’s central bank announced on 10 September 2025 that it was raising its target for gold to 30% of official reserve assets. That arithmetically implies a smaller share in currencies, but it is not a statement about the dollar.

The survey evidence runs the other way. HSBC’s annual survey of reserve managers, published on 6 May 2026, drew responses from 101 central banks managing more than US$9.5 trillion. Of the 92 that answered the question, 25 were considering increasing their dollar exposure over the following year and 11 were considering decreasing it. More reserve managers are leaning towards the dollar than away from it.

What is claimed	What the published data shows
The dollar is at its lowest reserve share since 1995	The low point of the IMF’s series is 56.42%, in the fourth quarter of 2025. The series itself begins in 1995
Central banks bought record gold in 2025	They bought 863 tonnes, less than in 2022, 2023 or 2024
Reserve managers are cutting their dollar holdings	More surveyed central banks are considering increasing dollar exposure than reducing it
Money is moving into the renminbi	Its share is 2.11% and has fallen since early 2022
Gold keeps setting records	The price is about a quarter below its January 2026 peak
The dollar’s fall reflects decisions by central banks	Exchange rate movements explained 92% of one recent quarterly fall

What any of this means for a company rather than a central bank

A business does not hold reserves and does not need a view on the international monetary system. What it needs is narrower. It must be able to receive money in the currency a customer pays in, hold it until it is needed, and move it without asking permission. That requirement exists whether or not the dollar’s share falls another point.

The reserve data is useful to a company in one narrow way. It establishes that the major currencies are all actively used and none is being abandoned. A business selling across several markets will therefore keep receiving several currencies. Converting each receipt into a single home currency on arrival is a decision with a cost, and it is usually made by default rather than deliberately.

The relevant question is therefore operational. Where can a company hold several currencies at once, move between them at market rates, and pay a supplier in a third currency without a filing. That is a banking and regulatory question, and it has a short list of answers.

Singapore has had no exchange control since 1978

Singapore liberalised exchange control with effect from June 1978. The instrument is a notice issued by the Monetary Authority on 25 May 1978, exempting transactions from the operative provisions of the Exchange Control Act. The Act itself remains on the statute book, which is a point of accuracy worth keeping. There is no restriction on a company holding, receiving or remitting foreign currency.

The absence is structural rather than discretionary. Singapore conducts monetary policy through the exchange rate rather than through interest rates, managing the Singapore dollar against a trade-weighted basket within an undisclosed band. A country that targets its exchange rate and permits free capital movement gives up control of domestic interest rates. That is the trade Singapore has made, and open capital flows are the side of it that a business experiences.

One restriction does exist and should be stated rather than glossed. The Monetary Authority limits Singapore dollar credit facilities to non-resident financial institutions where the proceeds could be used to speculate against the currency. It constrains lending to that specific class of counterparty. It does not touch an operating company holding or moving foreign currency.

Singapore’s own official foreign reserves stood at S$553.57 billion in September 2026, released on 7 October 2026, the highest level since February 2022. The Monetary Authority tightened policy slightly in April 2026 and again on 27 July 2026, steepening the path of appreciation. That is policy rather than forecast, and a currency on a steepened appreciation path is a relevant fact for anyone deciding where to hold a balance.

Where internationally sourced money is being managed

The asset management figures say more about direction of travel than the reserve data does. Singapore’s industry held S$6.7 trillion at the end of 2025, up 10.1%, with net inflows of S$376 billion. About three quarters of that money came from outside Singapore and about 88% of it is invested outside Singapore.

Singapore is also putting public money behind its own equity market, which is a different kind of signal. A programme announced in February 2025 was raised to S$6.5 billion in February 2026. It has allocated about S$5.4 billion to fourteen managers in three batches, the most recent on 30 September 2026. The distinction between allocated and deployed matters, and the amounts actually invested have not been published.

Two qualifications belong with the asset management figures. The 2025 growth was driven largely by market performance rather than new mandates, and the alternatives picture was weak, with real estate assets falling 16%. The number of licensed managers grew under 2% against 10% asset growth, which is not the shape of an industry adding firms.

What this means for treasurers, owners holding cash across markets, and anyone being sold a currency thesis

For a treasurer, the practical decision is about matching rather than forecasting. A company that earns in several currencies and spends in several currencies does not need a view on the dollar. It benefits simply from holding balances in the currencies it actually uses. The cost of converting every receipt on arrival is a real spread paid repeatedly, and it is usually invisible because nobody books it as a line item.

For an owner holding cash across markets, the question to ask of each banking relationship is narrow. Can the account hold the currencies the business receives, can money move between group entities without approval, and is there a filing attached to any of it. Those answers differ sharply between countries, and they differ more than tax rates do.

For anyone being shown a currency thesis by an adviser, the reserve numbers are a good test of whether the adviser has read the source. The dollar’s share is 56.70% and falling slowly. Most quarterly movement is the exchange rate. Gold buying is below its peak and the gold price is a quarter below its own. An adviser whose case depends on any of those three being otherwise is selling a story rather than reading the data.