The bill was gazetted in June 2026, has not passed, and would apply to a year of assessment that began in April 2025.

Hong Kong gazetted the Inland Revenue (Amendment) (Preferential Tax Regimes for Funds, Family-owned Investment Holding Vehicles and Carried Interest) Bill 2026 on 12 June 2026. It had its first reading on 24 June 2026. As at 11 October 2026 it has not passed. The bill sits with a committee of the Legislative Council, which met the Administration on 8 July 2026, and the Government expects it to pass during 2026.

Two things about it are being reported together and should be separated. The bill genuinely widens the concession, in several directions at once. It also attaches a substance requirement and a reporting obligation to the fund exemption for the first time. It would do so for a year of assessment that began on 1 April 2025.

Most coverage has described the substance test as new for family offices. It is not. Family-owned investment holding vehicles have had to meet exactly those conditions since the 2023 ordinance, and the bill relaxes how one of them is measured. The new burden lands on funds, which until now had no substance requirement at all.

What the bill widens

The clearest change is that funds with a single investor become eligible. The existing exemption has been in force since 1 April 2019. It is written around a statutory definition of a fund that requires multiple investors, which is why a fund of one has been excluded. The bill brings them in.

The list of eligible assets widens by eight classes. Loans and private credit, digital assets, overseas immovable property, carbon credits and emission instruments, insurance-linked securities and equity in non-corporate private entities are all added. Precious metals come in with a 20% cap and specified commodities with a 15% volume cap.

The 5% cap on incidental income is removed. That cap has been a persistent irritation, because a fund earning a small amount outside its qualifying transactions could lose the exemption on the lot. It is replaced by a list of excluded activities rather than a percentage, which is a more workable test. The treatment of special purpose entities is relaxed in the same direction.

What it conditions, and on whom

The fund exemption as it stands has no economic substance requirement and no reporting mechanism at all. The bill supplies both. That is the genuinely new obligation in the package, and it falls on funds rather than on family offices. A fund that has relied on the exemption since 2019 without employing anyone in Hong Kong is the entity whose position changes.

There is a real question about how it will work in practice. Advisers in Hong Kong report that it is not yet settled whether the requirements apply fund by fund. Nor is it settled whether core activities may be outsourced to a third party or an associate. The existing family office regime permits outsourcing to the eligible family office with monitoring. Whether the fund regime will do the same is expected rather than confirmed.

The retrospection is the part a principal should plan around. The bill applies to years of assessment beginning on or after 1 April 2025, which is a year already more than half elapsed. The revenue department issued a transitional notice on 12 June 2026. Taxpayers may file on the basis of the proposed exemption, must monitor developments, and must write in if returns need revising once the bill passes. That is a sensible administrative answer to an awkward position, and it is also an unusual thing to ask of a taxpayer.

The family office conditions have been in force since 2023

The HK$240 million threshold, the two employees and the HK$2 million of annual spending all took effect under an ordinance gazetted on 19 May 2023. They have applied from the 2022-23 year of assessment. They are not a new catch. The bill changes only how the threshold is measured. Loans from shareholders and from related special purpose entities are no longer deducted, which makes it easier to clear rather than harder.

Independent confirmation that those conditions were already law arrived from an unrelated direction. From 1 March 2026 Hong Kong’s investment migration scheme allows an applicant to hold qualifying investments through a private company. That company must be a family investment holding vehicle, and it must meet the same two employees, HK$2 million of spending and HK$240 million of managed assets.

The rate itself is unchanged, and worth seeing as an effective figure rather than a headline. Hong Kong charges 8.25% on the first HK$2 million of a corporation’s assessable profits and 16.5% above that. Only one company in a connected group may claim the lower tier in any year.

Singapore revised its own conditions on 31 July 2026

Singapore moved in the same window and in a different manner. The Monetary Authority issued a circular on 31 July 2026 that rewrote the conditions attaching to the fund tax incentive schemes. It split one set of rules into two, and the split corrects a confusion that has run through published commentary for years.

A fund managed by a single family office and a commercial fund managed by a third party now face materially different conditions under the same statutory provisions. The numbers most often quoted are the family office numbers, and applying them to a commercial fund overstates the entry requirement by a factor of four.

Under section 13O, a family office fund needs S$20 million of assets in designated investments at application and at the end of each period. A commercial fund needs S$5 million at application, with a grace period to the third year of assessment. Under section 13U both need S$50 million, but the surrounding conditions still differ.

The spending requirement runs on the same size bands and different amounts. A family office fund above S$2 billion must spend S$1 million a year locally. At least S$500,000 of that must be business spending, and the remainder may be charitable donations or grants recognised at twice the amount spent. A commercial fund of the same size must spend S$500,000. Two further conditions, a capital deployment requirement and a private banking account, apply to the family office track only.

Singapore is not immune to mid-stream change either

The comparison usually drawn is that Singapore publishes its conditions in advance and does not change them afterwards. The first half of that is defensible. The second half is not, and a reader who follows this closely will know it. Overstating the point is the fastest way to lose an audience that checks.

Three things in the 31 July 2026 circular cut against the claim. The removal of the annual asset maintenance test for commercial funds was made retroactive to 1 January 2025, nineteen months earlier. Family office funds holding awards granted under three earlier sets of conditions were moved onto the new ones, including the higher spending tiers and the banking requirement. And a commercial fund that fails to reach S$5 million by its third year of assessment has its award revoked from the date it started. That reopens years already filed.

The honest version of the comparison is narrower and still favours Singapore. Conditions arrive by dated circular, effective dates are generally prospective, and several of the tightening measures were deferred to the 2028 year of assessment. Hong Kong is legislating in June 2026 for a year that began in April 2025. The difference is in the orderliness of the process rather than in an absolute rule that one country never reaches backwards.

One asymmetry should also be acknowledged. Almost everything Hong Kong is doing retrospectively is in the taxpayer’s favour. Wider assets, a removed cap and a relaxed measurement all help. The only genuinely new burden being applied to a year already running is the substance and reporting requirement on funds.

Hong Kong is competing hard, and Singapore has said so

An article that presents Singapore’s position as unchallenged would be contradicted by Singapore’s own regulator. On 28 July 2026 the Managing Director of the Monetary Authority said it was in discussions with the industry about improving competitiveness. He named competition from places like Hong Kong. That is the regulator naming the pressure.

The pressure is real and funded. A study by Deloitte for Hong Kong’s investment promotion agency was published on 10 February 2026. It counted 3,384 single family offices operating in Hong Kong. Their direct operating spending was put at about HK$12.6 billion a year, employing more than 10,000 full-time professionals. The Government has set a target of helping more than 220 family offices establish or expand between 2026 and 2028.

Those counts are not comparable with Singapore’s. Hong Kong does not require a family office to register, so the 3,384 figure is an estimate of the whole population. Singapore reports more than 2,000 single family offices as at the end of 2025, but counts only those that have been awarded tax incentives. Offices operating in Singapore without an award are not in the number. One figure is an estimate of everything and the other is a floor.

Where the money actually sits

The assets under management figure is the one that compares cleanly between the two centres. It is published by the regulator rather than estimated by a consultancy, and it is measured the same way each year.

Singapore’s industry held S$6.7 trillion at the end of 2025, up 10.1% on the year, with net inflows of S$376 billion. Three quarters of that money is sourced from outside Singapore and about 88% of it is invested outside Singapore. The industry is a place where international money is managed rather than a pool of domestic savings.

Two details stop that being a victory lap. The growth was driven largely by market performance rather than by new mandates, and the alternatives picture was weak. Alternative assets grew 0.4%, private equity and venture capital grew 1%, and real estate fell 16%. The number of licensed managers rose under 2% against 10% asset growth, which is not the pattern of an industry adding firms.

What this means for families already in Hong Kong, families choosing now, and advisers with clients in both

For a family office already established in Hong Kong, the near-term question is a filing question rather than a structural one. The revenue department has invited taxpayers to file the 2025-26 year on the basis of the proposed exemption and to correct afterwards if the bill changes. That is workable, but a position taken now may have to be revisited. The file should record the basis on which the return was prepared. The substance conditions that apply to a family investment holding vehicle have not changed since 2023.

For a family choosing between the two now, the decision has stopped being about the headline rate. It is about which set of conditions the family can actually meet and sustain. Hong Kong asks for two employees, HK$2 million of local spending and HK$240 million under management. Singapore asks for a fund of S$20 million and two investment professionals, one of whom is not family. It also asks for local spending from S$200,000, a capital deployment requirement and a private banking account. Those are different shapes of commitment, and the right answer depends on where the family’s people and investments already are.

For an adviser with clients on both sides, the useful discipline is to date every condition. Hong Kong’s family office conditions date from 2023 and are being relaxed. Its fund substance requirement is proposed and would be retrospective. Singapore’s revised conditions date from 31 July 2026, with some elements deferred to 2028 and one made retroactive to the start of 2025. A client who is told only the current state of the rules, without the dates they arrived and the dates they bite, cannot plan around either country.

Division 296 became law on 13 March 2026 and applies from 1 July 2026. No company outside Australia reduces it by a dollar.

Australia has added a second layer of tax to large superannuation balances. Two Acts received Royal Assent on 13 March 2026, and the measure applies to income years starting on or after 1 July 2026. It had a difficult passage. An earlier version announced in February 2023 would have taxed unrealised gains, and it never passed. The Government redesigned it in October 2025, dropped the unrealised gains element and moved the start date back a year.

The commercial response being marketed to Australian owners is an offshore company, and it does not work. Division 296 measures two things only, both of them inside the superannuation system. A Singapore company is in neither. Money cannot lawfully leave superannuation before a condition of release. If the fund itself buys the shares, they stay in the balance and the calculation is unchanged.

That is worth stating plainly at the start, because the opposite is being sold by incorporation agents on both sides of the trade. What a Singapore company does for an Australian owner is a separate and narrower thing. This article sets out the tax as it actually works, and then what a Singapore structure can and cannot carry.

Division 296 is charged on earnings, not on the balance

The common description is that balances above A$3 million are taxed at 30% and balances above A$10 million at 40%. That is right about the headline and wrong about the base. The tax applies to earnings, and only to the proportion of those earnings attributable to the part of the balance above each threshold.

The mechanism works by proportion. An additional 15% applies to the share of earnings attributable to the balance above A$3 million. A further 10% applies to the share attributable to the balance above A$10 million. The fund’s own 15% sits underneath both. That is where the 30% and 40% headline figures come from, and they are the total rather than the additional charge.

A worked example makes it concrete. On a balance of A$12 million earning 6%, total earnings are A$720,000. Three quarters of the balance sits above A$3 million, so A$540,000 of those earnings attract the extra 15%, which is A$81,000. One sixth sits above A$10 million, so about A$120,000 attracts the further 10%, which is A$12,000. The Division 296 charge is therefore about A$93,000 for the year.

Expressed as a rate on all the earnings, that is 12.92% rather than 30% or 40%. The effective charge rises with the balance, from 3.75% at A$4 million to 17.75% at A$20 million. Both thresholds are indexed, in steps of A$150,000 and A$500,000, which removes the criticism that drove most of the opposition to the 2023 design.

Nobody is liable yet, and the first assessments arrive in 2028

Timing is the detail most commentary gets wrong, and it matters for planning. For the 2026-27 year the balance is tested at 30 June 2027 alone, which is a transitional concession. Nobody has a Division 296 liability as at October 2026, because the measurement date has not arrived.

The assessments come later still. Large funds have until the end of October to report balances, and self-managed funds may not lodge until the following May. The Australian Taxation Office has said the first assessments begin issuing in the later half of the 2027-28 income year. Payment is due 84 days after the notice.

One transitional concession deserves more attention than it gets. An eligible self-managed or small fund may elect to reset the cost base of its assets to market value at 30 June 2026. The election counts for Division 296 purposes only. For a fund holding assets with large accrued gains, that election is the single most consequential decision available, and the window for it has already opened.

A Singapore company does not reduce Division 296 by one dollar

There are four reasons for that, and each one is sufficient on its own. They are worth going through in order, because the first two defeat the idea outright and the second two defeat the fallback versions of it.

The first is that Division 296 is a function of the member’s total superannuation balance and the earnings reported by superannuation funds. A foreign company is neither a superannuation interest nor a superannuation fund, so money genuinely outside the system was never in the base to begin with.

The second is preservation. Superannuation benefits cannot be accessed until a condition of release is met. That generally means retirement at or after preservation age, reaching 65, death, permanent incapacity or terminal illness. A member below preservation age cannot move superannuation money into a foreign company at all. Attempting it is illegal early release, and the consequences are severe.

The third is what happens if the fund itself invests. A self-managed fund generally cannot acquire private company shares from a member or a related party. An investment in a related company is an in-house asset capped at 5% of the fund. Non-arm’s length dealings produce income taxed at the top marginal rate inside the fund. And decisively, the shares would still be a fund asset, still counted in the member’s balance, and the income still part of the Division 296 calculation.

The fourth reason applies outside superannuation entirely, and it is the one most offshore marketing ignores. It is also the reason an offshore company is usually the wrong answer for an Australian investor even where superannuation is not involved at all.

Outside superannuation, the controlled foreign company rules remove the usual advantage

An Australian tax resident is assessable on income from all sources, inside and outside Australia. The controlled foreign company rules then close the deferral that an offshore company would otherwise provide. Singapore is not on Australia’s list of seven listed countries, which contains Canada, France, Germany, Japan, New Zealand, the United Kingdom and the United States.

The test that decides the outcome is the active income test. A foreign company passes if less than 5% of its gross turnover is tainted income, which means at least 95% must be genuine active business income. Interest, dividends, rent, royalties and gains on shares are all tainted. A Singapore holding or investment company owned by an Australian resident will therefore fail that test. Its income is attributed to the resident and taxed at marginal rates in the year it is earned. That happens whether or not a dividend is paid.

The distribution is no better. A dividend from a Singapore company carries no Australian franking credit, because only Australian company tax generates franking credits. Singapore levies no withholding tax on dividends, so there is usually no foreign income tax offset either. The dividend therefore arrives with no credit of any kind and is taxed at the resident’s full marginal rate, up to 47%.

Leaving Australia triggers a deemed sale of almost everything

Residency is the only thing that changes the analysis, and changing it has a price. On ceasing to be an Australian resident, the individual is treated as having disposed of every asset at market value, except taxable Australian property. There is an election to defer, but it is all or nothing across every affected asset. The deferred assets are then treated as Australian property until sold, or until the person returns.

That election is often the worse choice. A later sale while non-resident loses the 50% capital gains discount for the non-resident period and attracts non-resident rates with no tax-free threshold. Triggering the deemed disposal on departure crystallises the gain while the person is still a resident and still entitled to the discount. That is the opposite of what most people assume.

Two things do not change on leaving. Division 296 contains no residency test, so an Australian expatriate who keeps a superannuation balance above the threshold remains liable. And a self-managed fund must continue to meet the tests for an Australian superannuation fund, including where central management and control sits. Members moving abroad for an extended period put the fund’s complying status at risk, and with it the concessional rate on everything inside it.

Three larger changes have been announced and not legislated

Division 296 is not the only thing moving in Australian tax. An owner planning for the next five years should know what is proposed as well as what is already law. Three larger measures sit in the announced column, and two of them would reshape how private wealth is held.

The 2026-27 Budget on 12 May 2026 announced the removal of the 50% capital gains discount for gains arising after 1 July 2027. It would be replaced by cost base indexation for assets held at least twelve months and a 30% minimum tax on real capital gains. It would apply to individuals, trusts and partnerships across all asset classes. Assets acquired before 20 September 1985 would be brought into the net for gains after that date. None of this is legislated.

The same Budget announced a 30% minimum tax on the taxable income of discretionary trusts from 1 July 2028, payable by the trustee. Franking credits would be applied by the trustee first, so they would not flow to beneficiaries, and company beneficiaries would receive no credit at all. Fixed trusts, superannuation funds, deceased estates and testamentary trusts existing at 12 May 2026 would be excluded. This too is announced only.

Separately, the High Court decided on 10 June 2026 that an unpaid present entitlement owed by a trust to a corporate beneficiary is not a loan. That takes it outside the deemed dividend rules. That is a win for a common structure, though the Tax Office has signalled it may use other provisions where those rules no longer reach.

What a Singapore company actually does for an Australian owner

The honest case is narrower than the marketing and more durable. A Singapore company that carries on a genuine business passes the active income test. It needs real people performing real functions, and at least 95% active income. Its retained active profits are not attributed to the Australian owner. They are taxed in Singapore at 17%, with exemptions that reduce the effective rate on smaller profits.

Beyond tax, it provides a regional base with treaty access, multi-currency banking and a labour market for regional roles. For an Australian business genuinely selling into Asia, that is a commercial decision that happens to have a reasonable tax outcome. For an Australian investor holding a portfolio, it is a worse outcome than doing nothing, with compliance costs on top.

One claim should be treated with suspicion wherever it appears. There is no dataset showing Australians relocating or restructuring offshore in response to Division 296. The closest official series is Australian Bureau of Statistics migration data for 2024-25. It shows Australian citizen departures falling rather than rising, and it covers a period ending before Division 296 was even law. Anyone citing a relocation wave is inventing it.

What this means for members near the threshold, owners weighing an offshore move, and advisers fielding the question

For a member whose balance is near or above A$3 million, the decisions worth making are inside the Australian system rather than outside it. The cost base reset election to 30 June 2026 market value is available now and is irreversible once the year closes. The measurement date is 30 June 2027, which means asset allocation and realisation timing in the current year both matter. Neither of those decisions involves a foreign company.

For an owner weighing a move abroad, residency is the only lever that works, and it carries its own bill. The deemed disposal on departure should be quantified before anything else, because it is usually the largest single number in the plan. Taking the charge rather than deferring it is frequently the better answer. And superannuation follows the person, so an expatriate with a large Australian balance remains inside Division 296.

For an adviser being asked whether a Singapore company helps, the answer separates cleanly into two parts. It does nothing for Division 296, and nothing for passive income while the client is an Australian resident. It does work for a genuine operating business in the region, where the active income test can be met and the structure stands on commercial grounds. Keeping those two answers apart is what protects the client and the adviser.

Indian parties were the third-largest group at Singapore’s arbitration centre in 2025. India’s own courts held 5.6 crore pending cases in March 2026.

An engineering components manufacturer in Pune ships to a distributor in Jakarta. The order is worth about US$1.4 million and the goods were accepted without complaint. Four months later the distributor has paid 30% and stopped. The contract was drafted by the manufacturer’s own lawyers, it is governed by Indian law, and it gives jurisdiction to the courts at Pune. The scenario is a composite rather than a client, and the contract terms are the ones most Indian exporters actually use.

Those two clauses are where the money is decided. A dispute governed by Indian law and heard in an Indian court must take its place in a system holding more than five crore pending cases. The same dispute, written to be arbitrated in Singapore, runs to a published timetable. It produces an award that is enforceable in India under a treaty both countries have signed.

The choice costs nothing to make at the drafting stage. It costs years to change afterwards, because the counterparty has no reason to agree to a new dispute clause once it is the one holding the money. By the time the first invoice goes unpaid, the terms are already fixed. Everything that follows is decided by two sentences written when both sides were still friendly.

Indian courts held 5.6 crore cases pending in March 2026

The official figures come from a parliamentary answer rather than from an estimate. As at 9 March 2026, the Supreme Court had 92,755 matters pending. The twenty-five High Courts held 63,61,340 between them. The district and subordinate courts held 4,95,62,229. The three levels together came to 5,60,16,324 cases.

Those totals cover every kind of case, and the data does not separate commercial matters from the rest. India publishes a civil and criminal split but no commercial category, so any figure claiming to isolate commercial suits should be treated carefully. What the national dashboard does publish is the age of what is waiting, and that is the more useful measure for a business deciding where to litigate.

A quarter of everything pending in the district courts has already been there more than five years, and 9% for more than ten. The High Courts are worse on both counts. Forty-three per cent of their pending cases are over five years old and 23% are over ten. Those are the courts an appeal reaches. A commercial matter that is appealed therefore enters a queue where almost a quarter of the cases have been waiting more than ten years.

India requires mediation before a commercial suit, and it settles under 2% of applications

The Commercial Courts Act of 2015 requires a party to attempt mediation before filing a commercial suit, unless urgent interim relief is needed. The process is given three months, extendable by two with consent. On paper it is a sensible filter that keeps straightforward claims out of court.

The government publishes what it actually achieves, and the figures are not encouraging. Since the scheme began in July 2018, 2,76,676 applications have been made. Of those, 2,30,751 were recorded as non-starters, which means the other side did not engage at all. Settlements came to 4,854, which is 1.75% of applications made.

The rate has never exceeded 3.1% in any year and has fallen in each of the last three. For the Pune manufacturer this is a three-month delay with a 98% chance of changing nothing. The distributor in Jakarta has every incentive to be one of the non-starters. The step has to be taken anyway, because a suit filed without it can be rejected.

The Commercial Courts Act sets a timetable, and no official data shows it being met

The Act does set real deadlines, and they are not unreasonable. A written defence must be filed within 120 days of service, after which the right to file one is lost. Arguments must close within six months of the first case management hearing. Judgment is due within ninety days of the close of arguments. An appeal must be filed within sixty days, and the appellate court is to endeavour to decide it within six months.

Whether any of that happens is not something India measures. No official average or median disposal time is published for commercial courts at any level. The only government-commissioned evaluation of the Act covered two states and used data ending in June 2021. It reported disposal rates of 0% in one of them and 11% in the other. It published no national timing figure at all.

The best available measurement comes from outside government. Two researchers examined the lifecycle of suits filed at the Bombay High Court between 2017 and 2024. They found a median of 1,580 days for commercial suits, against 1,117 days for ordinary ones. Commercial suits, in other words, took more than a year longer than the cases the specialist procedure was meant to improve on. The same work found that 40% of commercial cases had no hearing within three months of filing, and that close to half of all hearings were adjournments. That is one court and one research team rather than an official series, and it should be read as such.

There is no current international benchmark either. The World Bank discontinued the survey that used to produce a days-to-enforce-a-contract figure for India in 2021, and its replacement has not yet covered India. Any number in circulation purporting to measure this traces back to data gathered more than ten years ago.

A Singapore-seated arbitration runs to a published timetable

The alternative is not a different court. It is a different mechanism altogether, agreed in the contract before anything goes wrong, with its own rules and its own clock. The parties appoint the tribunal, the institution administers the timetable, and no queue has to be joined. The table below sets the two routes against each other on the stages that take the time.

Stage	Indian commercial suit	Singapore-seated arbitration
Before the claim can be brought	Mediation, three months, extendable by two	None required
Written defence	120 days from service, after which the right is lost	Set by the tribunal at the first case conference
Urgent protective relief	A separate application to the court	Emergency arbitrator appointed within 24 hours, decision within 14 days
Close of arguments	Within six months of the first case management hearing	Fixed by the tribunal within the overall time limit
Decision	Judgment within 90 days of the close of arguments	Three months from the tribunal being constituted on the streamlined procedure, six on the expedited one
What the measurements show	Median 1,580 days at one High Court, 2017 to 2024	Median 117 months to award, on a study published in 2016
Challenge	Appeal within 60 days, then a further process	Limited grounds only, decided by the Singapore courts

The Singapore rules were rewritten with effect from 1 January 2025 and the changes go to speed. A new streamlined procedure applies to disputes up to S$1 million and requires an award within three months of the tribunal being constituted. The expedited procedure threshold rose from S$6 million to S$10 million, with an award due within six months. For a claim of the size in this example, the expedited procedure is available.

The emergency provisions matter more than the headline timetable for an exporter chasing goods or money. An emergency arbitrator is appointed within twenty-four hours of the application. A protective order can be made within twenty-four hours of that appointment, and the full decision follows within fourteen days. Set against a court application in a system with five crore cases pending, that is a different order of responsiveness.

One honest qualification belongs here. The eleven-point-seven-month median comes from a study published in October 2016 covering 98 cases, and no equivalent study has been issued since. The three-month and six-month commitments in the current rules are the live figures. Both run from the constitution of the tribunal rather than from the day the dispute arises.

Indian parties have been among the top three users of Singapore arbitration for a decade

This is not a structure being proposed to Indian businesses for the first time. It is one they already use in large numbers, and the figures come from the arbitration centre’s own annual report rather than from anybody’s marketing material.

The centre handled 886 new cases in 2025, its second highest on record, with a total sum in dispute of US$14.53 billion. Parties came from 79 jurisdictions and 89% of cases were international. The figure includes 135 matters in which the centre acted only as appointing authority, so the administered caseload was 737.

India supplied 178 parties in 2025, placing it third behind mainland China and Vietnam. It was the largest single foreign user in both 2021 and 2022. The centre’s own president said in July 2026 that Indian parties had consistently remained among the top three over the past decade. Indian law was the third most frequently applied governing law in 2025. Indian arbitrators were the third most appointed, and four members of the centre’s court are from India. A liaison office opened in Delhi on 25 July 2026.

The practical point for a drafting lawyer is that none of this is exotic. An Indian company naming Singapore as the seat is doing what hundreds of Indian companies did last year. Indian counsel appear there regularly and Indian law is routinely applied.

A Singapore award is enforceable in India, and the Supreme Court narrowed the defences in March 2026

An award is only as good as the enforcement behind it. India is a party to the New York Convention and applies it on a reciprocal basis. An award qualifies only if it was made in a country India has notified. India notified Singapore by a notification dated 6 July 1999, so a Singapore award is enforceable in India as a matter of course.

The grounds for resisting enforcement were narrowed on 25 March 2026. The Supreme Court held that an argument already lost before the courts at the seat cannot be reopened in India. A party cannot revive it by relabelling it a public policy objection. Public policy itself remains available, and issues that genuinely turn on Indian law still fall to Indian courts. What the decision removes is the second bite, which was the most common delaying tactic.

A second decision, on 15 April 2026, confirmed that naming a seat fixes which country’s courts supervise the arbitration, and that naming a venue alone does not. For an Indian contract that is the difference between a Singapore clause that works and one that produces a jurisdictional fight before the merits are reached. The drafting distinction is a single word, and it decides which court hears every application.

Two long-running enforcement sagas show the worst case, and they are not typical

Honesty requires the counter-examples, because they are well known and a sceptical reader will raise them. Daiichi Sankyo obtained a Singapore-seated award of roughly ₹3,500 crore on 29 April 2016. The Delhi High Court allowed enforcement on 1 February 2018. As at 29 September 2026 the award was still being executed, more than ten years after it was made, with only a small fraction recovered.

Avitel followed a similar course. The arbitration centre issued its final award on 27 September 2014. The Bombay High Court decided the enforcement proceedings on 25 April 2023, and the Supreme Court confirmed the position on 4 March 2024. That is nine and a half years from award to final confirmation.

Both are among the most heavily contested enforcement proceedings in Indian legal history and neither is a typical outcome. India publishes no average time to enforce a foreign award, so no honest claim about the normal case can be made. What can be said is that these were the hardest fought examples of their kind. The 2026 Supreme Court decision removes one of the tactics used in them. And an award debtor with assets outside India can often be pursued somewhere the dispute is not reheard at all.

What India has announced and not yet done

Three items on the Indian side are worth knowing, because they are regularly described to clients as having already happened. All three are announcements rather than law, and the gap between the two is years rather than months.

The Arbitration Council of India, authorised by an amendment passed in 2019, has still not been constituted. The Ministry of Law and Justice confirmed this in a statement on 6 February 2026, seven years after the enabling provision. A draft Arbitration and Conciliation Amendment Bill was released for consultation on 18 October 2024, proposing emergency arbitrator recognition and an appellate arbitral tribunal. It has never been introduced in Parliament, and a 2026 session list confirms it was not introduced that year either.

India signed the Singapore Convention on Mediation on 7 August 2019, the day it opened, and has not ratified it. Singapore ratified on 25 February 2020 and the Convention entered into force on 12 September 2020. A mediated settlement reached in a cross-border dispute is therefore directly enforceable in Singapore and not in India. India did establish a Mediation Council of India on 27 August 2026 under its own domestic legislation. That is a real step, but it does not change the Convention position.

None of this means Indian arbitration does not work. It means the reforms most often cited as evidence of improvement are still drafts. A contract signed today has to work under the law as it stands rather than as it has been proposed.

What this means for exporters drafting contracts, businesses already in dispute, and groups selling across the region

For an exporter drafting a contract, the dispute clause is the cheapest clause to get right and the most expensive to get wrong. Naming Singapore as the seat, selecting institutional rules and choosing the governing law deliberately costs nothing at signature. The same change after a dispute has started requires the agreement of the party holding the money, which will not be given. Where the counterparty resists a Singapore seat, that resistance is itself information about how they expect the relationship to end.

For a business already in dispute under an Indian jurisdiction clause, the realistic question is about settlement value rather than about forum. The compulsory mediation step produces a settlement in under 2% of cases and will not help. The timetable in the Commercial Courts Act is real but unmeasured. The only published lifecycle study found a median of more than four years at one High Court. A claim worth pursuing should be priced with that duration and the cost of funds over it. The number that produces will often make an early commercial settlement look better than it did.

For a group selling across the region, the structural answer is to move the contracting party rather than only the clause. Where the Singapore company signs the regional customers, the contract is naturally Singapore law with a Singapore seat, and the dispute never touches the Indian system. That is the same arrangement that puts regional revenue and regional staff in Singapore for commercial reasons. The dispute resolution advantage comes with it rather than being bought separately.

On the OECD’s 2024 figures a cross-border case took India 46 months to close and Singapore 24. Behind that sit 603,647 appeals and no law setting a deadline to decide any of them.

Disputed tax is not an abstraction on a government balance sheet. It is working capital that a business has already earned and cannot use. Where the demand has been paid under protest, the cash has gone. Where it has not been paid, it sits on the balance sheet as a contingent liability that auditors, lenders and acquirers all have to price.

India publishes the size of that pool only when Parliament asks. The most recent answer put the total under appeal at more than ₹26 lakh crore as at 31 March 2025, across 603,647 pending appeals. That figure is the sum of the four forum-by-forum numbers given in the answer. It is not a total the government itself printed, and it is worth saying so.

Singapore’s revenue authority does not publish a comparable number, because there is no comparable backlog to report. What it publishes instead is an arrears rate of 0.66% across income tax, goods and services tax and property tax combined. The two measures are not the same thing and should not be set against each other. What they do show is which problem each administration considers worth counting.

India had 603,647 tax appeals pending at the end of March 2025

The amount under dispute has moved within a band rather than climbing steadily. It stood at ₹30.23 lakh crore at the end of the 2020-21 financial year. It fell sharply the following year and has climbed back since. The 2021-22 fall came almost entirely from the first appellate stage. The parliamentary answer does not explain it. It is better treated as an open question than as evidence of progress.

One component has moved in a single direction. The value of disputes before the High Courts was about ₹2.75 lakh crore at the end of March 2021. Four years later it was about ₹5.65 lakh crore. That is a doubling at the slowest and most expensive forum in the system. A matter that reaches a High Court has already passed through two earlier stages.

The flow figures published since then show why the stock does not fall. In the 2025-26 financial year the department disposed of 2.24 lakh first appeals and received 1.79 lakh new ones. First-stage pendency therefore ended the year at roughly 5.4 lakh, almost exactly where it stood five years earlier. The Finance Minister said at the time that the pace of clearing appeals had to match the pace of new filings. That is a fair description of the problem.

Nine in ten appeals sit at the first stage, and no law says when they must be decided

The distribution of the backlog matters more than its size, because the two measures point in different directions. The first appellate stage holds 89.4% of all pending appeals but 63.7% of the money. The High Courts hold 5.7% of the appeals and 21.5% of the money.

That split tells a business two different things. A large number of small disputes are queued at the first stage, where the decision is taken by a tax officer rather than a tribunal. A smaller number of much larger disputes have escalated into the courts, where they will take years and cost legal fees throughout.

No official figure exists for how long any of this takes. India publishes pendency and amounts, not durations. The Board sets internal disposal targets per officer, and those are not published either. Any average disposal time in circulation is a practitioner’s estimate rather than a measurement, and it should be treated as such. The defensible calculation is simpler. At the 2025-26 rate of disposal, clearing the standing first-stage pendency would take about two and a half years. That assumes not one new appeal is filed. In fact 1.79 lakh were.

The one like-for-like measure puts India at twice Singapore’s time

There is one place where India and Singapore are measured on the same basis, by the same body, in the same year. The OECD publishes how long each country takes to close a mutual agreement procedure case. That is the mechanism used when two tax authorities both claim the same profit. For a business with cross-border transactions, this is the dispute that actually bites.

On the 2024 figures a transfer pricing case took India an average of 45.73 months to close and Singapore 23.65 months. India sits 48% above the OECD average of 30.9 months, and Singapore sits 23% below it. On cases other than transfer pricing the gap is wider still, at 48.07 months against 21.34.

India’s position also moved in the wrong direction. In 2023 the same measure stood at 35.78 months for transfer pricing and 33.89 for other cases. One year later both had risen by roughly ten and fourteen months respectively. That is a deterioration on the OECD’s own numbers, and it happened while the inventory of cases was falling.

India’s biggest tax amnesty cleared under a tenth of the money

India has twice offered to settle disputed tax at a discount, in 2020 and again in 2024. The 2020 scheme has now been audited by the Comptroller and Auditor General. That audited result is the most reliable measure available of how much of the backlog can actually be settled. It records what was offered, what was taken up and what was collected.

Some 4.15 lakh cases covering ₹10.09 lakh crore were eligible. Applications came in for 1.32 lakh of them, covering ₹0.99 lakh crore, which is 9.8% of the money. Of the amount applied for, ₹71,924 crore was actually collected after interest and penalties were waived. The rest fell away. A scheme designed to clear the backlog therefore moved under a tenth of it.

A second scheme opened on 1 October 2024 on similar terms. No figure for declarations filed, disputes settled or revenue collected has been published in the two years since. No successor scheme was announced in the 2026 Budget. Any claim about what the 2024 scheme achieved should be treated as unsourced until the department publishes the result.

India pays 6% on refunds it delays and charges 12% on demands a business disputes

The asymmetry in interest rates is statutory, small and rarely mentioned. Where India holds money it should have refunded, it pays simple interest at half a per cent a month, which is 6% a year. Where a business has not paid a demand it is contesting, India charges 1% a month, which is 12% a year.

The volumes behind that are substantial. For the 2025-26 assessment year, 8.80 crore returns had been filed by early February 2026. Of those, 24.64 lakh had been pending processing for more than ninety days. The department reported better figures by July 2026, with about 1.47 lakh refund cases outstanding against 3.82 crore verified refund claims. No official figure exists for the rupee value of delayed refunds or for their age profile, which is itself worth noting.

For a business, the practical effect is that disputed tax is financed at an unfavourable spread in both directions. Money the department owes earns 6% a year. Money the business owes but disputes costs 12% a year. Neither rate reflects what the business actually pays for working capital, and neither is negotiable.

The new Income-tax Act left the appeal system almost exactly as it was

The Income-tax Act 2025 came into force on 1 April 2026 and replaced legislation that had stood since 1961. On dispute resolution it changed almost nothing of substance. The ladder runs from the assessing officer to the Commissioner, then the Appellate Tribunal, then the High Court and the Supreme Court, exactly as before. Sections were renumbered and powers were carried across.

Two points matter more than the renumbering. No statutory time limit was introduced for deciding a first appeal, which was the single most requested reform and did not happen. The transition provisions also keep every appeal pending on 1 April 2026 under the old Act. The entire backlog of 603,647 appeals is therefore litigated under legislation that has been replaced.

The Dispute Resolution Committee was carried over without change, including its thresholds. It is available only where the disputed variation is under ₹10 lakh and the returned income is under ₹50 lakh. Those limits keep it confined to very small matters, so it is not a route for a corporate dispute of any size.

Budget 2026 halved the deposit required to contest an assessment

The Finance Act 2026, which received assent on 30 March 2026, made one change that is commercially significant. The deposit required before appealing fell from 20% to 10% of the demand. It is now computed on the core tax rather than on the grossed-up amount. For a business contesting a large assessment, that halves the cash it must hand over to be heard.

Two smaller measures help as well. Assessment and penalty proceedings are now dealt with in a single common order rather than separately. The government justified that on the grounds that multiple proceedings hinder doing business. And no interest runs on a penalty amount while a first appeal is pending, whatever the eventual outcome.

The department has also become more selective about its own appeals. Monetary thresholds set in September 2024 stop it appealing below ₹60 lakh at the Tribunal. The limits are ₹2 crore at a High Court and ₹5 crore at the Supreme Court. The Supreme Court disposed of 573 direct tax cases that fell below the new limit, and about 6,000 departmental appeals were withdrawn in the 2025-26 financial year. The last official measurement of how often the department won was published in the Economic Survey of January 2018, on data from 2011 to 2015. It put the success rate below 30%.

An Indian advance ruling binds neither side. A Singapore ruling binds the revenue.

The cleanest contrast between the two systems is not about speed at all. It is about whether a business can fix its tax position in advance and rely on the answer. India and Singapore both offer a procedure for that, and only one of them produces a result anybody is bound by.

	India	Singapore
Appeals pending at the first stage	539,863 at 31 March 2025	Not published, and no backlog is reported
Time limit to decide a first objection or appeal	None in law	Six months, as a published service standard
Deposit required to contest an assessment	10% of the core tax demand, from 2026	None
Does an advance ruling bind the revenue authority	No	Yes
Months to close a cross-border transfer pricing case	45-7	23-7
What the revenue authority reports publicly	Disputed tax, when Parliament asks for it	An arrears rate of 0.66%

India replaced its Authority for Advance Rulings with a Board for Advance Rulings on 1 September 2021. The Board has two members, both serving income tax officers of senior rank, where the Authority included former judges. The decisive change came with it. A ruling of the Board binds neither the applicant nor the department, so a business can obtain one and still face the opposite position on assessment.

Whether the Board is functioning at all is now before a court. On 29 September 2026 the Delhi High Court directed the Union of India to file an affidavit disclosing the Board’s pendency and the status of member appointments. Counsel for the Union accepted difficulty in finding suitable candidates for the posts of Chairman and Vice-Chairman. No figure for applications pending or rulings issued has ever been published, which is why a court had to order the government to produce one.

Singapore runs the opposite arrangement. An advance ruling issued by the Inland Revenue Authority binds the Comptroller to apply the law in the manner set out in it. An objection to an assessment must be filed within two months, and the authority works to a published service standard of six months to decide it. Where a case is complex enough to need longer, the authority notifies an estimated completion date rather than leaving the taxpayer to wait.

India’s advance pricing programme is its success story, and it still takes three to four years

The fair comparison concedes what India does well. Its advance pricing agreement programme has grown steadily and now operates at record volume. It signed 174 agreements in the 2024-25 financial year and 219 in 2025-26, taking the cumulative total past a thousand. Singapore is among the countries with which India concludes bilateral agreements, so the route exists for exactly the structure this article describes.

The times are the qualification. A unilateral agreement took an average of 42.84 months to conclude in 2024-25, and a bilateral agreement 50.18 months. Both are slower than the global average of 39.6 months for advance pricing agreements. The 2026 Budget set a target of two years for a fast-tracked unilateral agreement for information technology services companies, which would roughly halve the current average. That is an enacted target rather than an achieved outcome, and it should be read as one.

Set against a Singapore advance ruling, which is measured in weeks and binds the revenue, the comparison is not close. But a business with substantial Indian transfer pricing exposure should use the Indian programme rather than avoid it. Certainty obtained in three years is still better than a dispute that runs for ten.

What this means for businesses with a demand outstanding, groups with cross-border transactions, and finance teams planning ahead

For a business with an Indian assessment outstanding, the arithmetic changed in March 2026 and is worth redoing. The deposit required to appeal is now 10% of the core tax rather than 20% of a larger base, which changes the cash cost of contesting. Against that, there is still no deadline by which the first appeal must be decided. The money deposited earns 6% a year if the business eventually wins. Deciding whether to contest or settle is a financing decision as much as a legal one.

For a group with cross-border transactions between India and the region, the mutual agreement procedure figures are the ones to plan around. On the most recent published data, a transfer pricing case takes an average of forty-six months to close. That average has been getting worse rather than better. The same transaction can often be arranged so the counterparty sits in a country that closes cases in half the time. The exposure is halved with it.

For a finance team planning ahead, the useful distinction is between countries where a position can be fixed in advance and countries where it cannot. Singapore’s binding advance ruling removes the question before it becomes a dispute. India’s non-binding ruling does not, and the body issuing them is currently explaining itself to a High Court. A business able to place a transaction on the side of the border where certainty is available should price that advantage properly. The alternative is measured in years rather than in fees.

The owner’s allowance is capped at US$250,000 a year and taxed at 20% on most uses. The company pays no such tax, but one late filing stops the next payment.

Two different sets of rules govern money leaving India, and they are constantly confused with each other. One applies to a person. The other applies to a company. Treating them as a single regime produces advice that is wrong in both directions, and it is the first thing a finance team should separate.

An individual resident in India may send up to US$250,000 abroad in a financial year under the Liberalised Remittance Scheme. Most of what that person sends is taxed at 20% as it leaves. An Indian company is not using that scheme at all, has no annual ceiling of that kind, and pays no such tax on its remittances. What the company faces instead is a sequence of filings before each payment. An enforcement lever then stops the next payment if any of them is late.

A Singapore company faces neither. Singapore abolished exchange control in 1978 and has not reintroduced it. There is no ceiling, no purpose code, no pre-payment certificate and no annual return on money held abroad. A Singapore company instructs its bank, the bank applies its own anti-money-laundering checks, and the money moves. That difference is the whole of this article.

A resident Indian may send US$250,000 abroad each year, and pays 20% on most of it

The remittance allowance has stood at US$250,000 per resident individual since 2015, and the 2026 Budget did not change it. The figure is denominated in United States dollars rather than rupees. Every resident individual holds one, including a minor, so a family of four adults has four allowances between them. One point is routinely misunderstood. The ceiling is a single pool covering both spending and investing, not a separate allowance for each.

The tax on top is where the real cost sits. India collects tax at source when the money leaves, at a rate that depends entirely on the stated purpose. The rates were last changed on 1 April 2026, when education and medical remittances came down from 5% to 2%. The rate that applies to almost everything a business owner does was left untouched at 20%.

Investing abroad, buying property abroad, making a gift and supporting a relative are all taxed at 20% above ₹10 lakh in a tax year. The threshold itself was raised from ₹7 lakh on 1 April 2025, which is a change many published guides still have not picked up. Education funded by a qualifying study loan is the only category that bears nothing at all. An overseas tour package is now taxed at 2% with no threshold whatsoever, so the first rupee is caught.

The 20% is recoverable, which makes it a financing cost rather than a tax

Tax collected at source is not a final charge. It is a prepayment, credited against the person’s income tax for the year and refunded if it exceeds what is owed. India also pays interest on a delayed refund. So the money is not lost, and any adviser who describes it as a 20% tax on sending money abroad is overstating the position.

What it does cost is time. Consider a remittance of ₹1 crore to fund an overseas investment. Tax of 20% applies to ₹90 lakh of that, which is ₹18 lakh collected at the moment of transfer. That sum is recovered when the return for the year is assessed, which can be a year or more later. At a cost of funds of 9%, carrying ₹18 lakh for twelve months costs about ₹1.62 lakh. On those assumptions the real cost is roughly 1.6% of the amount remitted, and it buys nothing.

That arithmetic is an illustration rather than a measurement, and the cost of funds is an assumption rather than a published figure. No Indian authority has ever quantified the working capital cost of the timing difference, so any number presented as official should be treated with suspicion. Two relieving measures notified in October 2024 help salaried remitters, by letting the credit be factored into payroll withholding instead of waiting for assessment.

An Indian company pays no remittance tax, but it files before every payment

The company position is different in kind. There is no ceiling on what an Indian company may pay abroad for goods, services or royalties, and no tax collected at source on those payments. What there is instead is a documentary sequence, and each step has to be completed before the bank will release the money.

The sequence begins with establishing whether the payment is taxable in India and at what rate. That requires a tax residence certificate from the recipient and a declaration of eligibility. Where the chargeable payments to non-residents exceed ₹5 lakh in a tax year, a practising chartered accountant must certify the rate in a prescribed form. The remitter then files its own declaration with the tax department electronically, before the payment is made.

Only then does the exchange control step begin. The company files a declaration with its authorised dealer bank, identifying the payment against a purpose code, with supporting documents. The withholding tax is deposited, a return is filed and a certificate is issued to the recipient. All of this precedes a transfer that a Singapore company would complete in a single banking instruction.

What has to happen before the money moves	India	Singapore
Establish whether tax is due, and at what rate	Required, with a residence certificate from the recipient	Not applicable
Accountant’s certificate on the rate	Required above ₹5 lakh of chargeable payments in that tax year	None
Declaration to the tax department before payment	Required, filed electronically	None
Declaration to the bank under exchange control	Required, against a purpose code, with documents	None
Deposit the withholding tax and file the return	Required	None
Annual return on money invested abroad	Required in every year, by the last day of December	None
The bank’s own anti-money-laundering checks	Required	Required
Can an earlier filing lapse block the next payment	Yes	No

Every form in that sequence was renumbered on 1 April 2026, when the Income-tax Act 2025 and the rules under it came into force. The pre-payment declaration and the accountant’s certificate now carry different numbers from the ones used for two decades. Nothing substantive changed, but every internal procedure, engagement letter and template that names the old forms is now out of date. That is an administrative cost with no corresponding benefit to anyone.

One late filing stops the next remittance

The enforcement mechanism attached to overseas investment is the part most finance teams underestimate, because it is not framed as a penalty. When an Indian investor puts money into a foreign company it files a form at the time of remittance. The bank then obtains a reference number for that investment. That number is frequently mistaken for a regulatory approval, and it is not one.

An annual performance report is then due for each foreign company, in every year, by the last day of December. It must be based on the foreign company’s audited accounts and certified by an auditor or a chartered accountant. A late filing can be regularised by paying a fee. That fee starts at ₹7,500 and rises with both the amount involved and the length of the delay, up to the full amount.

The real consequence is not the fee. Until the delay is regularised, the bank will not process any further remittance or financial commitment to that foreign company. A single missed annual report therefore stops the next funding round, the next capital injection and the next intercompany payment. Businesses discover this at the moment they most need the money to move, which is the worst possible time to find out.

Indians sent US$28.98 billion abroad last year, and the fastest-growing use was investment

The published totals do not support the idea that money is flooding out of India. Outward remittances by individuals peaked in the 2023-24 financial year and have fallen in each of the two years since. The 2025-26 total was US$28.98 billion, down 2% on the year before, which was itself down almost 7%.

What has changed is the composition rather than the total. Travel still accounts for the majority of the money, and spending on overseas education fell by 21% in a single year. The two categories that grew were investment in foreign equity and debt, up 56%, and the purchase of property abroad, up 64%.

Those two growing categories are precisely the ones taxed at 20%. The categories that received a rate cut in April 2026, education and medical treatment, are both shrinking. Whatever the policy intention, the relief landed on the uses that are declining and left the full rate on the uses that are rising.

India withholds tax when money leaves, and the treaty roughly halves it

Separately from all of the above, India taxes the payment itself in the recipient’s hands and collects it by withholding. The domestic rate on dividends, royalties and fees for technical services is 20%. Surcharge and cess push the effective rate to between 20.8% and 21.84%, depending on the recipient’s income band.

A Singapore recipient does materially better under the treaty, and the detail matters. Dividends are capped at 10%, but only where the beneficial owner is a company holding at least 25% of the shares. Everyone else pays 15%. Royalties and fees for technical services are both capped at 10%. The inclusion of technical service fees is a genuine advantage, and several of India’s other treaties do not offer it.

Interest is the exception that should be stated honestly. The treaty caps it at 10% only where the lender is a bank or a similar financial institution. Ordinary intercompany lending is capped at 15%, which is higher than Singapore’s own 15% domestic rate on interest paid out. The comparison is not one-sided on every line, and presenting it as though it were invites an easy rebuttal.

The treaty rate is not automatic, and India’s Supreme Court said so in January 2026

On 15 January 2026 the Supreme Court of India decided a long-running dispute involving an investment held through Mauritius and Singapore. It held that a tax residence certificate alone does not establish entitlement to treaty benefits, and that the authorities may examine where management and control actually sit. It also held that the general anti-avoidance rule can be applied to deny benefits that would otherwise have been grandfathered.

The practical consequence is straightforward. A Singapore company claiming the 10% dividend rate on Indian profits needs real decision-making in Singapore. It needs a genuine commercial reason to exist and directors who actually decide things. A certificate and a registered address no longer carry the point. Any structure sold on the treaty rate alone is now exposed to exactly the analysis the Court applied.

This cuts in the same direction as Singapore’s own rules. The conditions for a Singapore certificate of tax residence already require control and management to be exercised there. In practice that means a working executive director, or management by a Singapore firm. Both authorities are asking the same question, which makes the answer easier to give once and harder to fake.

India rewrote parts of its exchange control rulebook in 2026, and tightened one part

India has been consolidating its foreign exchange regulations through 2026, and the direction is genuinely towards simplification. New guarantee regulations notified in January 2026 replaced a set that had stood since 2000. Consolidated export and import regulations were notified in the same month. A draft set of foreign investment rules published on 21 July 2026 would replace the existing non-debt instrument rules entirely. The consultation on it closed on 31 August 2026, and the rules have not been notified since.

One change ran the other way, and it is worth knowing about. An amendment effective 1 October 2026 shortened the period allowed for bringing export proceeds back into India, from fifteen months to nine for goods and services. Exporters already on the caution list stay under the old timetable until they come off it. A business that routinely gives long credit to overseas customers now has less room than it had in September 2026.

The honest summary is that India is tidying its rulebook rather than dismantling it. The filings are being consolidated, not removed. The draft foreign investment rules are a draft, and treating them as current law would be a mistake. Nothing notified in 2026 changes the fact that an Indian company files before it pays and reports every year afterwards.

Singapore has had no exchange control since 1978

The comparison is short because there is almost nothing to describe. Singapore liberalised exchange control in 1978 and has not reinstated it. A Singapore company may move working capital to or from a group company in any currency, in any amount, in either direction, without permission from anyone. There is no annual ceiling, no purpose code and no pre-transfer certificate.

Nothing is withheld when a dividend leaves Singapore, to any shareholder, anywhere. That follows from the one-tier system in place since 1 January 2003, under which tax paid by the company is final. Interest paid to a non-resident is withheld at 15% and royalties at 10%, so Singapore is not a jurisdiction with no withholding at all. On dividends, which is what matters to an investor taking money out, the rate is nil.

One Singapore rule is regularly misquoted and should be put in its place. Physical currency and bearer instruments worth S$20,000 or more must be reported when carried across the border. That is a cash-smuggling control and it expressly excludes electronic and bank transfers. Anyone citing it as evidence that Singapore limits transfers to S$20,000 has read it backwards.

What this means for owners funding a business abroad, finance teams paying overseas suppliers, and groups holding cash in the region

For an owner funding a business abroad from personal money, the allowance and the 20% are both planning problems rather than obstacles. The allowance is per person and per financial year. Capitalising a foreign company properly is therefore a question of how many people and how many years, decided in advance. The 20% is recoverable, so the real cost is the carry on it until assessment. Both are worth modelling before the first transfer rather than after it.

For a finance team paying overseas suppliers, the filings are manageable and the reporting lapse is not. The forms renumbered on 1 April 2026, so every internal template and checklist needs reviewing against the current rules. The annual return on any overseas investment is the one deadline that carries a stop attached. It falls on the last day of December each year, regardless of how convenient that is. A calendar entry and a named owner inside the business are a cheaper control than an emergency regularisation.

For a group holding cash in the region, the question is where the money needs to be rather than where it happens to land. Money that sits in a Singapore company moves between regional entities without permission. It pays no Singapore withholding on its way to investors, and it is available at the speed of a bank instruction. Money routed through India for the same journey needs a filing before it leaves. It also needs a report in every year it stays out. That difference is operational before it is fiscal, and it compounds every time the business needs to move quickly.

India permits a founder-owned trading company and an Indian-company-owned subsidiary. It blocks the Singapore parent placed above an Indian business.

An Indian software company in Bengaluru sells to customers in Singapore, Jakarta and Dubai. Every invoice is raised in India and every payment is converted into rupees. Every contract is governed by Indian law and heard in an Indian court. The founder wants the regional side of that business to sit in a Singapore company instead, and her investors have asked for the same thing. Three advisers have told her this is difficult. Each of them described a structure she does not actually need.

India’s rules separate two things that are usually discussed as one. A resident Indian may own a Singapore company that carries on a genuine business of its own. An Indian company may own a Singapore subsidiary. What a resident Indian may not own is a Singapore company that in turn owns the Indian company he controls. The first two structures let Singapore receive regional income, keep it and eventually be sold. The third has been closed since August 2022.

That distinction is easy to miss, because both tests sit in one sentence of the rules. A resident Indian may invest in a foreign company only where it is a genuine operating business. It must also have no subsidiary under his control. Being a real trading company satisfies the first test. It does nothing at all for the second. A Singapore company with real regional customers and real regional staff still fails once the Indian company is placed underneath it.

Control is defined at a low threshold, which is why that second test catches almost every founder. A shareholder has control once he holds 10% or more of the votes. He also has it if he can appoint most of the directors. A founder who owns the majority of a Singapore parent clears that bar several times over.

A resident Indian may own a Singapore company directly, provided it owns nothing in India

This structure works because the Singapore company sits beside the Indian business rather than above it. The Singapore company signs its own regional customers and employs its own regional staff. It holds the cash those customers pay, and no Indian company sits underneath it. India does not object to a resident owning a business abroad. It objects only when that foreign business owns something back in India.

The Singapore company has to be a genuine operating business. That means a trading company, a services company or a distributor. A vehicle that only holds shares does not qualify, because the rules permit investment only into a company that actually carries on business. The requirement fits how most regional expansions are built in any case.

The money to establish the company travels under India’s personal remittance allowance. That allowance is US$250,000 for each resident individual in each financial year. Every resident adult holds a separate allowance, so two co-founders can commit twice that amount between them. The funds have to belong to the individual rather than being borrowed, and this matters where the Singapore company is a recognised start-up. Capitalising the company properly is therefore a question of planning across people and years, not of applying for permission.

The reporting is light in substance and unforgiving on timing. The bank files a form when the money leaves India and obtains a reference number for the investment. One annual return is then due at the end of December each year. That reference number is often mistaken for an approval, and it is not one. A late filing freezes further investment in the company until the reporting is brought up to date.

One limit on this structure can never be crossed. The Singapore company owned personally by the founder must not acquire the Indian company at any point. From the moment it does, the founder holds a foreign company with an Indian company beneath it. That is the exact arrangement the rules forbid. The directions also stop a founder who first invested without control from acquiring control later.

An Indian company may put up to four times its net worth into a Singapore subsidiary

Where the business has outgrown a personal allowance, the Indian company invests instead of the founder. An Indian company may commit up to four times its net worth to companies outside India, measured on its last audited balance sheet. The personal allowance does not apply to it. The ceiling is therefore set by the size of the Indian business rather than by a fixed figure. Reserve Bank approval is needed only above one billion United States dollars of commitment in a financial year. India also lets the foreign company invest back into India, up to two layers of subsidiaries deep, and a group holding one Singapore company sits well inside that limit.

This is the structure that suits a company with real regional revenue and a long future in India. The Singapore subsidiary signs the regional customers, holds the regional cash and employs the regional team. It pays a dividend up to the Indian parent whenever the group wants the money there. Singapore charges no withholding tax when that dividend leaves. India taxes it on arrival and gives credit for the Singapore tax already paid. The cash can therefore stay in Singapore for as long as the business needs it.

The trade-off is about who owns the value rather than about tax. The Singapore company sits beneath the Indian company, so selling the Singapore business is a sale made by the Indian company. That is fine where the plan is to build and sell one group. It is the wrong shape where foreign investors want to own the regional business directly. The question is far cheaper to settle before the structure is built than afterwards.

Regional customers can contract with the Singapore company and pay it directly

No money has to leave India for the Singapore company to be paid. A customer in Jakarta signs its contract with the Singapore company and pays the Singapore company. That payment never touches the Indian banking system. No Indian approval is involved and no Indian tax arises on the receipt. This is how Singapore comes to hold regional income, and it is simpler than most descriptions of it suggest.

What decides whether the arrangement survives scrutiny is where the work behind that revenue is done. If Singapore staff win the customer, serve the customer and support the customer, the profit belongs to the Singapore company. If the Bengaluru team does all three and Singapore only issues the invoice, India will say the profit was Indian throughout. India would be right to say it. The invoice follows the work, and no amount of paperwork reverses that order.

The structure is therefore a commercial decision before it is a tax decision. A regional sales team and a regional head who genuinely decides things are what make the revenue Singapore’s. Those are the same people the company would hire anyway once it is serious about the region.

Singapore taxes company profit at 17%, and charges nothing when money leaves

Singapore’s corporate tax rate is 17%. A standing exemption removes the first S$102,500 of chargeable income from tax each year. A company earning S$500,000 therefore pays an effective rate of about 13.5%, and one earning S$200,000 pays about 8.3%. A rebate of 50% of tax payable for the 2026 year of assessment, capped at S$40,000, was announced on 7 April 2026. That rebate sits in a bill before Parliament and had not passed into law by the end of September 2026.

One exemption that founders expect does not apply to a holding structure. The start-up exemption is more generous than the standing one and runs for a company’s first three years. It specifically excludes investment holding companies. A Singapore company that trades qualifies in the ordinary way, and a Singapore company that only owns shares does not. This is another reason the operating structure suits an early-stage business better.

Singapore charges no withholding tax on dividends paid out of the country. That applies to every shareholder, wherever they are resident. It follows from the one-tier tax system introduced on 1 January 2003, under which tax paid by the company is final. A foreign investor taking cash out of a Singapore company loses nothing at the Singapore border. Few competing jurisdictions offer that without attaching conditions.

Since 2024, the exemption on selling a foreign asset depends on having real staff in Singapore

Singapore does not tax capital gains, but a rule that took effect on 1 January 2024 qualifies that. A Singapore company that sells a foreign asset and brings the proceeds into Singapore is taxed on the gain. The charge applies unless the company has adequate economic substance in Singapore. Shares in an Indian company are a foreign asset, because the company issuing them is incorporated outside Singapore. The only realistic way out is to meet the substance test.

Economic substance has a specific meaning here, and it is not about documentation. The company’s operations must be managed and performed in Singapore. It must have adequate people and premises there to do so. Work may be outsourced to a third party or to another group company, provided that work is carried out in Singapore. The older safe harbour for selling a long-held shareholding does not rescue a company that fails this test. The major accounting firms all read the new rule as overriding it.

The conditions for a Singapore certificate of tax residence point the same way. A foreign-owned holding company qualifies only where control and management are exercised in Singapore. In practice that means a director based there in a genuine executive role, a senior executive based there, or management by a related Singapore company. A registered address and a nominee director satisfy neither test.

India taxes a share sale only where the company’s value comes mainly from India

This is where the two permitted structures produce genuinely different outcomes. India treats shares in a foreign company as Indian property where that company derives its value substantially from Indian assets. Two conditions must both be met. The Indian assets must exceed ₹10 crore in value, and they must represent at least half the value of everything the company owns. A Singapore company placed above an Indian business meets both without any effort.

A Singapore company that earns its value from the region fails both tests, and India has no claim on the sale of its shares. Where the worth is built on regional customers, regional contracts and regional staff, the Indian assets are neither large enough nor proportionally significant enough. That is the commercial case for building the regional business inside Singapore from the start. The value created in Singapore stays outside India’s reach for as long as it is genuinely created there.

The founder’s own position is a separate question. An individual resident in India is taxed on gains arising anywhere in the world. A gain on selling his Singapore shares is therefore taxable in India for as long as he lives there. If he has genuinely ceased to be resident, that charge does not arise. Residence turns on where a person actually lives and intends to live, not on the visa he holds. A paper change of residence comes apart years later, when a buyer examines it.

A resident founder cannot own a Singapore company that owns the Indian one

The blocked structure is the one most founders ask about, so it is worth setting out why it fails. It fails on the investment rules first, for the reasons given at the start of this article. A resident Indian may invest only in a foreign company with no subsidiary under his control. Shares in an unlisted Singapore company count as direct investment however small the stake, and control begins at 10% of the votes. A Singapore parent above an Indian business has a subsidiary from the day it is formed.

The rules also require shares acquired outside those limits to be sold within six months, so a structure assembled in breach cannot be left in place and explained later. The swap exception does not assist. It permits an exchange of shares only on a merger, a demerger or a liquidation, and a commercial exchange between a founder and an investor is none of those.

The structure fails on tax as well, and it failed there several years earlier. India and Singapore amended the capital gains article of their treaty by a protocol signed on 30 December 2016, in force on 27 February 2017. Before that amendment, a Singapore company could sell shares in an Indian company without paying tax in either country. The treaty reserved the gain to Singapore, and Singapore does not tax capital gains. India now taxes that gain at its full domestic rate.

Shares acquired before 1 April 2017 kept the old treatment and stay outside the Indian charge. Shares acquired in the two years from 1 April 2017 attracted half of India’s domestic rate under a transitional rule. Shares acquired from 1 April 2019 attract the full rate of 12.5%, with surcharge and cess applied on top. A company formed in 2026 sits entirely inside that third period. The protection founders are still being sold has been worth nothing for seven years.

Transferring the Indian shares up to a foreign parent would also carry an immediate cost. Handing Indian shares to a foreign company in exchange for its shares is a sale for Indian tax purposes. India computes the tax on market value rather than on what was originally paid. The founder receives shares rather than cash, but the tax has to be settled in cash in the year of the swap. That bill falls due before any new investment has been drawn against it.

The holding period decides how large that cost would be. Unlisted shares become long-term assets after 24 months. Before that point a resident individual is taxed at 39.00% and a non-resident company at 38.22%. After it, the same parties are taxed at 14.95% and 13.65%. The difference is close to three times the tax on identical value, which makes the timing of any restructuring more consequential than the choice of structure.

Moving a holding company back to India has cost two well-known companies published amounts

Several Indian companies have unwound offshore holding structures, and the cost is public. PhonePe’s return from Singapore in 2023 carried a reported tax charge of about ₹7,800 crore. Its investors bore that charge on the increase in value since their original investment. Reuters reported the figure on 5 January 2023 and the parent confirmed it in substance. Groww’s move back from the United States carried a reported ₹1,340 crore, though published accounts disagree about whether India or the United States collected it.

India has simplified the return journey three times since 2024 without reducing its cost. Rules notified in September 2024 removed the National Company Law Tribunal from the process. Rules notified in September 2025 widened the range of companies entitled to use that faster route. An amendment on 29 May 2026 aligned the exchange control paperwork with it.

None of those changes touched the approvals or the tax. The application still goes to the Central Government through the office of the Regional Director. Prior approval of the Reserve Bank of India remains expressly required for both companies. No tax relief has been introduced. The cost is therefore calculated on what the business is worth at the time of the move, not on what it was worth when the structure was built.

Delaware taxes the exit gain at 21%, and Mauritius lost its treaty advantage in 2016

Both jurisdictions are proposed constantly, usually by advisers who have not revisited either position since 2016. Delaware remains the default for founders raising American capital, and Mauritius is still presented as the traditional route into India. Neither improves on Singapore for a company whose operations are in Asia. The table below compares the three on the tests that decide the outcome.

Delaware is chosen for reasons that have nothing to do with tax, and those reasons are sound. Its Division of Corporations reported 2,287,728 registered entities at the end of 2025 and 334,461 new formations during that year. Settled case law and standard venture documents let a financing round be papered within days. On tax the position is worse than Singapore’s. A 21% federal charge falls on the exit gain, and the relief American groups rely on produces almost nothing where the Indian subsidiary has no accumulated profits.

Mauritius is the most frequently misdescribed of the three. Its treaty with India was amended by a protocol signed on 10 May 2016, on the same timetable and in the same terms as the Singapore protocol. Gains on Indian shares acquired from 1 April 2017 became taxable in India, at the full rate since 1 April 2019. The headline rate of 15% also understates the burden, because a 2% social responsibility levy, a 2% climate levy and a 5% contribution are charged alongside it.

India led the world on the number of listings in 2026 and raised a fraction of the money

Indian companies completed more initial public offerings than any other market in the first half of 2026. The amount raised tells a different story from the count. EY recorded 102 Indian listings raising about US$4 billion, against 85 American listings raising about US$128 billion. India leads the world on the number of transactions and sits far behind on their size. A company planning a very large raise should weigh the second figure rather than the first.

A company intending to sell its regional arm should note a separate point about buyers. The acquirers of an Asian business are mostly in Asia, and they prefer to buy a company they can own and run directly. A Singapore company holding the regional business is straightforward for such a buyer to acquire, and a regional business sitting inside an Indian company is not. That often matters more to the outcome than the tax comparison does.

One widely quoted statistic deserves to be retired. The only sourced study of where Indian companies place their holding companies used data as at 31 March 2022. It found 34 of 90 Indian unicorns registered outside India, with 25 in the United States and 8 in Singapore. Nothing comparable has been published since. Every percentage now circulating comes from firms selling incorporation services.

What this means for founders building regionally, founders already offshore, and investors pricing a round

For a founder building a regional business while resident in India, Singapore has always been available. The only real question is which of the two permitted structures fits. A founder who may move abroad should own the Singapore company personally and keep it separate from the Indian one. The same applies where investors want to own the regional business directly. A company with established regional revenue should let the Indian company invest instead. The choice is far cheaper to make at the outset than to correct later.

For a founder whose structure is already offshore, three questions deserve an answer this quarter. The first is where the key management and commercial decisions are actually taken, because that decides whether the foreign company is Indian for tax. The second is whether there is enough genuine substance in Singapore to keep a future sale of a foreign asset outside the 2024 charge. The third is what it would cost to unwind the structure today. That figure is priced on what the business is worth now, not on what it cost to build, and the published examples show how quickly it grows.

For an investor pricing a round into an Indian business held through Singapore, the question that matters most is where the Singapore company’s value comes from. If its worth comes mainly from the Indian business, India will tax the sale of its shares whoever sells them and wherever they live. If its worth comes from regional customers and regional staff, India has no claim on that sale. That single question deserves more attention in diligence than the structure chart receives. The answer is found in the revenue breakdown rather than in the legal file.

Active business income escapes annual attribution. The bill arrives on repatriation, at full marginal rates.

Consider a Canadian resident in his forties. He owns a fast-moving consumer goods wholesale and distribution business in Ontario. He buys branded household and personal care lines from third-party suppliers, and sells them to retailers and regional distributors. Demand from Southeast Asia has grown to the point where he wants a regional base. He is weighing Singapore against Dubai and Hong Kong. The scenario that follows is a composite rather than a client.

The question he asks first is whether a Singapore company lowers what he personally pays. The short answer is that it postpones Canadian tax rather than reducing it. Under one common structure it increases the total. The reason sits in two provisions of Canada’s Income Tax Act that most offshore marketing material leaves out.

What Canada taxes, and at what rate

Canada taxes its residents on worldwide income. Residence is a question of fact, weighed on the whole of a person’s circumstances. The Canada Revenue Agency treats three ties as almost always significant. Those are a dwelling in Canada, a spouse or common-law partner in Canada, and dependants in Canada. Secondary ties are assessed together rather than one by one. Separately, an individual who is not otherwise resident but spends 183 days or more in Canada in a calendar year is deemed resident for that whole year.

For 2026 the federal rate runs from 14% on the first C$58,523 of taxable income to 33% above C$258,482. Provincial tax sits on top of that. The combined top marginal rate on ordinary income is 53.53% in Ontario, 53.50% in British Columbia and 53.31% in Quebec. Alberta stops at 48.00%. Newfoundland and Labrador reaches 54.80%, but only above C$1,141,276. That makes it the wrong comparison for most owner-managers.

One widely repeated figure is now out of date. The proposed increase in the capital gains inclusion rate from one-half to two-thirds was deferred in January 2025. It was then cancelled on 21 March 2025. The measure was never enacted, so no repealing legislation was needed. The inclusion rate for 2026 is one-half. That puts the top capital gains rate at 26.76% in Ontario, exactly half the ordinary-income rate.

Reducing the Canadian bill before any offshore structure

Most of the available reduction sits inside Canada, and it is unglamorous. The small business deduction taxes the first C$500,000 of active business income at a combined 12.2% in Ontario, for a year ending before 1 July 2026. Ontario cut its lower rate on that date. The combined rate falls to roughly 11.7% blended for calendar 2026, and 11.2% from 2027. Two grinds erode the deduction. The limit reduces on a straight line where taxable capital employed in Canada exceeds C$10 million, and disappears at C$50 million. It also falls by C$5 for every C$1 of adjusted aggregate investment income above C$50,000, reaching nil at C$150,000.

The lifetime capital gains exemption is the largest single item. It stands at C$1,275,000 for 2026. It had been fixed at C$1.25 million for dispositions from 25 June 2024, before indexation resumed. It applies to qualified small business corporation shares, and it is not automatic. At disposition, 90% or more of the company’s asset value must be used principally in an active business carried on primarily in Canada. Throughout the preceding 24 months, more than half must have been. A distributor holding surplus cash, guaranteed investment certificates or non-operating property can fail the first test. The 24-month look-back means cleaning that up takes planning rather than a signature.

Two related points. The Canadian Entrepreneurs’ Incentive announced in Budget 2024 is not proceeding. Budget 2025 confirmed that by omission on 4 November 2025, so it should not appear in any plan. Separately, the tax on split income restricts paying dividends to family members at their own rates. The excluded shares route out of it is more available to this business than to a professional practice. It requires an individual aged 25 or over to hold, directly, 10% or more of both the votes and the value. The corporation must earn less than 90% of its income from services, and must not be a professional corporation. A goods wholesaler meets those last two conditions. Shares held through a family trust or a holding company fail the first.

Why the Singapore company defers rather than reduces

A Singapore company wholly owned by a Canadian resident is a controlled foreign affiliate. Its income is tested against the foreign accrual property income rules. Those rules attribute certain income to the Canadian shareholder each year, whether or not it is distributed. Income from a genuine active business is not in that base at all. It is excluded structurally rather than by exemption, so a real distribution business earns real deferral. Passive income is treated differently. Interest on the Singapore company’s cash balances, and foreign exchange gains on non-business holdings, are attributed annually. That point is routinely missed.

The deferral ends at repatriation, and that is where the structure disappoints. A dividend from a non-resident corporation to a Canadian individual is included in income at 100%, as ordinary income. There is no gross-up and no dividend tax credit. Both apply only to dividends from corporations resident in Canada. There is no deduction for the underlying Singapore corporate tax either, because the mechanism that provides one is open only to a corporation resident in Canada. And Singapore does not withhold on dividends, so there is no foreign tax credit to claim.

The arithmetic follows. C$100 of profit taxed in Singapore at 17% leaves C$83. Paid straight to the owner and taxed at Ontario’s 53.53%, it leaves C$38.57. The same C$100 earned in Canada at the small business rate and then distributed leaves C$46.15. At the general rate it leaves C$44.58. On repatriated profits the Singapore company is the worst of the three outcomes, not the best.

What a Canadian holding company changes

The provision that fixes this is section 113 of the Income Tax Act. It is restricted on its face to a corporation resident in Canada. Active business earnings of a foreign affiliate accumulate as exempt surplus, provided the affiliate is resident in a country with which Canada has a comprehensive tax treaty. Singapore qualifies, on a convention signed on 6 March 1976. A dividend paid out of that exempt surplus to a Canadian holding company attracts a full offsetting deduction. It therefore arrives free of Canadian corporate tax.

Inserting that holding company lifts the same C$100 from C$38.57 to C$43.37 in the owner’s hands. It also preserves the deferral until he chooses to take money personally. It still does not beat earning the profit in Canada on tax alone. What it does is let the Singapore company stand on its commercial merits, without an avoidable tax penalty attached. The difference between the two Singapore routes is roughly 12% of after-tax cash. It comes entirely from the presence or absence of a Canadian company in the chain.

The three ways the structure fails outright

The first failure is residence. A company incorporated outside Canada is resident in Canada if its central management and control is exercised there. Canada’s deemed-residence rule for corporations covers only companies incorporated in Canada. A Singapore company is therefore tested at common law, and the authorities look at where decisions are actually taken rather than where board minutes say they were. A sole shareholder who is also the controlling director, deciding from Ontario, produces a Canadian-resident company. The Canada–Singapore treaty does not rescue this. Its tie-breaker for companies asks the two tax authorities to try to agree, rather than deciding the question. Singapore applies the same control and management test for its own residence, so the same facts can also cost the company its Singapore treaty access.

The second is the supply chain. A specific rule recharacterises income from the sale of property as attributed income. It bites where the cost of that property is relevant in computing income from a business carried on in Canada, and where the goods were not produced in the affiliate’s own country. A safe harbour applies where more than 90% of the affiliate’s sales revenue comes from arm’s-length customers. A Singapore company buying from unrelated suppliers and selling to unrelated Asian customers, with no Canadian operating company on either side, is generally outside the rule. A Singapore company that buys from or sells to the Canadian business is the classic failure case. It also brings Canada’s transfer pricing rules with it. Those rules were rewritten for taxation years beginning after 4 November 2025, and the deadline to produce documentation on request fell from three months to 30 days.

The third is reporting. Shares of the Singapore company are specified foreign property. The exclusion for property used in an active business does not cover them. A foreign affiliate information return is due within 10 months of the year end. Penalties run to C$2,500 for simple failure, and to C$12,000 where the failure is knowing or grossly negligent. Those double where a demand to file has been served. The general anti-avoidance rule was also tightened for transactions after 2023. It now carries a 25% penalty on the denied benefit, and an economic substance test that counts against structures with little real activity.

Singapore, Dubai and Hong Kong compared

The comparison is closer than it is usually presented, and Singapore does not win every line. Two of the three offer a lower headline rate. One of the three charges no tax on goods at the border and no consumption tax at all. What separates them is less the rate than what has to be true, and stay true, for that rate to hold.

Where the Singapore case is weakest

On the Canadian treaty, Singapore is the worst of the three. The convention dates from 1976. It caps dividends at a flat 15%, with no reduced rate for a substantial shareholding. Both the UAE and Hong Kong treaties give 5% to a company controlling 10% or more of the voting power. Both also give 10% rather than 15% on interest and royalties. Singapore levies no dividend withholding tax of its own, so the practical bite falls on Canadian-source payments. The treaty text is still what it is.

The UAE headline of 9% is genuinely lower, and the free zone rate of 0% is real. It is much harder to reach than the marketing suggests. Distribution qualifies only where it is conducted in or from a Designated Zone. That is a fenced, customs-controlled area on a specific Cabinet list. Goods entering the country must also be imported through that zone. Most free zones sold to foreign owners are not Designated Zones. Finished consumer goods are not qualifying commodities either, so that separate route is closed. And non-qualifying revenue above the lower of AED 5 million or 5% of turnover costs the status entirely, rather than taxing the excess.

Hong Kong’s difficulty is different and sharper. Its territorial system exempts foreign-sourced profits. But for trading profits the tax authority’s practice is all or nothing. Where either the purchase or the sale contract is effected in Hong Kong, the presumption is full taxability, and apportionment does not apply to trading profits. A distributor with a Hong Kong office, Hong Kong staff taking orders and Hong Kong trade finance will struggle to keep an offshore claim standing. The substance needed to sustain that claim also pulls against the substance needed to obtain a Hong Kong residence certificate for treaty purposes.

What Singapore actually offers a goods business

Singapore’s advantage is not the headline rate. It is that the rate is predictable, and requires no claim to be sustained year after year. A distributor in Singapore pays a known percentage and files a return. A distributor relying on a Hong Kong offshore claim, or on qualifying free zone status in the UAE, is defending a position annually. The cost of losing that argument is the whole benefit, not a portion of it.

Partial tax exemption removes S$102,500 of chargeable income. That puts the effective rate at 8.3% on S$200,000 of profit, 13.5% on S$500,000 and 15.3% on S$1 million. A 50% corporate tax rebate for the 2026 year of assessment reduces it further. That rebate is capped at S$40,000 and was announced on 7 April 2026. A newly incorporated company meeting the shareholder conditions can also claim a larger start-up exemption for its first three years.

On the trade side, Singapore is the only one of the three that shares an agreement with Canada. Both are original parties to the Comprehensive and Progressive Agreement for Trans-Pacific Partnership, in force for both since 30 December 2018. Singapore is also a member of the Regional Comprehensive Economic Partnership. Hong Kong applied to join that agreement in January 2022, and an accession working group was approved on 21 September 2026. That is a procedural step rather than membership. Canada and the UAE concluded negotiations on a partnership agreement on 24 July 2026, with no signature or implementation date announced.

One caution on trade agreements. Preferential treatment attaches to goods that meet the rules of origin, not to the company that invoices them. A Singapore entity buying third-country goods and reselling them does not make those goods originating. The value lies elsewhere. It sits in preferential access for Canadian-origin and Singapore-origin goods, and in the regional cumulation rules that Singapore can use and Hong Kong currently cannot.

On banking, no jurisdiction publishes verifiable data on account opening outcomes for foreign-owned trading companies. That applies to the Monetary Authority of Singapore, the Hong Kong Monetary Authority and the UAE central bank alike. None publishes approval rates, rejection rates or timelines. Every specific claim of that kind in circulation traces back to marketing material with no published basis, and should be discounted accordingly.

What this means for owners staying in Canada, owners considering a move, and owners with an offshore company already

For an owner who intends to remain Canadian resident, the Singapore company should be justified commercially rather than fiscally. It earns deferral on genuine active business income for as long as profits stay offshore. That is worth having when the business is reinvesting. It does not reduce the eventual personal bill. Paid directly to the individual, it costs more than earning the same profit in Canada. If the company is going ahead, a Canadian holding company between the owner and the Singapore entity is what makes the arithmetic behave. That decision is cheapest to take at the outset, rather than by reorganising later.

For an owner considering leaving Canada, the departure charge is the number to establish first. Ceasing Canadian residence triggers a deemed disposition of most capital property at fair market value. The shares of the Canadian operating company are caught by it. Canadian real property and business property held through a Canadian permanent establishment are excluded, as are registered plans. Payment can be deferred without interest by election. Security is required where the federal tax on the deemed disposition exceeds C$16,500. That figure should be quantified before any move is planned. It is usually the largest single item, and it does not go away by leaving quietly.

For an owner who already runs goods through an offshore company, three checks are worth making this quarter. Establish where board decisions are actually taken, and whether the record supports it, because the residence test looks at conduct rather than paperwork. Establish whether any Canadian entity sits on either side of the offshore company in the supply chain. That is what turns trading profit into attributed income, and it brings transfer pricing documentation with a 30-day production deadline. And confirm the foreign affiliate and foreign property returns have been filed. Those penalties accrue whether or not any tax was ever owing.

Projected flows and measured flows are different things, and the gap is worth understanding before the next upgrade.

FTSE Russell reclassified Vietnam from Frontier to Secondary Emerging with effect from 21 September 2026. The change adds 27 Vietnamese stocks to its All-Cap index and six to its All-World index. Brokerages forecast inflows of between US$1.5 billion and US$4.4 billion. In the week to 18 September 2026, the last full week before the change took effect, foreign investors bought a net US$54.2 million of Vietnamese equities, according to Bloomberg. That was the largest weekly tally in about six weeks.

An index reclassification is the cleanest capital markets event available for this kind of comparison. The date is fixed well in advance. The constituent list is published. The buying by index-tracking funds is mechanical rather than discretionary. If forecasts and outcomes diverge this far under those conditions, they diverge further in every messier situation, which includes most private fundraising.

Foreign selling shrank to almost nothing before the upgrade

The more useful series is not the week of buying but the five months before it. Foreign investors were net sellers of Vietnamese equities in every month from May 2026. The scale of that selling then collapsed as the upgrade approached. The pattern is what a market looks like when it stops losing money rather than when it starts gaining it.

Net selling ran at 18,683 billion dong in May 2026. It fell to 14,276 billion in June 2026, 11,373 billion in July 2026 and 2,252 billion in August 2026. September 2026 to the 18th was still net negative, at 236 billion dong. Foreign investors have taken US$3.5 billion out of the market so far in 2026, after US$4.8 billion in the whole of 2025. Three consecutive net-buying sessions from 16 to 18 September 2026 totalled about 3,006 billion dong. On 18 September 2026 alone, 21 of the 27 stocks in the FTSE basket were net bought, worth 1,717 billion dong.

The reclassification did not reverse a year of selling. It stopped it. That is a smaller claim than the forecasts implied, and a more accurate description of what index inclusion does in practice. Inclusion changes who is obliged to hold a market. It does not by itself change what discretionary investors think of it, and the discretionary money is the larger pool.

What was forecast and what was measured

The forecasts published before the event span a wide range. They also measure different things from each other, which is part of why the range is so wide. Some estimate mechanical index buying, some estimate total foreign appetite, and one is a single firm describing its own intention.

SSI Securities estimated about US$240 million of passive net purchases in the first tranche. Its base case for the full implementation was about US$2.4 billion, with an optimistic case above US$4.4 billion. Vanguard’s head of investment management for Asia Pacific said on 18 September 2026 that the firm expects to invest US$2.5 billion in Vietnamese equities over the following year. Only the US$54.2 million is a measurement. The others are projections, or in Vanguard’s case a stated intention that has not yet been executed.

Two cautions apply to the measurement itself. Vietnamese exchange data for the three sessions to 18 September 2026 imply a larger figure than the weekly dollar total. That points to a different venue scope or cut-off between the two sources, and both should be quoted with their source attached. The full implementation is also phased across several tranches into 2027, so a single week measures the first instalment rather than the total. The honest reading is that the early evidence sits at the low end of the forecast range.

Southeast Asian startup funding rebounded to US$1.34 billion in August 2026

The private markets in the region moved over the same period, and the monthly series continues to swing violently. Three consecutive months have produced three completely different pictures of the same market. Each of them was reported as a trend at the time.

Startups in the region raised about US$1.34 billion in August 2026. That compares with roughly US$698 million in July 2026 and US$4.22 billion in June 2026. The three months vary by a factor of six. A founder benchmarking a raise against any one of them is using a number that will not describe the following month, and did not describe the preceding one either.

Four deals took 87% of the August total

The composition is the part that decides whether the rebound is relevant to any particular company. A total is made up of transactions, and transactions have sizes. When a few of them are very large, the total stops describing the typical case.

Four megadeals accounted for about 87% of the August 2026 total. Every other transaction in the region shared the remaining 13%. On a US$1.34 billion total, that is about US$174 million. That smaller figure is the pool from which an ordinary round is funded. It is the same pattern recorded in June 2026, when five deals took 93% of a much larger month. It is also the pattern visible in the region’s artificial intelligence funding across 2026.

One signal from outside the region is worth noting for anyone considering private credit as an alternative to equity. Morgan Stanley capped withdrawals from its North Haven Private Income Fund for a third consecutive quarter on 18 September 2026. Investors tendered 11.4% of shares against a 5% repurchase limit, with an expected effect on net asset value of about US$101 million. That is a United States wealth-management vehicle rather than an Asian one, and it says nothing directly about Asian credit supply. It does indicate that retail appetite for non-traded private credit is being tested, and vehicles of that kind are part of what ultimately funds credit managers in this region.

What this means for companies weighing a listing, founders raising now, and anyone quoting a forecast

For a company weighing a listing or an index-driven strategy, the Vietnam sequence is the best available evidence on what reclassification delivers. It stopped a run of selling that had lasted five months. It produced modest buying in the week before the effective date. It did not produce billions in a week, and the remaining tranches run into 2027. A business planning around a valuation uplift from index flows should build on the first-tranche estimate rather than the full-rollout case. The first tranche is the only part that has now been partly observed.

For a founder raising in Southeast Asia, the August 2026 rebound does not change the arithmetic. The relevant number is the 13% that was not taken by four deals, which works out at roughly US$174 million across every other company in the region for the month. Three months at US$4.22 billion, US$698 million and US$1.34 billion describe a market whose monthly totals are set by a handful of transactions. Plan on the deal count and the size distribution instead, and treat the monthly headline as noise.

For anyone quoting a forecast in a board paper or an investment memorandum, this episode is a clean test case. A brokerage range of US$1.5 billion to US$4.4 billion was published for an event with a fixed date, a published constituent list and mechanical buying. The first measurable week came in at US$54.2 million. Forecasts of this kind are estimates of a mechanical maximum, not predictions of behaviour, and they should be labelled as such wherever they appear. The figure that survives scrutiny is always the one with a measurement date attached to it.

Hong Kong proposed a 5% treasury rate. The Netherlands, Panama and Australia moved on migration, residence and trusts.

The Global Financial Centres Index published its fortieth edition on 16 September 2026. New York, London, Hong Kong and Singapore held the top four places, in that order, exactly as they did in March 2026. All four lost rating points. Four governments moved in the same week. Hong Kong proposed a 5% profits tax rate for corporate treasury centres. The Netherlands proposed a valuation step-up for companies moving in. Panama doubled one of its residence thresholds, and Australia closed consultation on a 30% minimum tax for discretionary trusts.

The ranking is the least informative of those five events. It measures perception through a survey of practitioners. The other four change what a structure costs, or where it can be put, and three of them carry dates that can be planned around. The ranking gets the coverage. The rule changes get the consequences.

The top four held their ranks and lost their ratings

The index is compiled by the Z/Yen Group with the China Development Institute. The fortieth edition researched 139 centres and placed 117 of them in the main index. It drew on 39,531 assessments from 6,147 respondents. Each centre is scored on a thousand-point scale, combining that survey with a set of instrumental factors such as infrastructure and regulation.

New York rated 761, London 757, Hong Kong 756 and Singapore 755. Shanghai, Tokyo, Seoul, Shenzhen, Dubai and Zurich followed in single-point steps down to 749. Twelve points separate first place from tenth. Singapore ranked second worldwide for finance and fifth for investment management. It ranked seventh for banking and seventh for trading. Its position therefore varies by four places depending on which activity is measured, which the single headline rank conceals.

Against March 2026, New York fell 6 points. London, Hong Kong and Singapore each fell 9. No rank changed. A fall across an entire leading group usually reflects a shift in the survey population or in sentiment, not movement between centres. The index is a perception measure. It is not a measurement of assets or of capital flows, and a nine-point move on a thousand-point scale should not be read as money leaving.

Hong Kong proposed a 5% rate for treasury and sourcing centres

Hong Kong’s Chief Executive delivered the 2026 Policy Address on 16 September 2026, the same day the index appeared. It commits the government to legislate a preferential profits tax rate for selected enterprises in key sectors. The rate would be 5%, or half the standard rate at 8.25%, for companies establishing corporate treasury centres and sourcing centres in Hong Kong. A corporate treasury centre is the entity that manages cash, funding and currency risk for a group.

Hong Kong’s standard profits tax is 16.5%, with the first HK$2 million of assessable profits taxed at 8.25%. A 5% rate would sit well below both, and below Singapore’s 17% headline corporate rate. The address set out three further commodity measures. A central clearing and settlement system for gold is to launch in the first quarter of 2027. A renminbi-denominated, physically settled gold futures contract is to be introduced at the exchange later in 2026. And a half-rate concession for eligible physical commodity trading is to go before the legislature in 2027.

All of these are announced rather than enacted, and each requires a bill that has not been introduced. The address also referred to preferential regimes for funds, single family offices and carried interest. Those were the subject of a separate bill gazetted on 12 June 2026. That bill was still before the legislature as at 20 September 2026, with no record of passage. Its concessions are therefore proposed rather than available. If it passes, the government has said they would apply from the 2025/26 year of assessment.

The Netherlands proposed a step-up for companies that move in

The Dutch government presented its 2027 Tax Plan on 15 September 2026. The measure that matters to holding structures is a step-up in Box 2, the part of the Dutch system that taxes substantial shareholdings. Where a company moves its place of effective management to the Netherlands, the acquisition price of its shares would be reset. The new price would be their fair market value on the date the company becomes Dutch tax resident.

The effect is that only gains accruing after migration fall into the Dutch charge. Gains built up while the company was resident elsewhere do not. That removes one of the standard objections to moving a holding company into a treaty jurisdiction. The objection is that the move itself imports a historic gain into a new tax net. The same plan holds the Box 3 rate at 36% for 2027, with tax-free capital of EUR 60,098 per person, up from EUR 59,357 in 2026. A replacement system taxing real returns is proposed for 1 January 2028 and remains a bill.

Panama doubled one of its residence thresholds

Panama replaced its qualified investor residency regime by Executive Decree No. 17 of 8 September 2026. It came into force on 16 September 2026, superseding the decree that had governed the programme since 2020. The regime grants residence in exchange for a qualifying investment, and the thresholds are the whole of it.

Property bought on first sale still qualifies at US$300,000. Property bought on the resale market now requires US$500,000, double the previous threshold. A securities market investment requires US$500,000 held for five years. A fixed-term bank deposit requires US$500,000 in a state bank, or US$750,000 in a private one. The decree also tightens valuation and appraisal requirements. It allows an applicant to switch between investment types within the five-year period, and permits dependants to be added after approval.

Australia closed consultation on a 30% trust tax on 18 September 2026

Australian Treasury released exposure draft legislation on 3 September 2026 for a minimum tax of 30% at trustee level on discretionary trusts, and consultation closed on 18 September 2026. The measure was announced in the 2026-27 Australian Budget and would start on 1 July 2028.

Non-corporate beneficiaries presently entitled to trust income would receive a non-refundable credit for tax paid at trustee level. Fixed trusts, managed investment trusts, widely held trusts, deceased estates and genuine testamentary trusts are all excluded. Trusts in existence on 1 July 2028 may elect to nominate fixed percentage entitlements and fall outside the charge. A temporary restructure rollover runs from 1 July 2027 to 30 June 2030. It allows assets to be moved out of discretionary trusts into companies or fixed trusts without an immediate tax charge. The measure is a draft and not law.

What this means for families in Hong Kong, families holding through trusts, and anyone choosing on rankings

For a family with Hong Kong exposure, the treasury centre proposal is worth tracking and not yet worth acting on. A 5% rate on treasury and sourcing activity would be among the lowest available anywhere for that function. It signals clearly where Hong Kong intends to compete. But it requires legislation that has not been introduced. The separate bill on funds, single family offices and carried interest has been before the legislature since June 2026 without passing. Structures should be built to work at current rates, with any concession treated as upside rather than as the basis of the plan.

For a family holding assets through a discretionary trust with Australian connections, the timetable is now visible and long. The charge would begin on 1 July 2028. The election for existing trusts attaches to trusts in place on that date. The restructure rollover opens a year earlier, on 1 July 2027. That sequence rewards deciding early, because the rollover window closes on 30 June 2030 and the election depends on a structure existing before the start date. Families with no Australian connection should still note the direction. A minimum tax at trustee level on discretionary trusts is the kind of measure that gets copied.

For anyone choosing a jurisdiction from a league table, the fortieth index shows the limits of that approach. Twelve points separate the first centre from the tenth. The top four have not changed order, and all four moved down together. Nothing in that tells a family where to put a holding company. The four rule changes in the same week do, because each one alters a specific cost, threshold or date. Singapore’s own position varies from second to seventh depending on which activity is measured. That is the most useful line in the report and the one least likely to be quoted.