
International tax structuring after the Mauritius cleanup: what works in 2026
The 2016 India-Mauritius Protocol ended the capital gains exemption. The MLI added the Principal Purpose Test. GAAR matured. The structures that genuinely work in 2026 are narrower and harder to set up than the structures founders read about in 2014.
India-Mauritius was the dominant inbound route into Indian equity for two decades. The 2016 Protocol to the DTAA phased out the capital gains exemption with effect from 1 April 2017. The grandfathering for investments made before then has now largely run out. The MLI brought in the Principal Purpose Test in 2020. GAAR matured into operational use by the tax department from 2018 onwards.
Indian-bound structures are still being set up. They are different from the structures of a decade ago. The choices that survive a 2026 review are narrower.
Where Mauritius still works
The capital gains exemption is gone for shares acquired after 1 April 2017. What remains in the India-Mauritius DTAA:
Dividend. 5% if shareholding is at least 10%, 15% otherwise (Article 10). Compares with 20% under Section 115A domestic rate (or 10% post-Budget 2024 amendment for closely held shareholders) — beneficial provision typically applies.
Interest. 7.5% on bank interest, where applicable. Compares with domestic withholding which depends on whether the lender is an NBFC, bank or other.
Royalty and FTS. 15% under Article 12, often beneficial to Section 115A's 10% — but only because the domestic rate is more favourable in this case; we use domestic, not treaty.
Other income. Generally taxable only in residence state under Article 22, which can apply to passive income streams that do not fit the specific articles.
Mauritius continues to be used for: (a) holding-company structures where the principal benefit is dividend repatriation, (b) inbound debt where the 7.5% interest rate applies, (c) hold-and-divest structures for pre-April 2017 grandfathered shares being exited now.
Singapore: similar trajectory, similar standing
The India-Singapore DTAA Protocol 2017 mirrored the Mauritius treatment of capital gains. The exemption ended on shares acquired after 1 April 2017. Singapore retains advantages in:
Treaty network. Singapore has a more substantial treaty network than Mauritius and is more credible as a regional operating hub. For groups that need a regional substance hub (board, executives, treasury), Singapore typically beats Mauritius.
Dividend rate. 10% if shareholding is at least 25%, 15% otherwise. Comparable to Mauritius.
Substance requirements. Singapore's Inland Revenue Authority increasingly applies substance tests for treaty benefits — minimum operating expenditure, local employees, board meetings held in Singapore. The cost of meeting these is higher than the cost of a Mauritius company, but the credibility is also higher.
Netherlands, Cyprus, UAE: the newer routings
Netherlands. Historically used for inbound investment with favourable capital gains treatment under the India-Netherlands DTAA. The MLI's Principal Purpose Test now applies. Substance requirements in the Netherlands are real — the Dutch authorities require operational substance for treaty access.
Cyprus. Niche use. India-Cyprus DTAA was renegotiated in 2016; capital gains rights moved to the source state for shares acquired after 1 April 2017, mirroring the Mauritius treatment. Cyprus retains some advantages for specific transactional structures.
UAE. The India-UAE DTAA has been used increasingly post 2023 for inbound holding structures. UAE has a 9% corporate tax from 2023 onwards (low but no longer zero), no withholding tax on dividend outbound, and a 0% capital gains rate on qualifying participation exemptions. The MLI applies. For groups with genuine UAE operations (regional headquarters, treasury, executives), the UAE has become a credible alternative to Singapore.
The Principal Purpose Test under the MLI
India is a signatory to the OECD Multilateral Instrument. The MLI overlays existing DTAAs with anti-abuse provisions including the Principal Purpose Test (PPT).
PPT denies treaty benefits where it is reasonable to conclude that obtaining the benefit was one of the principal purposes of any arrangement or transaction. The threshold is one of the principal purposes, not the principal purpose.
The implication: a Mauritius or Singapore holding company set up purely for treaty access, with no operational substance, will fail the PPT. The treaty benefit gets denied. The transaction is taxed as if no treaty existed.
How to defend: build substance.
Local board. A majority of directors resident in the treaty jurisdiction, with documented decision-making authority.
Local operations. Office space, employees, operating expenditure. Not a registered-agent address.
Local commercial purpose. The structure serves a real commercial purpose beyond tax — regional headquarters, treasury management, IP holding with development substance.
Commercial timing. The structure was set up well before any specific transaction it now benefits from. Setting up a Mauritius holdco three months before an exit, then claiming treaty benefit on the exit, is a textbook PPT fail.
GAAR: where the department is now active
General Anti-Avoidance Rules (Chapter X-A of the Income-tax Act, Sections 95-102) became operational from 1 April 2017. GAAR allows the tax authority to declare an arrangement to be an impermissible avoidance arrangement and re-characterise it.
The test for impermissibility: the arrangement (a) creates rights or obligations not at arm's length, (b) results directly or indirectly in misuse or abuse of the Income-tax Act, (c) lacks commercial substance, or (d) is entered into in a manner not normally employed for bona fide business purposes.
GAAR has a tax benefit threshold of ₹3 crore in a financial year. Transactions below this are outside GAAR. Above it, the assessing officer can refer the matter to the GAAR Panel.
Where the department has won GAAR matters: structures with no operational substance, transactions structured purely to access treaty benefit, transactions with circular flows of consideration, transactions involving newly incorporated entities timed to a transaction.
Where taxpayers have won: structures with genuine commercial purpose, structures set up in advance of any specific transaction, structures with operational substance.
What works in 2026
Three structural patterns we see surviving review.
First, the operating hub structure. A Singapore or UAE entity that genuinely operates — regional HQ, executives, treasury, board. Used for groups with non-India revenue or operations in the region. Treaty benefits available because the substance is real.
Second, the IP holding structure. An IP-holding entity in a treaty jurisdiction with genuine DEMPE functions — IP development team, IP management, licensing decisions made locally. Requires real personnel in the IP jurisdiction.
Third, the passive holding structure for legacy investments. A Mauritius or Singapore entity that holds pre-April 2017 grandfathered Indian shares. Operating for the limited purpose of holding and eventually divesting those shares. The structure is grandfathered; new investments through the same vehicle are not.
What does not work
Newly-incorporated treaty-jurisdiction shell companies with no substance, set up shortly before a specific transaction to access treaty benefits. The PPT and GAAR both bite. We do not recommend these.
The era of light-substance treaty-shopping ended in 2017. The structures that work in 2026 require investment in substance — board, executives, operations — that costs USD 250,000-1,000,000 a year. For groups whose tax saving justifies the substance cost, the structures work. For groups whose saving does not, the structures are net negative.
What we do at engagement
A structural review for inbound capital allocations above USD 25 million. Output: a memo identifying the optimal jurisdiction given the investment thesis, the substance requirements to support it, and a comparison against direct India investment.
For founder-owned global groups: an annual substance audit of the holding structure. The substance that was sufficient in 2020 is rarely sufficient in 2026.
Read the structure for what it is, not what it was set up to be. The cleanup since 2017 is now substantively complete; the next layer of changes is gradual rather than dramatic. The structures that work are the structures that have substance.
References

