GIFT City: A Foreign Jurisdiction Inside India

Two deeming provisions, both narrower than their reputation

The centre sits inside a Special Economic Zone at Gandhinagar, notified as a Multi Services zone on 18 August 2011 over about 105.44 hectares. The zone exists under the Special Economic Zones Act, 2005 (Act No. 28 of 2005, assented 23 June 2005), and section 18 of that Act is the provision that lets the Central Government approve an International Financial Services Centre inside such a zone, one per zone.

Everything people find counter-intuitive about the place comes from two legal fictions, and both are narrower than the summaries suggest. It is worth reading them in their own words, because the words do the limiting.

The customs fiction. Section 53(1) of the SEZ Act provides that a zone shall be "deemed to be a territory outside the customs territory of India for the purposes of undertaking the authorized operations". Three limits sit inside that one sentence: it is a deeming rather than a fact, it reaches only the customs territory, and it applies only to the authorised operations. The same Act says at section 1(2) that it extends to the whole of India, and defines the Domestic Tariff Area at section 2(i) as the whole of India excluding the zones. The zone is carved out of the domestic tariff area. It is not carved out of India.

The exchange-control fiction. The Foreign Exchange Management (International Financial Services Centre) Regulations, 2015, made under section 47 of the Foreign Exchange Management Act, 1999 and notified as FEMA.339/2015-RB on 2 March 2015 (gazetted as G.S.R. 218(E) on 23 March 2015), do the work everybody attributes to the whole zone. Regulation 3 provides that any financial institution or branch of a financial institution set up in the centre "and permitted/recognised as such" shall be treated as a person resident outside India.

One regulator where there are normally four

The domestic market splits supervision across four bodies, a division our guide to what SEBI actually does sets out in detail: SEBI's jurisdiction begins at the securities-market boundary and stops there, with banking at the Reserve Bank, insurance at the IRDAI and pensions at the PFRDA. Inside the centre that boundary is redrawn entirely.

The International Financial Services Centres Authority Act, 2019 (Act No. 50 of 2019) received Presidential assent on 19 December 2019 and was published in the Gazette of India, Extraordinary, Part II Section 1, No. 74, of 20 December 2019. The Authority was established on 27 April 2020. Section 13(1) is the transfer provision, and it operates "notwithstanding anything contained in any other law": all powers exercisable by an appropriate regulator listed in the First Schedule are, inside the centre, exercised by the Authority instead. The First Schedule names exactly four, with the statutes each surrenders: the Reserve Bank of India, whose column includes the Banking Regulation Act 1949, the Payment and Settlement Systems Act 2007 and the Foreign Exchange Management Act 1999 itself; SEBI, with the Securities Contracts (Regulation) Act 1956, the SEBI Act 1992 and the Depositories Act 1996; the IRDAI; and the PFRDA.

The mechanism is unusually thorough, and worth noting because it shows this was not done by circular. The Second Schedule inserts a disabling section into each parent statute in turn, twelve of them, each in the same words: the powers of the original regulator shall not extend to an International Financial Services Centre and shall be exercisable by the Authority. Section 44A was written into FEMA, section 28C into the SEBI Act, section 57A into the RBI Act. Parliament did not delegate the boundary. It legislated it, statute by statute.

Foreign currency is a statutory requirement, not a convention

Section 20 of the same Act states the rule in one line: "Every transaction of financial services in an International Financial Services Centre shall be in such foreign currency as may be specified by regulations in consultation with the Central Government." The specified list has been published as a schedule to the Authority's banking regulations. It ran to eleven currencies in the consolidated text as at 14 July 2023, the Authority's own board paper of 22 September 2025 refers to fifteen, and that paper proposes replacing the closed list altogether with any freely convertible foreign currency. Whether that amendment has been gazetted could not be confirmed, and it is recorded as unverified below.

The rupee is not banished, but its permitted uses are administrative rather than commercial. A banking unit may hold a rupee account to defray administrative and statutory expenses. Rupee-denominated business is contemplated, but the regulation permitting it requires settlement in a specified foreign currency, which is a different thing from trading in rupees. And for a single detail that captures the whole arrangement better than any summary: section 13(6) of the Authority's Act provides that all sums realised by way of penalties or fines shall be credited to the Consolidated Fund of India, in Indian rupees. Transact in dollars; pay your fines to the Indian exchequer in rupees. The India-ness never actually leaves.

What actually trades there

As at the Authority's register, last updated on the day this page was written, the centre hosts two stock exchanges, two clearing corporations and one depository. It is worth correcting a widely repeated claim here: those two exchanges did not merge. Talks ran from late 2022 and were called off in May 2024, and both continue to operate separately. What consolidated in 2023 was different: a dollar-denominated futures contract on a benchmark Indian index migrated to the centre from an offshore venue in July 2023, and it is now the dominant instrument traded there. Alongside the two stock exchanges sits a separately regulated bullion exchange, launched on 29 July 2022.

The trading day is the most visibly foreign thing about the place. One exchange runs exactly twenty-two hours across two sessions, from 04:30 to 17:00 and from just after 17:00 to 02:30. The other runs just under twenty hours inside a window from 06:30 to 02:45 the following morning, with a short break between sessions. The dominant index contract is sized at two US dollars per index point, ticks in half-dollars, and settles in cash. A domestic trader looking at those specifications is looking at an instrument denominated, sized and quoted in a currency they do not hold.

On scale, the Authority's own quarterly bulletin for the January to March 2026 quarter, published on 1 June 2026, records 1,147 final licences, registrations and authorisations as at March 2026, thirty-five operational banking units, banking assets of about 111 billion US dollars, and average monthly turnover across the exchanges of about 112 billion US dollars in that quarter. One caution on that first figure, because it is routinely misquoted: those are registrations, not distinct companies, and one entity may hold several. The Authority published a separate unique-entity count in earlier bulletins and stopped doing so, so no current figure for the number of firms exists.

The arithmetic nobody publishes: what the same exposure costs on each side

Almost every account of the IFSC leads with what is absent. No securities transaction tax on the trade. No Indian stamp duty. No goods and services tax on the trading service. Set against a domestic charge stack that a retail trader has been trained to resent, the list reads like an argument. It is not, and the reason is that the list is incomplete. It omits the one cost that only exists on the offshore side, and that cost is larger than everything on the list put together.

To see it, take a single exposure and price the same round trip twice. The worked example below uses an illustrative 10,00,000 rupees of index exposure, which at an illustrative rate of 86.00 rupees to the dollar is about 11,628 dollars. One entry, one exit. Every rupee figure that follows is illustrative and computed from the run described in the method note, not quoted from any provider. Our comparison of Indian and US markets for an Indian resident makes the general observation that conversion usually dominates a cross-border cost stack; the job here is to compute how far, on a venue where the trading charges really are close to nil.

On the domestic side the components are individually verifiable, and small. Securities transaction tax on an index futures contract is charged on the sell leg at 0.05 percent under section 159 of the Finance Act 2026 (Act No. 4 of 2026), which amended serial 4 of the section 98 table in the Finance (No. 2) Act 2004 with effect from 1 April 2026. Stamp duty is 0.002 percent on the buy leg under Article 56A(d) of Schedule I to the Indian Stamp Act 1899, inserted by section 21 of the Finance Act 2019. The regulator's own turnover fee is 0.0001 percent under Regulation 41(1) of the SEBI (Stock Brokers) Regulations 2026, gazetted on 7 January 2026. Goods and services tax at 18 percent applies to brokerage, exchange charges and the turnover fee, but not to the transaction tax or the stamp duty, which the intermediary merely remits. Add an illustrative flat brokerage and an illustrative exchange charge and the whole stack comes to about 617 rupees, or 0.060 percent of the exposure.

Now price the offshore route. The trading charges genuinely are lower. But before any of them apply, rupees have to become dollars, and afterwards dollars have to become rupees again. Each conversion is priced not as a fee but as a spread, which is why it so often escapes the comparison: it does not appear as a line on a statement, it appears as a slightly worse rate than the one on the screen. At an illustrative 50 basis points each way, that is 5,000 rupees going out and 5,000 rupees coming back. Add two wire charges and the modest offshore trading charges and the total is about 13,206 rupees, or 1.320 percent.

The gap is about 12,589 rupees on this exposure, and the offshore route costs roughly 21.4 times the domestic one. But the ratio is not the interesting part. The interesting part is the composition: 75.7 percent of the offshore total is the two currency conversions and nothing else. Every tax exemption on the page, added together, is smaller than the spread you pay to reach the venue where those exemptions apply.

It is also worth being precise about what kind of number the conversion spread is. It is not published, it is not uniform, and it is not a regulated maximum. It varies by provider, by amount, by time of day and by how much attention the customer is paying. That is why the figure includes a sensitivity strip rather than a single answer: at the low end the offshore round trip costs 8 times the domestic stack, at the high end 54 times. A reader who wants to know their own number has to ask for the rate they will actually receive, compare it against the interbank mid, and treat the difference as the cost. Nobody will volunteer it.

The toll is one-time, which cuts both ways

There is a fair objection to everything above. The conversion is paid on the way in and the way out, not on every trade. Somebody who remits once and then trades the same balance repeatedly amortises it. That is true, and it is worth computing rather than waving away.

So the arithmetic has a break-even, and it sits at roughly 31.6 round trips on the same remitted capital. Below that the domestic venue is cheaper. Above it the offshore venue is. That would be a clean and rather encouraging conclusion, except that it depends entirely on being allowed to leave the money there between trades, and that is precisely the condition the access rules govern. Amortisation is not a property of the arithmetic. It is a property of the rulebook, and it belongs in the next section rather than this one.

The exposure you did not choose: quantifying the currency leg

The currency has appeared so far only as a cost. It is also a risk, and the risk is structurally different from every other risk on this page because it is not the one the trader intended to take. Somebody who buys an index exposure has decided that they want the index. Denominating that exposure in a foreign currency attaches a second position to the first, one that nobody sized, nobody chose and nobody is being paid to hold. That is the structural point of the whole jurisdiction, and it is almost never quantified.

So quantify it. The simulation below is seeded and reproducible, runs 400,000 paths, and models two independent legs: the asset, in dollars, and the rupee-dollar rate. Both legs are given zero expected return, deliberately and by construction, so that nothing the model produces can be read as a forecast about either the asset or the currency. The only thing it can say is how much the outcome moves about, and how much of that movement each leg is responsible for.

In the base case the underlying carries an illustrative annualised volatility of 15 percent and the currency an illustrative 5 percent, with no correlation between them. The dollar outcome over a year has a standard deviation of 15.1 percent. The rupee outcome the resident actually receives has a standard deviation of 15.9 percent. The currency has widened the distribution by 5.5 percent and accounts for 9.9 percent of the variance.

Three things fall out of that run, and the second and third are the ones worth carrying away.

First, the share is inversely tied to the volatility of the underlying. A currency contributing 9.9 percent of the variance of a volatile index position contributes 19.8 percent of a calmer one at 10 percent volatility, 50.0 percent at 5 percent, and 73.5 percent at 3 percent. The crossover is exact and easy to remember: when the asset and the currency have the same volatility, the currency is half the risk. This is why foreign-currency exposure is a footnote for someone taking a large directional position and the dominant consideration for someone parking money in something stable. The quieter the asset, the louder the currency.

Second, time does not help. There is a widespread intuition that currency effects average out over a long holding period, and the simulation says they do not. The currency's share of variance is 9.9 percent at one month and 9.5 percent at five years. Both legs grow with the square root of time at the same rate, so their ratio is essentially fixed. You cannot wait out a currency exposure; you can only hedge it, and hedging it costs money that has to come out of whatever the venue was supposed to save you.

Third, and least comfortably for the argument this page is making, the currency is not automatically a bad thing. Everything above assumes the two legs are uncorrelated. They need not be. If the rupee tends to weaken when the underlying falls, the currency leg cushions the rupee outcome rather than amplifying it. At a correlation of negative 0.5 the simulation puts the rupee standard deviation at 13.2 percent against a dollar figure of 15.1 percent, which is to say the currency has reduced total dispersion, and its share of variance goes negative at about 7.0 percent. At a correlation of positive 0.5 it goes the other way, to 18.2 percent and a 18.9 percent share. So the honest statement is not that currency exposure is harmful. It is that currency exposure is a second position whose sign you do not control and cannot know in advance, taken on top of the one you meant to take.

The last comparison is the one that puts the earlier arithmetic in proportion. One standard deviation of the currency leg over a year, on the illustrative 10,00,000 rupee exposure, is about 50,000 rupees. The entire offshore cost stack computed in the previous section, both conversions, both wires and all the trading charges, is about 13,206 rupees. The unhedged, uncompensated, unchosen risk is roughly 3.8 times the size of the cost everyone argues about. A page that debated basis points and then said nothing about the currency would have measured the small thing carefully and ignored the large one.

The tax story, and why every version of it you will read is citing a repealed Act

Two things have to be established before any tax sentence on this subject can be trusted. The first is which statute is speaking. The second is who the provision is addressed to. Almost every published account of GIFT City gets the first wrong by accident and the second wrong by omission, and the second is the one that decides whether any of it is useful to the reader.

Start with the statute, because it moved. The Income-tax Act, 2025 (Act No. 30 of 2025) received Presidential assent on 21 August 2025 and was published in the Gazette of India, Extraordinary, Part II Section 1, No. 35, of that date. Section 1(3) brought it into force on 1 April 2026, and section 536(1) says in six words that the Income-tax Act, 1961 is hereby repealed. The Ministry of Finance confirmed commencement in a press release of 1 April 2026, and the department's own portal now states that the 1961 Act stands repealed from that date.

The practical consequence is that the section numbers in circulation are stale. The 2025 Act reorganised 819 sections into 536, and the IFSC provisions moved with everything else. The relief for a unit in the centre was section 80LA; it is now section 147. The capital-gains provision everyone quotes was section 47(viiab); it is now section 70(1)(r). The exemptions at sections 10(4D), 10(4E) and 10(4F) are now rows in Schedule VI. None of those old numbers is wrong for a pre-2026 year, and all of them are wrong as a description of the law in force. A reader who searches this topic will land almost exclusively on pages citing the repealed numbering.

The word that decides everything: non-resident

Now the second question, which is the one this page exists to answer. The headline that draws people to this subject is that transfers on an exchange in the centre are outside capital gains. Read the provision. Section 70(1)(r) of the 2025 Act applies to a transfer of a capital asset made by a non-resident on a recognised stock exchange located in an International Financial Services Centre, where the consideration is paid or payable in foreign currency. The repealed section 47(viiab) said the same thing in the same words. Both statutes, in identical terms, put the words made by a non-resident into the operative clause.

Schedule VI of the new Act is laid out as a table whose third column is headed "eligible persons", which makes the point impossible to miss. Every row reads either "Non-resident" or "Any specified fund", and a specified fund is defined in the Schedule's notes as one whose units are held by non-residents, with a narrow tolerance of up to five percent for a holder who became resident after subscribing. The relief at section 147 is addressed to an offshore banking unit or to a unit of the centre, meaning a licensed business operating there, not a customer of one.

What a resident does get, stated fairly

There is a real benefit, it is smaller than the headline, and almost nobody reports it. India's concessional rates on listed equity gains are conditional on securities transaction tax having been paid on the trade, and no such tax is collected on an exchange in the centre. Left alone, that condition would disqualify the trade from the concessional rate entirely. The 2025 Act deals with it explicitly: section 196(3) disapplies the transaction-tax condition for a trade on a recognised stock exchange in an IFSC where the consideration is in foreign currency, and section 198(4) does the same for the long-term rate. A resident therefore keeps the concessional short-term and long-term treatment on qualifying equity assets without the transaction tax having been paid. That is a genuine and verifiable relaxation. It is a relaxation of a condition, not a removal of the tax.

Two further points belong here precisely because they are unresolved, and a page that presented a tidy answer would be misleading you.

The first concerns derivatives. Under the 2025 Act, exchange-traded derivatives escape the speculative-transaction label and are treated as ordinary business income, which matters a great deal for set-off and carry-forward. But the definition at section 2(33) reaches a transaction carried out through an intermediary registered under section 12 of the SEBI Act, 1992. Intermediaries in the centre are authorised by the unified authority under its own 2019 statute, not registered with SEBI under section 12. Whether a derivative trade there therefore satisfies the definition is a question the drafting does not obviously answer, and no authority resolving it could be found. It is flagged in the ledger below as unverified rather than guessed at.

The second concerns the currency itself. Whether the gain or loss on the rupee-to-dollar-and-back leg is computed separately or absorbed into the transaction is not addressed by any provision, circular or ruling that could be located. Given the size of that leg, established earlier on this page, it is not a small gap. It is also exactly the kind of question on which an individual needs advice specific to their facts rather than a general page.

The conclusion, written plainly rather than diplomatically

Add up what has actually been established. The exemptions that make this jurisdiction famous are addressed, in the operative words of both the repealed and the current Income-tax Act, to a non-resident or to a licensed unit. A resident keeps a relaxation of one condition and nothing more. Reaching the venue costs roughly 21.4 times the domestic charge stack on a single round trip, of which 75.7 percent is a currency spread that nobody publishes. The route in runs through a remittance scheme with a cap, a tax collected at source, and conditions on what the money may be used for once it arrives. And holding a foreign-currency position attaches an exposure whose one standard deviation is about 3.8 times the entire cost of getting there, carrying, on the modelling used here, no expected return to compensate for it.

So the honest answer is that for a typical resident retail individual, this is not a useful venue today. Not because it is badly built, and not because the rules are unreasonable, but because almost nothing about it was designed with that person in mind. The arithmetic only turns favourable at heavy repeated use of a standing offshore balance, and the access conditions are precisely what make a standing offshore balance difficult. That is not an accident of drafting. It is the design working as intended.

It is worth being equally plain about the other half, because a page that only demolished would be as unbalanced as the brochures it is correcting. This jurisdiction is a serious and largely successful piece of financial infrastructure. It gives foreign capital a way to transact in Indian-adjacent markets under one rulebook, in a currency it already holds, without the exchange-control machinery that governs everything else. It has given aircraft and ship leasing, fund management, bullion and offshore banking a domestic home that previously sat in Singapore, Dubai and Mauritius. Those are real achievements and the reliefs that support them are coherently targeted. The mistake is not in the venue. The mistake is in reading a set of provisions written for non-residents as though they were an offer to residents.

What would change the answer? Three things, and they are worth naming so that a reader can watch for them rather than re-reading brochures. A relaxation of the permitted-purpose and idle-funds conditions, which would let a standing balance sit offshore and let the entry toll amortise. A meaningful narrowing of the retail conversion spread, which is the single input that dominates the entire cost comparison. And a provision extending any part of the exemption structure to residents, which would require Parliament rather than a circular, and which would sit awkwardly with the policy of taxing residents on income from whatever source derived. Until at least the first two move, the arithmetic on this page is the arithmetic.

The general lesson generalises past this one venue. A structural advantage stated as a list of things that are absent is almost always incomplete, because the list is drawn up by people who benefit from your reading it. The useful question is never "what does this remove?" It is "what does this add that was not there before, who is the provision addressed to, and what does it cost me to reach it?" Working that out for yourself, on a page of arithmetic rather than a page of claims, is the habit rather than the answer, and it is the method we teach.

Every rupee figure and every statistic above comes from a single deterministic script, seeded at 20260815 so that it reproduces identically on each run. Nothing on the page is a quotation from any provider, any platform or any price list, and no figure has been carried across from another source.

The charge comparison prices one round trip on an illustrative 10,00,000 rupees of index exposure, converted at an illustrative 86.00 rupees to the dollar. That rate is an arithmetic anchor and nothing more; it is not a market quotation and no conclusion on the page depends on its level. The domestic side uses the statutory rates cited in the ledger table above, plus an explicitly illustrative allowance for brokerage and for the equity-derivatives exchange transaction charge, which is flagged as unverified in that table. The offshore side uses an illustrative conversion spread of 50 basis points each way, an illustrative flat wire charge on each leg, and illustrative offshore trading charges. Because the conversion spread is the input that dominates the answer, it is swept from 10 to 150 basis points rather than asserted at one value.

The currency simulation runs 400,000 paths of a two-factor model: an asset leg denominated in dollars and a rupee-dollar leg, combined multiplicatively to give the rupee outcome an Indian resident would actually receive. Both legs are given zero expected return by construction. That choice is deliberate and it is what allows the exercise to be published at all: the model can say nothing whatever about what any asset or any currency will do, only about how far outcomes spread and which leg is responsible for the spread. Volatilities are stated as illustrative inputs, and the results are shown across a range of them rather than at a single setting, because the share attributable to the currency turns out to depend on that input more than on anything else.

All results are illustrative and simulated. They are not a track record, not a forecast, and not an indication of what any account would produce. The tax and regulatory statements are mechanism descriptions with the instrument, number and date attached, and several are marked unverified in the ledger precisely because they could not be confirmed from a primary source at the time of writing. Rates and thresholds change; the reader should confirm any figure they intend to rely on against the current text of the instrument itself.

• Tax on trading in India, the domestic side →
• The real cost of an Indian trade, charge by charge →
• Currency derivatives and the underlying-exposure rule →
• Where Indian retail participation is actually heading →