License or Sell? What the Treaty Decides, and What Pillar Two Undoes

How the India–Switzerland tax treaty routes a cross-border transfer of a trademark or patent, what that means in money, and why the 15% global minimum tax changes the answer for large groups. With a model you can run yourself.
In one paragraph
When a Swiss company lets an Indian business use its brand, India may withhold 10% of each royalty. When the Swiss company instead sells the Indian rights for a fixed price, the India–Switzerland treaty gives the taxing right to Switzerland alone. That looks like a clear win for selling. In money terms it often is not: for a normally taxed seller, the choice mainly moves tax revenue between the two countries. Selling saves real tax mainly when the Indian withholding tax would otherwise be lost. For groups subject to the OECD's 15% minimum tax (Pillar Two, or "GloBE"), even that saving largely disappears.
1. Where this starts
This article builds on two pieces by Pramod Kumar, former Vice President of the Income Tax Appellate Tribunal: his LinkedIn post A Treaty That Chose to Speak and his longer article What Was Sold, and How It Was Paid For (both October 2026). Both discuss the decision of the Mumbai bench of the Tribunal in Novartis AG (ITA 1027/Mum/2023, 12 October 2026). The Tribunal treated a complete assignment of Indian trademark registrations, limited to India and paid for with fixed sums, as the sale of distinct property rather than a royalty. Under the India–Switzerland treaty that classification decides which country may tax.
Kumar's post ends with a practical question: should owners of intangibles then sell rather than license? His answer names the situations in which a sale helps. This article puts numbers on those situations and then asks what Pillar Two does to them. My reading of the treaty and its Protocol follows his post and article, which he offered as food for thought rather than as a view on the outcome. The arithmetic, the transfer-pricing and Pillar Two analysis, and any errors are mine.
For non-specialists: renting versus selling
Owning a brand is a bit like owning a house. You can rent it out: you stay the owner and receive rent every year. For a brand that rent is called a royalty, and the rental contract is a licence. Or you can sell it: you receive one price today and you are no longer the owner. The tax rules treat these two very differently, and that is what this article is about.
2. How the treaty sorts the transaction
A royalty article covers payments for the use of a trademark or patent. Whether it also covers a payment for the trademark itself is, under most of India's treaties, a matter of interpretation. The India–Switzerland treaty settles it in its Protocol. Gains from alienating the right or property go to Article 7 (business profits) or Article 13 (capital gains). Only gains that are contingent on profits, productivity or use may be taxed under Article 12, at up to 10% of the gross amount. The Protocol does not call those contingent gains royalties; it lets India tax them under the royalty article.
Article 13(6), the residual capital-gains rule, then gives a fixed-price gain to the seller's country of residence, unless the property belongs to an Indian permanent establishment. Kumar's post sets out this routing in a clear flowchart, which I do not repeat here. In his words, the deed decides whether there is an alienation; the payment terms decide where it is taxed.
The contrast matters for the scenarios below. Under the India–US and India–UK treaties a fixed-price gain also falls outside the royalty article, but the capital-gains article leaves it to domestic law, so India may still tax it.
3. First question: who is the buyer?
Before asking how the treaty classifies a sale, I would ask a more basic question, and as a former tax inspector it is the first one I asked myself on reading the Novartis order: is the buyer related to the seller?
If it is, the price itself becomes the issue. A sale of intangible property between associated enterprises is an international transaction under India's transfer-pricing rules (sections 92 to 92F of the Income Tax Act, 1961), and Switzerland will apply the arm's length principle on its side. The two countries then look at the same price from opposite ends:
- India tests the buyer. A purchased trademark is depreciable in India. A price set too high inflates the Indian subsidiary's depreciation for years, so the Indian transfer-pricing officer will look for an overpayment.
- Switzerland tests the seller. A price set too low moves the brand's future value out of the Swiss tax base, so the Swiss authority will look for an underpayment.
Each side has an interest in the opposite adjustment, which is a recipe for double taxation and a mutual agreement procedure. The OECD Transfer Pricing Guidelines (Chapter VI) approach the price through the options realistically available to each party: the seller compares the price with continuing to license, the buyer with taking a licence. That is why the model below uses the present value of the royalties as the sale price. Where the value of a brand is highly uncertain at the time of the transfer, the guidance on hard-to-value intangibles allows a tax administration to use actual results to test the price afterwards; for an established brand that risk is smaller.
Kumar describes the case that does not stand on firm ground as one with three features together: a related buyer, a price based on projected royalties, and control retained by the seller. Taken together, that is a fair description of a weak case. Taken on its own, a price based on projected royalties is a normal basis for an arm's length price between related parties. In my view, the feature that matters most is the retained control, because it goes to whether anything was sold at all.
In Novartis this question was answered early. The three buyers (USV, Cipla and Eris) were described as independent purchasers, and the order does not suggest otherwise. A price agreed between unrelated parties is the best evidence of an arm's length price. One loose end remains, which Kumar also notes: the order mentions earlier co-marketing arrangements involving Indian group entities, which it does not name. The next step is mine: when a sale ends such an arrangement, an inspector will ask whether the group entity that loses its role should have been compensated. The OECD Guidelines deal with that question under business restructurings (Chapter IX).
4. Why compare in "present value"
Money received today is worth more than the same amount in ten years, because today's money can already be invested. To compare ten years of royalties with one sale price paid today, every future amount is converted back to its value today. That is the present value.
The base case used throughout: Indian sales of the branded products of 100 million a year, a 5% royalty (so 5.0 million a year) for 10 years, and a 10% required return. Ten payments of 5.0 add up to 50, but they are worth about 30.7 million today. That is also the most an independent buyer would pay to buy out the royalties, so it is used as the fixed sale price. All figures below are in millions, currency-neutral, and in present value.
5. Five scenarios under ordinary (domestic) tax rules
Assumptions: Swiss effective profit tax about 13% (for example Basel), Indian corporate tax 25.17%, Indian capital gains tax for a foreign company about 13.65% (12.5% plus surcharge and cess), UK corporation tax 25%, India–UK treaty royalty rate 15%. A self-created brand has a nil tax cost.
| Scenario | Licence | Sale | Difference | Who gains |
|---|---|---|---|---|
| S1 Profitable Swiss owner | 3.99 | 3.99 | 0.00 | Neutral for the owner; tax moves from India to Switzerland |
| S2 Swiss owner with tax losses | 3.07 | 0.00 | −3.07 | Sale: India's 10% would otherwise be a final cost |
| S3a UK owner, self-created brand | 4.61 | 4.19 | −0.41 | Barely matters: India taxes the gain too |
| S3b UK owner, brand bought for 25 | 4.61 | 0.78 | −3.83 | Sale: India taxes only the net gain |
| S4 Swiss parent sells to Indian subsidiary (group view) | 3.99 | 6.20 | +2.21 | Licence: after a sale India taxes the brand income at 25% |
| S5 Risk: treaty benefit denied under GAAR | 3.99 | 8.19 | +4.19 | Double taxation until resolved by mutual agreement |
Total tax in present value, millions. Difference = sale minus licence; negative means the sale is cheaper. In S5 the columns are "sale with treaty protection" and "sale with protection denied". In S3 the UK owner is assumed to have tax losses (a setting in the model).
S1: same cake, different slices
Under a licence India withholds 10% of each royalty. Switzerland taxes the royalty at about 13% but credits the 10%, so Switzerland collects only about 3%. Total: 13%. Under a sale Switzerland taxes the whole gain at 13% and India gets nothing. Total: again 13%. For the owner the two routes cost the same, about 4.0 million. What changes is the split: under a licence India receives about three quarters of the tax; under a sale it receives none.
S2: when the Indian tax "leaks"
Kumar notes that a sale helps where Indian tax on the gross royalty would not have been fully credited at home. This scenario shows how much that can be. A credit only helps if there is home-country tax to set it against. A Swiss biotech company with tax losses pays no Swiss tax, so India's 10% withholding becomes a final cost: 3.07 million in present value, about 10% of the brand's value. A fixed-price sale is untaxed in India, and the Swiss gain is absorbed by the losses. The same leakage arises when the Swiss tax on the net royalty is below 10% of the gross royalty, for example under the Swiss patent box (patents only, not trademarks) or when the royalty carries high costs.
S3: a UK owner, a different treaty
The India–UK treaty leaves capital gains on such property to domestic law, so India taxes the sale as well. For a self-created brand the whole price is gain (India: 13.65% × 30.7 = 4.19), little better than the royalty route. If the brand was bought earlier for 25, only the 5.7 increase is taxed (0.78). This is the case Kumar describes in his post: a sale helps when the asset was acquired at a cost, because India keeps the taxing right but taxes a net gain rather than a gross royalty. For a self-created mark, as he notes in his article, a sale mostly brings the tax forward.
S4: inside a group the arithmetic reverses
Kumar observes that for groups the arithmetic often reverses, because the brand's future returns stay in India. In numbers: if a Swiss parent sells the Indian brand rights to its own Indian subsidiary, the subsidiary loses its royalty deduction and the brand income is taxed in India at 25.17% instead of about 13% in Switzerland. Tax depreciation on the purchased brand (25% declining balance) compensates only partly. Over ten years the group pays about 2.2 million more, and the gap continues every year afterwards. This is also the scenario in which the transfer-pricing questions of section 3 arise.
S5: the downside
If the tax benefit is challenged and denied, India taxes the gain and Switzerland may refuse a credit, so both tax it: 8.2 million instead of 4.0 until the authorities resolve it. As Kumar points out in his article, the India–Switzerland treaty has no principal purpose test; the challenge would come through India's general anti-avoidance rule (GAAR). His post is clear about where the line runs. An independent buyer, a commercial reason to sell (such as leaving a market) and a seller who really lets go are on firm ground. A sale to a related company, priced on expected royalties, with the seller still in control, is not.
6. Enter Pillar Two: the 15% floor
For non-specialists: what is GloBE?
Since 2024 many countries apply an OECD agreement under which large multinational groups (consolidated revenue of at least €750 million) must pay at least 15% tax in every country where they operate. If the effective rate in a country is lower, the difference is collected as a "top-up tax". Switzerland collects that top-up itself, through a domestic minimum tax (a QDMTT), so the money stays in Switzerland. Think of a swimming pool with a raised floor: however low the rate, you never sink below 15%.
Two GloBE mechanics matter here:
- Foreign withholding counts. The 10% India withholds on a royalty is treated as tax paid by the Swiss recipient ("covered taxes"). So is Indian tax on a capital gain.
- Low-substance IP earns no carve-out. The substance-based income exclusion rewards payroll and tangible assets. A brand-holding entity has little of either, so practically all of its income is tested against 15%.
India does not, as far as publicly known in mid-2026, apply GloBE rules itself; its 25.17% corporate rate is above 15% in any case.
Figure 1. Tax on the brand income as a share of that income. Without GloBE, a low Swiss rate makes the licence costly (India's 10% exceeds the Swiss tax it can be credited against) and the sale attractive. With GloBE both routes end at 15%: the floor absorbs the difference, and only the split between India and Switzerland remains.
The scenarios again, now for a group in scope
| Scenario | Difference, domestic | Difference, with GloBE | What GloBE does |
|---|---|---|---|
| S1 Profitable Swiss owner | 0.00 | 0.00 | Both routes rise to 15% (4.61). Licence: India 3.07, Switzerland 1.54. Sale: Switzerland 4.61. |
| S1-PB Patent-box rate 6% | −1.23 | 0.00 | The advantage of selling disappears entirely. |
| S2 Swiss owner with losses | −3.07 | −2.46 | Using losses counts as tax at 13%; the sale attracts a 2% top-up, the licence does not. |
| S3a/S3b UK owner | −0.41 / −3.83 | −0.41 / −3.83 | No change: UK rate and Indian tax already exceed 15%. |
| S4 Group sale to Indian subsidiary | +2.21 | +2.21 | Both routes get the same 2% Swiss top-up; the licence stays cheaper. |
| S5 Treaty benefit denied | +4.19 | +3.58 | The Indian tax lifts the Swiss rate above 15%, so no top-up: the downside shrinks a little. |
7. What this means
- For a normally taxed owner, the treaty decides who collects, not how much. Licence or sale costs about the same; the sale shifts the revenue from India to Switzerland.
- Selling saves real tax mainly where India's tax would otherwise be lost (losses, a low home rate) or, under the US and UK treaties, where the asset was bought at a cost. These are the situations Kumar names; the model shows their size.
- Inside a group, selling the brand to India is usually more expensive, because the brand's future income is then taxed at India's higher rate, as Kumar also observes. The price then becomes a transfer-pricing question on both sides.
- For groups above €750 million revenue, the 15% floor largely neutralises the choice. Kumar's post concludes that a sale tends to favour the taxpayer "where the seller's treaty gives residuary gains to the residence State alone and the price is fixed". In his longer article he adds that the Protocol cuts both ways: a shield for a fixed-price sale, a possible sword for an earn-out. The figures here add a third point: within Pillar Two the shield mainly decides which State collects, not how much the group pays.
- In my view, that supports the defence against an abuse challenge. If the minimum tax removes most of the tax benefit, it is harder to argue that obtaining one was the main purpose of the transaction. The choice looks commercial.
- A further point of my own: one new asymmetry. Pillar Two's subject-to-tax rule (STTR), where it applies between two treaty partners, lets the source State top up low-taxed intra-group royalties to 9%, but it does not reach capital gains. A fixed-price sale escapes it.
The same logic travels to any source country, Costa Rica included. First ask who the buyer is and whether the price is at arm's length. Then ask what was transferred and how it is paid for. Then ask which article of the applicable treaty, or domestic law where there is no treaty, allocates the result. Finally, ask whether the 15% minimum tax has already flattened the difference.
8. Run the numbers yourself
The model below reproduces the calculations in this article. Change any yellow field; results update immediately. Everything runs in your browser; nothing is sent anywhere.
Please note: after five minutes without changes, the model returns to the standard values used in this article. The bar above the fields shows how much time is left, and you can also reset straight away.
| Scenario | Licence | Sale | Diff. | Licence + GloBE | Sale + GloBE | Diff. |
|---|
Total tax in present value. Diff. = option 2 minus option 1; green = option 2 cheaper. S5 compares "sale with treaty" (option 1) with "sale, treaty denied" (option 2).
Who collects the tax?
9. Glossary
- Royalty / licence
- A recurring payment for using a brand or patent without owning it; the "rent". The contract granting use is a licence.
- Alienation
- The legal term for transferring ownership, such as a sale or assignment.
- Withholding tax
- Tax the payer's country (here India) deducts directly from a payment abroad.
- Foreign tax credit
- The home country deducts tax already paid abroad from its own tax, but never more than its own tax on that income.
- Leakage
- The part of foreign tax that cannot be credited and therefore becomes a real cost.
- Associated enterprises / arm's length price
- Companies in the same group, and the price that independent parties would have agreed in the same circumstances. Transfer-pricing rules require related parties to use it.
- Permanent establishment (PE)
- A branch or fixed place of business in the other country. Property belonging to it may be taxed there.
- Present value
- What future amounts are worth today, given a required rate of return.
- GAAR / PPT
- Anti-abuse rules: India's general anti-avoidance rule, and the principal purpose test that many treaties now contain ("was obtaining a tax benefit one of the main purposes?"). The India–Switzerland treaty has no PPT.
- MAP
- Mutual agreement procedure: the two tax authorities resolve double taxation between themselves.
- GloBE / Pillar Two
- The OECD global minimum tax: groups with revenue of at least €750 million pay at least 15% per country.
- QDMTT
- Qualified domestic minimum top-up tax: a country collects the top-up to 15% itself.
- Covered taxes / ETR
- Taxes that count towards the 15% test (including foreign withholding), and the effective tax rate they produce.
- STTR
- Subject-to-tax rule: a treaty rule letting the source country tax low-taxed intra-group royalties and similar payments up to 9%.
- SBIE
- Substance-based income exclusion: a GloBE carve-out based on payroll and tangible assets; negligible for a brand holder.
10. Assumptions and limitations
- Rates are indicative as known in mid-2026 and must be verified before use. Indian surcharge thresholds and cess are folded into one effective rate; short-term holding periods are not modelled.
- The sale price is taken as the present value of the royalties. The transfer-pricing questions in section 3 are discussed but not modelled; neither are adjustments to the price.
- Foreign tax credit is limited to the home-country tax on the net royalty; deduction of uncreditable withholding as an expense is ignored.
- Indian depreciation: 25% declining balance on the purchase price, half-year rule ignored; the tail of the depreciation is included in closed form.
- GloBE: the model tests the marginal effect of this transaction against 15% rather than the jurisdiction's blended rate; SBIE is set to nil. For losses (S2), using a loss counts as covered tax at the local rate, which assumes the losses arose in GloBE years or fall under the transition rules.
- Not modelled: GST and VAT, currency and inflation, formalities such as tax residency certificates and any withholding on the sale price, and the STTR.
- The residual value of the brand after the licence term (default nil) counts on both sides: in the licence option as further licence income, with Indian withholding and home-country tax; in the sale option as part of the price. Both options therefore cover the same income.
Sources. ITAT Mumbai, Novartis AG v. ACIT (IT), ITA 1027/Mum/2023, order of 12 October 2026. Pramod Kumar, A Treaty That Chose to Speak, LinkedIn post, October 2026, linkedin.com. Pramod Kumar, What Was Sold, and How It Was Paid For: The Novartis Order and Three Readings of the India-Switzerland Protocol, October 2026, lnkd.in/p/gQ_ekkut. OECD Transfer Pricing Guidelines (2022), Chapters VI and IX.
The reading of the treaty and its Protocol follows Pramod Kumar's post and article. The figures, the transfer-pricing section and the Pillar Two analysis, including the points on the STTR and on the defence against an abuse challenge, are the author's own, and so are any errors. This article is for analysis and discussion and is not tax advice. Outcomes depend entirely on the assumptions entered.
