Defending the Residual Profit Split Method in India When the TPO Proposes TNMM for an Intangible-Owning Entity

By CA Mithilesh Reddy, Founder and CEO, and CA Rajesh Vaishnav, Partner, Steadfast Business Consulting LLP | Transfer Pricing

Introduction

Transfer pricing methods divide into those that test one side of a transaction and those that look at both. A one sided method selects the less complex party, benchmarks its return against comparables, and leaves whatever remains to the other side. It is the workhorse of Indian transfer pricing because it is simpler to apply and easier to verify. It also carries an assumption that is not always true: that only one party to the transaction is doing anything a comparable cannot replicate.

Indian transfer pricing officers rarely accept the profit split method and commonly propose to substitute the transactional net margin method. Defending the choice is not an argument about which method is more favourable. It is an argument about whether the conditions in the rule are satisfied, and it stands or falls on the functional and value chain evidence.

This note deals with what happens when that assumption fails, and with defending the profit split method when a transfer pricing officer proposes to substitute a one sided method in its place. It covers the statutory conditions and their two gateways, the Indian authority and its limits, how economic ownership of an intangible is established, how a residual split is constructed and documented, the high margin problem, and why the case for rejecting the alternative method has to be made affirmatively rather than left implied. The functional evidence on which the whole analysis rests is dealt with in our note on FAR analysis for transfer pricing audit defence.

1. The statutory conditions

Rule 10B(1)(d) of the Income-tax Rules, 1962 (now rule 79(1)(d) of the Income-tax Rules, 2026) provides for the profit split method and identifies the circumstances in which it applies: transactions involving transfer of unique intangibles, or multiple international transactions which are so interrelated that they cannot be evaluated separately for determining the arm’s length price.

Those are the two gateways, and they are disjunctive. The method is available where unique intangibles are involved, and it is available where the transactions are inseparably interrelated, whether or not intangibles are present. A submission should identify which gateway it relies on, and preferably both where the facts support it.

Rule 10C (now rule 80 of the Income-tax Rules, 2026) sets out the factors governing the selection of the most appropriate method, including the nature and class of the transaction, the functions performed, assets employed and risks assumed, the availability and reliability of data, and the degree of comparability. The method is not chosen in the abstract; it is chosen because it is the one that can be applied most reliably on the available data.

Under the Income-tax Act, 2025 the governing provision is section 165 (section 92C of the 1961 Act, determination of arm’s length price), within Chapter X, sections 161 to 177. There is no current administrative guidance on the method. The Central Board’s Circular No. 2/2013 dated 26 March 2013 on the application of the profit split method was withdrawn by Circular No. 5/2013 dated 29 June 2013 and has not been replaced, so a submission cannot rely on Board guidance here and must argue from the rule and the authorities.

2. The Indian authority

The leading tribunal decision is Infogain India Pvt. Ltd. v. DCIT, pronounced by the Delhi Bench on 19 August 2015 for assessment year 2008-09. The Bench held that under Rule 10B(1)(d) the profit split method applies where a transaction involves unique intangibles or where interrelated international transactions cannot be evaluated separately. On the facts the Indian and United States operations were highly integrated and both contributed significantly, so the profit split method was the most appropriate method. Where comparables are unavailable, residual profit may be split on relative contribution measured through key value drivers, consistent with the guidance published by the OECD and the United Nations. A split accepted by the department in adjacent years was directed to be followed.

The decision should be cited for what it says. It did not hold that the profit split method must always displace the transactional net margin method wherever unique intangibles exist. Its reasoning rested principally on the integration of the two operations and the fact that both contributed materially. A submission that overstates the holding gives the Revenue an easy answer.

On the related question of comparability, CIT v. UT Starcom Inc. (India Branch), Delhi High Court, 25 September 2017, is useful in both directions. The Court found no substantial question of law and dismissed the Revenue’s appeal, affirming the Tribunal’s reasoning that a large risk taking company with significant intangibles and assets is not comparable to a limited risk captive. The corollary is that an entity which does own significant intangibles and does carry entrepreneurial risk cannot be benchmarked against routine service providers either, which is often exactly what a proposed transactional net margin analysis does.

3. Why a one sided method breaks down here

The transactional net margin method works by selecting the less complex party as the tested party, finding comparables for it, and setting its return. Everything else falls to the other party by residue. That is a sound approach where the other party is the only one doing anything unusual.

Consider instead a group in which an Indian entity develops and owns a proprietary software platform, and overseas entities distribute it in their markets. If the overseas distributors are routine, they can be tested and the residual accrues to India, which is the correct outcome. A transfer pricing officer who instead tests the Indian developer against routine software service comparables inverts the structure: it caps the developer’s return at a service margin and leaves the residual with distributors that did not create it.

That inversion is the argument. It is not that the profit split method gives a better answer. It is that applying a one sided method with the wrong tested party allocates the return from the intangible to the party that does not own it, which is the opposite of what the arm’s length principle requires.

4. Establishing economic ownership of the intangible

Everything turns on who owns the intangible in economic terms, and that is a DEMPE question: who performs and controls the development, enhancement, maintenance, protection and exploitation functions, who funds them, and who bears the risk.

Legal ownership is the starting point and nothing more. The evidence that matters is operational.

• Development. Where the engineering teams sit, headcount by location, the roadmap and who sets it, who approves releases. Version control and project management records are more persuasive than an organisation chart because they are contemporaneous and were not created for the file.
• Enhancement. Who decides what gets built next, and on whose commercial judgment.
• Maintenance. Who runs the platform, who resolves defects, who carries the operational burden.
• Protection. Who files and maintains the intellectual property, who funds enforcement.
• Exploitation. Who sets pricing and commercial terms, who decides which markets to enter.
• Funding and risk. Who bears development cost, and in particular who absorbs the cost of a project that fails. The party that carries failure risk is the party that should earn the upside.

The corresponding question has to be answered for the overseas entities. Showing that they perform routine distribution functions, do not develop the product, do not carry development risk and operate within commercial parameters set elsewhere is what establishes them as the routine parties.

5. Constructing the split

A residual profit split has two stages, and both have to be documented.

Stage one: the routine return. Each party is allocated a return for its ordinary functions, benchmarked against comparables performing those functions. For a distributor this is a distribution margin; for a service entity, a service margin. This stage is a conventional benchmarking exercise and should be done conventionally, with a documented search.

Stage two: the residual. What remains after routine returns is allocated by reference to relative contribution to the unique value drivers. The allocation key is where the analysis is most often thin, and it is where the officer will concentrate.

Keys commonly used include cumulative development spend, capitalised and amortised over an appropriate life; headcount or cost of the people performing DEMPE functions; the relative significance of identified value drivers weighted on a reasoned basis; and, where the facts permit, a bargaining analysis of what each party could have obtained from an independent counterparty.

Whichever key is used, three things must be shown: that it measures contribution to the residual rather than merely activity; that the data underlying it is reliable and auditable; and that the result is stable, in the sense that reasonable variation in the inputs does not change the conclusion materially. A sensitivity table is worth more than a paragraph of justification.

6. The high margin problem

An entity that owns a valuable platform and licenses it will often report margins far above anything in a routine comparable set. Where it also claims a unit based deduction, the combination attracts attention, and the officer’s working assumption is frequently that the margin is evidence of mispricing.

The answer is structural. In a model where one party owns the intangible and others distribute, a high margin at the owner is what the arm’s length principle predicts, not a departure from it. The comparison being drawn is with service providers that own nothing, and the margin differential is the return on the intangible. Demonstrating that is done by showing the development investment that produced the platform, the period over which it was made, the risk that it would produce nothing, and the fact that the same entity absorbed that risk.

Where a deduction is also claimed, the two issues should be kept separate in the submission. The availability of the deduction is a question under its own provision. It is not a reason to reallocate profit, and conflating the two in the response invites the officer to do the same.

7. A worked example

Assume, hypothetically, that a group owns a commercially valuable intangible developed and managed in India while overseas affiliates carry out distinct distribution functions. The question is whether the Indian entity makes unique contributions and the dealings are sufficiently integrated to support a profit split, and whether a one-sided comparable can reliably measure the Indian entity’s return. The selection must be supported by actual contracts, risk control and contribution records.

Gateway. Both limbs of Rule 10B(1)(d) are engaged. The platform is a unique intangible, and the development and distribution transactions are so interrelated that neither can be priced without reference to the other.

Economic ownership. Established through development headcount and location, roadmap governance records, the funding pattern across years, and evidence that failed projects were absorbed in India.

Routine returns. Overseas distributors benchmarked against independent distributors in their markets; a documented search for each.

Residual. Allocated on a key reflecting cumulative development contribution, with the computation shown and a sensitivity analysis provided.

Meeting the substitution. The submission has to show affirmatively why the transactional net margin method cannot be applied reliably here, not merely that the profit split method can. The reasons are that no reliable comparable exists for an entity owning a platform of this kind; that the proposed comparables perform services and own nothing, so the differential is the very return in dispute; and that a one sided method applied to the intangible owner allocates the residual to parties that did not create it. On comparability the reasoning in UT Starcom applies.

Consistency. Where the department has accepted a split in an adjacent year, that acceptance should be in the submission. Infogain directed that an accepted split be followed, and consistency arguments carry real weight at tribunal level.

8. The allocation key, in figures

The residual stage is where submissions are thinnest, so it is worth showing what an adequately documented key looks like. Take a group with combined profit of 100 from the platform business, an Indian developer that owns the platform, and two overseas distributors.

Stage one, routine returns. Each distributor is benchmarked against independent distributors in its market and allocated a routine return of 8, giving 16 in total. The Indian entity is allocated a routine return for the service element of its activity, benchmarked against independent software service providers, of 14. Routine returns therefore absorb 30, and the residual is 70.

Stage two, the residual. The residual is attributable to the unique contributions. Suppose cumulative development spend is capitalised over a five year life and the amortised balance by entity is 60 in India and 5 across the two distributors, reflecting local adaptation work they have funded. On that key the residual splits 60 to 5 (about 92.3 to 7.7), giving the Indian entity about 64.6 and the distributors about 5.4 between them.

Totals. India takes 14 plus about 64.6, or about 78.6. The distributors take 16 plus about 5.4, or about 21.4.

Sensitivity. The submission should then show what happens if the key moves. If the amortisation life is three years rather than five, or if a different measure of development contribution is used, how much does the allocation change? Where the answer is a few percentage points, the conclusion is stable and the file should say so. Where a plausible change in assumption moves the allocation materially, the key is not doing enough work and a second key should be tested alongside it.

What makes this defensible is not the elegance of the arithmetic. It is that every figure is traceable to an audited number or to a documented computation, and that the submission has already answered the question an officer will ask about each input.

9. The alternatives, and why they were rejected

A method selection analysis is not complete until it explains why the other methods were not used. For a platform licensing model the realistic alternatives are these, and each should be dealt with in a sentence or two rather than left unmentioned.

Comparable uncontrolled price. Available in principle where a comparable licence of comparable technology to an independent party can be found. In practice it fails for a proprietary platform because the comparability requirements for an intangible licence are demanding, and a licence of different technology on different terms is not a comparable price.

Resale price and cost plus. Both are one sided and both require a reliable gross margin comparison. They are unsuited to a model whose value sits in an intangible rather than in a distribution or service margin.

Transactional net margin method with the overseas distributors as tested parties. This is the alternative that should be taken most seriously, because it can be a correct approach. Where the distributors are genuinely routine and the Indian entity is the only party with unique contributions, testing the distributors and leaving the residual in India produces the same answer as a residual profit split, with less complexity. The honest position is that this is often the better method, and the case for a profit split is strongest where the overseas entities also contribute something unique, such as a market intangible they have developed and funded.

Transactional net margin method with the Indian developer as tested party. This is the method usually proposed by the Revenue, and it is the one that has to be rejected affirmatively rather than merely displaced. The objection is not that it produces an unfavourable number. It is that it treats the party owning the unique intangible as the routine party, which inverts the structure of the transaction.

10. Where these cases fail

First, the gateway is asserted rather than established. Stating that unique intangibles exist is not the same as identifying them, showing what makes them unique and showing who owns them economically.

Second, the allocation key is unsupported. A percentage without a derivation is the single most common weakness, and officers go to it first.

Third, the routine stage is skipped. A profit split that allocates total profit without first benchmarking a routine return for each party is not a residual profit split and should not be described as one.

Fourth, the method changes between years. Where the profit split method appears in a profitable year and a one sided method in a lean one, the selection looks outcome driven.

Fifth, the DEMPE evidence is documentary rather than operational. An agreement recording that the Indian entity owns the intangible is a starting point; the functional evidence is what decides it.

Sixth, the case for rejecting the alternative is not made. The officer does not have to prove the transactional net margin method is better; the taxpayer has to show it cannot be applied reliably.

Seventh, Infogain is overstated. Citing it for the proposition that unique intangibles always require a profit split invites the correction, and the correction weakens everything around it.

11. Documentation

1. A value chain analysis identifying each contribution and the party making it.
2. Identification of each intangible, what makes it unique, and the evidence of economic ownership against each DEMPE function.
3. The functional analysis for every party, including the overseas entities whose routine characterisation the analysis depends on.
4. The gateway analysis under Rule 10B(1)(d), identifying which limb or limbs apply and why.
5. The method selection analysis under Rule 10C, including why each other method was rejected and specifically why a one sided method cannot be applied reliably.
6. The routine return benchmarking for each party, with documented searches.
7. The residual allocation key, its derivation, the underlying data and a sensitivity analysis.
8. The comparability analysis supporting rejection of any comparable set proposed by the Revenue.
9. Evidence of consistency with earlier years, including any departmental acceptance.
10. The accountant’s report under section 172 of the 2025 Act (section 92E of the 1961 Act), furnished in Form No. 48 under rule 85 of the Income-tax Rules, 2026 (Form 3CEB for earlier years), describing the transaction consistently with the analysis.

Practitioner’s perspective

The profit split method is treated with suspicion in Indian practice because it is harder to verify than a one sided method and because it has sometimes been used to justify a predetermined allocation. That suspicion is a reason to document it well, not a reason to avoid it where the rule points to it.

In our experience the cases that succeed share three features. The gateway under Rule 10B(1)(d) is established on evidence rather than asserted. The allocation key is derived from auditable data and tested for sensitivity. And the submission makes the affirmative case that a one sided method cannot be applied reliably, rather than only defending the method chosen.

One further point on sequencing. Where a group expects to apply this method, the evidence should be captured as the business runs: roadmap governance, development cost by entity, decision records. Reconstructing DEMPE evidence three years later from memory and organisation charts is possible but it is never as persuasive as records that were kept because the business needed them.

The profit split method under rule 79 of the Income-tax Rules, 2026

For tax years under the Income-tax Act, 2025, the method is applied through rule 79(1)(d) of the Income-tax Rules, 2026, which implements section 165. As published on the Income Tax Department website, the rule states that the profit split method applies mainly where a transaction involves unique intangibles or unique and valuable contributions by each enterprise, or where multiple transactions are so interrelated that they cannot be evaluated separately. It then describes two approaches in terms: a contribution profit split, which divides combined net profit by relative contributions assessed on functions, assets and risks, and a residual profit split, which first allocates part of the combined profit to each enterprise for contributions that can be benchmarked against uncontrolled transactions and then divides the residual by relative contributions.

The earlier Rule 10B(1)(d) did not spell out the residual mechanics, which is why Infogain, decided under the old rule, remained the main Indian authority for a two-stage split. The explicit rule text gives a taxpayer a firmer footing for the two-stage structure described in this article, and it makes the routine-return stage a stated step rather than a matter of practice. It does not shift the burden: the gateway, economic ownership and the allocation key still have to be proved on evidence, and a transfer pricing officer may still say that a one-sided method is more reliable.

Method selection continues to be tested against rule 80, which asks which method is best suited to the facts and gives the most reliable measure of the arm’s length price, by reference to the nature of the transaction, the functions, assets and risks of each enterprise, the reliability of data, comparability and the reliability of adjustments. A file that walks through each of those factors for the profit split method and for the proposed one-sided alternative is the best protection against substitution.

Frequently asked questions

1. When does Rule 10B(1)(d) permit the profit split method?

Where the transaction involves transfer of unique intangibles, or where multiple international transactions are so interrelated that they cannot be evaluated separately for determining the arm’s length price. The two limbs are disjunctive.

2. What is the difference between a contribution split and a residual split?

A contribution analysis divides the combined profit by reference to relative contributions. A residual analysis first allocates a routine return to each party for its ordinary functions and then divides only what remains by reference to the unique contributions.

3. Does the existence of unique intangibles mean the profit split method must be used?

No. Infogain did not so hold. Its reasoning rested principally on the integration of the operations and the fact that both parties contributed materially. The method must still be shown to be the most appropriate on the facts and the available data.

4. Why can a one sided method be inappropriate here?

Because it tests the less complex party and leaves the residual to the other side by default. Where the officer tests the intangible owner against routine comparables, the method caps the owner’s return at a service margin and allocates the residual to parties that did not create it.

5. How is economic ownership of an intangible established?

By evidence against each DEMPE function: who develops, enhances, maintains, protects and exploits, who funds the activity and who bears the risk of failure. Legal ownership is the starting point only.

6. What allocation keys are acceptable for the residual?

Keys that measure contribution to the residual rather than mere activity, derived from reliable and auditable data. Cumulative development spend, DEMPE headcount or cost, and weighted value drivers are commonly used. Whichever is chosen, the derivation and a sensitivity analysis should be documented.

7. Does a high margin at the Indian entity indicate mispricing?

Not where the entity owns the intangible. In a model where one party develops and owns and others distribute, a high margin at the owner is what the arm’s length principle predicts. The differential against service comparables is the return on the intangible.

8. How does a deduction claim interact with the transfer pricing analysis?

They are separate questions and should be kept separate in the submission. The availability of a deduction is governed by its own provision and is not a reason to reallocate profit between associated enterprises.

9. Does consistency with earlier years help?

Yes. In Infogain the Tribunal directed that a split accepted by the department in adjacent years be followed. Departmental acceptance in any year should be in the submission.

10. Which provisions govern this under the Income-tax Act, 2025?

Chapter X, sections 161 to 177, in particular section 165 (section 92C, determination of arm’s length price), section 171 (section 92D, documentation) and section 172 (section 92E, the accountant’s report). Section 536, the repeal and savings provision, preserves the 1961 Act for earlier years and pending proceedings.

11. Is the residual profit split method accepted in India?

It is accepted where the facts support it, but it is not common. The Delhi Tribunal in Infogain India (19 August 2015) held the profit split method most appropriate for integrated operations where both entities contributed materially. Officers often propose TNMM instead, so acceptance depends on proof of the gateway, economic ownership of the intangible and an auditable allocation key.

12. What does rule 79 of the Income-tax Rules, 2026 say about the profit split method?

Rule 79(1)(d) applies the method mainly to unique intangibles or unique and valuable contributions, or to transactions so interrelated that they cannot be evaluated separately. It describes a contribution split and a residual split, the latter allocating benchmarkable returns first and dividing the residual by relative contributions. It succeeds Rule 10B(1)(d) for tax years under the Income-tax Act, 2025.

13. Can the tax officer replace the profit split method with TNMM?

The officer can propose it, but the choice must follow the most appropriate method factors in Rule 10C (rule 80 of the 2026 Rules). The taxpayer should show that a one-sided method cannot be applied reliably, for example because the proposed comparables own no intangibles and the tested party would be wrongly treated as the routine party.

Related reading from SBC

• When Is the Berry Ratio the Right Profit Level Indicator for a Limited Risk Contract Manufacturer in India?
• Contract R&D or Routine Software Development: How the Functional Profile Decides the Transfer Pricing Method in India
• Digital Platforms and India Tax Once the Equalisation Levy Is Gone: SEP, Treaty and PE Compared
• More articles on the SBC blog

Disclaimer

This article is for general information and does not constitute professional advice. It reflects the law as understood at the date of publication and does not take account of later developments. Statutory references give the Income-tax Act, 2025 section with the Income-tax Act, 1961 section in brackets; section 536 preserves the 1961 Act for proceedings pending on 31 March 2026 and for earlier tax years. The selection of the most appropriate method depends on the functional profile and the data available in each case. No reader should act on the basis of this article without taking advice on their own circumstances.