Income Tax Transfer Pricing 5 March 2026 · 16 min read

When the Revenue Cherry-Picks:
ITAT Delhi on the Limited Risk Model, Berry Ratio, and the Boundaries of the "Other Method"

A recent ITAT Delhi order doesn't just decide a case — it draws bright lines around four recurring transfer pricing and income tax controversies that affect every MNE operating in India.

A February 2026 order of the Delhi ITAT (ITA No. 442/DEL/2017, AY 2012-13) has produced what may be one of the most analytically rigorous transfer pricing decisions in recent years. While the facts involve a telecom subsidiary of a global group, the legal principles it articulates extend far beyond any single industry. Four distinct controversies were addressed — each with implications that practitioners across sectors need to understand.

I. The Limited Risk Model Is Not a Tax Avoidance Device — It Is a Business Reality

The central battleground was the Limited Risk Model (LRM) — a structure where a multinational group designates one entity as the "entrepreneur" (bearing strategic risk, owning IP, controlling R&D and global operations) and compensates operating subsidiaries at an assured return for their limited-risk, routine functions.

The Revenue's attack was fundamental: it characterised the LRM as a contractual fiction designed to cap the Indian entity's upside profits and shift residual returns offshore. The TPO's reasoning was blunt — no independent third party would agree to have its profits capped while bearing real operational risk. The contractual allocation of risk, in the TPO's view, could not override economic substance.

This is a challenge that practitioners encounter repeatedly. Revenue authorities often treat intra-group agreements with suspicion — as though any arrangement that limits an Indian entity's profit must be abusive. The Tribunal's response was measured but decisive.

What the Tribunal Actually Held

The Tribunal did not simply accept the LRM on faith. It undertook an independent examination of the inter-company service agreement, the compensation formula, the actual financial results, and the functional analysis of both the Indian entity and its overseas principal. Its findings rested on several pillars:

First, substance matched form. The Indian entity genuinely operated as a delivery arm — it "completed the communication loop" for services initiated overseas. It did not control global strategy, R&D, brand, or customer acquisition. The value it added was proportional to its operating expenses, not to the total revenue flowing through it. This is the classic "spoke" in a hub-and-spoke model.

Second, the LRM was applied globally and uniformly. Every operating company in the group — across multiple jurisdictions — was compensated under the same formula. The model was not a device tailored for India; it was the group's actual operating structure, accepted by tax authorities in other countries. The Tribunal observed:

"The LRM is not only accepted and adopted by the group in India as well as across the globe, all the OpCos are remunerated similarly and the same is accepted in their respective tax jurisdictions. Therefore, we should accept the uniform method adopted by the group across the globe."

Third — and this is the most interesting finding — the Indian entity was actually under-compensated. The agreed formula guaranteed the higher of 11% on net sales or 14% on value-added expenses. In practice, the entity received only the 14% on VAE (approximately ₹19.50 crores), when 11% on net third-party sales would have yielded ₹43.31 crores. The Tribunal directed an adjustment of ₹23.81 crores — not in the Revenue's favour, but to ensure the assessee received what the LRM itself promised.

This is a remarkable finding. The Revenue sought an upward adjustment of ₹83.50 crores, arguing that the LRM was abusive. The Tribunal found that the entity was actually owed more under its own agreement — just not as much as the Revenue claimed.

The Legal Principle

An LRM is defensible when (a) the functional analysis supports limited-risk characterisation, (b) the model is applied consistently across jurisdictions, and (c) the compensation formula is verifiable against actual results. But — and this is the warning — the Tribunal will independently verify whether the entity is being compensated per its own agreement. If the formula says "higher of X or Y" and you're only paying the lower, expect an adjustment — potentially in your favour, but an adjustment nonetheless.

II. Berry Ratio Has a Place — When Functions Are Proportional to Costs, Not Sales

The Revenue argued forcefully against the use of the Berry Ratio (gross profit / operating expenses) as the Profit Level Indicator (PLI) under TNMM. Relying on Sumitomo Corporation India v. CIT (2016) 387 ITR 611 (Del.), the DR submitted that the Berry Ratio is inappropriate where the assessee owns significant fixed assets or uses intangibles — because the value of such assets is not captured in operating costs.

The Tribunal distinguished Sumitomo on facts and applied the OECD Guidelines criteria for Berry Ratio appropriateness:

Where a subsidiary merely "adds value" to services flowing through a global network — without controlling pricing, customer relationships, or strategic direction — the Berry Ratio captures the economic reality better than sales-based margins. The entity's contribution is measured by what it spends to deliver, not what it charges.

The Legal Principle

The Berry Ratio is a valid PLI under TNMM when the tested party's value contribution is proportional to its operating costs and it does not perform entrepreneurial functions. The blanket objection that "Berry Ratio cannot be used because the entity owns fixed assets" does not hold where the entity's role is fundamentally that of a service delivery arm. The OECD Guidelines — particularly paragraphs 2.106 to 2.108 — provide the analytical framework, and the Tribunal applied it rigorously.

III. The "Other Method" Is Not a Blank Cheque

Perhaps the most consequential holding for everyday transfer pricing practice: the Tribunal's treatment of the Other Method under Rule 10AB.

The TPO rejected TNMM and applied the Other Method to determine the arm's length price. His reasoning: since the LRM adjustment was essentially an accounting entry shifting profits, it should be benchmarked by asking "what would an independent party do?" — and concluding that no independent party would accept profit-capping. Therefore, the LRM adjustment was reduced to NIL, creating an upward adjustment of ₹83.50 crores.

The problem? The TPO did not identify a single comparable uncontrolled transaction.

Rule 10AB is explicit — the Other Method requires comparison with "the price which has been charged or paid, or would have been charged or paid, for the same or similar uncontrolled transaction, with or between non-associated enterprises, under similar circumstances." It is not a licence to hypothesise about what a rational party "would" do in the abstract.

The Tribunal relied on a formidable line of authority:

The Legal Principle

The Other Method under Rule 10AB is not a residuary power to impose the TPO's subjective view of what arm's length looks like. It requires identification of comparable uncontrolled transactions — the same foundational requirement that applies to every prescribed method under Section 92C. A TPO who rejects TNMM without demonstrating its inadequacy, and then applies the Other Method without any comparable data, produces an adjustment that is legally unsustainable.

The Tribunal's observation deserves to be quoted in every TP proceeding where the Other Method is invoked without rigour:

"It cannot be a situation where the revenue cherry picks the results."

IV. Telecom Charges Are Not Royalty — And Retrospective Amendments Cannot Rewrite Treaties

The second major issue involved the disallowance of approximately ₹161 crores under Section 40(a)(i) for alleged non-deduction of tax on payments to overseas telecom operators. The Revenue's position: these payments for data transmission constituted "royalty" under Section 9(1)(vi), as amended by the Finance Act 2012 (with retrospective effect from 1976), and therefore attracted withholding obligations under Section 195.

This is a controversy that has consumed enormous judicial bandwidth over the past decade. The Finance Act 2012 expanded the definition of "royalty" under the Act to include payments for the use of any "process" — and the Revenue has aggressively argued that data transmission involves the use of a "process" embedded in telecom infrastructure.

The Settled Position

The Tribunal applied what is now settled law in the jurisdiction of the Delhi High Court:

1. Data transmission services are not royalty. The Delhi High Court in DIT v. New Skies Satellites BV and Asia Satellite Telecommunications Co. Ltd. held — after threadbare analysis — that providing bandwidth or data transmission capacity does not amount to granting the right to use a "process." The customer receives a service; it does not acquire any right in the underlying equipment or technology.

2. Retrospective amendments to domestic law cannot be read into treaties. Even if the Finance Act 2012 expanded the definition of "royalty" under the Act, that expansion cannot be imported into the India-US DTAA (or any other treaty). The treaty definition of royalty is autonomous and must be interpreted on its own terms. The Delhi High Court was explicit: the amendment "cannot be read in the DTAA."

3. Retrospective amendments cannot create retrospective withholding obligations. The Supreme Court in Engineering Analysis Centre of Excellence v. CIT [2021] held that even if a statutory amendment operates retrospectively, it cannot impose TDS obligations on transactions completed before the amendment. The payer cannot be penalised for not deducting tax based on a law that did not exist at the time of payment.

The Legal Principle

For payments covered by a DTAA, the treaty definition of "royalty" governs — domestic amendments (including retrospective ones) do not alter treaty obligations. This principle is settled at the High Court level in Delhi and reinforced by the Supreme Court's reasoning in Engineering Analysis Centre. Any Section 40(a)(i) disallowance premised on the expanded domestic definition of royalty, where a more restrictive treaty definition applies, is unsustainable.

V. A New Licence Does Not Mean a New Undertaking Under Section 80-IA

The final issue is deceptively simple but recurs with surprising frequency: when a telecom operator obtains a new regulatory licence (e.g., NLD/ILD) after the sunset date of 31 March 2005, does that constitute a "new undertaking" — disentitling it from Section 80-IA deductions?

The Revenue argued yes — a new licence means a new undertaking, and since the NLD/ILD licences were obtained in 2008, the sunset clause was not satisfied.

The Tribunal, following its own consistent decisions across multiple assessment years, held that the test is substantive, not formal:

The Legal Principle

Section 80-IA eligibility turns on the commencement of telecommunication services, not on the acquisition of regulatory licences. Where the fundamental nature of services remains unchanged and the new licence merely enhances or secures existing operations, no new undertaking is created. The test — drawn from CIT v. Associated Cement Companies Ltd. (118 ITR 406, Bombay HC) and CIT v. Adarsh Cold Storage (280 ITR 58, Allahabad HC) — is whether the new activity is independently capable of functioning without the base of the existing undertaking.

VI. The Consistency Doctrine: Radhaswami Satsang Still Applies

Running through the entire order is a thread that practitioners too often undervalue: the principle of consistency.

The LRM had been accepted by the TPO in all prior assessment years. The Section 80-IA position had been decided in the assessee's favour for four prior years. The Section 40(a)(i) issue was settled at the High Court level.

The Tribunal invoked Radhaswami Satsang v. CIT (193 ITR 321, SC): where a fundamental aspect permeating through different assessment years has been found as a fact one way, and the parties have allowed that position to be sustained by not challenging it, the position cannot be changed in a subsequent year.

This is not just a procedural point — it reflects a deeper principle of tax administration. The Revenue cannot accept a business model when it produces modest results and then attack the same model when it produces better results. As the Tribunal put it:

"It cannot be a situation where the revenue cherry picks the results."

Concluding Observations

This order is significant not because of its factual complexity — the facts of a telecom subsidiary's operations are relatively straightforward — but because of the clarity of the legal principles it articulates:

  1. LRM structures are legitimate where substantiated by functional analysis, global consistency, and verifiable compensation formulas.
  2. The Berry Ratio is appropriate where functions are proportional to costs, not sales — and the OECD criteria provide the analytical framework.
  3. The Other Method under Rule 10AB requires comparable uncontrolled transactions. It is not a residuary power for ad hoc determinations.
  4. Treaty definitions of royalty are autonomous. Domestic amendments — even retrospective ones — do not alter treaty obligations or create retrospective withholding duties.
  5. Regulatory licences do not determine Section 80-IA eligibility. The substance of services does.
  6. Consistency matters. The Revenue cannot flip positions across years without changed circumstances.

For practitioners dealing with transfer pricing disputes, Section 40(a)(i) disallowances, or Section 80-IA claims, this order provides a well-reasoned roadmap — both for structuring arguments and for understanding where the Tribunal draws its lines.

Order Reference: ITA No. 442/DEL/2017 (AY 2012-13) — ITAT Delhi Bench "I" — Order dated 25 February 2026.

The AS Lex · Tax & Legal Practice · theaslex.in