How royalty rate benchmarking sets the rate you can defend

Royalty rate benchmarking prices a licence between group companies against what unrelated parties pay for comparable rights. It produces a range, each rate carrying its own conditions, and that range holds only while the comparable set, the adjustments and the file behind them still answer the Indian provisions.

A study ages faster than the agreement it supports. A licence signed for five years gets tested years later against comparables since superseded. In Tupperware India the Delhi Tribunal recorded an officer who rejected the documentation and set his own rate; in Mikuni India the royalty was priced against four comparable agreements.

• A benchmark returns a range from comparable uncontrolled licences, and a group's rate is one position inside it.

• Royalty comparables are thinner and older than margin comparables: disclosure is voluntary, and each rate sits on its own base.

• Mikuni India priced a royalty under CUP against four agreements; in Tupperware India the officer discarded the file and searched again.

• Section 92D and Rule 10D want the file ready when the rate is chosen, since an officer can supply a rate if it is thin.

What royalty rate benchmarking is, and what one number cannot show

Benchmarking a royalty means finding what unrelated parties pay for the same rights, in the same markets on the same terms, and pricing the licence on that evidence. Group policy is a starting point at best, and the comparable agreements carry the case.

A single percentage settles none of the questions that decide whether a rate holds. Which intangible is licensed matters, since a trademark, patented technology and know-how are priced on different evidence. So do the field of use, the territory, the term, exclusivity and the right to sublicense.

Where royalty comparables come from, and why the set is thinner

Royalty comparables come from three places: licences the group itself has signed with unrelated parties, agreements that become public through a filing or a court record, and databases that collect disclosed rates with their licensing terms.

That pool is thinner and older than a margin set. Disclosure is voluntary and often partial, so a rate can travel without the territory, exclusivity or term that gave it meaning. Each rate sits on a base the agreement defines, and a licence filed years ago describes an intangible that has since been rebranded.

The sellers say as much. ktMINE treats rates as comparable only once each payment is mapped to a common base, from gross sales to per unit: "This common base also allows ktMINE to provide meaningful statistical analysis of agreements." RoyaltyStat concedes that industry rate statistics "should not be treated as a reliable benchmark on their own".

Selection is a decided question. In the Cadence Design Systems case the Tribunal examined which entities could sit in a set, and excluding high-brand-value entities from a comparable set is the note on that decision.

Reading the interquartile range against the rate you propose

A benchmark reports a first quartile, a median and a third quartile, with the count of agreements behind them. The reading puts the group's rate against those numbers and turns the gap into rupees.

Suppose a licensee has a royalty base of Rs 500 crore and proposes 4%. If the set runs from 1% to 5% with a median of 2.5%, the proposal sits in the upper half, and 1.5 percentage points above the median is Rs 7.5 crore a year. A premium that size can be defended when the licence gives something the comparables do not: exclusivity, a long term, or wider rights.

The range is not a safe harbour. A position inside it does not settle arm's length, because an officer can test whether the comparables themselves hold up, and a tolerance band answers a different question.

The adjustments that move the rate before comparison

A rate taken from a database and the rate a group proposes rarely describe the same grant. Exclusivity comes first: a sole licence in a territory commands more than a non-exclusive one, and a set mixing the two must price the difference or separate the licence. Territory follows, then the term, including any renewal clause.

Bundling breaks more files than any other adjustment. A payment described as a trademark royalty may also buy know-how, design or technical services, and only the trademark element can be measured against a trademark licence. The Delhi High Court took such a payment apart in the LG Electronics case, holding one-third of a USD 11 million ICC sponsorship payment to be royalty; how LG Electronics split the payment reads that decision.

Sublicensing is the last adjustment, and the easiest to miss, since a licensee that may appoint sub-licensees holds a right the comparables may not. The OECD's Chapter VI comparability factors for intangibles give the checklist, and RoyaltyRange's summary carries the test that settles most of them: whether the data "is sufficiently detailed to permit an evaluation of the specific features of intangibles".

What the file has to show: section 92, Rule 10B and Rule 10D

Section 92 charges the income from an international transaction at arm's length, and a royalty to a parent is such a transaction. Rule 10B governs the choice of the most appropriate method and the comparability factors recorded when it is applied. Rule 10D, read with section 92D, requires the documentation to exist alongside the transaction.

A file that answers those provisions holds the functional analysis of the licensee, the comparable set with the criteria that produced it, a record of the comparables rejected and why, the adjustment made for each difference priced, and the range with the group's position inside it. The rejection record is the part most often thin. The wider file, including the Master File and the country-by-country report, is set out in what the documentation must prove.

How a TPO tests the rate, inside the range and outside it

In the ITAT Jaipur decision in Mikuni India the Tribunal had before it a royalty of Rs 6,08,80,707 paid to the Japanese parent for the 2016-17 year, recorded four comparable agreements, and priced the royalty under the CUP method.

Tupperware India goes further. The ITAT Delhi decision of 17 April 2025 records the officer setting aside the documentation, including the agreements the taxpayer had chosen as comparables, and fixing an arm's length royalty of 2% of sales from a search he conducted himself. The royalty on a substantive basis came to INR 510,510,213.

A rate inside the range is not safe if the comparables can be taken apart, and a rate outside it invites substitution. The Delhi High Court in Sony India held that a royalty cannot be cut to nil without the most appropriate method and its comparable data; the reasoning is in the Sony India royalty ruling.

When to commission the study, and what to give the adviser

A benchmark needs rebuilding when the facts behind it move: a new licence, a new territory, a change in term or exclusivity, or a file that predates the years under assessment.

Bring the licence agreement, the functional analysis of the licensee, any earlier study and any order already on record; a benchmark that ignores a rate an officer has fixed reads as a repetition of the argument he rejected. Where that help is useful, you can commission a royalty and interest-rate benchmarking study, and we will start from the agreement you hold. If a call is easier, the contact page reaches us.

This note rests on sections 92, 92C and 92D and Rules 10B and 10D, on the ITAT Jaipur decision in Mikuni India of 23 January 2025 and the ITAT Delhi decision in Tupperware India of 17 April 2025, and on the OECD's Chapter VI comparability factors for intangibles. Research closed on 11 November 2026.