What to Check in a Transfer Pricing Study Report

A study report earns its place by what it records and by the date it exists on. Rule 84 of the Income-tax Rules, 2026 names thirteen items the record must contain, from the ownership structure to the working behind the margin; Rule 84(6) fixes the date it must exist on; and the tolerance notified for the year fixes how far the price charged may sit from the arm's length figure.

The rule stops short of judging the economics. A study can carry all thirteen items and still be worth refusing, on a comparable set that leaves brand value in or a method selected after the margin was known.

This note is the buyer's test of a document already in hand. It reads Rule 84 clause by clause, sets out the date the file must exist on and the three points of economics worth challenging, then lists the questions to put to the provider before the study starts.

What this note establishes

• Rule 84(1), clauses (a) to (m), names thirteen records the file must hold. Clauses (e), (i), (j) and (k) carry the arguments: functions and risks, the choice of method, the working with comparables, the assumptions behind the price.

• The record must exist on the specified date in Rule 84(6), and Rule 84(7) allows last year's file to stand only while the transaction and its underlying assumptions are unchanged.

• Retention runs nine years from the end of the relevant tax year under Rule 84(8), so a destruction schedule built on the earlier eight-year period no longer fits.

• For assessment year 2025-26 the notified tolerance is 1% of the price charged in wholesale trading and 3% in other cases, under CBDT Notification No. 157/2025.

What a transfer pricing study report is, and what it is not

A transfer pricing engagement produces two documents, and they serve different purposes. The study report is the record the Indian entity keeps and produces when asked, required by section 171(1) of the Income-tax Act, 2025 and built under Rule 84 of the 2026 Rules, notified on 20 March 2026 and in force from 1 April 2026. The accountant's report is the one filed with the return: Form 48 replaces Form 3CEB, and it is made under section 172 of the 2025 Act. The return carries the accountant's report; the study has to exist before that report can honestly be signed.

The duty sits on the entity that entered the transaction, so the Indian subsidiary, the branch or the LLP owns the file. A master file and a country-by-country report prepared by the parent sit alongside it and do not discharge it. Rule 84(2) lifts the sub-rule (1) obligation where the aggregate value of international transactions in the tax year, as recorded in the books, does not exceed one crore rupees, and sub-rule (3) still asks the assessee to substantiate that the income was computed in accordance with section 161.

A study is a working paper with a legal floor under it, and the floor is what the rest of this note tests. If your group holds one now and wants the argument read before the file is relied on, contact us.

The thirteen items the file must record

Rule 84(1) is not a summary requirement. Each clause is a separate record that a reviewer can ask to see, and the text of Rule 84 runs from (a) to (m).

Clauses (e), (i), (j) and (k) decide most arguments.

Clause (e) asks for “a description of the functions performed, risks assumed and assets employed or to be employed by the assessee and by the associated enterprises” (Rule 84(1)(e)), which is the sentence a functional analysis lives or dies on. The method clause, (i), asks for the methods considered, the one selected, the reasons for selecting it and how it was applied, and a reconstructed choice shows there. Under (j) sits the actual working, the comparable data and the adjustments made, which is the part a reviewer will try to reproduce. The last of the four, (k), covers the assumptions, policies and price negotiations that critically affected the price, and it is usually the shortest section of a thin study.

The date the study must exist: Rule 84(6) and the specified date

Rule 84(6) is one sentence, and it carries the whole timing test. The record “should, as far as possible, be contemporaneous and shall exist on the specified date referred to in section 173(d)”, in the words of sub-rule (6).

The timing test has two consequences. The file is a document of the year, built while the transaction is being priced. A study commissioned in the month an assessment notice arrives answers nothing, because the analysis may be right while the record still fails the sub-rule. Our note on India's last Form 3CEB season covers the same deadline from the filing side.

The other consequence is the relief in sub-rule (7). Where a transaction continues beyond one tax year, fresh documentation is not needed each year unless a significant change occurs in its nature or terms, in the underlying assumptions, or in any other factor affecting the price. If such a change occurs, fresh documentation has to be maintained and has to bring out its effect on the pricing. A group that rolls last year's file forward should be able to show that the test was applied transaction by transaction.

What changed on 1 April 2026: section 92D gives way to section 171

The citations in a study depend on the year it covers, and the transition has left two regimes in circulation at once.

For FY 2025-26, under the Income-tax Act, 1961, the record was required by section 92D and described in Rule 10D of the Income-tax Rules, 1962, which asked for eight years from the end of the relevant assessment year. From tax year 2026-27, section 171(1) of the 2025 Act and Rule 84 take over, with sub-rule (8) asking for nine years from the end of the relevant tax year.

The move matters beyond the extra year. The clock now starts from the tax year, so document destruction policies, retention settings on shared drives and any offshore shared-service arrangement that holds the working papers have to be re-based. Checklists published this year still quote Rule 10D and eight years from the end of the assessment year, as one 2026 guide does, and that is how a group ends up with the wrong destruction date on its own file.

The economics to test: functions, comparables and tolerance

The clauses can be checked with a pen. The economics cannot, and a file that passes on clauses can still fail here.

The functional analysis comes first. Read clause (e) against what the entity does month to month. A limited-risk distributor that the study describes as carrying inventory and market risk will produce a margin the transaction does not support, and the mismatch then sits in the file for nine years. Ask for the sentence in the study that describes this entity, as distinct from the group template it resembles.

The comparable set is second, and it turns on exclusions. The exclusion most often missed is brand value: the Delhi High Court excluded TCS E-Serve and Infosys BPO from Cadence's comparable set on that ground, as our note on the Cadence ruling sets out. A set that keeps a brand-led company in, with no record of how that was judged, hands the assessing officer a rejection to make.

The tolerance band is third. For assessment year 2025-26, CBDT Notification No. 157/2025 provides that where the variation between the arm's length price and the price at which the transaction was undertaken does not exceed 1% of that price in wholesale trading, or 3% in other cases, the price charged is deemed to be the arm's length price.

Wholesale trading is defined in the notification by a purchase-cost test, with the purchase cost of finished goods at 80% or more of total cost, and an inventory test, with average monthly closing inventory at 10% or less of sales. Inside the band the price charged is treated as the arm's length price for that year. Outside it the notification does not apply, and the file has to carry the argument. Our note on the AY 2025-26 tolerance band reads the notification against the practice it governs.

Three ways a thin study fails on its own terms

A study can be complete on clauses and still fail when a reader tries to reproduce it.

The first failure is a method chosen after the margin was known. Clause (i) asks for the methods considered, the one selected and the reasons for selecting it. If the selection memo is dated after the benchmarking output, or if the reasons describe the method's convenience rather than the transaction's facts, the reasoning has been written backwards. An officer who reads the dates will see it.

A comparable set with no written rejection criteria is the second. Clause (h) asks for the analysis performed to evaluate comparability, and that analysis is reproducible only if a reader can apply the same filters: the database, the search strategy, the industry codes, the size band, and the reason each rejected company was rejected. Where that record is missing, the set cannot be defended even when it is right.

The third is an adjustment schedule that stops where the numbers begin to differ. Clause (j) asks for the adjustments made to account for differences between the transaction and the comparables. Where the group's receivable days, payable days or inventory sit outside the comparable median, a working capital adjustment has to be computed and shown, or its absence explained. Silence in that section is the easiest thing for a reviewer to find.

Questions to put to the provider before the work starts

The questions are worth asking in writing before the engagement letter, because each answer changes what the study has to contain.

1. Which specified date will the file exist by, and under which year's rule? The answer should name a date in the compliance calendar and the rule the file is built on. A delivery date at the end of the engagement is not the same thing.

2. Which comparable database and rejection criteria will be used, and will the rejection log arrive with the draft? You should receive a schedule listing every company rejected, with the reason and the filter that removed it.

3. What happens if the tested margin falls outside the notified tolerance? The answer should say whether the price, the transaction terms or the documentation change, and it should be in writing before the return is filed.

The second question is the one that decides the file, and our comparable-company benchmarking work is built around it. Asked after the draft arrives, the same questions become a negotiation over a document the group has already paid for.

If your group is commissioning a study for its first year under the 2025 Act, or holds a delivered file and wants the economics tested, contact us. We will read the functional analysis, the comparable set and the adjustment schedule against the clauses above and tell you where the file would be attacked, with the research in this note closed on 2 November 2026.