The Arm’s Length Principle Is a Question, Not a Formula

What a benchmarking study can miss

Transfer pricing documentation has settled into a familiar shape. Select a method, run a database search, apply filters, compute a range of margins, observe that the tested party falls inside it, and conclude that the transaction is at arm’s length. The file is thick, the arithmetic is sound, and the conclusion is stated with confidence.

And then the transfer pricing officer asks a question the file does not answer: why would an independent enterprise have entered into this transaction at all?

A fair price does not justify every transaction

Two neighbouring households agree that one will lend the other a ladder each monsoon, and that the borrower will pay for it. To ask whether the payment is fair, you would compare it with what ladders rent for locally. That is a sensible comparison, and it is what a benchmarking study does.

But suppose the borrowing household owns three ladders already. Now no rental rate is defensible, because an independent household in that position would not have rented a ladder at any price. The comparison of rates was never the whole question. It presumed that the transaction itself was one an independent party would have undertaken.

Commercial rationale, risk and comparability

The arm’s length principle asks what independent enterprises, dealing at arm’s length in comparable circumstances, would have agreed. That question has two parts, and benchmarking only addresses the second.

The first part is whether the transaction, as structured, is commercially rational for both sides. This is where the analysis of functions, assets and risks does its real work. It is not a preliminary section to be completed before the numbers begin — it determines which party should be earning the return, because in an arm’s length arrangement return follows risk, and risk follows control over risk. An entity that bears a risk it has no capability to control, and no financial capacity to absorb, would not have accepted that risk from an independent counterparty.

The second part — the pricing — then proceeds on the basis established by the first. Method selection is a consequence of the functional analysis, not a free choice. A transactional net margin method applied to a party that in truth performs the entrepreneurial functions will produce a defensible-looking margin for the wrong entity. The range will be met and the conclusion will still be wrong.

This is also why comparability adjustments are not cosmetic. If the comparables bear working capital, credit risk or capacity utilisation materially different from the tested party, the range describes a different business. A wide range is usually evidence of weak comparability, not of a generous safe harbour.

Build a consistent transfer pricing narrative

Begin the file with the functional analysis and let it constrain everything after it. If the analysis says the Indian entity develops and controls the intangible, do not select a method that treats it as a routine service provider because the data is easier to find.

Test the commercial rationality of the arrangement explicitly, in writing. Intra-group services, management fees, guarantee fees and intangible royalties all attract the same challenge: would an independent party have paid for this, and what did it receive? Evidence of benefit received — not merely of cost incurred and allocated — is what carries that argument.

And treat the master file, local file and country-by-country report as one narrative rather than three filings. Where the global value chain described centrally contradicts the functional characterisation asserted locally, the inconsistency is the finding. It is found more often than any pricing error.