The benchmark range and median concept in transfer pricing refers to the statistical interval derived from a set of comparable uncontrolled transactions or entities, within which an arm’s length price or margin is considered to fall. Where a controlled transaction’s result falls outside that range, tax authorities may adjust it to a point within the range — typically the median. The legal foundation rests on the arm’s length standard in Article 9 of the OECD Model Tax Convention and corresponding domestic provisions, such as Article 57 of the French General Tax Code or analogous Romanian and Polish legislation. The concept recognises that the arm’s length principle does not produce a single price but rather a range of acceptable outcomes, and disputes arise precisely over how that range is constructed and which point within it governs any adjustment.

Disputes typically arise in two forms. First, tax authorities challenge the composition of the comparables set, arguing that the taxpayer’s database search — whether using Thomson Reuters LoanConnector, Bureau van Dijk, or similar tools — applied insufficiently rigorous selection criteria, producing a range that is too wide or unrepresentative. Second, where the controlled transaction falls outside an accepted range, the question becomes whether an adjustment should be made to the median or to the nearest point on the range. Taxpayers frequently argue, as in the Romanian case of A. Median S.R.L., that any point within the range satisfies the arm’s length standard, whereas authorities assert a mandatory adjustment to the median. The choice of transfer pricing method — CUP, TNMM, or cost-plus — also affects range construction, as illustrated by disputes in Kenya, Poland, and Italy.

The governing framework is Chapter III of the OECD Transfer Pricing Guidelines, particularly paragraphs 3.55 to 3.66, which address the arm’s length range and the use of interquartile or other statistical measures. Paragraph 3.62 specifically contemplates adjustment to the median where a point outside the range cannot be identified as more reliable. The 2022 OECD Guidelines maintain this approach without prescribing a single statistical measure, leaving room for domestic variation.

Courts examine whether the comparables search was systematic and reproducible, whether selection criteria were consistently applied, and whether rejected comparables were excluded on defensible functional or financial grounds. In Gas-Trader and Stream-Heat, Hungarian courts scrutinised the database methodology closely and found for the taxpayer where the authority’s own range construction was flawed. The contested legal question across jurisdictions is whether adjustment to the median is obligatory or merely permissible when a result falls outside the range.

These cases matter because benchmark range disputes arise in virtually every transfer pricing audit, making the methodology for constructing and applying a comparables range one of the most practically significant issues a transfer pricing practitioner will encounter.