Transfer pricing: operational guidance: examining transfer pricing reports: arm’s length range
A usual feature of transfer pricing reports is the inclusion of a range of results that companies put forward as comparables, presented as the arm’s length range into which the company fits. Further analysis may be of a statistical nature, concentrating on the interquartile range of these results (see INTM485120) and in cases where the actual results of the tested party are within this interquartile range, the report will typically conclude that the transfer pricing must be arm’s length.
Considering a range of results is appropriate because, however strict the comparability requirements, companies will have differing margins, finance costs, volume, markets, attitude to risk-taking and so on. Gathering evidence for an arm’s length price is considered in detail at INTM485000 onwards, but the point to be made in relation to transfer pricing reports is that even rigorous selection is likely to produce a range of possible arm’s length answers. This can be reduced by careful analysis, adjustment, and filtering out of inappropriate or anomalous companies. This should narrow the range to a more realistic and useable sample. A wide range may of course include the tested party, but if it would also include thousands of other companies in a variety of industries, it isn’t of much help.
The OECD lists a number of factors of comparability starting at 1.38 of the Transfer Pricing Guidelines, but recognises that a range of results is appropriate and that if the results of the tested party are within this range then it is reasonable to conclude that its results are arm’s length. However, while the OECD recognises that use of statistical tools such as the inter-quartile range might help to enhance the reliability of such a range, the range must be valid in the first place.
Case teams should be aware of the potential distinction between an arm’s length range as extracted from a careful analysis of suitable companies and a range which consists of the results of independent companies which are no more than superficially similar to the tested party. The latter is not an arm’s length range within the meaning of the Guidelines. Paragraph 3.56 of the Guidelines provides that where any result within the range can be determined to have a lesser degree of comparability with the tested transaction than others, it should be eliminated from the range. It is not unusual to see fairly long lists of companies bearing little obvious similarity to the tested party only being narrowed down “scientifically” by the use of statistical tools.
Benchmarking studies that return a wide range of results should be treated with some caution as it is likely that the comparables at the extreme end of the range may not be of the same degree of comparability. On the other hand, it is dangerous to assume that the closest comparables are those that cluster around the centre of the range. A small number of strong comparables is likely to give a more accurate result than a large number of weak ones.
Where the results of the tested party fall outside the range, the question arises as to where within the range its transfer prices should be adjusted; see INTM485120 for guidance on this point.