INTM485120 - Transfer pricing operational guidance: Evidence gathering: Searching for comparables: range of results

Usually a transfer pricing model will produce a range of possible results. Case teams must use all available evidence to conclude what the arm’s length price would have been. If the results of the tested party fall within an acceptable range of arm’s length prices, then no adjustment should be made. If the results of the tested party fall outside an acceptable range teams will need to agree how to revise the tax computation so that the arm’s length price replaces the actual transfer price.

Key issues to consider with a range of results

Generally there are three key issues to consider when looking at a wide range of arm’s length results:

  • Whether the results are reasonably comparable
  • How the range has been calculated (transfer pricing reports usually use an interquartile range)
  • Where in the range the arm’s length price should fall?

Case teams should review each company individually to determine whether it truly is comparable. Even after they have carried out that exercise there may be a number of potentially comparable companies, and the range of comparable results can be quite large. Teams should always consider all of the facts and circumstances to make an informed judgement on where in a range a tested party would be at arm’s length.

One of the most important tasks for teams when putting forward their calculations and transfer pricing model is being able to demonstrate that the range of results should be as narrow as possible to reflect the fact that functionally similar companies which are accurately comparable are likely to have similar results.

The use of an interquartile range

A transfer pricing report usually produces a list of companies proposed as comparables. The results are often summarised as an interquartile range. An interquartile range discards the results of the bottom quarter and top quarter of the results. The median is the mid-point of the interquartile range. The median will generally produce a different result to the average of the range being considered. In a case where all the comparables being used are more or less equally valid, and there is no reason why the tested company is any better performance wise than those comparables companies, then there is probably nothing wrong with using the interquartile range.

There is nothing in either TIOPA10/Part 4 or the OECD Transfer Pricing Guidelines that say that an interquartile range must be used, although paragraph 3.57 of the Guidelines states that the use of an interquartile range may enhance the reliability of a range in which non-quantifiable comparability defects remain as a result of the limitations in available information on the comparables used.

A potential problem with using the interquartile range is the discarding of more accurate comparables which fall within the full range but outside the inter-quartile range. This problem arises when some of the companies in the reported list are less reliable comparables than others. It is therefore important to carry out as robust a comparability analysis as is reasonably possible in arriving at the arm’s length range from which the inter-quartile range is derived.

If case teams are satisfied that the comparables are all highly reliable, then there is no need to restrict themselves to using an interquartile range. The task is to find the accurate comparables.

Factors to take into account when trying to narrow the range

To narrow the range of results case teams will have to consider the comparable companies put forward very carefully. This involves looking at the available information about the comparable companies, including what they say about themselves on their websites. Think about the following points:

  • Should any of the companies obviously be excluded? Are there any other companies which should be included? This may involve a search of commercial databases.
  • Is there a subset of comparables within the larger range? For example, consider a company carrying out contract R & D in the field of computer software. The transfer pricing report may contain 16 comparable companies carrying out contract R & D in the computer field (ranging from hardware, operating systems, communications, switching and software), but that there are 3 companies involved in just software R & D. Why not use just those 3 companies as a starting point? These companies should in theory be more comparable to the tested party.

Decide where, within a range of results, to set the transfer price

The OECD Guidelines say that if the results of the tested party fall within the arm’s length range of results (or the inter-quartile range where this is used to enhance the reliability), then no adjustment should be made. If they fall outside that range then the tax authority will need to determine where in the range the transfer price should be set (see paragraphs 3.60 to 3.62 of the Guidelines).

The reliability of the comparability analysis should be taken into account in deciding where, in a range, the transfer price should be set. If all points in the range are considered highly reliable, it could be argued that any point in the range satisfies the arm’s length principle. However, if unidentifiable or unquantifiable comparability defects remain (for example, due to limitations in information available on the comparable transactions) the use of measures such as the median, mean or weighted average and so on may be useful in deciding where to set the transfer price. In all cases, selecting the appropriate measure maximises the likelihood that the adjusted price falls within the true arm’s length range. Case teams should always look at a tested party in the round to consider what its results would be if it was independent. It might be that a well-run company, efficient and competitive, would profit more that one where the opposite was true. On other occasions, these factors might not materially affect profit in a commercial world. A company which is inefficient might still have a product that is profitable.