The ruling of the Spanish High Court (“Audiencia Nacional”) nº 1072/2019, of 6 March, has pronounced on two relevant aspects in the practical application of the Transactional Net Margin Method (TNMM):

• The use of multi-year averages in both the comparable entities and the tested party; and
• The most suitable point in the interquartile range at which to make an adjustment, if any.

Brief statement of facts and controversial issues

The taxpayer was part of a multinational group that operates in the home and office furniture distribution sector, and its role within the value chain is to act as a wholesale distributor in Spain.

The TNMM was used to verify the arm’s length value of product purchase and sale transactions. The return on sales (ROS) was selected as a profit level indicator (PLI), and a search for comparable independent companies was provided to the Spanish Tax Administration.

In tax year 2007, the taxpayer’s result was below the lower quartile (outside the interquartile range) of the results obtained by comparable independent companies, measured as the average of the 2003-2005 years. Given this situation, the Spanish Tax Authorities proceeded to carry out a transfer pricing adjustment that corrected the taxpayer’s operating result so that the ROS was in the median of the distribution of average results of comparable companies.

Use of multiple-year data

The first controversial issue was the taxpayer’s claim to use multiple years’ average data not only to conform to the arm’s length range obtained from the search for comparable independent companies, but also to the result of the tested party, that is to say, of the taxpayer’s controlled transactions which are subject to review. Specifically, the taxpayer presented the average of the results obtained in the 2006-2008 financial years, which was slightly above the lower quartile.

In view of the lack of specific provisions in the consolidated text of the Corporate Income Tax Law (CITL) and its regulations (also maintained in the legislation in force nowadays), both the Spanish Tax Administration and the Spanish High Court must rely on the recommendations contained in the OECD Transfer Pricing Guidelines, which, as indicated in the Statement of Reasons for Law 36/2006, on measures for the prevention of tax fraud (and in the CITL in force nowadays), must be used as an interpretative source of the Arm’s Length Principle and the Spanish regulations on related-party transactions. Specifically, paragraphs 3.75 to 3.79 of the Guidelines are cited in the Judgment, which deal specifically with situations in which the use of multi-year data is necessary to improve comparability.

Based on these recommendations, the Second Legal Basis of the Judgment establishes three application criteria:

• Ideally, data from «contemporaneous uncontrolled transactions» should be used, which reflects the behavior of independent parties in an economic environment identical to that in which the taxpayer’s controlled transactions took place;
• In certain cases and for certain reasons, » examining multiple year data is often useful in a comparability analysis, but it is not a systematic requirement»;
• The multiple-year data could be used in the comparability analysis, but once the arm’s length range is determined, the tested party’s result should be evaluated individually and in relation to the results of each tax year subject to examination.

Although the taxpayer alleges before the court certain economic circumstances regarding the related-party transactions and the levels of risk borne, in order to support the need to use multiple-year data of the results of the tested party as well, the evidence provided of such economic circumstances is not perceived as sufficiently solid to meet the taxpayer’s claim. Therefore, the Tax Administration is right on this point.

The most appropriate point in the range

Having concluded that the taxpayer’s results in 2007 were outside the interquartile range, the second controversial is the point in that range at which the transfer pricing adjustment must be made.

At this point, the Spanish Tax Administration automatically adjusted the taxpayer’s results to the median of the interquartile range, without further reasoning or argument as to the appropriateness or preference of this point with respect to other points within the arm’s length range.

Actually, the automatic adjustment to the median of the interquartile range comes from United States legislation, where it is mandatory under the Code of Federal Regulations (section 1.482-1):

“(3) Adjustment if taxpayer’s results are outside arm’s length range.

If the results of a controlled transaction fall outside the arm’s length range, the district director may make allocations that adjust the controlled taxpayer’s result to any point within the arm’s length range.

If the interquartile range is used to determine the arm’s length range, such adjustment will ordinarily be to the median of all the results. The median is the 50th percentile of the results, which is determined in a manner analogous to that described in paragraph (e)(2)(iii)(C) of this section (Interquartile range).

In other cases, an adjustment normally will be made to the arithmetic mean of all the results.

See §1.482–1(f)(2)(iii)(D) for determination of an adjustment when a controlled taxpayer’s result for a multiple year period falls outside an arm’s length range consisting of the average results of uncontrolled comparables over the same period.”

Since the Spanish transfer pricing regulations do not provide any guidance in this aspect, the Court again relies on the OECD’s Transfer Pricing Guidelines as a source of interpretative criteria on the application of the Arm’s Length Principle and the Transactional Net Margin Method. Indeed, paragraph 3.62 states that:

“In determining this point, where the range comprises results of relatively equal and high reliability, it could be argued that any point in the range satisfies the arm’s length principle. Where comparability defects remain as discussed at paragraph 3.57, it may be appropriate to use measures of central tendency to determine this point (for instance the median, the mean or weighted averages, etc., depending on the specific characteristics of the data set), in order to minimise the risk of error due to unknown or unquantifiable remaining comparability defects.”

As the Guidelines state, the Court emphasizes that, in order for the adjustment to be made to the median, there must be defects in comparability between the selected comparable companies and the tested party, and that these differences cannot be quantified for the purpose of correcting them by means of a comparability adjustment.

Since the previous reviewing body (the Central Economic-Administrative Tribunal) expressly ruled out such comparability defects in the search for comparable companies when evaluating the 2008 results, the Spanish High Court does not consider that there are reasons to substantiate the adjustment to the median, and therefore the general rule contained in the Guidelines applies: any point within the interquartile range is equally reliable and valid for making the adjustment (and for the benefit of the taxpayer, it is made at the closest point in the range).

Our assessment of the Spanish High Court Ruling

The ruling of the Spanish High Court of 6 March 2019 is crucial to establish the criteria for the practical application of the TNMM.

In the first place, as a matter of substance, both the TEAC resolution appealed and the Judgement of the High Court fully assume the use of the interquartile range as a valid and necessary statistical method to improve comparability in the searches of comparable companies carried out in databases. In fact, its use is not called into question at any time, even when the case refers to a fiscal year (2007) before the modification of the OECD Guidelines in 2010, which expressly included the use of «truncated ranges» by statistical methods to improve their reliability, or the most recent incorporation of this technique in the new Corporate Income Tax Regulation (art. 17.7 of Royal Decree 634/2015, of 10 July):

«When, despite the lack of sufficient data, it has been possible to determine a range of values that reasonably comply with the principle of free competition, taking into account the process of selecting comparables and the limitations of the information available, statistical measures may be used to minimise the risk of error caused by defects in comparability.”

The second fundamental issue to be highlighted is the full acceptance by the Spanish High Court of the OECD’s Transfer Pricing Guidelines as a valid interpretative source of the Spanish legal and regulatory rules on related-party transactions, especially regarding the technical details of the application of valuation methods, the complexity of which is not insignificant.

With respect to the litigious questions of the case, the High Court aligns its pronouncement with the criteria established by the OECD in its Transfer Pricing Guidelines, and therefore with the international standard in the matter:

1) The use of multiple-year data must be limited to data from comparable companies, in order to make a more reliable calculation of the arm’s length range. The results of the tested party should, in principle, refer to the tax year in which the controlled transactions subject to analysis were carried out.

This should not be interpreted as prohibiting the application of multiannual data to the tested party in all cases. However, the onus probandi of the circumstances justifying the appropriateness of using multiannual data to analyse the related-party transactions falls on the taxpayer.

2) In principle, all points in the arm’s length range would be equally reliable and valid for carrying out a transfer price adjustment.

In this case, the Tax Administration has the burden of proving the comparability defects of the sample of comparables in order to select a specific point or apply a statistical measure (median, arithmetic average) to reduce the risk of error.