THE ARM’S LENGTH PRINCIPLE – GOING BEYOND THE RANGE

A common misconception in Transfer Pricing is that the arm's length principle is a notion that is very closely linked to the tool that is the interquartile range. It is easy to see why such an error can occur, as the notions of arm's length principle and the arm's length price are often used interchangeably, the TNMM method is the preferred way of approximating "normal" profit levels, and the interquartile range is a very commonly used tool when applying a TNMM. However, there are two main points where this is a gross exaggeration, and the effects are compounded by the fact that tax authorities in some jurisdictions are turning such simplifications into law.

1. The arm's length principle is not about prices or profits

The arm's length principle (or rather, its opposite) is defined by the OECD Guidelines (2010) as such:

[Where] conditions are made or imposed between the two [associated] enterprises in their commercial or financial relations which differ from those which would be made between independent enterprises, then any profits which would, but for those conditions, have accrued to one of the enterprises, but, by reason of those conditions, have not so accrued, may be included in the profits of that enterprise and taxed accordingly.

As you can clearly see, nowhere in the definition does it mention that the focus of the arm's length principle should be on the margins that have been earned, but rather on the conditions behind these transactions, and whether those conditions are imposed on the entities that perform them. Only after that has been determined, can a correction of the profits obtained be performed. In practice, this falls prey to yet another oversimplification, in that the arm's length principle becomes primarily a contractual, functional, economic and strategic analysis. And while these are valuable tools in understanding the application of the arm's length principle, the underlying concept is much more complex and difficult to quantify.

Our interpretation is that, in essence, the arm's length principle is focused on detecting abnormalities and fraud rather than requiring a strict justification of a company's actions and inefficiencies. The principle hinges on proving that the actions of the company can be justifiable, rather than directly just testing the profits of the company and drawing conclusions of fraud from that. But since the principle itself is based more on the background than the text itself (as it is, again, a simplification or real world phenomena), both tax authorities and some taxpayers have manipulated it to meet their individual goals, ignoring the background.

As such, tax authorities have stretched it to mean that any decision made or profit earned must be thoroughly justified, most often by comparison to other companies on the market, which may not be part of a group (even if the tax authorities have audited and cleared them), and an imperfect comparison will automatically lead to the decision that fraud has been committed. On the other hand, unscrupulous taxpayers have stretched the definition in the other direction, by turning the arm's length principle criteria into strict rules the automatically and completely justify any absurdities in business behaviour, and setting prices strictly based on transfer pricing benchmarks. 

Due to these influences (although with a declared goal primarily in the direction of eliminating taxpayer abuse), in a more recent update of the BEPS Initiative, Actions 8-10, the OECD sets out to further clarify the meaning of the arm's length principle, and focuses on three main areas: allocation of risk (actual risk and financial capacity to assume risk), limiting the profit potential for funding transactions, and recharacterizing commercially irrational transactions. Again, the trend is towards analysing behaviour rather than prices, and only if that condition points towards foul play does it make sense to actually consider looking at the prices or profit margins. 

Finally, something important to mention is that even in the definition of the arm's length principle crafted by the OECD, it relies on certain assumptions that are never addressed. For example, the OECD bases the arm's length principle on the fact that a company will always know all the options on the market and the consequences of those choices, will choose to enter into a transaction in a strictly rational, optimal manner, and its motivation will always be transparent and demonstrable. Of course, reality has often proven that the evolution of companies is rarely linkable to such straightforward criteria, and often times the decisions made by the people in charge are affected by their own flaws and personal motivation, even if it results in a negative outcome or even bankruptcy. The OECD guide offers no guidance to account for such situations, and therefore the arm's length principle is fundamentally flawed.

2. The interquartile range is not always necessary

The interquartile range is merely a tool that is used to eliminate some of the differences between the tested entity and the results of the benchmarking study, and relies on the concept that, after a perfectly performed benchmarking study, profit is a direct correlation to the functional profile of the resulted entities, and therefore applying a central tendency measure will weed out the entities that are vastly different. 

Of course, in reality things are much more complicated, and simply retaining the most common profitability doesn't always account for arm's length deviations. The reasoning is partially sound, in that when determining the expected profitability in a field it is normal to look at a statistically controlled average, and then set transfer prices according to that, if the business model, strategy and market allows it. However, when adjusting transfer prices, such a narrowing of the profitability range doesn't really make sense, except in a protectionist way that ignores the actual market forces at play. 

The OECD guide agrees, stating that, should the selected comparables be similar enough to the tested party, all of them will be arm's length comparables (and central tendency measures need only be used if significant and unexplained deviations exist amongst the results):

3.62 [...] 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.

Thus, we have yet more proof that if the reasoning and methodology are correct, the interquartile range is not significant, but rather the entire range of comparable results can be used, despite the fact that many jurisdictions expressly demand it. Our suggestion is that more complex statistical tools should be used in this assessment, on a case by case basis, rather than applying a cookie-cutter solution of the inter-quartile range as is currently the case.

In conclusion, while the arm's length range is indeed often used, and sometimes a useful tool in proving adherence to the arm's length principle, it is not essential, and in no case should it supersede the more significant parts of understanding the business reasoning, conditions, strategy, bargaining power and freedom of choice that better separate fraudulent from legitimate transactions.